net: dsa: resolve forwarding path for dsa slave ports
[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>
31
1da177e4 32#include <linux/percpu.h>
4d5b78c0 33#include <linux/rculist.h>
bea3348e 34#include <linux/workqueue.h>
114cf580 35#include <linux/dynamic_queue_limits.h>
1da177e4 36
a050c33f 37#include <net/net_namespace.h>
7a6b6f51 38#ifdef CONFIG_DCB
2f90b865
AD
39#include <net/dcbnl.h>
40#endif
5bc1421e 41#include <net/netprio_cgroup.h>
e817f856 42#include <net/xdp.h>
a050c33f 43
a59e2ecb 44#include <linux/netdev_features.h>
77162022 45#include <linux/neighbour.h>
607ca46e 46#include <uapi/linux/netdevice.h>
61bd3857 47#include <uapi/linux/if_bonding.h>
e4c6734e 48#include <uapi/linux/pkt_cls.h>
59cc1f61 49#include <linux/hashtable.h>
a59e2ecb 50
115c1d6e 51struct netpoll_info;
313162d0 52struct device;
cc69837f 53struct ethtool_ops;
c1f19b51 54struct phy_device;
2f657a60 55struct dsa_port;
607259a6 56struct ip_tunnel_parm;
30e9bb84
AT
57struct macsec_context;
58struct macsec_ops;
c6e970a0 59
e679c9c1 60struct sfp_bus;
704232c2
JB
61/* 802.11 specific */
62struct wireless_dev;
98a18b6f
AA
63/* 802.15.4 specific */
64struct wpan_dev;
03c57747 65struct mpls_dev;
7c46a640
AD
66/* UDP Tunnel offloads */
67struct udp_tunnel_info;
cc4e3835
JK
68struct udp_tunnel_nic_info;
69struct udp_tunnel_nic;
a7862b45 70struct bpf_prog;
814abfab 71struct xdp_buff;
1da177e4 72
5198d545 73void synchronize_net(void);
f629d208
JP
74void netdev_set_default_ethtool_ops(struct net_device *dev,
75 const struct ethtool_ops *ops);
d07d7507 76
9a1654ba
JP
77/* Backlog congestion levels */
78#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
79#define NET_RX_DROP 1 /* packet dropped */
80
7151affe
TY
81#define MAX_NEST_DEV 8
82
572a9d7b
PM
83/*
84 * Transmit return codes: transmit return codes originate from three different
85 * namespaces:
86 *
87 * - qdisc return codes
88 * - driver transmit return codes
89 * - errno values
90 *
91 * Drivers are allowed to return any one of those in their hard_start_xmit()
92 * function. Real network devices commonly used with qdiscs should only return
93 * the driver transmit return codes though - when qdiscs are used, the actual
94 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
95 * higher layers. Virtual network devices transmit synchronously; in this case
96 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
97 * others are propagated to higher layers.
98 */
99
100/* qdisc ->enqueue() return codes. */
101#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
102#define NET_XMIT_DROP 0x01 /* skb dropped */
103#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 104#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 105
b9df3cb8
GR
106/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
107 * indicates that the device will soon be dropping packets, or already drops
108 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 109#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
110#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
111
dc1f8bf6 112/* Driver transmit return codes */
9a1654ba 113#define NETDEV_TX_MASK 0xf0
572a9d7b 114
dc1f8bf6 115enum netdev_tx {
572a9d7b 116 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
117 NETDEV_TX_OK = 0x00, /* driver took care of packet */
118 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
119};
120typedef enum netdev_tx netdev_tx_t;
121
9a1654ba
JP
122/*
123 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
124 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
125 */
126static inline bool dev_xmit_complete(int rc)
127{
128 /*
129 * Positive cases with an skb consumed by a driver:
130 * - successful transmission (rc == NETDEV_TX_OK)
131 * - error while transmitting (rc < 0)
132 * - error while queueing to a different device (rc & NET_XMIT_MASK)
133 */
134 if (likely(rc < NET_XMIT_MASK))
135 return true;
136
137 return false;
138}
139
1da177e4 140/*
5e82b4b2 141 * Compute the worst-case header length according to the protocols
1da177e4
LT
142 * used.
