selftests: forwarding: add basic QoS classification test for Ocelot switches
[linux-block.git] / include / linux / netdevice.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the Interfaces handler.
8 *
9 * Version: @(#)dev.h 1.0.10 08/12/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Corey Minyard <wf-rch!minyard@relay.EU.net>
14 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
1da177e4
LT
19 * Moved to /usr/include/linux for NET3
20 */
21#ifndef _LINUX_NETDEVICE_H
22#define _LINUX_NETDEVICE_H
23
d7fe0f24 24#include <linux/timer.h>
187f1882 25#include <linux/bug.h>
bea3348e 26#include <linux/delay.h>
60063497 27#include <linux/atomic.h>
53511453 28#include <linux/prefetch.h>
1da177e4
LT
29#include <asm/cache.h>
30#include <asm/byteorder.h>
625788b5 31#include <asm/local.h>
1da177e4 32
1da177e4 33#include <linux/percpu.h>
4d5b78c0 34#include <linux/rculist.h>
bea3348e 35#include <linux/workqueue.h>
114cf580 36#include <linux/dynamic_queue_limits.h>
1da177e4 37
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>
406f42fa 51#include <linux/rbtree.h>
9ba74e6c 52#include <net/net_trackers.h>
a59e2ecb 53
115c1d6e 54struct netpoll_info;
313162d0 55struct device;
cc69837f 56struct ethtool_ops;
c1f19b51 57struct phy_device;
2f657a60 58struct dsa_port;
607259a6 59struct ip_tunnel_parm;
30e9bb84
AT
60struct macsec_context;
61struct macsec_ops;
6264f58c
JK
62struct netdev_name_node;
63struct sd_flow_limit;
e679c9c1 64struct sfp_bus;
704232c2
JB
65/* 802.11 specific */
66struct wireless_dev;
98a18b6f
AA
67/* 802.15.4 specific */
68struct wpan_dev;
03c57747 69struct mpls_dev;
7c46a640
AD
70/* UDP Tunnel offloads */
71struct udp_tunnel_info;
cc4e3835
JK
72struct udp_tunnel_nic_info;
73struct udp_tunnel_nic;
a7862b45 74struct bpf_prog;
814abfab 75struct xdp_buff;
1da177e4 76
5198d545 77void synchronize_net(void);
f629d208
JP
78void netdev_set_default_ethtool_ops(struct net_device *dev,
79 const struct ethtool_ops *ops);
d07d7507 80
9a1654ba
JP
81/* Backlog congestion levels */
82#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
83#define NET_RX_DROP 1 /* packet dropped */
84
7151affe
TY
85#define MAX_NEST_DEV 8
86
572a9d7b
PM
87/*
88 * Transmit return codes: transmit return codes originate from three different
89 * namespaces:
90 *
91 * - qdisc return codes
92 * - driver transmit return codes
93 * - errno values
94 *
95 * Drivers are allowed to return any one of those in their hard_start_xmit()
96 * function. Real network devices commonly used with qdiscs should only return
97 * the driver transmit return codes though - when qdiscs are used, the actual
98 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
99 * higher layers. Virtual network devices transmit synchronously; in this case
100 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
101 * others are propagated to higher layers.
102 */
103
104/* qdisc ->enqueue() return codes. */
105#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
106#define NET_XMIT_DROP 0x01 /* skb dropped */
107#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 108#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 109
b9df3cb8
GR
110/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
111 * indicates that the device will soon be dropping packets, or already drops
112 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 113#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
114#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
115
dc1f8bf6 116/* Driver transmit return codes */
9a1654ba 117#define NETDEV_TX_MASK 0xf0
572a9d7b 118
dc1f8bf6 119enum netdev_tx {
572a9d7b 120 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
121 NETDEV_TX_OK = 0x00, /* driver took care of packet */
122 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
123};
124typedef enum netdev_tx netdev_tx_t;
125
9a1654ba
JP
126/*
127 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
128 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
129 */
130static inline bool dev_xmit_complete(int rc)
131{
132 /*
133 * Positive cases with an skb consumed by a driver:
134 * - successful transmission (rc == NETDEV_TX_OK)
135 * - error while transmitting (rc < 0)
136 * - error while queueing to a different device (rc & NET_XMIT_MASK)
137 */
138 if (likely(rc < NET_XMIT_MASK))
139 return true;
140
141 return false;
142}
143
1da177e4 144/*
5e82b4b2 145 * Compute the worst-case header length according to the protocols
1da177e4
LT
146 * used.
147 */
fe2918b0 148
c0eb4540
KS
149#if defined(CONFIG_HYPERV_NET)
150# define LL_MAX_HEADER 128
151#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
152# if defined(CONFIG_MAC80211_MESH)
153# define LL_MAX_HEADER 128
154# else
155# define LL_MAX_HEADER 96
156# endif
1da177e4 157#else
8388e3da 158# define LL_MAX_HEADER 32
1da177e4
LT
159#endif
160
d11ead75
BH
161#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
162 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
163#define MAX_HEADER LL_MAX_HEADER
164#else
165#define MAX_HEADER (LL_MAX_HEADER + 48)
166#endif
167
168/*
be1f3c2c
BH
169 * Old network device statistics. Fields are native words
170 * (unsigned long) so they can be read and written atomically.
1da177e4 171 */
fe2918b0 172
d94d9fee 173struct net_device_stats {
3cfde79c
BH
174 unsigned long rx_packets;
175 unsigned long tx_packets;
176 unsigned long rx_bytes;
177 unsigned long tx_bytes;
178 unsigned long rx_errors;
179 unsigned long tx_errors;
180 unsigned long rx_dropped;
181 unsigned long tx_dropped;
182 unsigned long multicast;
1da177e4 183 unsigned long collisions;
1da177e4 184 unsigned long rx_length_errors;
3cfde79c
BH
185 unsigned long rx_over_errors;
186 unsigned long rx_crc_errors;
187 unsigned long rx_frame_errors;
188 unsigned long rx_fifo_errors;
189 unsigned long rx_missed_errors;
1da177e4
LT
190 unsigned long tx_aborted_errors;
191 unsigned long tx_carrier_errors;
192 unsigned long tx_fifo_errors;
193 unsigned long tx_heartbeat_errors;
194 unsigned long tx_window_errors;
1da177e4
LT
195 unsigned long rx_compressed;
196 unsigned long tx_compressed;
197};
198
625788b5
ED
199/* per-cpu stats, allocated on demand.
200 * Try to fit them in a single cache line, for dev_get_stats() sake.
201 */
202struct net_device_core_stats {
6510ea97
SAS
203 unsigned long rx_dropped;
204 unsigned long tx_dropped;
205 unsigned long rx_nohandler;
0e55546b 206 unsigned long rx_otherhost_dropped;
6510ea97 207} __aligned(4 * sizeof(unsigned long));
1da177e4
LT
208
209#include <linux/cache.h>
210#include <linux/skbuff.h>
211
adc9300e 212#ifdef CONFIG_RPS
c5905afb 213#include <linux/static_key.h>
dc05360f
ED
214extern struct static_key_false rps_needed;
215extern struct static_key_false rfs_needed;
adc9300e
ED
216#endif
217
1da177e4
LT
218struct neighbour;
219struct neigh_parms;
220struct sk_buff;
221
f001fde5
JP
222struct netdev_hw_addr {
223 struct list_head list;
406f42fa 224 struct rb_node node;
f001fde5
JP
225 unsigned char addr[MAX_ADDR_LEN];
226 unsigned char type;
ccffad25
JP
227#define NETDEV_HW_ADDR_T_LAN 1
228#define NETDEV_HW_ADDR_T_SAN 2
8e1b3884
TY
229#define NETDEV_HW_ADDR_T_UNICAST 3
230#define NETDEV_HW_ADDR_T_MULTICAST 4
22bedad3 231 bool global_use;
4cd729b0 232 int sync_cnt;
8f8f103d 233 int refcount;
4543fbef 234 int synced;
f001fde5
JP
235 struct rcu_head rcu_head;
236};
237
31278e71
JP
238struct netdev_hw_addr_list {
239 struct list_head list;
240 int count;
406f42fa
GN
241
242 /* Auxiliary tree for faster lookup on addition and deletion */
243 struct rb_root tree;
31278e71
JP
244};
245
22bedad3
JP
246#define netdev_hw_addr_list_count(l) ((l)->count)
247#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
248#define netdev_hw_addr_list_for_each(ha, l) \
249 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 250
22bedad3
JP
251#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
252#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
253#define netdev_for_each_uc_addr(ha, dev) \
254 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 255
22bedad3
JP
256#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
257#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 258#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 259 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 260
d94d9fee 261struct hh_cache {
5b3dc2f3 262 unsigned int hh_len;
3644f0ce 263 seqlock_t hh_lock;
1da177e4
LT
264
265 /* cached hardware header; allow for machine alignment needs. */
266#define HH_DATA_MOD 16
267#define HH_DATA_OFF(__len) \
5ba0eac6 268 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
269#define HH_DATA_ALIGN(__len) \
270 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
271 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
272};
273
5e82b4b2 274/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
275 * Alternative is:
276 * dev->hard_header_len ? (dev->hard_header_len +
277 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
278 *
279 * We could use other alignment values, but we must maintain the
280 * relationship HH alignment <= LL alignment.
281 */
282#define LL_RESERVED_SPACE(dev) \
f5184d26 283 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 284#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
f5184d26 285 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 286
3b04ddde
SH
287struct header_ops {
288 int (*create) (struct sk_buff *skb, struct net_device *dev,
289 unsigned short type, const void *daddr,
95c96174 290 const void *saddr, unsigned int len);
3b04ddde 291 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 292 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
293 void (*cache_update)(struct hh_cache *hh,
294 const struct net_device *dev,
295 const unsigned char *haddr);
2793a23a 296 bool (*validate)(const char *ll_header, unsigned int len);
e78b2915 297 __be16 (*parse_protocol)(const struct sk_buff *skb);
3b04ddde
SH
298};
299
1da177e4 300/* These flag bits are private to the generic network queueing
5e82b4b2 301 * layer; they may not be explicitly referenced by any other
1da177e4
LT
302 * code.
303 */
304
d94d9fee 305enum netdev_state_t {
1da177e4
LT
306 __LINK_STATE_START,
307 __LINK_STATE_PRESENT,
1da177e4 308 __LINK_STATE_NOCARRIER,
b00055aa
SR
309 __LINK_STATE_LINKWATCH_PENDING,
310 __LINK_STATE_DORMANT,
eec517cd 311 __LINK_STATE_TESTING,
1da177e4
LT
312};
313
6312fe77
LR
314struct gro_list {
315 struct list_head list;
316 int count;
317};
318
bea3348e 319/*
d9f37d01
LR
320 * size of gro hash buckets, must less than bit number of
321 * napi_struct::gro_bitmask
bea3348e 322 */
07d78363 323#define GRO_HASH_BUCKETS 8
d9f37d01
LR
324
325/*
326 * Structure for NAPI scheduling similar to tasklet but with weighting
327 */
bea3348e
SH
328struct napi_struct {
329 /* The poll_list must only be managed by the entity which
330 * changes the state of the NAPI_STATE_SCHED bit. This means
331 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 332 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
333 * can remove from the list right before clearing the bit.
334 */
335 struct list_head poll_list;
336
337 unsigned long state;
338 int weight;
6f8b12d6 339 int defer_hard_irqs_count;
d9f37d01 340 unsigned long gro_bitmask;
bea3348e
SH
341 int (*poll)(struct napi_struct *, int);
342#ifdef CONFIG_NETPOLL
bea3348e 343 int poll_owner;
bea3348e 344#endif
5d38a079 345 struct net_device *dev;
6312fe77 346 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 347 struct sk_buff *skb;
323ebb61
EC
348 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
349 int rx_count; /* length of rx_list */
3b47d303 350 struct hrtimer timer;
404f7c9e 351 struct list_head dev_list;
af12fa6e
ET
352 struct hlist_node napi_hash_node;
353 unsigned int napi_id;
29863d41 354 struct task_struct *thread;
bea3348e
SH
355};
356
d94d9fee 357enum {
7fd3253a
BT
358 NAPI_STATE_SCHED, /* Poll is scheduled */
359 NAPI_STATE_MISSED, /* reschedule a napi */
360 NAPI_STATE_DISABLE, /* Disable pending */
361 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
362 NAPI_STATE_LISTED, /* NAPI added to system lists */
363 NAPI_STATE_NO_BUSY_POLL, /* Do not add in napi_hash, no busy polling */
364 NAPI_STATE_IN_BUSY_POLL, /* sk_busy_loop() owns this NAPI */
365 NAPI_STATE_PREFER_BUSY_POLL, /* prefer busy-polling over softirq processing*/
29863d41 366 NAPI_STATE_THREADED, /* The poll is performed inside its own thread*/
cb038357 367 NAPI_STATE_SCHED_THREADED, /* Napi is currently scheduled in threaded mode */
217f6974
ED
368};
369
370enum {
7fd3253a
BT
371 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
372 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
373 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
374 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
375 NAPIF_STATE_LISTED = BIT(NAPI_STATE_LISTED),
376 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
377 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
378 NAPIF_STATE_PREFER_BUSY_POLL = BIT(NAPI_STATE_PREFER_BUSY_POLL),
29863d41 379 NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
cb038357 380 NAPIF_STATE_SCHED_THREADED = BIT(NAPI_STATE_SCHED_THREADED),
bea3348e
SH
381};
382
5b252f0c 383enum gro_result {
d1c76af9
HX
384 GRO_MERGED,
385 GRO_MERGED_FREE,
386 GRO_HELD,
387 GRO_NORMAL,
25393d3f 388 GRO_CONSUMED,
d1c76af9 389};
5b252f0c 390typedef enum gro_result gro_result_t;
d1c76af9 391
8a4eb573
JP
392/*
393 * enum rx_handler_result - Possible return values for rx_handlers.
394 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
395 * further.
396 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
397 * case skb->dev was changed by rx_handler.
398 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 399 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
400 *
401 * rx_handlers are functions called from inside __netif_receive_skb(), to do
402 * special processing of the skb, prior to delivery to protocol handlers.
403 *
404 * Currently, a net_device can only have a single rx_handler registered. Trying
405 * to register a second rx_handler will return -EBUSY.
406 *
407 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
408 * To unregister a rx_handler on a net_device, use
409 * netdev_rx_handler_unregister().
410 *
411 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
412 * do with the skb.
413 *
5e82b4b2 414 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 415 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 416 * the skb to be delivered in some other way.
8a4eb573
JP
417 *
418 * If the rx_handler changed skb->dev, to divert the skb to another
419 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
420 * new device will be called if it exists.
421 *
5e82b4b2 422 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 423 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 424 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 425 *
5e82b4b2 426 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
427 * delivered, it should return RX_HANDLER_PASS.
428 *
429 * A device without a registered rx_handler will behave as if rx_handler
430 * returned RX_HANDLER_PASS.
431 */
432
433enum rx_handler_result {
434 RX_HANDLER_CONSUMED,
435 RX_HANDLER_ANOTHER,
436 RX_HANDLER_EXACT,
437 RX_HANDLER_PASS,
438};
439typedef enum rx_handler_result rx_handler_result_t;
440typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 441
f629d208 442void __napi_schedule(struct napi_struct *n);
bc9ad166 443void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 444
4d29515f 445static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
446{
447 return test_bit(NAPI_STATE_DISABLE, &n->state);
448}
449
7fd3253a
BT
450static inline bool napi_prefer_busy_poll(struct napi_struct *n)
451{
452 return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
453}
454
39e6c820 455bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
456
457/**
458 * napi_schedule - schedule NAPI poll
5e82b4b2 459 * @n: NAPI context
bea3348e
SH
460 *
461 * Schedule NAPI poll routine to be called if it is not already
462 * running.
463 */
464static inline void napi_schedule(struct napi_struct *n)
465{
466 if (napi_schedule_prep(n))
467 __napi_schedule(n);
468}
469
bc9ad166
ED
470/**
471 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 472 * @n: NAPI context
bc9ad166
ED
473 *
474 * Variant of napi_schedule(), assuming hard irqs are masked.
475 */
476static inline void napi_schedule_irqoff(struct napi_struct *n)
477{
478 if (napi_schedule_prep(n))
479 __napi_schedule_irqoff(n);
480}
481
bfe13f54 482/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 483static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
484{
485 if (napi_schedule_prep(napi)) {
486 __napi_schedule(napi);
4d29515f 487 return true;
bfe13f54 488 }
4d29515f 489 return false;
bfe13f54
RD
490}
491
364b6055 492bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
493/**
494 * napi_complete - NAPI processing complete
5e82b4b2 495 * @n: NAPI context
bea3348e
SH
496 *
497 * Mark NAPI processing as complete.
3b47d303 498 * Consider using napi_complete_done() instead.
364b6055 499 * Return false if device should avoid rearming interrupts.
bea3348e 500 */
364b6055 501static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
502{
503 return napi_complete_done(n, 0);
504}
bea3348e 505
5fdd2f0e
WW
506int dev_set_threaded(struct net_device *dev, bool threaded);
507
bea3348e
SH
508/**
509 * napi_disable - prevent NAPI from scheduling
5e82b4b2 510 * @n: NAPI context
bea3348e
SH
511 *
512 * Stop NAPI from being scheduled on this context.
513 * Waits till any outstanding processing completes.
514 */
3b47d303 515void napi_disable(struct napi_struct *n);
bea3348e 516
29863d41 517void napi_enable(struct napi_struct *n);
bea3348e 518
c264c3de
SH
519/**
520 * napi_synchronize - wait until NAPI is not running
5e82b4b2 521 * @n: NAPI context
c264c3de
SH
522 *
523 * Wait until NAPI is done being scheduled on this context.
524 * Waits till any outstanding processing completes but
525 * does not disable future activations.
526 */
527static inline void napi_synchronize(const struct napi_struct *n)
528{
facc432f
AB
529 if (IS_ENABLED(CONFIG_SMP))
530 while (test_bit(NAPI_STATE_SCHED, &n->state))
531 msleep(1);
532 else
533 barrier();
c264c3de 534}
c264c3de 535
6c5c9581
MK
536/**
537 * napi_if_scheduled_mark_missed - if napi is running, set the
538 * NAPIF_STATE_MISSED
539 * @n: NAPI context
540 *
541 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
542 * NAPI is scheduled.
543 **/
544static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
545{
546 unsigned long val, new;
547
548 do {
549 val = READ_ONCE(n->state);
550 if (val & NAPIF_STATE_DISABLE)
551 return true;
552
553 if (!(val & NAPIF_STATE_SCHED))
554 return false;
555
556 new = val | NAPIF_STATE_MISSED;
557 } while (cmpxchg(&n->state, val, new) != val);
558
559 return true;
560}
561
d94d9fee 562enum netdev_queue_state_t {
73466498
TH
563 __QUEUE_STATE_DRV_XOFF,
564 __QUEUE_STATE_STACK_XOFF,
c3f26a26 565 __QUEUE_STATE_FROZEN,
79d16385 566};
8e2f1a63
DB
567
568#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
569#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
570#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
571
572#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
573#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
574 QUEUE_STATE_FROZEN)
575#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
576 QUEUE_STATE_FROZEN)
577
73466498
TH
578/*
579 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
580 * netif_tx_* functions below are used to manipulate this flag. The
581 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
582 * queue independently. The netif_xmit_*stopped functions below are called
583 * to check if the queue has been stopped by the driver or stack (either
584 * of the XOFF bits are set in the state). Drivers should not need to call
585 * netif_xmit*stopped functions, they should only be using netif_tx_*.
586 */
79d16385 587
bb949fbd 588struct netdev_queue {
6a321cb3 589/*
5e82b4b2 590 * read-mostly part
6a321cb3 591 */
bb949fbd 592 struct net_device *dev;
0b688f24
ED
593 netdevice_tracker dev_tracker;
594
46e5da40 595 struct Qdisc __rcu *qdisc;
b0e1e646 596 struct Qdisc *qdisc_sleeping;
ccf5ff69 597#ifdef CONFIG_SYSFS
1d24eb48
TH
598 struct kobject kobj;
599#endif
f2cd2d3e
ED
600#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
601 int numa_node;
602#endif
c0ef079c
FW
603 unsigned long tx_maxrate;
604 /*
605 * Number of TX timeouts for this queue
606 * (/sys/class/net/DEV/Q/trans_timeout)
607 */
8160fb43 608 atomic_long_t trans_timeout;
ffcfe25b
AD
609
610 /* Subordinate device that the queue has been assigned to */
611 struct net_device *sb_dev;
661b8d1b 612#ifdef CONFIG_XDP_SOCKETS
1742b3d5 613 struct xsk_buff_pool *pool;
661b8d1b 614#endif
6a321cb3 615/*
5e82b4b2 616 * write-mostly part
6a321cb3
ED
617 */
618 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
619 int xmit_lock_owner;
9d21493b 620 /*
9b36627a 621 * Time (in jiffies) of last Tx
9d21493b
ED
622 */
623 unsigned long trans_start;
ccf5ff69 624
114cf580
TH
625 unsigned long state;
626
627#ifdef CONFIG_BQL
628 struct dql dql;
629#endif
e8a0464c 630} ____cacheline_aligned_in_smp;
bb949fbd 631
79134e6c 632extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 633extern int sysctl_devconf_inherit_init_net;
79134e6c 634
316cdaa1
MB
635/*
636 * sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
637 * == 1 : For initns only
638 * == 2 : For none.
639 */
79134e6c
ED
640static inline bool net_has_fallback_tunnels(const struct net *net)
641{
3753d977
MB
642 return !IS_ENABLED(CONFIG_SYSCTL) ||
643 !sysctl_fb_tunnels_only_for_init_net ||
644 (net == &init_net && sysctl_fb_tunnels_only_for_init_net == 1);
79134e6c
ED
645}
646
f2cd2d3e
ED
647static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
648{
649#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
650 return q->numa_node;
651#else
b236da69 652 return NUMA_NO_NODE;
f2cd2d3e
ED
653#endif
654}
655
656static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
657{
658#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
659 q->numa_node = node;
660#endif
661}
662
df334545 663#ifdef CONFIG_RPS
0a9627f2
TH
664/*
665 * This structure holds an RPS map which can be of variable length. The
666 * map is an array of CPUs.
667 */
668struct rps_map {
669 unsigned int len;
670 struct rcu_head rcu;
bb4cf02d 671 u16 cpus[];
0a9627f2 672};
60b778ce 673#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 674
fec5e652 675/*
c445477d
BH
676 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
677 * tail pointer for that CPU's input queue at the time of last enqueue, and
678 * a hardware filter index.
fec5e652
TH
679 */
680struct rps_dev_flow {
681 u16 cpu;
c445477d 682 u16 filter;
fec5e652
TH
683 unsigned int last_qtail;
684};
c445477d 685#define RPS_NO_FILTER 0xffff
fec5e652
TH
686
687/*
688 * The rps_dev_flow_table structure contains a table of flow mappings.
689 */
690struct rps_dev_flow_table {
691 unsigned int mask;
692 struct rcu_head rcu;
bb4cf02d 693 struct rps_dev_flow flows[];
fec5e652
TH
694};
695#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 696 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
697
698/*
699 * The rps_sock_flow_table contains mappings of flows to the last CPU
700 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
701 * Each entry is a 32bit value. Upper part is the high-order bits
702 * of flow hash, lower part is CPU number.