143 */
fe2918b0 144
c0eb4540
KS
145#if defined(CONFIG_HYPERV_NET)
146# define LL_MAX_HEADER 128
147#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
148# if defined(CONFIG_MAC80211_MESH)
149# define LL_MAX_HEADER 128
150# else
151# define LL_MAX_HEADER 96
152# endif
1da177e4 153#else
8388e3da 154# define LL_MAX_HEADER 32
1da177e4
LT
155#endif
156
d11ead75
BH
157#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
158 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
159#define MAX_HEADER LL_MAX_HEADER
160#else
161#define MAX_HEADER (LL_MAX_HEADER + 48)
162#endif
163
164/*
be1f3c2c
BH
165 * Old network device statistics. Fields are native words
166 * (unsigned long) so they can be read and written atomically.
1da177e4 167 */
fe2918b0 168
d94d9fee 169struct net_device_stats {
3cfde79c
BH
170 unsigned long rx_packets;
171 unsigned long tx_packets;
172 unsigned long rx_bytes;
173 unsigned long tx_bytes;
174 unsigned long rx_errors;
175 unsigned long tx_errors;
176 unsigned long rx_dropped;
177 unsigned long tx_dropped;
178 unsigned long multicast;
1da177e4 179 unsigned long collisions;
1da177e4 180 unsigned long rx_length_errors;
3cfde79c
BH
181 unsigned long rx_over_errors;
182 unsigned long rx_crc_errors;
183 unsigned long rx_frame_errors;
184 unsigned long rx_fifo_errors;
185 unsigned long rx_missed_errors;
1da177e4
LT
186 unsigned long tx_aborted_errors;
187 unsigned long tx_carrier_errors;
188 unsigned long tx_fifo_errors;
189 unsigned long tx_heartbeat_errors;
190 unsigned long tx_window_errors;
1da177e4
LT
191 unsigned long rx_compressed;
192 unsigned long tx_compressed;
193};
194
1da177e4
LT
195
196#include <linux/cache.h>
197#include <linux/skbuff.h>
198
adc9300e 199#ifdef CONFIG_RPS
c5905afb 200#include <linux/static_key.h>
dc05360f
ED
201extern struct static_key_false rps_needed;
202extern struct static_key_false rfs_needed;
adc9300e
ED
203#endif
204
1da177e4
LT
205struct neighbour;
206struct neigh_parms;
207struct sk_buff;
208
f001fde5
JP
209struct netdev_hw_addr {
210 struct list_head list;
211 unsigned char addr[MAX_ADDR_LEN];
212 unsigned char type;
ccffad25
JP
213#define NETDEV_HW_ADDR_T_LAN 1
214#define NETDEV_HW_ADDR_T_SAN 2
8e1b3884
TY
215#define NETDEV_HW_ADDR_T_UNICAST 3
216#define NETDEV_HW_ADDR_T_MULTICAST 4
22bedad3 217 bool global_use;
4cd729b0 218 int sync_cnt;
8f8f103d 219 int refcount;
4543fbef 220 int synced;
f001fde5
JP
221 struct rcu_head rcu_head;
222};
223
31278e71
JP
224struct netdev_hw_addr_list {
225 struct list_head list;
226 int count;
227};
228
22bedad3
JP
229#define netdev_hw_addr_list_count(l) ((l)->count)
230#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
231#define netdev_hw_addr_list_for_each(ha, l) \
232 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 233
22bedad3
JP
234#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
235#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
236#define netdev_for_each_uc_addr(ha, dev) \
237 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
6683ece3 238
22bedad3
JP
239#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
240#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 241#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 242 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
6683ece3 243
d94d9fee 244struct hh_cache {
5b3dc2f3 245 unsigned int hh_len;
3644f0ce 246 seqlock_t hh_lock;
1da177e4
LT
247
248 /* cached hardware header; allow for machine alignment needs. */
249#define HH_DATA_MOD 16
250#define HH_DATA_OFF(__len) \
5ba0eac6 251 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
252#define HH_DATA_ALIGN(__len) \
253 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
254 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
255};
256
5e82b4b2 257/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
258 * Alternative is:
259 * dev->hard_header_len ? (dev->hard_header_len +
260 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
261 *
262 * We could use other alignment values, but we must maintain the
263 * relationship HH alignment <= LL alignment.