567e4b79 703 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
704 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
705 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 706 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
707 */
708struct rps_sock_flow_table {
567e4b79 709 u32 mask;
93c1af6c 710
bb4cf02d 711 u32 ents[] ____cacheline_aligned_in_smp;
fec5e652 712};
567e4b79 713#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
714
715#define RPS_NO_CPU 0xffff
716
567e4b79
ED
717extern u32 rps_cpu_mask;
718extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
719
fec5e652
TH
720static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
721 u32 hash)
722{
723 if (table && hash) {
567e4b79
ED
724 unsigned int index = hash & table->mask;
725 u32 val = hash & ~rps_cpu_mask;
fec5e652 726
5e82b4b2 727 /* We only give a hint, preemption can change CPU under us */
567e4b79 728 val |= raw_smp_processor_id();
fec5e652 729
567e4b79
ED
730 if (table->ents[index] != val)
731 table->ents[index] = val;
fec5e652
TH
732 }
733}
734
c445477d 735#ifdef CONFIG_RFS_ACCEL
f629d208
JP
736bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
737 u16 filter_id);
c445477d 738#endif
a953be53 739#endif /* CONFIG_RPS */
c445477d 740
0a9627f2
TH
741/* This structure contains an instance of an RX queue. */
742struct netdev_rx_queue {
95d1d249 743 struct xdp_rxq_info xdp_rxq;
a953be53 744#ifdef CONFIG_RPS
6e3f7faf
ED
745 struct rps_map __rcu *rps_map;
746 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 747#endif
6e3f7faf 748 struct kobject kobj;
fe822240 749 struct net_device *dev;
80e8921b
ED
750 netdevice_tracker dev_tracker;
751
661b8d1b 752#ifdef CONFIG_XDP_SOCKETS
1742b3d5 753 struct xsk_buff_pool *pool;
661b8d1b 754#endif
0a9627f2 755} ____cacheline_aligned_in_smp;
a953be53
MD
756
757/*
758 * RX queue sysfs structures and functions.
759 */
760struct rx_queue_attribute {
761 struct attribute attr;
718ad681 762 ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
a953be53 763 ssize_t (*store)(struct netdev_rx_queue *queue,
718ad681 764 const char *buf, size_t len);
a953be53 765};
d314774c 766
044ab86d
AT
767/* XPS map type and offset of the xps map within net_device->xps_maps[]. */
768enum xps_map_type {
769 XPS_CPUS = 0,
770 XPS_RXQS,
771 XPS_MAPS_MAX,
772};
773
bf264145
TH
774#ifdef CONFIG_XPS
775/*
776 * This structure holds an XPS map which can be of variable length. The
777 * map is an array of queues.
778 */
779struct xps_map {
780 unsigned int len;
781 unsigned int alloc_len;
782 struct rcu_head rcu;
bb4cf02d 783 u16 queues[];
bf264145 784};
60b778ce 785#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
786#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
787 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
788
789/*
790 * This structure holds all XPS maps for device. Maps are indexed by CPU.
255c04a8 791 *
5478fcd0
AT
792 * We keep track of the number of cpus/rxqs used when the struct is allocated,
793 * in nr_ids. This will help not accessing out-of-bound memory.
794 *
255c04a8
AT
795 * We keep track of the number of traffic classes used when the struct is
796 * allocated, in num_tc. This will be used to navigate the maps, to ensure we're
797 * not crossing its upper bound, as the original dev->num_tc can be updated in
798 * the meantime.
bf264145
TH
799 */
800struct xps_dev_maps {
801 struct rcu_head rcu;
5478fcd0 802 unsigned int nr_ids;
255c04a8 803 s16 num_tc;
bb4cf02d 804 struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
bf264145 805};
80d19669
AN
806
807#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 808 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
809
810#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
811 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
812
bf264145
TH
813#endif /* CONFIG_XPS */
814
4f57c087
JF
815#define TC_MAX_QUEUE 16
816#define TC_BITMASK 15
817/* HW offloaded queuing disciplines txq count and offset maps */
818struct netdev_tc_txq {
819 u16 count;
820 u16 offset;
821};
822
68bad94e
NP
823#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
824/*
825 * This structure is to hold information about the device
826 * configured to run FCoE protocol stack.
827 */
828struct netdev_fcoe_hbainfo {
829 char manufacturer[64];
830 char serial_number[64];
831 char hardware_version[64];
832 char driver_version[64];
833 char optionrom_version[64];
834 char firmware_version[64];
835 char model[256];
836 char model_description[256];
837};
838#endif
839
02637fce 840#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 841
02637fce
JP
842/* This structure holds a unique identifier to identify some
843 * physical item (port for example) used by a netdevice.
66b52b0d 844 */
02637fce
JP
845struct netdev_phys_item_id {
846 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
847 unsigned char id_len;
848};
849
d754f98b
SF
850static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
851 struct netdev_phys_item_id *b)
852{
853 return a->id_len == b->id_len &&
854 memcmp(a->id, b->id, a->id_len) == 0;
855}
856
99932d4f 857typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
858 struct sk_buff *skb,
859 struct net_device *sb_dev);
99932d4f 860
ddb94eaf
PNA
861enum net_device_path_type {
862 DEV_PATH_ETHERNET = 0,
e4417d69 863 DEV_PATH_VLAN,
ec9d16ba 864 DEV_PATH_BRIDGE,
f6efc675 865 DEV_PATH_PPPOE,
0994d492 866 DEV_PATH_DSA,
a333215e 867 DEV_PATH_MTK_WDMA,
ddb94eaf
PNA
868};
869
870struct net_device_path {
871 enum net_device_path_type type;
872 const struct net_device *dev;
e4417d69
PNA
873 union {
874 struct {
875 u16 id;
876 __be16 proto;
f6efc675 877 u8 h_dest[ETH_ALEN];
e4417d69 878 } encap;
bcf2766b
FF
879 struct {
880 enum {
881 DEV_PATH_BR_VLAN_KEEP,
882 DEV_PATH_BR_VLAN_TAG,
883 DEV_PATH_BR_VLAN_UNTAG,
26267bf9 884 DEV_PATH_BR_VLAN_UNTAG_HW,
bcf2766b
FF
885 } vlan_mode;
886 u16 vlan_id;
887 __be16 vlan_proto;
888 } bridge;
0994d492
FF
889 struct {
890 int port;
891 u16 proto;
892 } dsa;
a333215e
FF
893 struct {
894 u8 wdma_idx;
895 u8 queue;
896 u16 wcid;
897 u8 bss;
898 } mtk_wdma;
e4417d69 899 };
ddb94eaf
PNA
900};
901
902#define NET_DEVICE_PATH_STACK_MAX 5
bcf2766b 903#define NET_DEVICE_PATH_VLAN_MAX 2
ddb94eaf
PNA
904
905struct net_device_path_stack {
906 int num_paths;
907 struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
908};
909
910struct net_device_path_ctx {
911 const struct net_device *dev;
912 const u8 *daddr;
bcf2766b
FF
913
914 int num_vlans;
915 struct {
916 u16 id;
917 __be16 proto;
918 } vlan[NET_DEVICE_PATH_VLAN_MAX];
ddb94eaf
PNA
919};
920
2572ac53 921enum tc_setup_type {
575ed7d3 922 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 923 TC_SETUP_CLSU32,
5b33f488 924 TC_SETUP_CLSFLOWER,
ade9b658 925 TC_SETUP_CLSMATCHALL,
332ae8e2 926 TC_SETUP_CLSBPF,
8c4083b3 927 TC_SETUP_BLOCK,
8521db4c 928 TC_SETUP_QDISC_CBS,
602f3baf 929 TC_SETUP_QDISC_RED,
7fdb61b4 930 TC_SETUP_QDISC_PRIO,
f971b132 931 TC_SETUP_QDISC_MQ,
25db26a9 932 TC_SETUP_QDISC_ETF,
98b0e5f6 933 TC_SETUP_ROOT_QDISC,
890d8d23 934 TC_SETUP_QDISC_GRED,
9c66d156 935 TC_SETUP_QDISC_TAPRIO,
c29f74e0 936 TC_SETUP_FT,
d35eb52b 937 TC_SETUP_QDISC_ETS,
ef6aadcc 938 TC_SETUP_QDISC_TBF,
aaca9408 939 TC_SETUP_QDISC_FIFO,
d03b195b 940 TC_SETUP_QDISC_HTB,
8cbfe939 941 TC_SETUP_ACT,
16e5cc64
JF
942};
943
f4e63525
JK
944/* These structures hold the attributes of bpf state that are being passed
945 * to the netdevice through the bpf op.
a7862b45 946 */
f4e63525 947enum bpf_netdev_command {
a7862b45
BB
948 /* Set or clear a bpf program used in the earliest stages of packet
949 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
950 * is responsible for calling bpf_prog_put on any old progs that are
951 * stored. In case of error, the callee need not release the new prog
952 * reference, but on success it takes ownership and must bpf_prog_put
953 * when it is no longer used.
954 */
955 XDP_SETUP_PROG,
ee5d032f 956 XDP_SETUP_PROG_HW,
ab3f0063 957 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
958 BPF_OFFLOAD_MAP_ALLOC,
959 BPF_OFFLOAD_MAP_FREE,
1742b3d5 960 XDP_SETUP_XSK_POOL,
a7862b45
BB
961};
962
cae1927c 963struct bpf_prog_offload_ops;
ddf9f970 964struct netlink_ext_ack;
74515c57 965struct xdp_umem;
75ccae62 966struct xdp_dev_bulk_queue;
aa8d3a71 967struct bpf_xdp_link;
ddf9f970 968
7f0a8382
AN
969enum bpf_xdp_mode {
970 XDP_MODE_SKB = 0,
971 XDP_MODE_DRV = 1,
972 XDP_MODE_HW = 2,
973 __MAX_XDP_MODE
974};
975
976struct bpf_xdp_entity {
977 struct bpf_prog *prog;
aa8d3a71 978 struct bpf_xdp_link *link;
7f0a8382 979};
ddf9f970 980
f4e63525
JK
981struct netdev_bpf {
982 enum bpf_netdev_command command;
a7862b45
BB
983 union {
984 /* XDP_SETUP_PROG */
ddf9f970 985 struct {
32d60277 986 u32 flags;
ddf9f970
JK
987 struct bpf_prog *prog;
988 struct netlink_ext_ack *extack;
989 };
a3884572
JK
990 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
991 struct {
992 struct bpf_offloaded_map *offmap;
993 };
1742b3d5 994 /* XDP_SETUP_XSK_POOL */
74515c57 995 struct {
1742b3d5 996 struct xsk_buff_pool *pool;
f8ebfaf6 997 u16 queue_id;
74515c57 998 } xsk;
a7862b45
BB
999 };
1000};
16e5cc64 1001
9116e5e2
MK
1002/* Flags for ndo_xsk_wakeup. */
1003#define XDP_WAKEUP_RX (1 << 0)
1004#define XDP_WAKEUP_TX (1 << 1)
1005
d77e38e6
SK
1006#ifdef CONFIG_XFRM_OFFLOAD
1007struct xfrmdev_ops {
1008 int (*xdo_dev_state_add) (struct xfrm_state *x);
1009 void (*xdo_dev_state_delete) (struct xfrm_state *x);
1010 void (*xdo_dev_state_free) (struct xfrm_state *x);
1011 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
1012 struct xfrm_state *x);
50bd870a 1013 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
1014};
1015#endif
1016
6c557001
FW
1017struct dev_ifalias {
1018 struct rcu_head rcuhead;
1019 char ifalias[];
1020};
1021
b473b0d2 1022struct devlink;
da68b4ad 1023struct tlsdev_ops;
b473b0d2 1024
93642e14
JP
1025struct netdev_net_notifier {
1026 struct list_head list;
1027 struct notifier_block *nb;
1028};
1029
d314774c
SH
1030/*
1031 * This structure defines the management hooks for network devices.
00829823
SH
1032 * The following hooks can be defined; unless noted otherwise, they are
1033 * optional and can be filled with a null pointer.
d314774c
SH
1034 *
1035 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
1036 * This function is called once when a network device is registered.
1037 * The network device can use this for any late stage initialization
1038 * or semantic validation. It can fail with an error code which will
1039 * be propagated back to register_netdev.
d314774c
SH
1040 *
1041 * void (*ndo_uninit)(struct net_device *dev);
1042 * This function is called when device is unregistered or when registration
1043 * fails. It is not called if init fails.
1044 *
1045 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 1046 * This function is called when a network device transitions to the up
d314774c
SH
1047 * state.
1048 *
1049 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 1050 * This function is called when a network device transitions to the down
d314774c
SH
1051 * state.
1052 *
dc1f8bf6
SH
1053 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1054 * struct net_device *dev);
00829823 1055 * Called when a packet needs to be transmitted.
e79d8429
RR
1056 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
1057 * the queue before that can happen; it's for obsolete devices and weird
1058 * corner cases, but the stack really does a non-trivial amount
1059 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 1060 * Required; cannot be NULL.
00829823 1061 *
1a2a1444
DM
1062 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1063 * struct net_device *dev
1064 * netdev_features_t features);
1065 * Called by core transmit path to determine if device is capable of
1066 * performing offload operations on a given packet. This is to give
1067 * the device an opportunity to implement any restrictions that cannot
1068 * be otherwise expressed by feature flags. The check is called with
1069 * the set of features that the stack has calculated and it returns
1070 * those the driver believes to be appropriate.
cdba756f 1071 *
f663dd9a 1072 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 1073 * struct net_device *sb_dev);
5e82b4b2 1074 * Called to decide which queue to use when device supports multiple
00829823
SH
1075 * transmit queues.
1076 *
d314774c
SH
1077 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1078 * This function is called to allow device receiver to make
5e82b4b2 1079 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
1080 *
1081 * void (*ndo_set_rx_mode)(struct net_device *dev);
1082 * This function is called device changes address list filtering.
01789349 1083 * If driver handles unicast address filtering, it should set
5e82b4b2 1084 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
1085 *
1086 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1087 * This function is called when the Media Access Control address
37b607c5 1088 * needs to be changed. If this interface is not defined, the
5e82b4b2 1089 * MAC address can not be changed.
d314774c
SH
1090 *
1091 * int (*ndo_validate_addr)(struct net_device *dev);
1092 * Test if Media Access Control address is valid for the device.
1093 *
1094 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
3d9d00bd
AB
1095 * Old-style ioctl entry point. This is used internally by the
1096 * appletalk and ieee802154 subsystems but is no longer called by
1097 * the device ioctl handler.
1098 *
1099 * int (*ndo_siocbond)(struct net_device *dev, struct ifreq *ifr, int cmd);
1100 * Used by the bonding driver for its device specific ioctls:
1101 * SIOCBONDENSLAVE, SIOCBONDRELEASE, SIOCBONDSETHWADDR, SIOCBONDCHANGEACTIVE,
1102 * SIOCBONDSLAVEINFOQUERY, and SIOCBONDINFOQUERY
d314774c 1103 *
a7605370
AB
1104 * * int (*ndo_eth_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
1105 * Called for ethernet specific ioctls: SIOCGMIIPHY, SIOCGMIIREG,
1106 * SIOCSMIIREG, SIOCSHWTSTAMP and SIOCGHWTSTAMP.
1107 *
d314774c
SH
1108 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1109 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1110 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1111 * interface (PCI) for low level management.
1112 *
1113 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1114 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1115 * of a device.
d314774c 1116 *
0290bd29 1117 * void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
5e82b4b2 1118 * Callback used when the transmitter has not made any progress
d314774c
SH
1119 * for dev->watchdog ticks.
1120 *
bc1f4470 1121 * void (*ndo_get_stats64)(struct net_device *dev,
1122 * struct rtnl_link_stats64 *storage);
d308e38f 1123 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1124 * Called when a user wants to get the network device usage
be1f3c2c 1125 * statistics. Drivers must do one of the following:
3cfde79c
BH
1126 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1127 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1128 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1129 * (which should normally be dev->stats) and return a pointer to
1130 * it. The structure may be changed asynchronously only if each
1131 * field is written atomically.
1132 * 3. Update dev->stats asynchronously and atomically, and define
1133 * neither operation.
d314774c 1134 *
3df5b3c6 1135 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1136 * Return true if this device supports offload stats of this attr_id.
1137 *
1138 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1139 * void *attr_data)
1140 * Get statistics for offload operations by attr_id. Write it into the
1141 * attr_data pointer.
1142 *
5d632cb7 1143 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1144 * If device supports VLAN filtering this function is called when a
80d5c368 1145 * VLAN id is registered.
d314774c 1146 *
5d632cb7 1147 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1148 * If device supports VLAN filtering this function is called when a
80d5c368 1149 * VLAN id is unregistered.
d314774c
SH
1150 *
1151 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1152 *
1153 * SR-IOV management functions.
1154 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1155 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1156 * u8 qos, __be16 proto);
ed616689
SC
1157 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1158 * int max_tx_rate);
5f8444a3 1159 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1160 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1161 * int (*ndo_get_vf_config)(struct net_device *dev,
1162 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1163 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1164 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1165 * struct nlattr *port[]);
01a3d796
VZ
1166 *
1167 * Enable or disable the VF ability to query its RSS Redirection Table and
1168 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1169 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1170 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1171 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1172 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1173 * void *type_data);
6a4bc2b4
FF
1174 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1175 * This is always called from the stack with the rtnl lock held and netif
1176 * tx queues stopped. This allows the netdevice to perform queue
1177 * management safely.
c445477d 1178 *
e9bce845
YZ
1179 * Fiber Channel over Ethernet (FCoE) offload functions.
1180 * int (*ndo_fcoe_enable)(struct net_device *dev);
1181 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1182 * so the underlying device can perform whatever needed configuration or
1183 * initialization to support acceleration of FCoE traffic.
1184 *
1185 * int (*ndo_fcoe_disable)(struct net_device *dev);
1186 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1187 * so the underlying device can perform whatever needed clean-ups to
1188 * stop supporting acceleration of FCoE traffic.
1189 *
1190 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1191 * struct scatterlist *sgl, unsigned int sgc);
1192 * Called when the FCoE Initiator wants to initialize an I/O that
1193 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1194 * perform necessary setup and returns 1 to indicate the device is set up
1195 * successfully to perform DDP on this I/O, otherwise this returns 0.
1196 *
1197 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1198 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1199 * indicated by the FC exchange id 'xid', so the underlying device can
1200 * clean up and reuse resources for later DDP requests.
1201 *
1202 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1203 * struct scatterlist *sgl, unsigned int sgc);
1204 * Called when the FCoE Target wants to initialize an I/O that
1205 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1206 * perform necessary setup and returns 1 to indicate the device is set up
1207 * successfully to perform DDP on this I/O, otherwise this returns 0.
1208 *
68bad94e
NP
1209 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1210 * struct netdev_fcoe_hbainfo *hbainfo);
1211 * Called when the FCoE Protocol stack wants information on the underlying
1212 * device. This information is utilized by the FCoE protocol stack to
1213 * register attributes with Fiber Channel management service as per the
1214 * FC-GS Fabric Device Management Information(FDMI) specification.
1215 *
e9bce845
YZ
1216 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1217 * Called when the underlying device wants to override default World Wide
1218 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1219 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1220 * protocol stack to use.
1221 *
c445477d
BH
1222 * RFS acceleration.
1223 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1224 * u16 rxq_index, u32 flow_id);
1225 * Set hardware filter for RFS. rxq_index is the target queue index;
1226 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1227 * Return the filter ID on success, or a negative error code.
fbaec0ea 1228 *
8b98a70c 1229 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1230 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1231 * Called to make another netdev an underling.
1232 *
1233 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1234 * Called to release previously enslaved netdev.
5455c699 1235 *
cff9f12b
MG
1236 * struct net_device *(*ndo_get_xmit_slave)(struct net_device *dev,
1237 * struct sk_buff *skb,
1238 * bool all_slaves);
1239 * Get the xmit slave of master device. If all_slaves is true, function
1240 * assume all the slaves can transmit.
1241 *
5455c699 1242 * Feature/offload setting functions.
1a2a1444
DM
1243 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1244 * netdev_features_t features);
1245 * Adjusts the requested feature flags according to device-specific
1246 * constraints, and returns the resulting flags. Must not modify
1247 * the device state.
1248 *
c8f44aff 1249 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1250 * Called to update device configuration to new features. Passed
1251 * feature set might be less than what was returned by ndo_fix_features()).
1252 * Must return >0 or -errno if it changed dev->features itself.
1253 *
edc7d573 1254 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1255 * struct net_device *dev,
87b0984e
PM
1256 * const unsigned char *addr, u16 vid, u16 flags,
1257 * struct netlink_ext_ack *extack);
77162022 1258 * Adds an FDB entry to dev for addr.
1690be63
VY
1259 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1260 * struct net_device *dev,
f6f6424b 1261 * const unsigned char *addr, u16 vid)
77162022 1262 * Deletes the FDB entry from dev coresponding to addr.
1306d536
NA
1263 * int (*ndo_fdb_del_bulk)(struct ndmsg *ndm, struct nlattr *tb[],
1264 * struct net_device *dev,
1265 * u16 vid,
1266 * struct netlink_ext_ack *extack);
77162022 1267 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1268 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1269 * int *idx)
77162022
JF
1270 * Used to add FDB entries to dump requests. Implementers should add
1271 * entries to skb and update idx with the number of entries.
e5a55a89 1272 *
ad41faa8 1273 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1274 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1275 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1276 * struct net_device *dev, u32 filter_mask,
1277 * int nlflags)
ad41faa8
ND
1278 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1279 * u16 flags);
4bf84c35
JP
1280 *
1281 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1282 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1283 * which do not represent real hardware may define this to allow their
1284 * userspace components to manage their virtual carrier state. Devices
1285 * that determine carrier state from physical hardware properties (eg
1286 * network cables) or protocol-dependent mechanisms (eg
1287 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1288 *
1289 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1290 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1291 * Called to get ID of physical port of this device. If driver does
1292 * not implement this, it is assumed that the hw is not able to have
1293 * multiple net devices on single physical port.
53cf5275 1294 *
d6abc596
FF
1295 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1296 * struct netdev_phys_item_id *ppid)
1297 * Called to get the parent ID of the physical port of this device.
1298 *
a6cc0cfa
JF
1299 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1300 * struct net_device *dev)
1301 * Called by upper layer devices to accelerate switching or other
1302 * station functionality into hardware. 'pdev is the lowerdev
1303 * to use for the offload and 'dev' is the net device that will
1304 * back the offload. Returns a pointer to the private structure
1305 * the upper layer will maintain.
1306 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1307 * Called by upper layer device to delete the station created
1308 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1309 * the station and priv is the structure returned by the add
1310 * operation.
822b3b2e
JF
1311 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1312 * int queue_index, u32 maxrate);
1313 * Called when a user wants to set a max-rate limitation of specific
1314 * TX queue.
a54acb3a
ND
1315 * int (*ndo_get_iflink)(const struct net_device *dev);
1316 * Called to get the iflink value of this device.
fc4099f1
PS
1317 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1318 * This function is used to get egress tunnel information for given skb.
1319 * This is useful for retrieving outer tunnel header parameters while
1320 * sampling packet.
871b642a
PA
1321 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1322 * This function is used to specify the headroom that the skb must
1323 * consider when allocation skb during packet reception. Setting
1324 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1325 * forward. Setting a negative value resets the rx headroom to the
871b642a 1326 * default value.
f4e63525 1327 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1328 * This function is used to set or query state related to XDP on the
f4e63525
JK
1329 * netdevice and manage BPF offload. See definition of
1330 * enum bpf_netdev_command for details.
42b33468
JDB
1331 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1332 * u32 flags);
735fc405
JDB
1333 * This function is used to submit @n XDP packets for transmit on a
1334 * netdevice. Returns number of frames successfully transmitted, frames
1335 * that got dropped are freed/returned via xdp_return_frame().
1336 * Returns negative number, means general error invoking ndo, meaning
1337 * no frames were xmit'ed and core-caller will free all frames.
879af96f
JM
1338 * struct net_device *(*ndo_xdp_get_xmit_slave)(struct net_device *dev,
1339 * struct xdp_buff *xdp);
1340 * Get the xmit slave of master device based on the xdp_buff.
9116e5e2
MK
1341 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1342 * This function is used to wake up the softirq, ksoftirqd or kthread
1343 * responsible for sending and/or receiving packets on a specific
1344 * queue id bound to an AF_XDP socket. The flags field specifies if
1345 * only RX, only Tx, or both should be woken up using the flags
1346 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
5dc37bb9
JP
1347 * struct devlink_port *(*ndo_get_devlink_port)(struct net_device *dev);
1348 * Get devlink port instance associated with a given netdev.
b473b0d2
JK
1349 * Called with a reference on the netdevice and devlink locks only,
1350 * rtnl_lock is not held.
607259a6
CH
1351 * int (*ndo_tunnel_ctl)(struct net_device *dev, struct ip_tunnel_parm *p,
1352 * int cmd);
1353 * Add, change, delete or get information on an IPv4 tunnel.
9aa1206e
DB
1354 * struct net_device *(*ndo_get_peer_dev)(struct net_device *dev);
1355 * If a device is paired with a peer device, return the peer instance.
1356 * The caller must be under RCU read context.
ddb94eaf
PNA
1357 * int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx, struct net_device_path *path);
1358 * Get the forwarding path to reach the real device from the HW destination address
d314774c
SH
1359 */
1360struct net_device_ops {
1361 int (*ndo_init)(struct net_device *dev);
1362 void (*ndo_uninit)(struct net_device *dev);
1363 int (*ndo_open)(struct net_device *dev);
1364 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1365 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1366 struct net_device *dev);
1367 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1368 struct net_device *dev,
1369 netdev_features_t features);
00829823 1370 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1371 struct sk_buff *skb,
a350ecce 1372 struct net_device *sb_dev);
d314774c
SH
1373 void (*ndo_change_rx_flags)(struct net_device *dev,
1374 int flags);
d314774c 1375 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1376 int (*ndo_set_mac_address)(struct net_device *dev,
1377 void *addr);
d314774c 1378 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1379 int (*ndo_do_ioctl)(struct net_device *dev,
1380 struct ifreq *ifr, int cmd);
a7605370
AB
1381 int (*ndo_eth_ioctl)(struct net_device *dev,
1382 struct ifreq *ifr, int cmd);
3d9d00bd
AB
1383 int (*ndo_siocbond)(struct net_device *dev,
1384 struct ifreq *ifr, int cmd);
ad7eab2a
AB
1385 int (*ndo_siocwandev)(struct net_device *dev,
1386 struct if_settings *ifs);
b9067f5d
AB
1387 int (*ndo_siocdevprivate)(struct net_device *dev,
1388 struct ifreq *ifr,
1389 void __user *data, int cmd);
d314774c
SH
1390 int (*ndo_set_config)(struct net_device *dev,
1391 struct ifmap *map);
00829823
SH
1392 int (*ndo_change_mtu)(struct net_device *dev,
1393 int new_mtu);
1394 int (*ndo_neigh_setup)(struct net_device *dev,
1395 struct neigh_parms *);
0290bd29
MT
1396 void (*ndo_tx_timeout) (struct net_device *dev,
1397 unsigned int txqueue);
d314774c 1398
bc1f4470 1399 void (*ndo_get_stats64)(struct net_device *dev,
1400 struct rtnl_link_stats64 *storage);
3df5b3c6 1401 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1402 int (*ndo_get_offload_stats)(int attr_id,
1403 const struct net_device *dev,
1404 void *attr_data);
d314774c
SH
1405 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1406
8e586137 1407 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1408 __be16 proto, u16 vid);
8e586137 1409 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1410 __be16 proto, u16 vid);
d314774c 1411#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1412 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1413 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1414 struct netpoll_info *info);
0e34e931 1415 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1416#endif
95c26df8
WM
1417 int (*ndo_set_vf_mac)(struct net_device *dev,
1418 int queue, u8 *mac);
1419 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1420 int queue, u16 vlan,
1421 u8 qos, __be16 proto);
ed616689
SC
1422 int (*ndo_set_vf_rate)(struct net_device *dev,
1423 int vf, int min_tx_rate,
1424 int max_tx_rate);
5f8444a3
GR
1425 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1426 int vf, bool setting);
dd461d6a
HS
1427 int (*ndo_set_vf_trust)(struct net_device *dev,
1428 int vf, bool setting);
95c26df8
WM
1429 int (*ndo_get_vf_config)(struct net_device *dev,
1430 int vf,
1431 struct ifla_vf_info *ivf);
1d8faf48
RE
1432 int (*ndo_set_vf_link_state)(struct net_device *dev,
1433 int vf, int link_state);
3b766cd8
EBE
1434 int (*ndo_get_vf_stats)(struct net_device *dev,
1435 int vf,
1436 struct ifla_vf_stats
1437 *vf_stats);
57b61080
SF
1438 int (*ndo_set_vf_port)(struct net_device *dev,
1439 int vf,
1440 struct nlattr *port[]);
1441 int (*ndo_get_vf_port)(struct net_device *dev,
1442 int vf, struct sk_buff *skb);
30aad417
DG
1443 int (*ndo_get_vf_guid)(struct net_device *dev,
1444 int vf,
1445 struct ifla_vf_guid *node_guid,
1446 struct ifla_vf_guid *port_guid);
cc8e27cc
EC
1447 int (*ndo_set_vf_guid)(struct net_device *dev,
1448 int vf, u64 guid,
1449 int guid_type);
01a3d796
VZ
1450 int (*ndo_set_vf_rss_query_en)(
1451 struct net_device *dev,
1452 int vf, bool setting);
16e5cc64 1453 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1454 enum tc_setup_type type,
de4784ca 1455 void *type_data);
d11ead75 1456#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1457 int (*ndo_fcoe_enable)(struct net_device *dev);
1458 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1459 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1460 u16 xid,
1461 struct scatterlist *sgl,
1462 unsigned int sgc);
1463 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1464 u16 xid);
6247e086
YZ
1465 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1466 u16 xid,
1467 struct scatterlist *sgl,
1468 unsigned int sgc);
68bad94e
NP
1469 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1470 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1471#endif
1472
d11ead75 1473#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1474#define NETDEV_FCOE_WWNN 0
1475#define NETDEV_FCOE_WWPN 1
1476 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1477 u64 *wwn, int type);
4d288d57 1478#endif
3c9c36bc 1479
c445477d
BH
1480#ifdef CONFIG_RFS_ACCEL
1481 int (*ndo_rx_flow_steer)(struct net_device *dev,
1482 const struct sk_buff *skb,
1483 u16 rxq_index,
1484 u32 flow_id);
1485#endif
fbaec0ea 1486 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1487 struct net_device *slave_dev,
1488 struct netlink_ext_ack *extack);
fbaec0ea
JP
1489 int (*ndo_del_slave)(struct net_device *dev,
1490 struct net_device *slave_dev);
cff9f12b
MG
1491 struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
1492 struct sk_buff *skb,
1493 bool all_slaves);
719a402c
TT
1494 struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
1495 struct sock *sk);
c8f44aff
MM
1496 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1497 netdev_features_t features);
5455c699 1498 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1499 netdev_features_t features);
503eebc2
JP
1500 int (*ndo_neigh_construct)(struct net_device *dev,
1501 struct neighbour *n);
1502 void (*ndo_neigh_destroy)(struct net_device *dev,
1503 struct neighbour *n);
77162022
JF
1504
1505 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1506 struct nlattr *tb[],
77162022 1507 struct net_device *dev,
6b6e2725 1508 const unsigned char *addr,
f6f6424b 1509 u16 vid,
87b0984e
PM
1510 u16 flags,
1511 struct netlink_ext_ack *extack);
77162022 1512 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1513 struct nlattr *tb[],
77162022 1514 struct net_device *dev,
f6f6424b
JP
1515 const unsigned char *addr,
1516 u16 vid);
1306d536
NA
1517 int (*ndo_fdb_del_bulk)(struct ndmsg *ndm,
1518 struct nlattr *tb[],
1519 struct net_device *dev,
1520 u16 vid,
1521 struct netlink_ext_ack *extack);
77162022
JF
1522 int (*ndo_fdb_dump)(struct sk_buff *skb,
1523 struct netlink_callback *cb,
1524 struct net_device *dev,
5d5eacb3 1525 struct net_device *filter_dev,
d297653d 1526 int *idx);
5b2f94b2
RP
1527 int (*ndo_fdb_get)(struct sk_buff *skb,
1528 struct nlattr *tb[],
1529 struct net_device *dev,
1530 const unsigned char *addr,
1531 u16 vid, u32 portid, u32 seq,
1532 struct netlink_ext_ack *extack);
e5a55a89 1533 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1534 struct nlmsghdr *nlh,
2fd527b7
PM
1535 u16 flags,
1536 struct netlink_ext_ack *extack);
e5a55a89
JF
1537 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1538 u32 pid, u32 seq,
6cbdceeb 1539 struct net_device *dev,
46c264da
ND
1540 u32 filter_mask,
1541 int nlflags);
407af329 1542 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1543 struct nlmsghdr *nlh,
1544 u16 flags);
4bf84c35
JP
1545 int (*ndo_change_carrier)(struct net_device *dev,
1546 bool new_carrier);
66b52b0d 1547 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1548 struct netdev_phys_item_id *ppid);
d6abc596
FF
1549 int (*ndo_get_port_parent_id)(struct net_device *dev,
1550 struct netdev_phys_item_id *ppid);
db24a904
DA
1551 int (*ndo_get_phys_port_name)(struct net_device *dev,
1552 char *name, size_t len);
a6cc0cfa
JF
1553 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1554 struct net_device *dev);
1555 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1556 void *priv);
1557
822b3b2e
JF
1558 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1559 int queue_index,
1560 u32 maxrate);
a54acb3a 1561 int (*ndo_get_iflink)(const struct net_device *dev);
fc4099f1
PS
1562 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1563 struct sk_buff *skb);
871b642a
PA
1564 void (*ndo_set_rx_headroom)(struct net_device *dev,
1565 int needed_headroom);
f4e63525
JK
1566 int (*ndo_bpf)(struct net_device *dev,
1567 struct netdev_bpf *bpf);
735fc405 1568 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1569 struct xdp_frame **xdp,
1570 u32 flags);
879af96f
JM
1571 struct net_device * (*ndo_xdp_get_xmit_slave)(struct net_device *dev,
1572 struct xdp_buff *xdp);
9116e5e2
MK
1573 int (*ndo_xsk_wakeup)(struct net_device *dev,
1574 u32 queue_id, u32 flags);
5dc37bb9 1575 struct devlink_port * (*ndo_get_devlink_port)(struct net_device *dev);
607259a6
CH
1576 int (*ndo_tunnel_ctl)(struct net_device *dev,
1577 struct ip_tunnel_parm *p, int cmd);
9aa1206e 1578 struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
ddb94eaf
PNA
1579 int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
1580 struct net_device_path *path);
d314774c
SH
1581};
1582
7aa98047 1583/**
270f3385 1584 * enum netdev_priv_flags - &struct net_device priv_flags
7aa98047
LR
1585 *
1586 * These are the &struct net_device, they are only set internally
1587 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1588 * userspace; this means that the order of these flags can change
7aa98047
LR
1589 * during any kernel release.
1590 *
1591 * You should have a pretty good reason to be extending these flags.
1592 *
1593 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1594 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1595 * @IFF_BONDING: bonding master or slave
7aa98047 1596 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1597 * @IFF_WAN_HDLC: WAN HDLC device
1598 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1599 * release skb->dst
1600 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1601 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1602 * @IFF_MACVLAN_PORT: device used as macvlan port
1603 * @IFF_BRIDGE_PORT: device used as bridge port
1604 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1605 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1606 * @IFF_UNICAST_FLT: Supports unicast filtering
1607 * @IFF_TEAM_PORT: device used as team port
1608 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1609 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1610 * change when it's running
1611 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1612 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1613 * underlying stacked devices
007979ea 1614 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1615 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1616 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1617 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1618 * @IFF_TEAM: device is a team device
d4ab4286 1619 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1620 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1621 * entity (i.e. the master device for bridged veth)
3c175784 1622 * @IFF_MACSEC: device is a MACsec device
f5426250 1623 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1624 * @IFF_FAILOVER: device is a failover master device
1625 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1626 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8065a779 1627 * @IFF_LIVE_RENAME_OK: rename is allowed while device is up and running
c2ff53d8
XZ
1628 * @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1629 * skb_headlen(skb) == 0 (data starts from frag0)
2106efda 1630 * @IFF_CHANGE_PROTO_DOWN: device supports setting carrier via IFLA_PROTO_DOWN
7aa98047
LR
1631 */
1632enum netdev_priv_flags {
1633 IFF_802_1Q_VLAN = 1<<0,
1634 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1635 IFF_BONDING = 1<<2,
1636 IFF_ISATAP = 1<<3,
1637 IFF_WAN_HDLC = 1<<4,
1638 IFF_XMIT_DST_RELEASE = 1<<5,
1639 IFF_DONT_BRIDGE = 1<<6,
1640 IFF_DISABLE_NETPOLL = 1<<7,
1641 IFF_MACVLAN_PORT = 1<<8,
1642 IFF_BRIDGE_PORT = 1<<9,
1643 IFF_OVS_DATAPATH = 1<<10,
1644 IFF_TX_SKB_SHARING = 1<<11,
1645 IFF_UNICAST_FLT = 1<<12,
1646 IFF_TEAM_PORT = 1<<13,
1647 IFF_SUPP_NOFCS = 1<<14,
1648 IFF_LIVE_ADDR_CHANGE = 1<<15,
1649 IFF_MACVLAN = 1<<16,
1650 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1651 IFF_L3MDEV_MASTER = 1<<18,
1652 IFF_NO_QUEUE = 1<<19,
1653 IFF_OPENVSWITCH = 1<<20,
1654 IFF_L3MDEV_SLAVE = 1<<21,
1655 IFF_TEAM = 1<<22,
1656 IFF_RXFH_CONFIGURED = 1<<23,
1657 IFF_PHONY_HEADROOM = 1<<24,
1658 IFF_MACSEC = 1<<25,
f5426250 1659 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1660 IFF_FAILOVER = 1<<27,
1661 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1662 IFF_L3MDEV_RX_HANDLER = 1<<29,
8065a779 1663 IFF_LIVE_RENAME_OK = 1<<30,
c2ff53d8 1664 IFF_TX_SKB_NO_LINEAR = 1<<31,
2106efda 1665 IFF_CHANGE_PROTO_DOWN = BIT_ULL(32),
7aa98047
LR
1666};
1667
1668#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1669#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1670#define IFF_BONDING IFF_BONDING
7aa98047 1671#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1672#define IFF_WAN_HDLC IFF_WAN_HDLC
1673#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1674#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1675#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1676#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1677#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1678#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1679#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1680#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1681#define IFF_TEAM_PORT IFF_TEAM_PORT
1682#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1683#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1684#define IFF_MACVLAN IFF_MACVLAN
02875878 1685#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1686#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1687#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1688#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1689#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1690#define IFF_TEAM IFF_TEAM
d4ab4286 1691#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
2463e073 1692#define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
3c175784 1693#define IFF_MACSEC IFF_MACSEC
f5426250 1694#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1695#define IFF_FAILOVER IFF_FAILOVER
1696#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
d5256083 1697#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
8065a779 1698#define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
c2ff53d8 1699#define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
7aa98047 1700
4e096a18
OR
1701/* Specifies the type of the struct net_device::ml_priv pointer */
1702enum netdev_ml_priv_type {
1703 ML_PRIV_NONE,
1704 ML_PRIV_CAN,
1705};
1706
536721b1
KK
1707/**
1708 * struct net_device - The DEVICE structure.
d651983d
MCC
1709 *
1710 * Actually, this whole structure is a big mistake. It mixes I/O
1711 * data with strictly "high-level" data, and it has to know about
1712 * almost every data structure used in the INET module.
536721b1
KK
1713 *
1714 * @name: This is the first field of the "visible" part of this structure
1715 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1716 * of the interface.
536721b1 1717 *
ff927412 1718 * @name_node: Name hashlist node
536721b1
KK
1719 * @ifalias: SNMP alias
1720 * @mem_end: Shared memory end
1721 * @mem_start: Shared memory start
1722 * @base_addr: Device I/O address
1723 * @irq: Device IRQ number
1724 *
1725 * @state: Generic network queuing layer state, see netdev_state_t
1726 * @dev_list: The global list of network devices
5e82b4b2
BH
1727 * @napi_list: List entry used for polling NAPI devices
1728 * @unreg_list: List entry when we are unregistering the
1729 * device; see the function unregister_netdev
1730 * @close_list: List entry used when we are closing the device
62d885fe
BP
1731 * @ptype_all: Device-specific packet handlers for all protocols
1732 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1733 *
1734 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1735 * @features: Currently active device features
1736 * @hw_features: User-changeable features
1737 *
1738 * @wanted_features: User-requested features
1739 * @vlan_features: Mask of features inheritable by VLAN devices
1740 *
1741 * @hw_enc_features: Mask of features inherited by encapsulating devices
1742 * This field indicates what encapsulation
1743 * offloads the hardware is capable of doing,
1744 * and drivers will need to set them appropriately.
1745 *
1746 * @mpls_features: Mask of features inheritable by MPLS
a1fa83bd 1747 * @gso_partial_features: value(s) from NETIF_F_GSO\*
536721b1
KK
1748 *
1749 * @ifindex: interface index
5e82b4b2 1750 * @group: The group the device belongs to
536721b1
KK
1751 *
1752 * @stats: Statistics struct, which was left as a legacy, use
1753 * rtnl_link_stats64 instead
1754 *
625788b5 1755 * @core_stats: core networking counters,
536721b1 1756 * do not use this in drivers
9e55e5d3
FF
1757 * @carrier_up_count: Number of times the carrier has been up
1758 * @carrier_down_count: Number of times the carrier has been down
536721b1 1759 *
536721b1
KK
1760 * @wireless_handlers: List of functions to handle Wireless Extensions,
1761 * instead of ioctl,
1762 * see <net/iw_handler.h> for details.
1763 * @wireless_data: Instance data managed by the core of wireless extensions
1764 *
1765 * @netdev_ops: Includes several pointers to callbacks,
1766 * if one wants to override the ndo_*() functions
1767 * @ethtool_ops: Management operations
a1fa83bd 1768 * @l3mdev_ops: Layer 3 master device operations
f997c55c
AA
1769 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1770 * discovery handling. Necessary for e.g. 6LoWPAN.
a1fa83bd
RD
1771 * @xfrmdev_ops: Transformation offload operations
1772 * @tlsdev_ops: Transport Layer Security offload operations
d476059e 1773 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1774 * of Layer 2 headers.
1775 *
1776 * @flags: Interface flags (a la BSD)
1777 * @priv_flags: Like 'flags' but invisible to userspace,
1778 * see if.h for the definitions
1779 * @gflags: Global flags ( kept as legacy )
1780 * @padded: How much padding added by alloc_netdev()
1781 * @operstate: RFC2863 operstate
1782 * @link_mode: Mapping policy to operstate
1783 * @if_port: Selectable AUI, TP, ...
1784 * @dma: DMA channel
1785 * @mtu: Interface MTU value
61e84623
JW
1786 * @min_mtu: Interface Minimum MTU value
1787 * @max_mtu: Interface Maximum MTU value
536721b1 1788 * @type: Interface hardware type
2793a23a 1789 * @hard_header_len: Maximum hardware header length.
217e6fa2 1790 * @min_header_len: Minimum hardware header length
536721b1
KK
1791 *
1792 * @needed_headroom: Extra headroom the hardware may need, but not in all
1793 * cases can this be guaranteed
1794 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1795 * cases can this be guaranteed. Some cases also use
1796 * LL_MAX_HEADER instead to allocate the skb
1797 *
1798 * interface address info:
1799 *
1800 * @perm_addr: Permanent hw address
1801 * @addr_assign_type: Hw address assignment type
1802 * @addr_len: Hardware address length
5343da4c
TY
1803 * @upper_level: Maximum depth level of upper devices.
1804 * @lower_level: Maximum depth level of lower devices.
8626a0c8 1805 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1806 * @dev_id: Used to differentiate devices that share
1807 * the same link layer address
1808 * @dev_port: Used to differentiate devices that share
1809 * the same function
1810 * @addr_list_lock: XXX: need comments on this one
a1fa83bd 1811 * @name_assign_type: network interface name assignment type
5e82b4b2 1812 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1813 * has been enabled due to the need to listen to
1814 * additional unicast addresses in a device that
1815 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1816 * @uc: unicast mac addresses
1817 * @mc: multicast mac addresses
1818 * @dev_addrs: list of device hw addresses
1819 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1820 * @promiscuity: Number of times the NIC is told to work in
1821 * promiscuous mode; if it becomes 0 the NIC will
1822 * exit promiscuous mode
536721b1
KK
1823 * @allmulti: Counter, enables or disables allmulticast mode
1824 *
1825 * @vlan_info: VLAN info
1826 * @dsa_ptr: dsa specific data
1827 * @tipc_ptr: TIPC specific data
1828 * @atalk_ptr: AppleTalk link
1829 * @ip_ptr: IPv4 specific data
1830 * @dn_ptr: DECnet specific data
1831 * @ip6_ptr: IPv6 specific data
1832 * @ax25_ptr: AX.25 specific data
1833 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
a1fa83bd
RD
1834 * @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
1835 * device struct
1836 * @mpls_ptr: mpls_dev struct pointer
583be982 1837 * @mctp_ptr: MCTP specific data
536721b1 1838 *
536721b1
KK
1839 * @dev_addr: Hw address (before bcast,
1840 * because most packets are unicast)
1841 *
1842 * @_rx: Array of RX queues
1843 * @num_rx_queues: Number of RX queues
1844 * allocated at register_netdev() time
1845 * @real_num_rx_queues: Number of RX queues currently active in device
a1fa83bd
RD
1846 * @xdp_prog: XDP sockets filter program pointer
1847 * @gro_flush_timeout: timeout for GRO layer in NAPI
5c45a918
MCC
1848 * @napi_defer_hard_irqs: If not zero, provides a counter that would
1849 * allow to avoid NIC hard IRQ, on busy queues.