264 */
265#define LL_RESERVED_SPACE(dev) \
f5184d26 266 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 267#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
f5184d26 268 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 269
3b04ddde
SH
270struct header_ops {
271 int (*create) (struct sk_buff *skb, struct net_device *dev,
272 unsigned short type, const void *daddr,
95c96174 273 const void *saddr, unsigned int len);
3b04ddde 274 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 275 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
276 void (*cache_update)(struct hh_cache *hh,
277 const struct net_device *dev,
278 const unsigned char *haddr);
2793a23a 279 bool (*validate)(const char *ll_header, unsigned int len);
e78b2915 280 __be16 (*parse_protocol)(const struct sk_buff *skb);
3b04ddde
SH
281};
282
1da177e4 283/* These flag bits are private to the generic network queueing
5e82b4b2 284 * layer; they may not be explicitly referenced by any other
1da177e4
LT
285 * code.
286 */
287
d94d9fee 288enum netdev_state_t {
1da177e4
LT
289 __LINK_STATE_START,
290 __LINK_STATE_PRESENT,
1da177e4 291 __LINK_STATE_NOCARRIER,
b00055aa
SR
292 __LINK_STATE_LINKWATCH_PENDING,
293 __LINK_STATE_DORMANT,
eec517cd 294 __LINK_STATE_TESTING,
1da177e4
LT
295};
296
297
298/*
5e82b4b2 299 * This structure holds boot-time configured netdevice settings. They
fe2918b0 300 * are then used in the device probing.
1da177e4
LT
301 */
302struct netdev_boot_setup {
303 char name[IFNAMSIZ];
304 struct ifmap map;
305};
306#define NETDEV_BOOT_SETUP_MAX 8
307
f629d208 308int __init netdev_boot_setup(char *str);
1da177e4 309
6312fe77
LR
310struct gro_list {
311 struct list_head list;
312 int count;
313};
314
bea3348e 315/*
d9f37d01
LR
316 * size of gro hash buckets, must less than bit number of
317 * napi_struct::gro_bitmask
bea3348e 318 */
07d78363 319#define GRO_HASH_BUCKETS 8
d9f37d01
LR
320
321/*
322 * Structure for NAPI scheduling similar to tasklet but with weighting
323 */
bea3348e
SH
324struct napi_struct {
325 /* The poll_list must only be managed by the entity which
326 * changes the state of the NAPI_STATE_SCHED bit. This means
327 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 328 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
329 * can remove from the list right before clearing the bit.