536721b1
KK
1850 *
1851 * @rx_handler: handler for received packets
1852 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1853 * @miniq_ingress: ingress/clsact qdisc specific data for
1854 * ingress processing
536721b1 1855 * @ingress_queue: XXX: need comments on this one
2f5e70c8 1856 * @nf_hooks_ingress: netfilter hooks executed for ingress packets
536721b1
KK
1857 * @broadcast: hw bcast address
1858 *
14ffbbb8
TG
1859 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1860 * indexed by RX queue number. Assigned by driver.
1861 * This must only be set if the ndo_rx_flow_steer
1862 * operation is defined
1863 * @index_hlist: Device index hash chain
1864 *
536721b1
KK
1865 * @_tx: Array of TX queues
1866 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1867 * @real_num_tx_queues: Number of TX queues currently active in device
1868 * @qdisc: Root qdisc from userspace point of view
1869 * @tx_queue_len: Max frames per queue allowed
1870 * @tx_global_lock: XXX: need comments on this one
a1fa83bd 1871 * @xdp_bulkq: XDP device bulk queue
044ab86d 1872 * @xps_maps: all CPUs/RXQs maps for XPS device
536721b1
KK
1873 *
1874 * @xps_maps: XXX: need comments on this one
46209401
JP
1875 * @miniq_egress: clsact qdisc specific data for
1876 * egress processing
42df6e1d 1877 * @nf_hooks_egress: netfilter hooks executed for egress packets
a1fa83bd 1878 * @qdisc_hash: qdisc hash table
536721b1 1879 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1880 * the watchdog (see dev_watchdog())
536721b1
KK
1881 * @watchdog_timer: List of timers
1882 *
eb02d39a 1883 * @proto_down_reason: reason a netdev interface is held down
536721b1 1884 * @pcpu_refcnt: Number of references to this device
add2d736 1885 * @dev_refcnt: Number of references to this device
4d92b95f 1886 * @refcnt_tracker: Tracker directory for tracked references to this device
536721b1 1887 * @todo_list: Delayed register/unregister
536721b1
KK
1888 * @link_watch_list: XXX: need comments on this one
1889 *
1890 * @reg_state: Register/unregister state machine
1891 * @dismantle: Device is going to be freed
1892 * @rtnl_link_state: This enum represents the phases of creating
1893 * a new link
1894 *
cf124db5
DM
1895 * @needs_free_netdev: Should unregister perform free_netdev?
1896 * @priv_destructor: Called from unregister
536721b1
KK
1897 * @npinfo: XXX: need comments on this one
1898 * @nd_net: Network namespace this network device is inside
1899 *
1900 * @ml_priv: Mid-layer private
4e096a18 1901 * @ml_priv_type: Mid-layer private type
536721b1
KK
1902 * @lstats: Loopback statistics
1903 * @tstats: Tunnel statistics
1904 * @dstats: Dummy statistics
1905 * @vstats: Virtual ethernet statistics
1906 *
1907 * @garp_port: GARP
1908 * @mrp_port: MRP
1909 *
4a0cb83b
JK
1910 * @dm_private: Drop monitor private
1911 *
536721b1
KK
1912 * @dev: Class/net/name entry
1913 * @sysfs_groups: Space for optional device, statistics and wireless
1914 * sysfs groups
1915 *
1916 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1917 * @rtnl_link_ops: Rtnl_link_ops
1918 *
1919 * @gso_max_size: Maximum size of generic segmentation offload
1920 * @gso_max_segs: Maximum number of segments that can be passed to the
1921 * NIC for GSO
1922 *
1923 * @dcbnl_ops: Data Center Bridging netlink ops
1924 * @num_tc: Number of traffic classes in the net device
1925 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1926 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1927 *
1928 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1929 *
1930 * @priomap: XXX: need comments on this one
1931 * @phydev: Physical device may attach itself
1932 * for hardware timestamping
e679c9c1 1933 * @sfp_bus: attached &struct sfp_bus structure.
1a33e10e 1934 *
1a33e10e 1935 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
536721b1 1936 *
d746d707
AK
1937 * @proto_down: protocol port state information can be sent to the
1938 * switch driver and used to set the phys state of the
1939 * switch port.
1940 *
61941143
HK
1941 * @wol_enabled: Wake-on-LAN is enabled
1942 *
29863d41
WW
1943 * @threaded: napi threaded mode is enabled
1944 *
93642e14
JP
1945 * @net_notifier_list: List of per-net netdev notifier block
1946 * that follow this device when it is moved
1947 * to another network namespace.
1948 *
30e9bb84
AT
1949 * @macsec_ops: MACsec offloading ops
1950 *
cc4e3835
JK
1951 * @udp_tunnel_nic_info: static structure describing the UDP tunnel
1952 * offload capabilities of the device
1953 * @udp_tunnel_nic: UDP tunnel offload state
ffa59b0b 1954 * @xdp_state: stores info on attached XDP BPF programs
cc4e3835 1955 *
b62e3317 1956 * @nested_level: Used as a parameter of spin_lock_nested() of
a93bdcb9
MCC
1957 * dev->addr_list_lock.
1958 * @unlink_list: As netif_addr_lock() can be called recursively,
1959 * keep a list of interfaces to be deleted.
eac1b93c
CL
1960 * @gro_max_size: Maximum size of aggregated packet in generic
1961 * receive offload (GRO)
a93bdcb9 1962 *
d07b26f5 1963 * @dev_addr_shadow: Copy of @dev_addr to catch direct writes.
63f13937 1964 * @linkwatch_dev_tracker: refcount tracker used by linkwatch.
f12bf6f3 1965 * @watchdog_dev_tracker: refcount tracker used by watchdog.
b2309a71
ED
1966 * @dev_registered_tracker: tracker for reference held while
1967 * registered
9309f97a 1968 * @offload_xstats_l3: L3 HW stats for this netdevice.
d07b26f5 1969 *
1da177e4
LT
1970 * FIXME: cleanup struct net_device such that network protocol info
1971 * moves out.
1972 */
1973
d94d9fee 1974struct net_device {
1da177e4 1975 char name[IFNAMSIZ];
ff927412 1976 struct netdev_name_node *name_node;
6c557001 1977 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1978 /*
1979 * I/O specific fields
1980 * FIXME: Merge these and struct ifmap into one
1981 */
536721b1
KK
1982 unsigned long mem_end;
1983 unsigned long mem_start;
1984 unsigned long base_addr;
1da177e4
LT
1985
1986 /*
536721b1
KK
1987 * Some hardware also needs these fields (state,dev_list,
1988 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1989 * part of the usual set specified in Space.c.
1990 */
1991
1da177e4
LT
1992 unsigned long state;
1993
7562f876 1994 struct list_head dev_list;
bea3348e 1995 struct list_head napi_list;
44a0873d 1996 struct list_head unreg_list;
5cde2829 1997 struct list_head close_list;
7866a621
SN
1998 struct list_head ptype_all;
1999 struct list_head ptype_specific;
2f268f12 2000
2f268f12
VF
2001 struct {
2002 struct list_head upper;
2003 struct list_head lower;
2004 } adj_list;
2005
28af22c6
JDB
2006 /* Read-mostly cache-line for fast-path access */
2007 unsigned int flags;
2106efda 2008 unsigned long long priv_flags;
28af22c6
JDB
2009 const struct net_device_ops *netdev_ops;
2010 int ifindex;
2011 unsigned short gflags;
2012 unsigned short hard_header_len;
2013
2014 /* Note : dev->mtu is often read without holding a lock.
2015 * Writers usually hold RTNL.
2016 * It is recommended to use READ_ONCE() to annotate the reads,
2017 * and to use WRITE_ONCE() to annotate the writes.
2018 */
2019 unsigned int mtu;
2020 unsigned short needed_headroom;
2021 unsigned short needed_tailroom;
2022
c8f44aff 2023 netdev_features_t features;
c8f44aff 2024 netdev_features_t hw_features;
c8f44aff 2025 netdev_features_t wanted_features;
c8f44aff 2026 netdev_features_t vlan_features;
6a674e9c 2027 netdev_features_t hw_enc_features;
0d89d203 2028 netdev_features_t mpls_features;
802ab55a 2029 netdev_features_t gso_partial_features;
04ed3e74 2030
28af22c6
JDB
2031 unsigned int min_mtu;
2032 unsigned int max_mtu;
2033 unsigned short type;
2034 unsigned char min_header_len;
2035 unsigned char name_assign_type;
2036
7a66bbc9 2037 int group;
1da177e4 2038
28af22c6 2039 struct net_device_stats stats; /* not used by modern drivers */
015f0688 2040
625788b5 2041 struct net_device_core_stats __percpu *core_stats;
1da177e4 2042
b2d3bcfa
DD
2043 /* Stats to monitor link on/off, flapping */
2044 atomic_t carrier_up_count;
2045 atomic_t carrier_down_count;
2046
b86e0280 2047#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
2048 const struct iw_handler_def *wireless_handlers;
2049 struct iw_public_data *wireless_data;
b86e0280 2050#endif
76fd8593 2051 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
2052#ifdef CONFIG_NET_L3_MASTER_DEV
2053 const struct l3mdev_ops *l3mdev_ops;
2054#endif
f997c55c
AA
2055#if IS_ENABLED(CONFIG_IPV6)
2056 const struct ndisc_ops *ndisc_ops;
2057#endif
1da177e4 2058
9cb0d21d 2059#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
2060 const struct xfrmdev_ops *xfrmdev_ops;
2061#endif
2062
a5c37c63
IL
2063#if IS_ENABLED(CONFIG_TLS_DEVICE)
2064 const struct tlsdev_ops *tlsdev_ops;
2065#endif
2066
3b04ddde
SH
2067 const struct header_ops *header_ops;
2068
536721b1
KK
2069 unsigned char operstate;
2070 unsigned char link_mode;
b00055aa 2071
536721b1
KK
2072 unsigned char if_port;
2073 unsigned char dma;
bdc220da 2074
1da177e4 2075 /* Interface address info. */
536721b1
KK
2076 unsigned char perm_addr[MAX_ADDR_LEN];
2077 unsigned char addr_assign_type;
2078 unsigned char addr_len;
5343da4c
TY
2079 unsigned char upper_level;
2080 unsigned char lower_level;
1fc70edb 2081
a0a9663d 2082 unsigned short neigh_priv_len;
536721b1
KK
2083 unsigned short dev_id;
2084 unsigned short dev_port;
28af22c6
JDB
2085 unsigned short padded;
2086
ccffad25 2087 spinlock_t addr_list_lock;
28af22c6 2088 int irq;
1fc70edb 2089
536721b1
KK
2090 struct netdev_hw_addr_list uc;
2091 struct netdev_hw_addr_list mc;
2092 struct netdev_hw_addr_list dev_addrs;
2093
4c3d5e7b
ED
2094#ifdef CONFIG_SYSFS
2095 struct kset *queues_kset;
1fc70edb
TY
2096#endif
2097#ifdef CONFIG_LOCKDEP
2098 struct list_head unlink_list;
4c3d5e7b 2099#endif
9d45abe1
WC
2100 unsigned int promiscuity;
2101 unsigned int allmulti;
1fc70edb
TY
2102 bool uc_promisc;
2103#ifdef CONFIG_LOCKDEP
2104 unsigned char nested_level;
2105#endif
1da177e4 2106
1da177e4 2107
5e82b4b2 2108 /* Protocol-specific pointers */
65ac6a5f 2109
d11ead75 2110#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 2111 struct vlan_info __rcu *vlan_info;
65ac6a5f 2112#endif
34a430d7 2113#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 2114 struct dsa_port *dsa_ptr;
37cb0620
YX
2115#endif
2116#if IS_ENABLED(CONFIG_TIPC)
536721b1 2117 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 2118#endif
d6c6d0bb 2119#if IS_ENABLED(CONFIG_ATALK)
536721b1 2120 void *atalk_ptr;
89e58148 2121#endif
536721b1 2122 struct in_device __rcu *ip_ptr;
330c7272 2123#if IS_ENABLED(CONFIG_DECNET)
536721b1 2124 struct dn_dev __rcu *dn_ptr;
330c7272 2125#endif
536721b1 2126 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 2127#if IS_ENABLED(CONFIG_AX25)
536721b1 2128 void *ax25_ptr;
19ff13f2 2129#endif
536721b1 2130 struct wireless_dev *ieee80211_ptr;
98a18b6f 2131 struct wpan_dev *ieee802154_ptr;
03c57747
RS
2132#if IS_ENABLED(CONFIG_MPLS_ROUTING)
2133 struct mpls_dev __rcu *mpls_ptr;
2134#endif
583be982
JK
2135#if IS_ENABLED(CONFIG_MCTP)
2136 struct mctp_dev __rcu *mctp_ptr;
2137#endif
1da177e4 2138
9356b8fc 2139/*
cd13539b 2140 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 2141 */
9356b8fc 2142 /* Interface address info used in eth_type_trans() */
adeef3e3 2143 const unsigned char *dev_addr;
f001fde5 2144
0a9627f2 2145 struct netdev_rx_queue *_rx;
0a9627f2 2146 unsigned int num_rx_queues;
62fe0b40 2147 unsigned int real_num_rx_queues;
0a9627f2 2148
7acedaf5 2149 struct bpf_prog __rcu *xdp_prog;
3b47d303 2150 unsigned long gro_flush_timeout;
6f8b12d6 2151 int napi_defer_hard_irqs;
eac1b93c
CL
2152#define GRO_MAX_SIZE 65536
2153 unsigned int gro_max_size;
61391cde 2154 rx_handler_func_t __rcu *rx_handler;
2155 void __rcu *rx_handler_data;
e8a0464c 2156
4cda01e8 2157#ifdef CONFIG_NET_CLS_ACT
46209401 2158 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 2159#endif
24824a09 2160 struct netdev_queue __rcu *ingress_queue;
e687ad60 2161#ifdef CONFIG_NETFILTER_INGRESS
960632ec 2162 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 2163#endif
d2788d34 2164
536721b1 2165 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
2166#ifdef CONFIG_RFS_ACCEL
2167 struct cpu_rmap *rx_cpu_rmap;
2168#endif
2169 struct hlist_node index_hlist;
cd13539b
ED
2170
2171/*
2172 * Cache lines mostly used on transmit path
2173 */
e8a0464c
DM
2174 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
2175 unsigned int num_tx_queues;
fd2ea0a7 2176 unsigned int real_num_tx_queues;
5891cd5e 2177 struct Qdisc __rcu *qdisc;
0cd29503 2178 unsigned int tx_queue_len;
c3f26a26 2179 spinlock_t tx_global_lock;
75ccae62
THJ
2180
2181 struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
cd13539b 2182
bf264145 2183#ifdef CONFIG_XPS
044ab86d 2184 struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
bf264145 2185#endif
1f211a1b 2186#ifdef CONFIG_NET_CLS_ACT
46209401 2187 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 2188#endif
42df6e1d
LW
2189#ifdef CONFIG_NETFILTER_EGRESS
2190 struct nf_hook_entries __rcu *nf_hooks_egress;
2191#endif
0c4f691f 2192
75ccae62
THJ
2193#ifdef CONFIG_NET_SCHED
2194 DECLARE_HASHTABLE (qdisc_hash, 4);
2195#endif
9356b8fc 2196 /* These may be needed for future network-power-down code. */
9356b8fc 2197 struct timer_list watchdog_timer;
75ccae62 2198 int watchdog_timeo;
9356b8fc 2199
829eb208
RP
2200 u32 proto_down_reason;
2201
1da177e4 2202 struct list_head todo_list;
919067cc
ED
2203
2204#ifdef CONFIG_PCPU_DEV_REFCNT
75ccae62 2205 int __percpu *pcpu_refcnt;
919067cc
ED
2206#else
2207 refcount_t dev_refcnt;
2208#endif
4d92b95f 2209 struct ref_tracker_dir refcnt_tracker;
1da177e4 2210
e014debe 2211 struct list_head link_watch_list;
572a103d 2212
1da177e4 2213 enum { NETREG_UNINITIALIZED=0,
b17a7c17 2214 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
2215 NETREG_UNREGISTERING, /* called unregister_netdevice */
2216 NETREG_UNREGISTERED, /* completed unregister todo */
2217 NETREG_RELEASED, /* called free_netdev */
937f1ba5 2218 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
2219 } reg_state:8;
2220
536721b1 2221 bool dismantle;
a2835763
PM
2222
2223 enum {
2224 RTNL_LINK_INITIALIZED,
2225 RTNL_LINK_INITIALIZING,
2226 } rtnl_link_state:16;
1da177e4 2227
cf124db5
DM
2228 bool needs_free_netdev;
2229 void (*priv_destructor)(struct net_device *dev);
1da177e4 2230
1da177e4 2231#ifdef CONFIG_NETPOLL
5fbee843 2232 struct netpoll_info __rcu *npinfo;
1da177e4 2233#endif
eae792b7 2234
0c5c9fb5 2235 possible_net_t nd_net;
4a1c5371 2236
4951704b 2237 /* mid-layer private */
4e096a18
OR
2238 void *ml_priv;
2239 enum netdev_ml_priv_type ml_priv_type;
2240
a7855c78 2241 union {
536721b1 2242 struct pcpu_lstats __percpu *lstats;
8f84985f 2243 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2244 struct pcpu_dstats __percpu *dstats;
a7855c78 2245 };
536721b1 2246
fb585b44 2247#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2248 struct garp_port __rcu *garp_port;
fb585b44
TK
2249#endif
2250#if IS_ENABLED(CONFIG_MRP)
febf018d 2251 struct mrp_port __rcu *mrp_port;
fb585b44 2252#endif
b26ef81c
ED
2253#if IS_ENABLED(CONFIG_NET_DROP_MONITOR)
2254 struct dm_hw_stat_delta __rcu *dm_private;
2255#endif
5e82b4b2 2256 struct device dev;
0c509a6c 2257 const struct attribute_group *sysfs_groups[4];
a953be53 2258 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2259
38f7b870 2260 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2261
82cc1a7a
PWJ
2262 /* for setting kernel sock attribute on TCP connection setup */
2263#define GSO_MAX_SIZE 65536
2264 unsigned int gso_max_size;
30b678d8
BH
2265#define GSO_MAX_SEGS 65535
2266 u16 gso_max_segs;
743b03a8 2267
7a6b6f51 2268#ifdef CONFIG_DCB
32953543 2269 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2270#endif
ffcfe25b 2271 s16 num_tc;
5e82b4b2
BH
2272 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2273 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2274
d11ead75 2275#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2276 unsigned int fcoe_ddp_xid;
5bc1421e 2277#endif
86f8515f 2278#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2279 struct netprio_map __rcu *priomap;
4d288d57 2280#endif
5e82b4b2 2281 struct phy_device *phydev;
e679c9c1 2282 struct sfp_bus *sfp_bus;
1a33e10e 2283 struct lock_class_key *qdisc_tx_busylock;
5e82b4b2 2284 bool proto_down;
61941143 2285 unsigned wol_enabled:1;
29863d41 2286 unsigned threaded:1;
93642e14
JP
2287
2288 struct list_head net_notifier_list;
30e9bb84
AT
2289
2290#if IS_ENABLED(CONFIG_MACSEC)
2291 /* MACsec management functions */
2292 const struct macsec_ops *macsec_ops;
2293#endif
cc4e3835
JK
2294 const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
2295 struct udp_tunnel_nic *udp_tunnel_nic;
7f0a8382
AN
2296
2297 /* protected by rtnl_lock */
2298 struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
d07b26f5
JK
2299
2300 u8 dev_addr_shadow[MAX_ADDR_LEN];
63f13937 2301 netdevice_tracker linkwatch_dev_tracker;
f12bf6f3 2302 netdevice_tracker watchdog_dev_tracker;
b2309a71 2303 netdevice_tracker dev_registered_tracker;
9309f97a 2304 struct rtnl_hw_stats64 *offload_xstats_l3;
1da177e4 2305};
43cb76d9 2306#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2307
b5cdae32
DM
2308static inline bool netif_elide_gro(const struct net_device *dev)
2309{
2310 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2311 return true;
2312 return false;
2313}
2314
1da177e4 2315#define NETDEV_ALIGN 32
1da177e4 2316
4f57c087
JF
2317static inline
2318int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2319{
2320 return dev->prio_tc_map[prio & TC_BITMASK];
2321}
2322
2323static inline
2324int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2325{
2326 if (tc >= dev->num_tc)
2327 return -EINVAL;
2328
2329 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2330 return 0;
2331}
2332
8d059b0f 2333int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2334void netdev_reset_tc(struct net_device *dev);
2335int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2336int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2337
2338static inline
2339int netdev_get_num_tc(struct net_device *dev)
2340{
2341 return dev->num_tc;
2342}
2343
f468f21b
TT
2344static inline void net_prefetch(void *p)
2345{
2346 prefetch(p);
2347#if L1_CACHE_BYTES < 128
2348 prefetch((u8 *)p + L1_CACHE_BYTES);
2349#endif
2350}
2351
2352static inline void net_prefetchw(void *p)
2353{
2354 prefetchw(p);
2355#if L1_CACHE_BYTES < 128
2356 prefetchw((u8 *)p + L1_CACHE_BYTES);
2357#endif
2358}
2359
ffcfe25b
AD
2360void netdev_unbind_sb_channel(struct net_device *dev,
2361 struct net_device *sb_dev);
2362int netdev_bind_sb_channel_queue(struct net_device *dev,
2363 struct net_device *sb_dev,
2364 u8 tc, u16 count, u16 offset);
2365int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2366static inline int netdev_get_sb_channel(struct net_device *dev)
2367{
2368 return max_t(int, -dev->num_tc, 0);
2369}
2370
e8a0464c
DM
2371static inline
2372struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2373 unsigned int index)
2374{
2375 return &dev->_tx[index];
2376}
2377
10c51b56
DB
2378static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2379 const struct sk_buff *skb)
2380{
2381 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2382}
2383
e8a0464c
DM
2384static inline void netdev_for_each_tx_queue(struct net_device *dev,
2385 void (*f)(struct net_device *,
2386 struct netdev_queue *,
2387 void *),
2388 void *arg)
2389{
2390 unsigned int i;
2391
2392 for (i = 0; i < dev->num_tx_queues; i++)
2393 f(dev, &dev->_tx[i], arg);
2394}
2395
1a33e10e
CW
2396#define netdev_lockdep_set_classes(dev) \
2397{ \
2398 static struct lock_class_key qdisc_tx_busylock_key; \
1a33e10e 2399 static struct lock_class_key qdisc_xmit_lock_key; \
845e0ebb 2400 static struct lock_class_key dev_addr_list_lock_key; \
1a33e10e
CW
2401 unsigned int i; \
2402 \
2403 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
845e0ebb
CW
2404 lockdep_set_class(&(dev)->addr_list_lock, \
2405 &dev_addr_list_lock_key); \
1a33e10e
CW
2406 for (i = 0; i < (dev)->num_tx_queues; i++) \
2407 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2408 &qdisc_xmit_lock_key); \
2409}
2410
b71b5837
PA
2411u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2412 struct net_device *sb_dev);
4bd97d51
PA
2413struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2414 struct sk_buff *skb,
2415 struct net_device *sb_dev);
8c4c49df 2416
871b642a
PA
2417/* returns the headroom that the master device needs to take in account
2418 * when forwarding to this dev
2419 */
2420static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2421{
2422 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2423}
2424
2425static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2426{
2427 if (dev->netdev_ops->ndo_set_rx_headroom)
2428 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2429}
2430
2431/* set the device rx headroom to the dev's default */
2432static inline void netdev_reset_rx_headroom(struct net_device *dev)
2433{
2434 netdev_set_rx_headroom(dev, -1);
2435}
2436
4e096a18
OR
2437static inline void *netdev_get_ml_priv(struct net_device *dev,
2438 enum netdev_ml_priv_type type)
2439{
2440 if (dev->ml_priv_type != type)
2441 return NULL;
2442
2443 return dev->ml_priv;
2444}
2445
2446static inline void netdev_set_ml_priv(struct net_device *dev,
2447 void *ml_priv,
2448 enum netdev_ml_priv_type type)
2449{
2450 WARN(dev->ml_priv_type && dev->ml_priv_type != type,
2451 "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
2452 dev->ml_priv_type, type);
2453 WARN(!dev->ml_priv_type && dev->ml_priv,
2454 "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
2455
2456 dev->ml_priv = ml_priv;
2457 dev->ml_priv_type = type;
2458}
2459
c346dca1
YH
2460/*
2461 * Net namespace inlines
2462 */
2463static inline
2464struct net *dev_net(const struct net_device *dev)
2465{
c2d9ba9b 2466 return read_pnet(&dev->nd_net);
c346dca1
YH
2467}
2468
2469static inline
f5aa23fd 2470void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2471{
0c5c9fb5 2472 write_pnet(&dev->nd_net, net);
c346dca1
YH
2473}
2474
bea3348e
SH
2475/**
2476 * netdev_priv - access network device private data
2477 * @dev: network device
2478 *
2479 * Get network device private data
2480 */
6472ce60 2481static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2482{
1ce8e7b5 2483 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2484}
2485
1da177e4
LT
2486/* Set the sysfs physical device reference for the network logical device
2487 * if set prior to registration will cause a symlink during initialization.
2488 */
43cb76d9 2489#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2490
384912ed 2491/* Set the sysfs device type for the network logical device to allow
3f79410c 2492 * fine-grained identification of different network device types. For
5e82b4b2 2493 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2494 */
2495#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2496
82dc3c63
ED
2497/* Default NAPI poll() weight
2498 * Device drivers are strongly advised to not use bigger value
2499 */
2500#define NAPI_POLL_WEIGHT 64
2501
3b582cc1 2502/**
5e82b4b2 2503 * netif_napi_add - initialize a NAPI context
3b582cc1 2504 * @dev: network device
5e82b4b2 2505 * @napi: NAPI context
3b582cc1
SH
2506 * @poll: polling function
2507 * @weight: default weight
2508 *
5e82b4b2
BH
2509 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2510 * *any* of the other NAPI-related functions.
3b582cc1 2511 */
d565b0a1
HX
2512void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2513 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2514
d64b5e85 2515/**
5e82b4b2 2516 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2517 * @dev: network device
5e82b4b2 2518 * @napi: NAPI context
d64b5e85
ED
2519 * @poll: polling function
2520 * @weight: default weight
2521 *
2522 * This variant of netif_napi_add() should be used from drivers using NAPI
2523 * to exclusively poll a TX queue.
2524 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2525 */
2526static inline void netif_tx_napi_add(struct net_device *dev,
2527 struct napi_struct *napi,
2528 int (*poll)(struct napi_struct *, int),
2529 int weight)
2530{
2531 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2532 netif_napi_add(dev, napi, poll, weight);
2533}
2534
5198d545
JK
2535/**
2536 * __netif_napi_del - remove a NAPI context
2537 * @napi: NAPI context
2538 *
2539 * Warning: caller must observe RCU grace period before freeing memory
2540 * containing @napi. Drivers might want to call this helper to combine
2541 * all the needed RCU grace periods into a single one.
2542 */
2543void __netif_napi_del(struct napi_struct *napi);
2544
d8156534 2545/**
5e82b4b2
BH
2546 * netif_napi_del - remove a NAPI context
2547 * @napi: NAPI context
d8156534 2548 *
5e82b4b2 2549 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2550 */
5198d545
JK
2551static inline void netif_napi_del(struct napi_struct *napi)
2552{
2553 __netif_napi_del(napi);
2554 synchronize_net();
2555}
d565b0a1 2556
1da177e4 2557struct packet_type {
f2ccd8fa 2558 __be16 type; /* This is really htons(ether_type). */
fa788d98 2559 bool ignore_outgoing;
f2ccd8fa 2560 struct net_device *dev; /* NULL is wildcarded here */
f1d9268e 2561 netdevice_tracker dev_tracker;
f2ccd8fa
DM
2562 int (*func) (struct sk_buff *,
2563 struct net_device *,
2564 struct packet_type *,
2565 struct net_device *);
17266ee9
EC
2566 void (*list_func) (struct list_head *,
2567 struct packet_type *,
2568 struct net_device *);
c0de08d0
EL
2569 bool (*id_match)(struct packet_type *ptype,
2570 struct sock *sk);
47934e06 2571 struct net *af_packet_net;
1da177e4
LT
2572 void *af_packet_priv;
2573 struct list_head list;
2574};
2575
f191a1d1 2576struct offload_callbacks {
576a30eb 2577 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2578 netdev_features_t features);
d4546c25
DM
2579 struct sk_buff *(*gro_receive)(struct list_head *head,
2580 struct sk_buff *skb);
299603e8 2581 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2582};
2583
2584struct packet_offload {
2585 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2586 u16 priority;
f191a1d1
VY
2587 struct offload_callbacks callbacks;
2588 struct list_head list;
1da177e4
LT
2589};
2590
5e82b4b2 2591/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2592struct pcpu_sw_netstats {
2593 u64 rx_packets;
2594 u64 rx_bytes;
2595 u64 tx_packets;
2596 u64 tx_bytes;
2597 struct u64_stats_sync syncp;
9a5ee462 2598} __aligned(4 * sizeof(u64));
52bb6677
LR
2599
2600struct pcpu_lstats {
fd2f4737
ED
2601 u64_stats_t packets;
2602 u64_stats_t bytes;
52bb6677 2603 struct u64_stats_sync syncp;
9a5ee462 2604} __aligned(2 * sizeof(u64));
8f84985f 2605
de7d5084
ED
2606void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2607
451b05f4
FF
2608static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
2609{
2610 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2611
2612 u64_stats_update_begin(&tstats->syncp);
2613 tstats->rx_bytes += len;
2614 tstats->rx_packets++;
2615 u64_stats_update_end(&tstats->syncp);
2616}
2617
d3fd6548
HK
2618static inline void dev_sw_netstats_tx_add(struct net_device *dev,
2619 unsigned int packets,
2620 unsigned int len)
2621{
2622 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2623
2624 u64_stats_update_begin(&tstats->syncp);
2625 tstats->tx_bytes += len;
2626 tstats->tx_packets += packets;
2627 u64_stats_update_end(&tstats->syncp);
2628}
2629
dd5382a0
ED
2630static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
2631{
2632 struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
2633
2634 u64_stats_update_begin(&lstats->syncp);
fd2f4737
ED
2635 u64_stats_add(&lstats->bytes, len);
2636 u64_stats_inc(&lstats->packets);
dd5382a0
ED
2637 u64_stats_update_end(&lstats->syncp);
2638}
2639
aabc92bb
PNA
2640#define __netdev_alloc_pcpu_stats(type, gfp) \
2641({ \
2642 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2643 if (pcpu_stats) { \
2644 int __cpu; \
2645 for_each_possible_cpu(__cpu) { \
2646 typeof(type) *stat; \
2647 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2648 u64_stats_init(&stat->syncp); \
2649 } \
2650 } \
2651 pcpu_stats; \
1c213bd2
WC
2652})
2653
aabc92bb 2654#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2655 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2656
81b01894
HK
2657#define devm_netdev_alloc_pcpu_stats(dev, type) \
2658({ \
2659 typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
2660 if (pcpu_stats) { \
2661 int __cpu; \
2662 for_each_possible_cpu(__cpu) { \
2663 typeof(type) *stat; \
2664 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2665 u64_stats_init(&stat->syncp); \
2666 } \
2667 } \
2668 pcpu_stats; \
2669})
2670
764f5e54
JP
2671enum netdev_lag_tx_type {
2672 NETDEV_LAG_TX_TYPE_UNKNOWN,
2673 NETDEV_LAG_TX_TYPE_RANDOM,
2674 NETDEV_LAG_TX_TYPE_BROADCAST,
2675 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2676 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2677 NETDEV_LAG_TX_TYPE_HASH,
2678};
2679
f44aa9ef
JH
2680enum netdev_lag_hash {
2681 NETDEV_LAG_HASH_NONE,
2682 NETDEV_LAG_HASH_L2,
2683 NETDEV_LAG_HASH_L34,
2684 NETDEV_LAG_HASH_L23,
2685 NETDEV_LAG_HASH_E23,
2686 NETDEV_LAG_HASH_E34,
7b8fc010 2687 NETDEV_LAG_HASH_VLAN_SRCMAC,
f44aa9ef
JH
2688 NETDEV_LAG_HASH_UNKNOWN,
2689};
2690
764f5e54
JP
2691struct netdev_lag_upper_info {
2692 enum netdev_lag_tx_type tx_type;
f44aa9ef 2693 enum netdev_lag_hash hash_type;
764f5e54
JP
2694};
2695
fb1b2e3c
JP
2696struct netdev_lag_lower_state_info {
2697 u8 link_up : 1,
2698 tx_enabled : 1;
2699};
2700
1da177e4
LT
2701#include <linux/notifier.h>
2702
ede2762d
KT
2703/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2704 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2705 * adding new types.
dcfe1421 2706 */
ede2762d
KT
2707enum netdev_cmd {
2708 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2709 NETDEV_DOWN,
2710 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2711 detected a hardware crash and restarted
2712 - we can use this eg to kick tcp sessions
2713 once done */
ede2762d
KT
2714 NETDEV_CHANGE, /* Notify device state change */
2715 NETDEV_REGISTER,
2716 NETDEV_UNREGISTER,
2717 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2718 NETDEV_CHANGEADDR, /* notify after the address change */
2719 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2720 NETDEV_GOING_DOWN,
2721 NETDEV_CHANGENAME,
2722 NETDEV_FEAT_CHANGE,
2723 NETDEV_BONDING_FAILOVER,
2724 NETDEV_PRE_UP,
2725 NETDEV_PRE_TYPE_CHANGE,
2726 NETDEV_POST_TYPE_CHANGE,
2727 NETDEV_POST_INIT,
ede2762d
KT
2728 NETDEV_RELEASE,
2729 NETDEV_NOTIFY_PEERS,
2730 NETDEV_JOIN,
2731 NETDEV_CHANGEUPPER,
2732 NETDEV_RESEND_IGMP,
2733 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2734 NETDEV_CHANGEINFODATA,
2735 NETDEV_BONDING_INFO,
2736 NETDEV_PRECHANGEUPPER,
2737 NETDEV_CHANGELOWERSTATE,
2738 NETDEV_UDP_TUNNEL_PUSH_INFO,
2739 NETDEV_UDP_TUNNEL_DROP_INFO,
2740 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2741 NETDEV_CVLAN_FILTER_PUSH_INFO,
2742 NETDEV_CVLAN_FILTER_DROP_INFO,
2743 NETDEV_SVLAN_FILTER_PUSH_INFO,
2744 NETDEV_SVLAN_FILTER_DROP_INFO,
9309f97a
PM
2745 NETDEV_OFFLOAD_XSTATS_ENABLE,
2746 NETDEV_OFFLOAD_XSTATS_DISABLE,
2747 NETDEV_OFFLOAD_XSTATS_REPORT_USED,
2748 NETDEV_OFFLOAD_XSTATS_REPORT_DELTA,
ede2762d
KT
2749};
2750const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2751
f629d208
JP
2752int register_netdevice_notifier(struct notifier_block *nb);
2753int unregister_netdevice_notifier(struct notifier_block *nb);
a30c7b42
JP
2754int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2755int unregister_netdevice_notifier_net(struct net *net,
2756 struct notifier_block *nb);
93642e14
JP
2757int register_netdevice_notifier_dev_net(struct net_device *dev,
2758 struct notifier_block *nb,
2759 struct netdev_net_notifier *nn);
2760int unregister_netdevice_notifier_dev_net(struct net_device *dev,
2761 struct notifier_block *nb,
2762 struct netdev_net_notifier *nn);
351638e7
JP
2763
2764struct netdev_notifier_info {
51d0c047
DA
2765 struct net_device *dev;
2766 struct netlink_ext_ack *extack;
351638e7
JP
2767};
2768
af7d6cce
SD
2769struct netdev_notifier_info_ext {
2770 struct netdev_notifier_info info; /* must be first */
2771 union {
2772 u32 mtu;
2773 } ext;
2774};
2775
be9efd36
JP
2776struct netdev_notifier_change_info {
2777 struct netdev_notifier_info info; /* must be first */
2778 unsigned int flags_changed;
2779};
2780
0e4ead9d
JP
2781struct netdev_notifier_changeupper_info {
2782 struct netdev_notifier_info info; /* must be first */
2783 struct net_device *upper_dev; /* new upper dev */
2784 bool master; /* is upper dev master */
5e82b4b2 2785 bool linking; /* is the notification for link or unlink */
29bf24af 2786 void *upper_info; /* upper dev info */
0e4ead9d
JP
2787};
2788
04d48266
JP
2789struct netdev_notifier_changelowerstate_info {
2790 struct netdev_notifier_info info; /* must be first */
2791 void *lower_state_info; /* is lower dev state */
2792};
2793
1570415f
PM
2794struct netdev_notifier_pre_changeaddr_info {
2795 struct netdev_notifier_info info; /* must be first */
2796 const unsigned char *dev_addr;
2797};
2798
9309f97a
PM
2799enum netdev_offload_xstats_type {
2800 NETDEV_OFFLOAD_XSTATS_TYPE_L3 = 1,
2801};
2802
2803struct netdev_notifier_offload_xstats_info {
2804 struct netdev_notifier_info info; /* must be first */
2805 enum netdev_offload_xstats_type type;
2806
2807 union {
2808 /* NETDEV_OFFLOAD_XSTATS_REPORT_DELTA */
2809 struct netdev_notifier_offload_xstats_rd *report_delta;
2810 /* NETDEV_OFFLOAD_XSTATS_REPORT_USED */
2811 struct netdev_notifier_offload_xstats_ru *report_used;
2812 };
2813};
2814
2815int netdev_offload_xstats_enable(struct net_device *dev,
2816 enum netdev_offload_xstats_type type,
2817 struct netlink_ext_ack *extack);
2818int netdev_offload_xstats_disable(struct net_device *dev,
2819 enum netdev_offload_xstats_type type);
2820bool netdev_offload_xstats_enabled(const struct net_device *dev,
2821 enum netdev_offload_xstats_type type);
2822int netdev_offload_xstats_get(struct net_device *dev,
2823 enum netdev_offload_xstats_type type,
2824 struct rtnl_hw_stats64 *stats, bool *used,
2825 struct netlink_ext_ack *extack);
2826void
2827netdev_offload_xstats_report_delta(struct netdev_notifier_offload_xstats_rd *rd,
2828 const struct rtnl_hw_stats64 *stats);
2829void
2830netdev_offload_xstats_report_used(struct netdev_notifier_offload_xstats_ru *ru);
2831void netdev_offload_xstats_push_delta(struct net_device *dev,
2832 enum netdev_offload_xstats_type type,
2833 const struct rtnl_hw_stats64 *stats);
2834
75538c2b
CW
2835static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2836 struct net_device *dev)
2837{
2838 info->dev = dev;
51d0c047 2839 info->extack = NULL;
75538c2b
CW
2840}
2841
351638e7
JP
2842static inline struct net_device *
2843netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2844{
2845 return info->dev;
2846}
2847
51d0c047
DA
2848static inline struct netlink_ext_ack *
2849netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2850{
2851 return info->extack;
2852}
2853
f629d208 2854int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2855
2856
1da177e4
LT
2857extern rwlock_t dev_base_lock; /* Device list lock */
2858
881d966b
EB
2859#define for_each_netdev(net, d) \
2860 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2861#define for_each_netdev_reverse(net, d) \
2862 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2863#define for_each_netdev_rcu(net, d) \
2864 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2865#define for_each_netdev_safe(net, d, n) \
2866 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2867#define for_each_netdev_continue(net, d) \
2868 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
afa0df59
JP
2869#define for_each_netdev_continue_reverse(net, d) \
2870 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
2871 dev_list)
254245d2 2872#define for_each_netdev_continue_rcu(net, d) \
2873 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2874#define for_each_netdev_in_bond_rcu(bond, slave) \
2875 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2876 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2877#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2878
a050c33f
DL
2879static inline struct net_device *next_net_device(struct net_device *dev)
2880{
2881 struct list_head *lh;
2882 struct net *net;
2883
c346dca1 2884 net = dev_net(dev);
a050c33f
DL
2885 lh = dev->dev_list.next;
2886 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2887}
2888
ce81b76a
ED
2889static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2890{
2891 struct list_head *lh;
2892 struct net *net;
2893
2894 net = dev_net(dev);
ccf43438 2895 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2896 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2897}
2898
a050c33f
DL
2899static inline struct net_device *first_net_device(struct net *net)
2900{
2901 return list_empty(&net->dev_base_head) ? NULL :
2902 net_device_entry(net->dev_base_head.next);
2903}
7562f876 2904
ccf43438
ED
2905static inline struct net_device *first_net_device_rcu(struct net *net)
2906{
2907 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2908
2909 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2910}
2911
f629d208 2912int netdev_boot_setup_check(struct net_device *dev);
f629d208
JP
2913struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2914 const char *hwaddr);
2915struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
f629d208
JP
2916void dev_add_pack(struct packet_type *pt);
2917void dev_remove_pack(struct packet_type *pt);
2918void __dev_remove_pack(struct packet_type *pt);
2919void dev_add_offload(struct packet_offload *po);
2920void dev_remove_offload(struct packet_offload *po);
f629d208 2921
a54acb3a 2922int dev_get_iflink(const struct net_device *dev);
fc4099f1 2923int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
ddb94eaf
PNA
2924int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
2925 struct net_device_path_stack *stack);
6c555490
WC
2926struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2927 unsigned short mask);
f629d208
JP
2928struct net_device *dev_get_by_name(struct net *net, const char *name);
2929struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2930struct net_device *__dev_get_by_name(struct net *net, const char *name);
75ea27d0 2931bool netdev_name_in_use(struct net *net, const char *name);
f629d208 2932int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 2933int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 2934void dev_close(struct net_device *dev);
2935void dev_close_many(struct list_head *head, bool unlink);
f629d208 2936void dev_disable_lro(struct net_device *dev);
0c4b51f0 2937int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2938u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 2939 struct net_device *sb_dev);
a4ea8a3d 2940u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 2941 struct net_device *sb_dev);
36ccdf85 2942
c526fd8f 2943int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev);
36ccdf85
BT
2944int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
2945
c526fd8f
PB
2946static inline int dev_queue_xmit(struct sk_buff *skb)
2947{
2948 return __dev_queue_xmit(skb, NULL);
2949}
2950
2951static inline int dev_queue_xmit_accel(struct sk_buff *skb,
2952 struct net_device *sb_dev)
2953{
2954 return __dev_queue_xmit(skb, sb_dev);
2955}
2956
36ccdf85
BT
2957static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
2958{
2959 int ret;
2960
2961 ret = __dev_direct_xmit(skb, queue_id);
2962 if (!dev_xmit_complete(ret))
2963 kfree_skb(skb);
2964 return ret;
2965}
2966
f629d208
JP
2967int register_netdevice(struct net_device *dev);
2968void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2969void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2970static inline void unregister_netdevice(struct net_device *dev)
2971{
2972 unregister_netdevice_queue(dev, NULL);
2973}
2974
f629d208
JP
2975int netdev_refcnt_read(const struct net_device *dev);
2976void free_netdev(struct net_device *dev);
74d332c1 2977void netdev_freemem(struct net_device *dev);
f629d208 2978int init_dummy_netdev(struct net_device *dev);
937f1ba5 2979
cff9f12b
MG
2980struct net_device *netdev_get_xmit_slave(struct net_device *dev,
2981 struct sk_buff *skb,
2982 bool all_slaves);
719a402c
TT
2983struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
2984 struct sock *sk);
f629d208
JP
2985struct net_device *dev_get_by_index(struct net *net, int ifindex);
2986struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2987struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2988struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208 2989int dev_restart(struct net_device *dev);
86911732 2990
5f114163 2991
0c4e8581
SH
2992static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
2993 unsigned short type,
3b04ddde 2994 const void *daddr, const void *saddr,
95c96174 2995 unsigned int len)
0c4e8581 2996{
f1ecfd5d 2997 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 2998 return 0;
3b04ddde
SH
2999
3000 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
3001}
3002
b95cce35
SH
3003static inline int dev_parse_header(const struct sk_buff *skb,
3004 unsigned char *haddr)
3005{
3006 const struct net_device *dev = skb->dev;
3007
1b83336b 3008 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 3009 return 0;
3b04ddde 3010 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
3011}
3012
e78b2915
MM
3013static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
3014{
3015 const struct net_device *dev = skb->dev;
3016
3017 if (!dev->header_ops || !dev->header_ops->parse_protocol)
3018 return 0;
3019 return dev->header_ops->parse_protocol(skb);
3020}
3021
2793a23a
WB
3022/* ll_header must have at least hard_header_len allocated */
3023static inline bool dev_validate_header(const struct net_device *dev,
3024 char *ll_header, int len)
3025{
3026 if (likely(len >= dev->hard_header_len))
3027 return true;
217e6fa2
WB
3028 if (len < dev->min_header_len)
3029 return false;
2793a23a
WB
3030
3031 if (capable(CAP_SYS_RAWIO)) {
3032 memset(ll_header + len, 0, dev->hard_header_len - len);
3033 return true;
3034 }
3035
3036 if (dev->header_ops && dev->header_ops->validate)
3037 return dev->header_ops->validate(ll_header, len);
3038
3039 return false;
3040}
3041
d5496990
EB
3042static inline bool dev_has_header(const struct net_device *dev)
3043{
3044 return dev->header_ops && dev->header_ops->create;
3045}
3046
1da177e4 3047/*
5e82b4b2 3048 * Incoming packets are placed on per-CPU queues
1da177e4 3049 */
d94d9fee 3050struct softnet_data {
1da177e4 3051 struct list_head poll_list;
6e7676c1 3052 struct sk_buff_head process_queue;
1da177e4 3053
dee42870 3054 /* stats */
cd7b5396
DM
3055 unsigned int processed;
3056 unsigned int time_squeeze;
cd7b5396 3057 unsigned int received_rps;
fd793d89 3058#ifdef CONFIG_RPS
88751275 3059 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
3060#endif
3061#ifdef CONFIG_NET_FLOW_LIMIT
3062 struct sd_flow_limit __rcu *flow_limit;
3063#endif
3064 struct Qdisc *output_queue;
3065 struct Qdisc **output_queue_tailp;
3066 struct sk_buff *completion_queue;
f53c7239
SK
3067#ifdef CONFIG_XFRM_OFFLOAD
3068 struct sk_buff_head xfrm_backlog;
3069#endif
97cdcf37
FW
3070 /* written and read only by owning cpu: */
3071 struct {
3072 u16 recursion;
3073 u8 more;
2f1e85b1
TZ
3074#ifdef CONFIG_NET_EGRESS
3075 u8 skip_txqueue;
3076#endif
97cdcf37 3077 } xmit;
4cdb1e2e 3078#ifdef CONFIG_RPS
501e7ef5
ED
3079 /* input_queue_head should be written by cpu owning this struct,
3080 * and only read by other cpus. Worth using a cache line.
3081 */
3082 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3083
3084 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3085 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3086 struct softnet_data *rps_ipi_next;
3087 unsigned int cpu;
76cc8b13 3088 unsigned int input_queue_tail;
1e94d72f 3089#endif
95c96174 3090 unsigned int dropped;
0a9627f2 3091 struct sk_buff_head input_pkt_queue;
bea3348e 3092 struct napi_struct backlog;
99bbc707 3093
68822bdf
ED
3094 /* Another possibly contended cache line */
3095 spinlock_t defer_lock ____cacheline_aligned_in_smp;
3096 int defer_count;
3097 struct sk_buff *defer_list;
3098 call_single_data_t defer_csd;
1da177e4
LT
3099};
3100
76cc8b13 3101static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3102{
3103#ifdef CONFIG_RPS
76cc8b13
TH
3104 sd->input_queue_head++;
3105#endif
3106}
3107
3108static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3109 unsigned int *qtail)
3110{
3111#ifdef CONFIG_RPS
3112 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3113#endif
3114}
3115
0a9627f2 3116DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3117
97cdcf37
FW
3118static inline int dev_recursion_level(void)
3119{
28b05b92 3120 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3121}
3122
fb7861d1 3123#define XMIT_RECURSION_LIMIT 8
97cdcf37
FW
3124static inline bool dev_xmit_recursion(void)
3125{
3126 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3127 XMIT_RECURSION_LIMIT);
3128}
3129
3130static inline void dev_xmit_recursion_inc(void)
3131{
3132 __this_cpu_inc(softnet_data.xmit.recursion);
3133}
3134
3135static inline void dev_xmit_recursion_dec(void)
3136{
3137 __this_cpu_dec(softnet_data.xmit.recursion);
3138}
3139
f629d208 3140void __netif_schedule(struct Qdisc *q);
46e5da40 3141void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3142
fd2ea0a7
DM
3143static inline void netif_tx_schedule_all(struct net_device *dev)
3144{
3145 unsigned int i;
3146
3147 for (i = 0; i < dev->num_tx_queues; i++)
3148 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3149}
3150
f9a7cbbf 3151static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3152{
73466498 3153 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3154}
3155
bea3348e
SH
3156/**
3157 * netif_start_queue - allow transmit
3158 * @dev: network device
3159 *
3160 * Allow upper layers to call the device hard_start_xmit routine.
3161 */
1da177e4
LT
3162static inline void netif_start_queue(struct net_device *dev)
3163{
e8a0464c 3164 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3165}
3166
fd2ea0a7
DM
3167static inline void netif_tx_start_all_queues(struct net_device *dev)
3168{
3169 unsigned int i;
3170
3171 for (i = 0; i < dev->num_tx_queues; i++) {
3172 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3173 netif_tx_start_queue(txq);
3174 }
3175}
3176
46e5da40 3177void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3178
d29f749e
DJ
3179/**
3180 * netif_wake_queue - restart transmit
3181 * @dev: network device
3182 *
3183 * Allow upper layers to call the device hard_start_xmit routine.
3184 * Used for flow control when transmit resources are available.
3185 */
79d16385
DM
3186static inline void netif_wake_queue(struct net_device *dev)
3187{
e8a0464c 3188 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3189}
3190
fd2ea0a7
DM
3191static inline void netif_tx_wake_all_queues(struct net_device *dev)
3192{
3193 unsigned int i;
3194
3195 for (i = 0; i < dev->num_tx_queues; i++) {
3196 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3197 netif_tx_wake_queue(txq);
3198 }
3199}
3200
f9a7cbbf 3201static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3202{
73466498 3203 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3204}
3205
bea3348e
SH
3206/**
3207 * netif_stop_queue - stop transmitted packets
3208 * @dev: network device
3209 *
3210 * Stop upper layers calling the device hard_start_xmit routine.
3211 * Used for flow control when transmit resources are unavailable.
3212 */
1da177e4
LT
3213static inline void netif_stop_queue(struct net_device *dev)
3214{
e8a0464c 3215 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3216}
3217
a2029240 3218void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3219
4d29515f 3220static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3221{
73466498 3222 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3223}
3224
bea3348e
SH
3225/**
3226 * netif_queue_stopped - test if transmit queue is flowblocked
3227 * @dev: network device
3228 *
3229 * Test if transmit queue on device is currently unable to send.
3230 */
4d29515f 3231static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3232{
e8a0464c 3233 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3234}
3235
4d29515f 3236static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3237{
73466498
TH
3238 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3239}
3240
8e2f1a63
DB
3241static inline bool
3242netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3243{
3244 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3245}
3246
8e2f1a63
DB
3247static inline bool
3248netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3249{
3250 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3251}
3252
f57bac3c
VM
3253/**
3254 * netdev_queue_set_dql_min_limit - set dql minimum limit
3255 * @dev_queue: pointer to transmit queue
3256 * @min_limit: dql minimum limit
3257 *
3258 * Forces xmit_more() to return true until the minimum threshold
3259 * defined by @min_limit is reached (or until the tx queue is
3260 * empty). Warning: to be use with care, misuse will impact the
3261 * latency.
3262 */
3263static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
3264 unsigned int min_limit)
3265{
3266#ifdef CONFIG_BQL
3267 dev_queue->dql.min_limit = min_limit;
3268#endif
3269}
3270
53511453
ED
3271/**
3272 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3273 * @dev_queue: pointer to transmit queue
3274 *
3275 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3276 * to give appropriate hint to the CPU.
53511453
ED
3277 */
3278static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3279{
3280#ifdef CONFIG_BQL
3281 prefetchw(&dev_queue->dql.num_queued);
3282#endif
3283}
3284
3285/**
3286 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3287 * @dev_queue: pointer to transmit queue
3288 *
3289 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3290 * to give appropriate hint to the CPU.
53511453
ED
3291 */
3292static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3293{
3294#ifdef CONFIG_BQL
3295 prefetchw(&dev_queue->dql.limit);
3296#endif
3297}
3298
c5d67bd7
TH
3299static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3300 unsigned int bytes)
3301{
114cf580
TH
3302#ifdef CONFIG_BQL
3303 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3304
3305 if (likely(dql_avail(&dev_queue->dql) >= 0))
3306 return;
3307
3308 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3309
3310 /*
3311 * The XOFF flag must be set before checking the dql_avail below,
3312 * because in netdev_tx_completed_queue we update the dql_completed
3313 * before checking the XOFF flag.
3314 */
3315 smp_mb();
3316
3317 /* check again in case another CPU has just made room avail */
3318 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3319 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3320#endif
c5d67bd7
TH
3321}
3322
3e59020a
ED
3323/* Variant of netdev_tx_sent_queue() for drivers that are aware
3324 * that they should not test BQL status themselves.
3325 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3326 * skb of a batch.
3327 * Returns true if the doorbell must be used to kick the NIC.
3328 */
3329static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3330 unsigned int bytes,
3331 bool xmit_more)
3332{
3333 if (xmit_more) {
3334#ifdef CONFIG_BQL
3335 dql_queued(&dev_queue->dql, bytes);
3336#endif
3337 return netif_tx_queue_stopped(dev_queue);
3338 }
3339 netdev_tx_sent_queue(dev_queue, bytes);
3340 return true;
3341}
3342
0042d0c8
FF
3343/**
3344 * netdev_sent_queue - report the number of bytes queued to hardware
3345 * @dev: network device
3346 * @bytes: number of bytes queued to the hardware device queue
3347 *
3348 * Report the number of bytes queued for sending/completion to the network
3349 * device hardware queue. @bytes should be a good approximation and should
3350 * exactly match netdev_completed_queue() @bytes
3351 */
c5d67bd7
TH
3352static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3353{
3354 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3355}
3356
620344c4
HK
3357static inline bool __netdev_sent_queue(struct net_device *dev,
3358 unsigned int bytes,
3359 bool xmit_more)
3360{
3361 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3362 xmit_more);
3363}
3364
c5d67bd7 3365static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3366 unsigned int pkts, unsigned int bytes)
c5d67bd7 3367{
114cf580 3368#ifdef CONFIG_BQL
b37c0fbe
AD
3369 if (unlikely(!bytes))
3370 return;
3371
3372 dql_completed(&dev_queue->dql, bytes);
3373
3374 /*
3375 * Without the memory barrier there is a small possiblity that
3376 * netdev_tx_sent_queue will miss the update and cause the queue to
3377 * be stopped forever
3378 */
3379 smp_mb();
3380
f3acd33d 3381 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3382 return;
3383
3384 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3385 netif_schedule_queue(dev_queue);
114cf580 3386#endif
c5d67bd7
TH
3387}
3388
0042d0c8
FF
3389/**
3390 * netdev_completed_queue - report bytes and packets completed by device
3391 * @dev: network device
3392 * @pkts: actual number of packets sent over the medium
3393 * @bytes: actual number of bytes sent over the medium
3394 *
3395 * Report the number of bytes and packets transmitted by the network device
3396 * hardware queue over the physical medium, @bytes must exactly match the
3397 * @bytes amount passed to netdev_sent_queue()
3398 */
c5d67bd7 3399static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3400 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3401{
3402 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3403}
3404
3405static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3406{
114cf580 3407#ifdef CONFIG_BQL
5c490354 3408 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3409 dql_reset(&q->dql);
3410#endif
c5d67bd7
TH
3411}
3412
0042d0c8
FF
3413/**
3414 * netdev_reset_queue - reset the packets and bytes count of a network device
3415 * @dev_queue: network device
3416 *
3417 * Reset the bytes and packet count of a network device and clear the
3418 * software flow control OFF bit for this network device
3419 */
c5d67bd7
TH
3420static inline void netdev_reset_queue(struct net_device *dev_queue)
3421{
3422 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3423}
3424
b9507bda
DB
3425/**
3426 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3427 * @dev: network device
3428 * @queue_index: given tx queue index
3429 *
3430 * Returns 0 if given tx queue index >= number of device tx queues,
3431 * otherwise returns the originally passed tx queue index.
3432 */
3433static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3434{
3435 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3436 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3437 dev->name, queue_index,
3438 dev->real_num_tx_queues);
3439 return 0;
3440 }
3441
3442 return queue_index;
3443}
3444
bea3348e
SH
3445/**
3446 * netif_running - test if up
3447 * @dev: network device
3448 *
3449 * Test if the device has been brought up.
3450 */
4d29515f 3451static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3452{
3453 return test_bit(__LINK_STATE_START, &dev->state);
3454}
3455
f25f4e44 3456/*
5e82b4b2 3457 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3458 * stop, and a check if it's stopped. All other device management is
3459 * done at the overall netdevice level.
3460 * Also test the device if we're multiqueue.
3461 */
bea3348e
SH
3462
3463/**
3464 * netif_start_subqueue - allow sending packets on subqueue
3465 * @dev: network device
3466 * @queue_index: sub queue index
3467 *
3468 * Start individual transmit queue of a device with multiple transmit queues.
3469 */
f25f4e44
PWJ
3470static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3471{
fd2ea0a7 3472 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3473
3474 netif_tx_start_queue(txq);
f25f4e44
PWJ
3475}
3476
bea3348e
SH
3477/**
3478 * netif_stop_subqueue - stop sending packets on subqueue
3479 * @dev: network device
3480 * @queue_index: sub queue index
3481 *
3482 * Stop individual transmit queue of a device with multiple transmit queues.
3483 */
f25f4e44
PWJ
3484static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3485{
fd2ea0a7 3486 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3487 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3488}
3489
bea3348e 3490/**
270f3385 3491 * __netif_subqueue_stopped - test status of subqueue
bea3348e
SH
3492 * @dev: network device
3493 * @queue_index: sub queue index
3494 *
3495 * Check individual transmit queue of a device with multiple transmit queues.
3496 */
4d29515f
DM
3497static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3498 u16 queue_index)
f25f4e44 3499{
fd2ea0a7 3500 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3501
3502 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3503}
3504
270f3385
MCC
3505/**
3506 * netif_subqueue_stopped - test status of subqueue
3507 * @dev: network device
3508 * @skb: sub queue buffer pointer
3509 *
3510 * Check individual transmit queue of a device with multiple transmit queues.
3511 */
4d29515f
DM
3512static inline bool netif_subqueue_stopped(const struct net_device *dev,
3513 struct sk_buff *skb)
668f895a
PE
3514{
3515 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3516}
bea3348e 3517
738b35cc
FF
3518/**
3519 * netif_wake_subqueue - allow sending packets on subqueue
3520 * @dev: network device
3521 * @queue_index: sub queue index
3522 *
3523 * Resume individual transmit queue of a device with multiple transmit queues.
3524 */
3525static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3526{
3527 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3528
3529 netif_tx_wake_queue(txq);
3530}
f25f4e44 3531
537c00de 3532#ifdef CONFIG_XPS
53af53ae 3533int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3534 u16 index);
80d19669 3535int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
044ab86d 3536 u16 index, enum xps_map_type type);
80d19669
AN
3537
3538/**
3539 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3540 * @j: CPU/Rx queue index
3541 * @mask: bitmask of all cpus/rx queues
3542 * @nr_bits: number of bits in the bitmask
3543 *
3544 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3545 */
3546static inline bool netif_attr_test_mask(unsigned long j,
3547 const unsigned long *mask,
3548 unsigned int nr_bits)
3549{
3550 cpu_max_bits_warn(j, nr_bits);
3551 return test_bit(j, mask);
3552}
3553
3554/**
3555 * netif_attr_test_online - Test for online CPU/Rx queue
3556 * @j: CPU/Rx queue index
3557 * @online_mask: bitmask for CPUs/Rx queues that are online
3558 * @nr_bits: number of bits in the bitmask
3559 *
3560 * Returns true if a CPU/Rx queue is online.
3561 */
3562static inline bool netif_attr_test_online(unsigned long j,
3563 const unsigned long *online_mask,
3564 unsigned int nr_bits)
3565{
3566 cpu_max_bits_warn(j, nr_bits);
3567
3568 if (online_mask)
3569 return test_bit(j, online_mask);
3570
3571 return (j < nr_bits);
3572}
3573
3574/**
3575 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3576 * @n: CPU/Rx queue index
3577 * @srcp: the cpumask/Rx queue mask pointer
3578 * @nr_bits: number of bits in the bitmask
3579 *
3580 * Returns >= nr_bits if no further CPUs/Rx queues set.
3581 */
3582static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3583 unsigned int nr_bits)
3584{
3585 /* -1 is a legal arg here. */
3586 if (n != -1)
3587 cpu_max_bits_warn(n, nr_bits);
3588
3589 if (srcp)
3590 return find_next_bit(srcp, nr_bits, n + 1);
3591
3592 return n + 1;
3593}
3594
3595/**
a1fa83bd 3596 * netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
80d19669
AN
3597 * @n: CPU/Rx queue index
3598 * @src1p: the first CPUs/Rx queues mask pointer
3599 * @src2p: the second CPUs/Rx queues mask pointer
3600 * @nr_bits: number of bits in the bitmask
3601 *
3602 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3603 */
3604static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3605 const unsigned long *src2p,
3606 unsigned int nr_bits)
3607{
3608 /* -1 is a legal arg here. */
3609 if (n != -1)
3610 cpu_max_bits_warn(n, nr_bits);
3611
3612 if (src1p && src2p)
3613 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3614 else if (src1p)
3615 return find_next_bit(src1p, nr_bits, n + 1);
3616 else if (src2p)
3617 return find_next_bit(src2p, nr_bits, n + 1);
3618
3619 return n + 1;
3620}
537c00de
AD
3621#else
3622static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3623 const struct cpumask *mask,
537c00de
AD
3624 u16 index)
3625{
3626 return 0;
3627}
c9fbb2d2
KK
3628
3629static inline int __netif_set_xps_queue(struct net_device *dev,
3630 const unsigned long *mask,
044ab86d 3631 u16 index, enum xps_map_type type)
c9fbb2d2
KK
3632{
3633 return 0;
3634}
537c00de
AD
3635#endif
3636
bea3348e
SH
3637/**
3638 * netif_is_multiqueue - test if device has multiple transmit queues
3639 * @dev: network device
3640 *
3641 * Check if device has multiple transmit queues
bea3348e 3642 */
4d29515f 3643static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3644{
a02cec21 3645 return dev->num_tx_queues > 1;
f25f4e44 3646}
1da177e4 3647
f629d208 3648int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3649
a953be53 3650#ifdef CONFIG_SYSFS
f629d208 3651int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3652#else
3653static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3654 unsigned int rxqs)
62fe0b40 3655{
c29c2ebd 3656 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3657 return 0;
3658}
3659#endif
271e5b7d
JK
3660int netif_set_real_num_queues(struct net_device *dev,
3661 unsigned int txq, unsigned int rxq);
62fe0b40 3662
65073a67
DB
3663static inline struct netdev_rx_queue *
3664__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3665{
3666 return dev->_rx + rxq;
3667}
3668
a953be53
MD
3669#ifdef CONFIG_SYSFS
3670static inline unsigned int get_netdev_rx_queue_index(
3671 struct netdev_rx_queue *queue)
3672{
3673 struct net_device *dev = queue->dev;
3674 int index = queue - dev->_rx;
3675
3676 BUG_ON(index >= dev->num_rx_queues);
3677 return index;
3678}
3679#endif
3680
f629d208 3681int netif_get_num_default_rss_queues(void);
16917b87 3682
e6247027
ED
3683enum skb_free_reason {
3684 SKB_REASON_CONSUMED,
3685 SKB_REASON_DROPPED,
3686};
3687
3688void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3689void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3690
e6247027
ED
3691/*
3692 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3693 * interrupt context or with hardware interrupts being disabled.
afa79d08 3694 * (in_hardirq() || irqs_disabled())
e6247027
ED
3695 *
3696 * We provide four helpers that can be used in following contexts :
3697 *
3698 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3699 * replacing kfree_skb(skb)
3700 *
3701 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3702 * Typically used in place of consume_skb(skb) in TX completion path
3703 *
3704 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3705 * replacing kfree_skb(skb)
3706 *
3707 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3708 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3709 */
e6247027
ED
3710static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3711{
3712 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3713}
3714
3715static inline void dev_consume_skb_irq(struct sk_buff *skb)
3716{
3717 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3718}
3719
3720static inline void dev_kfree_skb_any(struct sk_buff *skb)
3721{
3722 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3723}
3724
3725static inline void dev_consume_skb_any(struct sk_buff *skb)
3726{
3727 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3728}
1da177e4 3729
fe21cb91
KKD
3730u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp,
3731 struct bpf_prog *xdp_prog);
7c497478
JW
3732void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3733int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208 3734int netif_rx(struct sk_buff *skb);
baebdf48
SAS
3735int __netif_rx(struct sk_buff *skb);
3736
04eb4489 3737int netif_receive_skb(struct sk_buff *skb);
1c601d82 3738int netif_receive_skb_core(struct sk_buff *skb);
587652bb 3739void netif_receive_skb_list_internal(struct list_head *head);
f6ad8c1b 3740void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3741gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3742void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3743struct sk_buff *napi_get_frags(struct napi_struct *napi);
3744gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3745struct packet_offload *gro_find_receive_by_type(__be16 type);
3746struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3747
3748static inline void napi_free_frags(struct napi_struct *napi)
3749{
3750 kfree_skb(napi->skb);
3751 napi->skb = NULL;
3752}
3753
24b27fc4 3754bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
3755int netdev_rx_handler_register(struct net_device *dev,
3756 rx_handler_func_t *rx_handler,
3757 void *rx_handler_data);
3758void netdev_rx_handler_unregister(struct net_device *dev);
3759
3760bool dev_valid_name(const char *name);
d0efb162
PC
3761static inline bool is_socket_ioctl_cmd(unsigned int cmd)
3762{
3763 return _IOC_TYPE(cmd) == SOCK_IOC_TYPE;
3764}
29c49648
AB
3765int get_user_ifreq(struct ifreq *ifr, void __user **ifrdata, void __user *arg);
3766int put_user_ifreq(struct ifreq *ifr, void __user *arg);
44c02a2c 3767int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
a554bf96 3768 void __user *data, bool *need_copyout);
876f0bf9 3769int dev_ifconf(struct net *net, struct ifconf __user *ifc);
a554bf96 3770int dev_ethtool(struct net *net, struct ifreq *ifr, void __user *userdata);
f629d208 3771unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
3772int __dev_change_flags(struct net_device *dev, unsigned int flags,
3773 struct netlink_ext_ack *extack);
567c5e13
PM
3774int dev_change_flags(struct net_device *dev, unsigned int flags,
3775 struct netlink_ext_ack *extack);
cb178190
DM
3776void __dev_notify_flags(struct net_device *, unsigned int old_flags,
3777 unsigned int gchanges);
f629d208 3778int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 3779int dev_get_alias(const struct net_device *, char *, size_t);
0854fa82
AV
3780int __dev_change_net_namespace(struct net_device *dev, struct net *net,
3781 const char *pat, int new_ifindex);
3782static inline
eeb85a14 3783int dev_change_net_namespace(struct net_device *dev, struct net *net,
0854fa82
AV
3784 const char *pat)
3785{
3786 return __dev_change_net_namespace(dev, net, pat, 0);
3787}
f51048c3 3788int __dev_set_mtu(struct net_device *, int);
f629d208 3789int dev_set_mtu(struct net_device *, int);
d59cdf94
PM
3790int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
3791 struct netlink_ext_ack *extack);
3a37a963
PM
3792int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
3793 struct netlink_ext_ack *extack);
3b23a32a
CW
3794int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
3795 struct netlink_ext_ack *extack);
3796int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
d6abc596
FF
3797int dev_get_port_parent_id(struct net_device *dev,
3798 struct netdev_phys_item_id *ppid, bool recurse);
3799bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
f53c7239 3800struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
3801struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3802 struct netdev_queue *txq, int *ret);
d67b9cd2 3803
aa8d3a71 3804int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
879af96f 3805u8 dev_xdp_prog_count(struct net_device *dev);
7f0a8382 3806u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
aa8d3a71 3807
a0265d28 3808int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3809int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
5f7d5728 3810int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3811bool is_skb_forwardable(const struct net_device *dev,
3812 const struct sk_buff *skb);
1da177e4 3813
5f7d5728
JDB
3814static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
3815 const struct sk_buff *skb,
3816 const bool check_mtu)
3817{
3818 const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
3819 unsigned int len;
3820
3821 if (!(dev->flags & IFF_UP))
3822 return false;
3823
3824 if (!check_mtu)
3825 return true;
3826
3827 len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
3828 if (skb->len <= len)
3829 return true;
3830
3831 /* if TSO is enabled, we don't care about the length as the packet
3832 * could be forwarded without being segmented before
3833 */
3834 if (skb_is_gso(skb))
3835 return true;
3836
3837 return false;
3838}
3839
6510ea97 3840struct net_device_core_stats __percpu *netdev_core_stats_alloc(struct net_device *dev);
625788b5 3841
6510ea97 3842static inline struct net_device_core_stats __percpu *dev_core_stats(struct net_device *dev)
625788b5
ED
3843{
3844 /* This READ_ONCE() pairs with the write in netdev_core_stats_alloc() */
3845 struct net_device_core_stats __percpu *p = READ_ONCE(dev->core_stats);
3846
3847 if (likely(p))
6510ea97 3848 return p;
625788b5
ED
3849
3850 return netdev_core_stats_alloc(dev);
3851}
3852
3853#define DEV_CORE_STATS_INC(FIELD) \
3854static inline void dev_core_stats_##FIELD##_inc(struct net_device *dev) \
3855{ \
6510ea97 3856 struct net_device_core_stats __percpu *p; \
fc93db15 3857 \
fc93db15 3858 p = dev_core_stats(dev); \
625788b5 3859 if (p) \
6510ea97 3860 this_cpu_inc(p->FIELD); \
625788b5
ED
3861}
3862DEV_CORE_STATS_INC(rx_dropped)
3863DEV_CORE_STATS_INC(tx_dropped)
3864DEV_CORE_STATS_INC(rx_nohandler)
794c24e9 3865DEV_CORE_STATS_INC(rx_otherhost_dropped)
625788b5 3866
4e3264d2 3867static __always_inline int ____dev_forward_skb(struct net_device *dev,
5f7d5728
JDB
3868 struct sk_buff *skb,
3869 const bool check_mtu)
4e3264d2
MKL
3870{
3871 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
5f7d5728 3872 unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
625788b5 3873 dev_core_stats_rx_dropped_inc(dev);
4e3264d2
MKL
3874 kfree_skb(skb);
3875 return NET_RX_DROP;
3876 }
3877
ff70202b 3878 skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
4e3264d2
MKL
3879 skb->priority = 0;
3880 return 0;
3881}
3882
9f9a742d 3883bool dev_nit_active(struct net_device *dev);
74b20582
DA
3884void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
3885
4c6c11ea 3886static inline void __dev_put(struct net_device *dev)
1da177e4 3887{
b37a4668 3888 if (dev) {
919067cc 3889#ifdef CONFIG_PCPU_DEV_REFCNT
b37a4668 3890 this_cpu_dec(*dev->pcpu_refcnt);
919067cc 3891#else
b37a4668 3892 refcount_dec(&dev->dev_refcnt);
919067cc 3893#endif
b37a4668 3894 }
1da177e4
LT
3895}
3896
4c6c11ea 3897static inline void __dev_hold(struct net_device *dev)
15333061 3898{
b37a4668 3899 if (dev) {
919067cc 3900#ifdef CONFIG_PCPU_DEV_REFCNT
b37a4668 3901 this_cpu_inc(*dev->pcpu_refcnt);
919067cc 3902#else
b37a4668 3903 refcount_inc(&dev->dev_refcnt);
919067cc 3904#endif
b37a4668 3905 }
15333061 3906}
1da177e4 3907
4c6c11ea
ED
3908static inline void __netdev_tracker_alloc(struct net_device *dev,
3909 netdevice_tracker *tracker,
3910 gfp_t gfp)
3911{
3912#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
3913 ref_tracker_alloc(&dev->refcnt_tracker, tracker, gfp);
3914#endif
3915}
3916
3917/* netdev_tracker_alloc() can upgrade a prior untracked reference
3918 * taken by dev_get_by_name()/dev_get_by_index() to a tracked one.
3919 */
4d92b95f
ED
3920static inline void netdev_tracker_alloc(struct net_device *dev,
3921 netdevice_tracker *tracker, gfp_t gfp)
3922{
3923#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4c6c11ea
ED
3924 refcount_dec(&dev->refcnt_tracker.no_tracker);
3925 __netdev_tracker_alloc(dev, tracker, gfp);
4d92b95f
ED
3926#endif
3927}
3928
3929static inline void netdev_tracker_free(struct net_device *dev,
3930 netdevice_tracker *tracker)
3931{
3932#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
3933 ref_tracker_free(&dev->refcnt_tracker, tracker);
3934#endif
3935}
3936
3937static inline void dev_hold_track(struct net_device *dev,
3938 netdevice_tracker *tracker, gfp_t gfp)
3939{
3940 if (dev) {
4c6c11ea
ED
3941 __dev_hold(dev);
3942 __netdev_tracker_alloc(dev, tracker, gfp);
4d92b95f
ED
3943 }
3944}
3945
3946static inline void dev_put_track(struct net_device *dev,
3947 netdevice_tracker *tracker)
3948{
3949 if (dev) {
3950 netdev_tracker_free(dev, tracker);
4c6c11ea 3951 __dev_put(dev);
4d92b95f
ED
3952 }
3953}
3954
4c6c11ea
ED
3955/**
3956 * dev_hold - get reference to device
3957 * @dev: network device
3958 *
3959 * Hold reference to device to keep it from being freed.
3960 * Try using dev_hold_track() instead.
3961 */
3962static inline void dev_hold(struct net_device *dev)
3963{
3964 dev_hold_track(dev, NULL, GFP_ATOMIC);
3965}
3966
3967/**
3968 * dev_put - release reference to device
3969 * @dev: network device
3970 *
3971 * Release reference to device to allow it to be freed.
3972 * Try using dev_put_track() instead.
3973 */
3974static inline void dev_put(struct net_device *dev)
3975{
3976 dev_put_track(dev, NULL);
3977}
3978
9038c320
ED
3979static inline void dev_replace_track(struct net_device *odev,
3980 struct net_device *ndev,
3981 netdevice_tracker *tracker,
3982 gfp_t gfp)
3983{
9038c320 3984 if (odev)
9280ac2e
ED
3985 netdev_tracker_free(odev, tracker);
3986
4c6c11ea
ED
3987 __dev_hold(ndev);
3988 __dev_put(odev);
9280ac2e 3989
9038c320 3990 if (ndev)
4c6c11ea 3991 __netdev_tracker_alloc(ndev, tracker, gfp);
9038c320
ED
3992}
3993
1da177e4
LT
3994/* Carrier loss detection, dial on demand. The functions netif_carrier_on
3995 * and _off may be called from IRQ context, but it is caller
3996 * who is responsible for serialization of these calls.
b00055aa
SR
3997 *
3998 * The name carrier is inappropriate, these functions should really be
3999 * called netif_lowerlayer_*() because they represent the state of any
4000 * kind of lower layer not just hardware media.
1da177e4 4001 */
f629d208 4002void linkwatch_fire_event(struct net_device *dev);
1da177e4 4003
bea3348e
SH
4004/**
4005 * netif_carrier_ok - test if carrier present
4006 * @dev: network device
4007 *
4008 * Check if carrier is present on device
4009 */
4d29515f 4010static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
4011{
4012 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
4013}
4014
f629d208 4015unsigned long dev_trans_start(struct net_device *dev);
9d21493b 4016
f629d208 4017void __netdev_watchdog_up(struct net_device *dev);
1da177e4 4018
f629d208 4019void netif_carrier_on(struct net_device *dev);
f629d208 4020void netif_carrier_off(struct net_device *dev);
490dceca 4021void netif_carrier_event(struct net_device *dev);
1da177e4 4022
bea3348e
SH
4023/**
4024 * netif_dormant_on - mark device as dormant.
4025 * @dev: network device
4026 *
4027 * Mark device as dormant (as per RFC2863).
4028 *
4029 * The dormant state indicates that the relevant interface is not
4030 * actually in a condition to pass packets (i.e., it is not 'up') but is
4031 * in a "pending" state, waiting for some external event. For "on-
4032 * demand" interfaces, this new state identifies the situation where the
4033 * interface is waiting for events to place it in the up state.
bea3348e 4034 */
b00055aa
SR
4035static inline void netif_dormant_on(struct net_device *dev)
4036{
4037 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
4038 linkwatch_fire_event(dev);
4039}
4040
bea3348e
SH
4041/**
4042 * netif_dormant_off - set device as not dormant.
4043 * @dev: network device
4044 *
4045 * Device is not in dormant state.
4046 */
b00055aa
SR
4047static inline void netif_dormant_off(struct net_device *dev)
4048{
4049 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
4050 linkwatch_fire_event(dev);
4051}
4052
bea3348e 4053/**
8ecbc40a 4054 * netif_dormant - test if device is dormant
bea3348e
SH
4055 * @dev: network device
4056 *
8ecbc40a 4057 * Check if device is dormant.
bea3348e 4058 */
4d29515f 4059static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
4060{
4061 return test_bit(__LINK_STATE_DORMANT, &dev->state);
4062}
4063
4064
eec517cd
AL
4065/**
4066 * netif_testing_on - mark device as under test.
4067 * @dev: network device
4068 *
4069 * Mark device as under test (as per RFC2863).
4070 *
4071 * The testing state indicates that some test(s) must be performed on
4072 * the interface. After completion, of the test, the interface state
4073 * will change to up, dormant, or down, as appropriate.
4074 */
4075static inline void netif_testing_on(struct net_device *dev)
4076{
4077 if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
4078 linkwatch_fire_event(dev);
4079}
4080
4081/**
4082 * netif_testing_off - set device as not under test.
4083 * @dev: network device
4084 *
4085 * Device is not in testing state.
4086 */
4087static inline void netif_testing_off(struct net_device *dev)
4088{
4089 if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
4090 linkwatch_fire_event(dev);
4091}
4092
4093/**
4094 * netif_testing - test if device is under test
4095 * @dev: network device
4096 *
4097 * Check if device is under test
4098 */
4099static inline bool netif_testing(const struct net_device *dev)
4100{
4101 return test_bit(__LINK_STATE_TESTING, &dev->state);
4102}
4103
4104
bea3348e
SH
4105/**
4106 * netif_oper_up - test if device is operational
4107 * @dev: network device
4108 *
4109 * Check if carrier is operational
4110 */
4d29515f 4111static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 4112{
b00055aa
SR
4113 return (dev->operstate == IF_OPER_UP ||
4114 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
4115}
4116
bea3348e
SH
4117/**
4118 * netif_device_present - is device available or removed
4119 * @dev: network device
4120 *
4121 * Check if device has not been removed from system.
4122 */
7a126a43 4123static inline bool netif_device_present(const struct net_device *dev)
1da177e4
LT
4124{
4125 return test_bit(__LINK_STATE_PRESENT, &dev->state);
4126}
4127
f629d208 4128void netif_device_detach(struct net_device *dev);
1da177e4 4129
f629d208 4130void netif_device_attach(struct net_device *dev);
1da177e4
LT
4131
4132/*
4133 * Network interface message level settings
4134 */
1da177e4
LT
4135
4136enum {
6a94b8cc
MK
4137 NETIF_MSG_DRV_BIT,
4138 NETIF_MSG_PROBE_BIT,
4139 NETIF_MSG_LINK_BIT,
4140 NETIF_MSG_TIMER_BIT,
4141 NETIF_MSG_IFDOWN_BIT,
4142 NETIF_MSG_IFUP_BIT,
4143 NETIF_MSG_RX_ERR_BIT,
4144 NETIF_MSG_TX_ERR_BIT,
4145 NETIF_MSG_TX_QUEUED_BIT,
4146 NETIF_MSG_INTR_BIT,
4147 NETIF_MSG_TX_DONE_BIT,
4148 NETIF_MSG_RX_STATUS_BIT,
4149 NETIF_MSG_PKTDATA_BIT,
4150 NETIF_MSG_HW_BIT,
4151 NETIF_MSG_WOL_BIT,
4152
4153 /* When you add a new bit above, update netif_msg_class_names array
4154 * in net/ethtool/common.c
4155 */
4156 NETIF_MSG_CLASS_COUNT,
1da177e4 4157};
6a94b8cc
MK
4158/* Both ethtool_ops interface and internal driver implementation use u32 */
4159static_assert(NETIF_MSG_CLASS_COUNT <= 32);
4160
4161#define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
4162#define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
4163
4164#define NETIF_MSG_DRV __NETIF_MSG(DRV)
4165#define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
4166#define NETIF_MSG_LINK __NETIF_MSG(LINK)
4167#define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
4168#define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
4169#define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
4170#define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
4171#define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
4172#define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
4173#define NETIF_MSG_INTR __NETIF_MSG(INTR)
4174#define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
4175#define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
4176#define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
4177#define NETIF_MSG_HW __NETIF_MSG(HW)
4178#define NETIF_MSG_WOL __NETIF_MSG(WOL)
1da177e4
LT
4179
4180#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
4181#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
4182#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
4183#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
4184#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
4185#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
4186#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
4187#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
4188#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
4189#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
4190#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
4191#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
4192#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
4193#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
4194#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
4195
4196static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
4197{
4198 /* use default */
4199 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
4200 return default_msg_enable_bits;
4201 if (debug_value == 0) /* no output */
4202 return 0;
4203 /* set low N bits */
f4d7b3e2 4204 return (1U << debug_value) - 1;
1da177e4
LT
4205}
4206
c773e847 4207static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 4208{
c773e847 4209 spin_lock(&txq->_xmit_lock);
7a10d8c8
ED
4210 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4211 WRITE_ONCE(txq->xmit_lock_owner, cpu);
22dd7495
JHS
4212}
4213
5a717f4f
MT
4214static inline bool __netif_tx_acquire(struct netdev_queue *txq)
4215{
4216 __acquire(&txq->_xmit_lock);
4217 return true;
4218}
4219
4220static inline void __netif_tx_release(struct netdev_queue *txq)
4221{
4222 __release(&txq->_xmit_lock);
4223}
4224
fd2ea0a7
DM
4225static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
4226{
4227 spin_lock_bh(&txq->_xmit_lock);
7a10d8c8
ED
4228 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4229 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
fd2ea0a7
DM
4230}
4231
4d29515f 4232static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 4233{
4d29515f 4234 bool ok = spin_trylock(&txq->_xmit_lock);
7a10d8c8
ED
4235
4236 if (likely(ok)) {
4237 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4238 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
4239 }
c3f26a26
DM
4240 return ok;
4241}
4242
4243static inline void __netif_tx_unlock(struct netdev_queue *txq)
4244{
7a10d8c8
ED
4245 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4246 WRITE_ONCE(txq->xmit_lock_owner, -1);
c3f26a26
DM
4247 spin_unlock(&txq->_xmit_lock);
4248}
4249
4250static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
4251{
7a10d8c8
ED
4252 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4253 WRITE_ONCE(txq->xmit_lock_owner, -1);
c3f26a26
DM
4254 spin_unlock_bh(&txq->_xmit_lock);
4255}
4256
5337824f
ED
4257/*
4258 * txq->trans_start can be read locklessly from dev_watchdog()
4259 */
08baf561
ED
4260static inline void txq_trans_update(struct netdev_queue *txq)
4261{
4262 if (txq->xmit_lock_owner != -1)
5337824f
ED
4263 WRITE_ONCE(txq->trans_start, jiffies);
4264}
4265
4266static inline void txq_trans_cond_update(struct netdev_queue *txq)
4267{
4268 unsigned long now = jiffies;
4269
4270 if (READ_ONCE(txq->trans_start) != now)
4271 WRITE_ONCE(txq->trans_start, now);
08baf561
ED
4272}
4273
ba162f8e
FW
4274/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
4275static inline void netif_trans_update(struct net_device *dev)
4276{
9b36627a
FW
4277 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
4278
5337824f 4279 txq_trans_cond_update(txq);
ba162f8e
FW
4280}
4281
d29f749e
DJ
4282/**
4283 * netif_tx_lock - grab network device transmit lock
4284 * @dev: network device
d29f749e
DJ
4285 *
4286 * Get network device transmit lock
4287 */
dab8fe32 4288void netif_tx_lock(struct net_device *dev);
932ff279
HX
4289
4290static inline void netif_tx_lock_bh(struct net_device *dev)
4291{
e8a0464c
DM
4292 local_bh_disable();
4293 netif_tx_lock(dev);
932ff279
HX
4294}
4295
dab8fe32 4296void netif_tx_unlock(struct net_device *dev);
932ff279
HX
4297
4298static inline void netif_tx_unlock_bh(struct net_device *dev)
4299{
e8a0464c
DM
4300 netif_tx_unlock(dev);
4301 local_bh_enable();
932ff279
HX
4302}
4303
c773e847 4304#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4305 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4306 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4307 } else { \
4308 __netif_tx_acquire(txq); \
22dd7495
JHS
4309 } \
4310}
4311
5efeac44
EB
4312#define HARD_TX_TRYLOCK(dev, txq) \
4313 (((dev->features & NETIF_F_LLTX) == 0) ? \
4314 __netif_tx_trylock(txq) : \
5a717f4f 4315 __netif_tx_acquire(txq))
5efeac44 4316
c773e847 4317#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4318 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4319 __netif_tx_unlock(txq); \
5a717f4f
MT
4320 } else { \
4321 __netif_tx_release(txq); \
22dd7495
JHS
4322 } \
4323}
4324
1da177e4
LT
4325static inline void netif_tx_disable(struct net_device *dev)
4326{
fd2ea0a7 4327 unsigned int i;
c3f26a26 4328 int cpu;
fd2ea0a7 4329
c3f26a26
DM
4330 local_bh_disable();
4331 cpu = smp_processor_id();
3aa6bce9 4332 spin_lock(&dev->tx_global_lock);
fd2ea0a7
DM
4333 for (i = 0; i < dev->num_tx_queues; i++) {
4334 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4335
4336 __netif_tx_lock(txq, cpu);
fd2ea0a7 4337 netif_tx_stop_queue(txq);
c3f26a26 4338 __netif_tx_unlock(txq);
fd2ea0a7 4339 }
3aa6bce9 4340 spin_unlock(&dev->tx_global_lock);
c3f26a26 4341 local_bh_enable();
1da177e4
LT
4342}
4343
e308a5d8
DM
4344static inline void netif_addr_lock(struct net_device *dev)
4345{
1fc70edb 4346 unsigned char nest_level = 0;
e308a5d8 4347
1fc70edb
TY
4348#ifdef CONFIG_LOCKDEP
4349 nest_level = dev->nested_level;
4350#endif
4351 spin_lock_nested(&dev->addr_list_lock, nest_level);
845e0ebb
CW
4352}
4353
e308a5d8
DM
4354static inline void netif_addr_lock_bh(struct net_device *dev)
4355{
1fc70edb
TY
4356 unsigned char nest_level = 0;
4357
4358#ifdef CONFIG_LOCKDEP
4359 nest_level = dev->nested_level;
4360#endif
4361 local_bh_disable();
4362 spin_lock_nested(&dev->addr_list_lock, nest_level);
e308a5d8
DM
4363}
4364
4365static inline void netif_addr_unlock(struct net_device *dev)
4366{
4367 spin_unlock(&dev->addr_list_lock);
4368}
4369
4370static inline void netif_addr_unlock_bh(struct net_device *dev)
4371{
4372 spin_unlock_bh(&dev->addr_list_lock);
4373}
4374
f001fde5 4375/*
31278e71 4376 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4377 * rcu_read_lock held.
4378 */
4379#define for_each_dev_addr(dev, ha) \
31278e71 4380 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4381
1da177e4
LT
4382/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4383
f629d208 4384void ether_setup(struct net_device *dev);
1da177e4
LT
4385
4386/* Support for loadable net-drivers */
f629d208 4387struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4388 unsigned char name_assign_type,
f629d208
JP
4389 void (*setup)(struct net_device *),
4390 unsigned int txqs, unsigned int rxqs);
c835a677
TG
4391#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4392 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4393
c835a677
TG
4394#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4395 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4396 count)
36909ea4 4397
f629d208
JP
4398int register_netdev(struct net_device *dev);
4399void unregister_netdev(struct net_device *dev);
f001fde5 4400
cd16627f
BG
4401int devm_register_netdev(struct device *dev, struct net_device *ndev);
4402
22bedad3 4403/* General hardware address lists handling functions */
f629d208
JP
4404int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4405 struct netdev_hw_addr_list *from_list, int addr_len);
4406void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4407 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4408int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4409 struct net_device *dev,
4410 int (*sync)(struct net_device *, const unsigned char *),
4411 int (*unsync)(struct net_device *,
4412 const unsigned char *));
e7946760
IK
4413int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4414 struct net_device *dev,
4415 int (*sync)(struct net_device *,
4416 const unsigned char *, int),
4417 int (*unsync)(struct net_device *,
4418 const unsigned char *, int));
4419void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4420 struct net_device *dev,
4421 int (*unsync)(struct net_device *,
4422 const unsigned char *, int));
670e5b8e
AD
4423void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4424 struct net_device *dev,
4425 int (*unsync)(struct net_device *,
4426 const unsigned char *));
f629d208 4427void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4428
f001fde5 4429/* Functions used for device addresses handling */
adeef3e3
JK
4430void dev_addr_mod(struct net_device *dev, unsigned int offset,
4431 const void *addr, size_t len);
4432
48eab831 4433static inline void
40af35fd 4434__dev_addr_set(struct net_device *dev, const void *addr, size_t len)
48eab831 4435{
adeef3e3 4436 dev_addr_mod(dev, 0, addr, len);
48eab831
JK
4437}
4438
4439static inline void dev_addr_set(struct net_device *dev, const u8 *addr)
4440{
4441 __dev_addr_set(dev, addr, dev->addr_len);
4442}
4443
f629d208
JP
4444int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4445 unsigned char addr_type);
4446int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4447 unsigned char addr_type);
a748ee24
JP
4448
4449/* Functions used for unicast addresses handling */
f629d208
JP
4450int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4451int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4452int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4453int dev_uc_sync(struct net_device *to, struct net_device *from);
4454int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4455void dev_uc_unsync(struct net_device *to, struct net_device *from);
4456void dev_uc_flush(struct net_device *dev);
4457void dev_uc_init(struct net_device *dev);
f001fde5 4458
670e5b8e
AD
4459/**
4460 * __dev_uc_sync - Synchonize device's unicast list
4461 * @dev: device to sync
4462 * @sync: function to call if address should be added
4463 * @unsync: function to call if address should be removed
4464 *
4465 * Add newly added addresses to the interface, and release
4466 * addresses that have been deleted.
5e82b4b2 4467 */
670e5b8e
AD
4468static inline int __dev_uc_sync(struct net_device *dev,
4469 int (*sync)(struct net_device *,
4470 const unsigned char *),
4471 int (*unsync)(struct net_device *,
4472 const unsigned char *))
4473{
4474 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4475}
4476
4477/**
e793c0f7 4478 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4479 * @dev: device to sync
4480 * @unsync: function to call if address should be removed
4481 *
4482 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4483 */
670e5b8e
AD
4484static inline void __dev_uc_unsync(struct net_device *dev,
4485 int (*unsync)(struct net_device *,
4486 const unsigned char *))
4487{
4488 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4489}
4490
22bedad3 4491/* Functions used for multicast addresses handling */
f629d208
JP
4492int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4493int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4494int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4495int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4496int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4497int dev_mc_sync(struct net_device *to, struct net_device *from);
4498int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4499void dev_mc_unsync(struct net_device *to, struct net_device *from);
4500void dev_mc_flush(struct net_device *dev);
4501void dev_mc_init(struct net_device *dev);
f001fde5 4502
670e5b8e
AD
4503/**
4504 * __dev_mc_sync - Synchonize device's multicast list
4505 * @dev: device to sync
4506 * @sync: function to call if address should be added
4507 * @unsync: function to call if address should be removed
4508 *
4509 * Add newly added addresses to the interface, and release
4510 * addresses that have been deleted.
5e82b4b2 4511 */
670e5b8e
AD
4512static inline int __dev_mc_sync(struct net_device *dev,
4513 int (*sync)(struct net_device *,
4514 const unsigned char *),
4515 int (*unsync)(struct net_device *,
4516 const unsigned char *))
4517{
4518 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4519}
4520
4521/**
e793c0f7 4522 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4523 * @dev: device to sync
4524 * @unsync: function to call if address should be removed
4525 *
4526 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4527 */
670e5b8e
AD
4528static inline void __dev_mc_unsync(struct net_device *dev,
4529 int (*unsync)(struct net_device *,
4530 const unsigned char *))
4531{
4532 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4533}
4534
4417da66 4535/* Functions used for secondary unicast and multicast support */
f629d208 4536void dev_set_rx_mode(struct net_device *dev);
f629d208
JP
4537int dev_set_promiscuity(struct net_device *dev, int inc);
4538int dev_set_allmulti(struct net_device *dev, int inc);
4539void netdev_state_change(struct net_device *dev);
7061eb8c 4540void __netdev_notify_peers(struct net_device *dev);
f629d208
JP
4541void netdev_notify_peers(struct net_device *dev);
4542void netdev_features_change(struct net_device *dev);
1da177e4 4543/* Load a device via the kmod */
f629d208
JP
4544void dev_load(struct net *net, const char *name);
4545struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4546 struct rtnl_link_stats64 *storage);
4547void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4548 const struct net_device_stats *netdev_stats);
44fa32f0
HK
4549void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
4550 const struct pcpu_sw_netstats __percpu *netstats);
a1839426 4551void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
eeda3fd6 4552
1da177e4 4553extern int netdev_max_backlog;
3d48b53f
MT
4554extern int dev_rx_weight;
4555extern int dev_tx_weight;
323ebb61 4556extern int gro_normal_batch;
9ff162a8 4557
1fc70edb
TY
4558enum {
4559 NESTED_SYNC_IMM_BIT,
4560 NESTED_SYNC_TODO_BIT,
4561};
4562
4563#define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
4564#define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
4565
4566#define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
4567#define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
4568
eff74233 4569struct netdev_nested_priv {
1fc70edb 4570 unsigned char flags;
eff74233
TY
4571 void *data;
4572};
4573
f629d208 4574bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4575struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4576 struct list_head **iter);
8b5be856 4577
44a40855
VY
4578/* iterate through upper list, must be called under RCU read lock */
4579#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4580 for (iter = &(dev)->adj_list.upper, \
4581 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4582 updev; \
4583 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4584
1a3f060c
DA
4585int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4586 int (*fn)(struct net_device *upper_dev,
eff74233
TY
4587 struct netdev_nested_priv *priv),
4588 struct netdev_nested_priv *priv);
1a3f060c
DA
4589
4590bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4591 struct net_device *upper_dev);
4592
25cc72a3
IS
4593bool netdev_has_any_upper_dev(struct net_device *dev);
4594
f629d208
JP
4595void *netdev_lower_get_next_private(struct net_device *dev,
4596 struct list_head **iter);
4597void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4598 struct list_head **iter);
31088a11
VF
4599
4600#define netdev_for_each_lower_private(dev, priv, iter) \
4601 for (iter = (dev)->adj_list.lower.next, \
4602 priv = netdev_lower_get_next_private(dev, &(iter)); \
4603 priv; \
4604 priv = netdev_lower_get_next_private(dev, &(iter)))
4605
4606#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4607 for (iter = &(dev)->adj_list.lower, \
4608 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4609 priv; \
4610 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4611
4085ebe8
VY
4612void *netdev_lower_get_next(struct net_device *dev,
4613 struct list_head **iter);
7ce856aa 4614
4085ebe8 4615#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4616 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4617 ldev = netdev_lower_get_next(dev, &(iter)); \
4618 ldev; \
4619 ldev = netdev_lower_get_next(dev, &(iter)))
4620
7151affe 4621struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
7ce856aa 4622 struct list_head **iter);
1a3f060c
DA
4623int netdev_walk_all_lower_dev(struct net_device *dev,
4624 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4625 struct netdev_nested_priv *priv),
4626 struct netdev_nested_priv *priv);
1a3f060c
DA
4627int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4628 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4629 struct netdev_nested_priv *priv),
4630 struct netdev_nested_priv *priv);
1a3f060c 4631
f629d208 4632void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4633void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4634struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4635struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4636int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4637 struct netlink_ext_ack *extack);
f629d208 4638int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4639 struct net_device *upper_dev,
42ab19ee
DA
4640 void *upper_priv, void *upper_info,
4641 struct netlink_ext_ack *extack);
f629d208
JP
4642void netdev_upper_dev_unlink(struct net_device *dev,
4643 struct net_device *upper_dev);
32b6d34f
TY
4644int netdev_adjacent_change_prepare(struct net_device *old_dev,
4645 struct net_device *new_dev,
4646 struct net_device *dev,
4647 struct netlink_ext_ack *extack);
4648void netdev_adjacent_change_commit(struct net_device *old_dev,
4649 struct net_device *new_dev,
4650 struct net_device *dev);
4651void netdev_adjacent_change_abort(struct net_device *old_dev,
4652 struct net_device *new_dev,
4653 struct net_device *dev);
5bb025fa 4654void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4655void *netdev_lower_dev_get_private(struct net_device *dev,
4656 struct net_device *lower_dev);
04d48266
JP
4657void netdev_lower_state_changed(struct net_device *lower_dev,
4658 void *lower_state_info);
960fb622
ED
4659
4660/* RSS keys are 40 or 52 bytes long */
4661#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4662extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4663void netdev_rss_key_fill(void *buffer, size_t len);
4664
f629d208 4665int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4666int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4667int skb_csum_hwoffload_help(struct sk_buff *skb,
4668 const netdev_features_t features);
4669
f629d208
JP
4670struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4671 netdev_features_t features, bool tx_path);
23c7f8d7
SK
4672struct sk_buff *skb_eth_gso_segment(struct sk_buff *skb,
4673 netdev_features_t features, __be16 type);
f629d208
JP
4674struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4675 netdev_features_t features);
12b0004d 4676
61bd3857
MS
4677struct netdev_bonding_info {
4678 ifslave slave;
4679 ifbond master;
4680};
4681
4682struct netdev_notifier_bonding_info {
4683 struct netdev_notifier_info info; /* must be first */
4684 struct netdev_bonding_info bonding_info;
4685};
4686
4687void netdev_bonding_info_change(struct net_device *dev,
4688 struct netdev_bonding_info *bonding_info);
4689
6b08d6c1
MK
4690#if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
4691void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
4692#else
4693static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
4694 const void *data)
4695{
4696}
4697#endif
4698
12b0004d
CW
4699static inline
4700struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4701{
4702 return __skb_gso_segment(skb, features, true);
4703}
53d6471c 4704__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4705
4706static inline bool can_checksum_protocol(netdev_features_t features,
4707 __be16 protocol)
4708{
c8cd0989
TH
4709 if (protocol == htons(ETH_P_FCOE))
4710 return !!(features & NETIF_F_FCOE_CRC);
4711
4712 /* Assume this is an IP checksum (not SCTP CRC) */
4713
4714 if (features & NETIF_F_HW_CSUM) {
4715 /* Can checksum everything */
4716 return true;
4717 }
4718
4719 switch (protocol) {
4720 case htons(ETH_P_IP):
4721 return !!(features & NETIF_F_IP_CSUM);
4722 case htons(ETH_P_IPV6):
4723 return !!(features & NETIF_F_IPV6_CSUM);
4724 default:
4725 return false;
4726 }
ec5f0615 4727}
12b0004d 4728
fb286bb2 4729#ifdef CONFIG_BUG
7fe50ac8 4730void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4731#else
7fe50ac8
CW
4732static inline void netdev_rx_csum_fault(struct net_device *dev,
4733 struct sk_buff *skb)
fb286bb2
HX
4734{
4735}
4736#endif
1da177e4 4737/* rx skb timestamps */
f629d208
JP
4738void net_enable_timestamp(void);
4739void net_disable_timestamp(void);
1da177e4 4740
4798248e 4741static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4742 struct sk_buff *skb, struct net_device *dev,
4743 bool more)
4798248e 4744{
6b16f9ee 4745 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4746 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4747}
4748
97cdcf37
FW
4749static inline bool netdev_xmit_more(void)
4750{
4751 return __this_cpu_read(softnet_data.xmit.more);
4752}
4753
10b3ad8c 4754static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4755 struct netdev_queue *txq, bool more)
4798248e
DM
4756{
4757 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4758 netdev_tx_t rc;
4798248e 4759
fa2dbdc2 4760 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4761 if (rc == NETDEV_TX_OK)
4762 txq_trans_update(txq);
4763
4764 return rc;
4798248e
DM
4765}
4766
b793dc5c 4767int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4768 const void *ns);
b793dc5c 4769void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4770 const void *ns);
58292cbe 4771
737aec57 4772extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4773
f629d208 4774const char *netdev_drivername(const struct net_device *dev);
6579e57b 4775
da08143b
MK
4776static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4777 netdev_features_t f2)
4778{
c8cd0989
TH
4779 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4780 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4781 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4782 else
b6a0e72a 4783 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4784 }
da08143b 4785
c8cd0989 4786 return f1 & f2;
da08143b
MK
4787}
4788
c8f44aff
MM
4789static inline netdev_features_t netdev_get_wanted_features(
4790 struct net_device *dev)
5455c699
MM
4791{
4792 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4793}
c8f44aff
MM
4794netdev_features_t netdev_increment_features(netdev_features_t all,
4795 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4796
4797/* Allow TSO being used on stacked device :
4798 * Performing the GSO segmentation before last device
4799 * is a performance improvement.
4800 */
4801static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4802 netdev_features_t mask)
4803{
4804 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4805}
4806
6cb6a27c 4807int __netdev_update_features(struct net_device *dev);
5455c699 4808void netdev_update_features(struct net_device *dev);
afe12cc8 4809void netdev_change_features(struct net_device *dev);
7f353bf2 4810
fc4a7489
PM
4811void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4812 struct net_device *dev);
4813
e38f3025
TM
4814netdev_features_t passthru_features_check(struct sk_buff *skb,
4815 struct net_device *dev,
4816 netdev_features_t features);
c1e756bf 4817netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 4818
4d29515f 4819static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4820{
7b748340 4821 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4822
4823 /* check flags correspondence */
4824 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
4825 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4826 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4827 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4828 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4829 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4830 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4831 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4832 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4833 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4834 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4835 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4836 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4837 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4838 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 4839 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 4840 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 4841 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
3b335832 4842 BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
0345e186 4843
d6b4991a 4844 return (features & feature) == feature;
576a30eb
HX
4845}
4846
4d29515f 4847static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4848{
278b2513 4849 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4850 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4851}
4852
8b86a61d 4853static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4854 netdev_features_t features)
7967168c 4855{
fc741216 4856 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
4857 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
4858 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
4859}
4860
82cc1a7a
PWJ
4861static inline void netif_set_gso_max_size(struct net_device *dev,
4862 unsigned int size)
4863{
4b66d216
ED
4864 /* dev->gso_max_size is read locklessly from sk_setup_caps() */
4865 WRITE_ONCE(dev->gso_max_size, size);
82cc1a7a
PWJ
4866}
4867
6d872df3
ED
4868static inline void netif_set_gso_max_segs(struct net_device *dev,
4869 unsigned int segs)
4870{
4871 /* dev->gso_max_segs is read locklessly from sk_setup_caps() */
4872 WRITE_ONCE(dev->gso_max_segs, segs);
4873}
4874
eac1b93c
CL
4875static inline void netif_set_gro_max_size(struct net_device *dev,
4876 unsigned int size)
4877{
4878 /* This pairs with the READ_ONCE() in skb_gro_receive() */
4879 WRITE_ONCE(dev->gro_max_size, size);
4880}
4881
7a7ffbab
WCC
4882static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
4883 int pulled_hlen, u16 mac_offset,
4884 int mac_len)
4885{
4886 skb->protocol = protocol;
4887 skb->encapsulation = 1;
4888 skb_push(skb, pulled_hlen);
4889 skb_reset_transport_header(skb);
4890 skb->mac_header = mac_offset;
4891 skb->network_header = skb->mac_header + mac_len;
4892 skb->mac_len = mac_len;
4893}
4894
3c175784
SD
4895static inline bool netif_is_macsec(const struct net_device *dev)
4896{
4897 return dev->priv_flags & IFF_MACSEC;
4898}
4899
b618aaa9 4900static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
4901{
4902 return dev->priv_flags & IFF_MACVLAN;
4903}
4904
b618aaa9 4905static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
4906{
4907 return dev->priv_flags & IFF_MACVLAN_PORT;
4908}
4909
b618aaa9 4910static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 4911{
4912 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
4913}
4914
b618aaa9 4915static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
4916{
4917 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
4918}
4919
3bdc0eba
BG
4920static inline bool netif_supports_nofcs(struct net_device *dev)
4921{
4922 return dev->priv_flags & IFF_SUPP_NOFCS;
4923}
4924
d5256083
DB
4925static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
4926{
4927 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
4928}
4929
007979ea 4930static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 4931{
007979ea 4932 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
4933}
4934
fee6d4c7
DA
4935static inline bool netif_is_l3_slave(const struct net_device *dev)
4936{
4937 return dev->priv_flags & IFF_L3MDEV_SLAVE;
4938}
4939
0894ae3f
JP
4940static inline bool netif_is_bridge_master(const struct net_device *dev)
4941{
4942 return dev->priv_flags & IFF_EBRIDGE;
4943}
4944
28f9ee22
VY
4945static inline bool netif_is_bridge_port(const struct net_device *dev)
4946{
4947 return dev->priv_flags & IFF_BRIDGE_PORT;
4948}
4949
35d4e172
JP
4950static inline bool netif_is_ovs_master(const struct net_device *dev)
4951{
4952 return dev->priv_flags & IFF_OPENVSWITCH;
4953}
4954
5be66141
JP
4955static inline bool netif_is_ovs_port(const struct net_device *dev)
4956{
4957 return dev->priv_flags & IFF_OVS_DATAPATH;
4958}
4959
df23bb18
SB
4960static inline bool netif_is_any_bridge_port(const struct net_device *dev)
4961{
4962 return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
4963}
4964
b618aaa9 4965static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
4966{
4967 return dev->priv_flags & IFF_TEAM;
4968}
4969
b618aaa9 4970static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
4971{
4972 return dev->priv_flags & IFF_TEAM_PORT;
4973}
4974
b618aaa9 4975static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
4976{
4977 return netif_is_bond_master(dev) || netif_is_team_master(dev);
4978}
4979
b618aaa9 4980static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
4981{
4982 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
4983}
4984
d4ab4286
KJ
4985static inline bool netif_is_rxfh_configured(const struct net_device *dev)
4986{
4987 return dev->priv_flags & IFF_RXFH_CONFIGURED;
4988}
4989
30c8bd5a
SS
4990static inline bool netif_is_failover(const struct net_device *dev)
4991{
4992 return dev->priv_flags & IFF_FAILOVER;
4993}
4994
4995static inline bool netif_is_failover_slave(const struct net_device *dev)
4996{
4997 return dev->priv_flags & IFF_FAILOVER_SLAVE;
4998}
4999
02875878
ED
5000/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
5001static inline void netif_keep_dst(struct net_device *dev)
5002{
5003 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
5004}
5005
18d3df3e
PA
5006/* return true if dev can't cope with mtu frames that need vlan tag insertion */
5007static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
5008{
5009 /* TODO: reserve and use an additional IFF bit, if we get more users */
01992152 5010 return netif_is_macsec(dev);
18d3df3e
PA
5011}
5012
505d4f73 5013extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 5014
571ba423
JP
5015/* Logging, debugging and troubleshooting/diagnostic helpers. */
5016
5017/* netdev_printk helpers, similar to dev_printk */
5018
5019static inline const char *netdev_name(const struct net_device *dev)
5020{
c6f854d5
VF
5021 if (!dev->name[0] || strchr(dev->name, '%'))
5022 return "(unnamed net_device)";
571ba423
JP
5023 return dev->name;
5024}
5025
8397ed36
DA
5026static inline bool netdev_unregistering(const struct net_device *dev)
5027{
5028 return dev->reg_state == NETREG_UNREGISTERING;
5029}
5030
ccc7f496
VF
5031static inline const char *netdev_reg_state(const struct net_device *dev)
5032{
5033 switch (dev->reg_state) {
5034 case NETREG_UNINITIALIZED: return " (uninitialized)";
5035 case NETREG_REGISTERED: return "";
5036 case NETREG_UNREGISTERING: return " (unregistering)";
5037 case NETREG_UNREGISTERED: return " (unregistered)";
5038 case NETREG_RELEASED: return " (released)";
5039 case NETREG_DUMMY: return " (dummy)";
5040 }
5041
5042 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
5043 return " (unknown)";
5044}
5045
ce3fdb69 5046__printf(3, 4) __cold
6ea754eb
JP
5047void netdev_printk(const char *level, const struct net_device *dev,
5048 const char *format, ...);
ce3fdb69 5049__printf(2, 3) __cold
6ea754eb 5050void netdev_emerg(const struct net_device *dev, const char *format, ...);
ce3fdb69 5051__printf(2, 3) __cold
6ea754eb 5052void netdev_alert(const struct net_device *dev, const char *format, ...);
ce3fdb69 5053__printf(2, 3) __cold
6ea754eb 5054void netdev_crit(const struct net_device *dev, const char *format, ...);
ce3fdb69 5055__printf(2, 3) __cold
6ea754eb 5056void netdev_err(const struct net_device *dev, const char *format, ...);
ce3fdb69 5057__printf(2, 3) __cold
6ea754eb 5058void netdev_warn(const struct net_device *dev, const char *format, ...);
ce3fdb69 5059__printf(2, 3) __cold
6ea754eb 5060void netdev_notice(const struct net_device *dev, const char *format, ...);
ce3fdb69 5061__printf(2, 3) __cold
6ea754eb 5062void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 5063
375ef2b1
GP
5064#define netdev_level_once(level, dev, fmt, ...) \
5065do { \
7071732c 5066 static bool __section(".data.once") __print_once; \
375ef2b1
GP
5067 \
5068 if (!__print_once) { \
5069 __print_once = true; \
5070 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
5071 } \
5072} while (0)
5073
5074#define netdev_emerg_once(dev, fmt, ...) \
5075 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
5076#define netdev_alert_once(dev, fmt, ...) \
5077 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
5078#define netdev_crit_once(dev, fmt, ...) \
5079 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
5080#define netdev_err_once(dev, fmt, ...) \
5081 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
5082#define netdev_warn_once(dev, fmt, ...) \
5083 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
5084#define netdev_notice_once(dev, fmt, ...) \
5085 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
5086#define netdev_info_once(dev, fmt, ...) \
5087 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
5088
8909c9ad
VK
5089#define MODULE_ALIAS_NETDEV(device) \
5090 MODULE_ALIAS("netdev-" device)
5091
ceabef7d
OZ
5092#if defined(CONFIG_DYNAMIC_DEBUG) || \
5093 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
571ba423
JP
5094#define netdev_dbg(__dev, format, args...) \
5095do { \
ffa10cb4 5096 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 5097} while (0)
b558c96f
JC
5098#elif defined(DEBUG)
5099#define netdev_dbg(__dev, format, args...) \
5100 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
5101#else
5102#define netdev_dbg(__dev, format, args...) \
5103({ \
5104 if (0) \
5105 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
5106})
5107#endif
5108
5109#if defined(VERBOSE_DEBUG)
5110#define netdev_vdbg netdev_dbg
5111#else
5112
5113#define netdev_vdbg(dev, format, args...) \
5114({ \
5115 if (0) \
5116 netdev_printk(KERN_DEBUG, dev, format, ##args); \
5117 0; \
5118})
5119#endif
5120
5121/*
5122 * netdev_WARN() acts like dev_printk(), but with the key difference
5123 * of using a WARN/WARN_ON to get the message out, including the
5124 * file/line information and a backtrace.
5125 */
5126#define netdev_WARN(dev, format, args...) \
e1cfe3d0 5127 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 5128 netdev_reg_state(dev), ##args)
571ba423 5129
72dd831e 5130#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 5131 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
5132 netdev_reg_state(dev), ##args)
5133
b3d95c5c
JP
5134/* netif printk helpers, similar to netdev_printk */
5135
5136#define netif_printk(priv, type, level, dev, fmt, args...) \
5137do { \
5138 if (netif_msg_##type(priv)) \
5139 netdev_printk(level, (dev), fmt, ##args); \
5140} while (0)
5141
f45f4321
JP
5142#define netif_level(level, priv, type, dev, fmt, args...) \
5143do { \
5144 if (netif_msg_##type(priv)) \
5145 netdev_##level(dev, fmt, ##args); \
5146} while (0)
5147
b3d95c5c 5148#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 5149 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 5150#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 5151 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 5152#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 5153 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 5154#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 5155 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 5156#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 5157 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 5158#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 5159 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 5160#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 5161 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 5162
ceabef7d
OZ
5163#if defined(CONFIG_DYNAMIC_DEBUG) || \
5164 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
b3d95c5c
JP
5165#define netif_dbg(priv, type, netdev, format, args...) \
5166do { \
5167 if (netif_msg_##type(priv)) \
b5fb0a03 5168 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 5169} while (0)
0053ea9c
JP
5170#elif defined(DEBUG)
5171#define netif_dbg(priv, type, dev, format, args...) \
5172 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
5173#else
5174#define netif_dbg(priv, type, dev, format, args...) \
5175({ \
5176 if (0) \
5177 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
5178 0; \
5179})
5180#endif
5181
f617f276
EC
5182/* if @cond then downgrade to debug, else print at @level */
5183#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
5184 do { \
5185 if (cond) \
5186 netif_dbg(priv, type, netdev, fmt, ##args); \
5187 else \
5188 netif_ ## level(priv, type, netdev, fmt, ##args); \
5189 } while (0)
5190
b3d95c5c 5191#if defined(VERBOSE_DEBUG)
bcfcc450 5192#define netif_vdbg netif_dbg
b3d95c5c
JP
5193#else
5194#define netif_vdbg(priv, type, dev, format, args...) \
5195({ \
5196 if (0) \
a4ed89cb 5197 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
5198 0; \
5199})
5200#endif
571ba423 5201
900ff8c6
CW
5202/*
5203 * The list of packet types we will receive (as opposed to discard)
5204 * and the routines to invoke.
5205 *
5206 * Why 16. Because with 16 the only overlap we get on a hash of the
5207 * low nibble of the protocol value is RARP/SNAP/X.25.
5208 *
900ff8c6 5209 * 0800 IP
900ff8c6
CW
5210 * 0001 802.3
5211 * 0002 AX.25
5212 * 0004 802.2
5213 * 8035 RARP
5214 * 0005 SNAP
5215 * 0805 X.25
5216 * 0806 ARP
5217 * 8137 IPX
5218 * 0009 Localtalk
5219 * 86DD IPv6
5220 */
5221#define PTYPE_HASH_SIZE (16)
5222#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
5223
744b8376
VO
5224extern struct list_head ptype_all __read_mostly;
5225extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
5226
4de83b88
MB
5227extern struct net_device *blackhole_netdev;
5228
385a154c 5229#endif /* _LINUX_NETDEVICE_H */