330 */
331 struct list_head poll_list;
332
333 unsigned long state;
334 int weight;
6f8b12d6 335 int defer_hard_irqs_count;
d9f37d01 336 unsigned long gro_bitmask;
bea3348e
SH
337 int (*poll)(struct napi_struct *, int);
338#ifdef CONFIG_NETPOLL
bea3348e 339 int poll_owner;
bea3348e 340#endif
5d38a079 341 struct net_device *dev;
6312fe77 342 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 343 struct sk_buff *skb;
323ebb61
EC
344 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
345 int rx_count; /* length of rx_list */
3b47d303 346 struct hrtimer timer;
404f7c9e 347 struct list_head dev_list;
af12fa6e
ET
348 struct hlist_node napi_hash_node;
349 unsigned int napi_id;
29863d41 350 struct task_struct *thread;
bea3348e
SH
351};
352
d94d9fee 353enum {
7fd3253a
BT
354 NAPI_STATE_SCHED, /* Poll is scheduled */
355 NAPI_STATE_MISSED, /* reschedule a napi */
356 NAPI_STATE_DISABLE, /* Disable pending */
357 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
358 NAPI_STATE_LISTED, /* NAPI added to system lists */
359 NAPI_STATE_NO_BUSY_POLL, /* Do not add in napi_hash, no busy polling */
360 NAPI_STATE_IN_BUSY_POLL, /* sk_busy_loop() owns this NAPI */
361 NAPI_STATE_PREFER_BUSY_POLL, /* prefer busy-polling over softirq processing*/
29863d41 362 NAPI_STATE_THREADED, /* The poll is performed inside its own thread*/
217f6974
ED
363};
364
365enum {
7fd3253a
BT
366 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
367 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
368 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
369 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
370 NAPIF_STATE_LISTED = BIT(NAPI_STATE_LISTED),
371 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
372 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
373 NAPIF_STATE_PREFER_BUSY_POLL = BIT(NAPI_STATE_PREFER_BUSY_POLL),
29863d41 374 NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
bea3348e
SH
375};
376
5b252f0c 377enum gro_result {
d1c76af9
HX
378 GRO_MERGED,
379 GRO_MERGED_FREE,
380 GRO_HELD,
381 GRO_NORMAL,
25393d3f 382 GRO_CONSUMED,
d1c76af9 383};
5b252f0c 384typedef enum gro_result gro_result_t;
d1c76af9 385
8a4eb573
JP
386/*
387 * enum rx_handler_result - Possible return values for rx_handlers.
388 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
389 * further.
390 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
391 * case skb->dev was changed by rx_handler.
392 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 393 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
394 *
395 * rx_handlers are functions called from inside __netif_receive_skb(), to do
396 * special processing of the skb, prior to delivery to protocol handlers.
397 *
398 * Currently, a net_device can only have a single rx_handler registered. Trying
399 * to register a second rx_handler will return -EBUSY.
400 *
401 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
402 * To unregister a rx_handler on a net_device, use
403 * netdev_rx_handler_unregister().
404 *
405 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
406 * do with the skb.
407 *
5e82b4b2 408 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 409 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 410 * the skb to be delivered in some other way.
8a4eb573
JP
411 *
412 * If the rx_handler changed skb->dev, to divert the skb to another
413 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
414 * new device will be called if it exists.
415 *
5e82b4b2 416 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 417 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 418 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 419 *
5e82b4b2 420 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
421 * delivered, it should return RX_HANDLER_PASS.
422 *
423 * A device without a registered rx_handler will behave as if rx_handler
424 * returned RX_HANDLER_PASS.
425 */
426
427enum rx_handler_result {
428 RX_HANDLER_CONSUMED,
429 RX_HANDLER_ANOTHER,
430 RX_HANDLER_EXACT,
431 RX_HANDLER_PASS,
432};
433typedef enum rx_handler_result rx_handler_result_t;
434typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 435
f629d208 436void __napi_schedule(struct napi_struct *n);
bc9ad166 437void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 438
4d29515f 439static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
440{
441 return test_bit(NAPI_STATE_DISABLE, &n->state);
442}
443
7fd3253a
BT
444static inline bool napi_prefer_busy_poll(struct napi_struct *n)
445{
446 return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
447}
448
39e6c820 449bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
450
451/**
452 * napi_schedule - schedule NAPI poll
5e82b4b2 453 * @n: NAPI context
bea3348e
SH
454 *
455 * Schedule NAPI poll routine to be called if it is not already
456 * running.
457 */
458static inline void napi_schedule(struct napi_struct *n)
459{
460 if (napi_schedule_prep(n))
461 __napi_schedule(n);
462}
463
bc9ad166
ED
464/**
465 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 466 * @n: NAPI context
bc9ad166
ED
467 *
468 * Variant of napi_schedule(), assuming hard irqs are masked.
469 */
470static inline void napi_schedule_irqoff(struct napi_struct *n)
471{
472 if (napi_schedule_prep(n))
473 __napi_schedule_irqoff(n);
474}
475
bfe13f54 476/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 477static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
478{
479 if (napi_schedule_prep(napi)) {
480 __napi_schedule(napi);
4d29515f 481 return true;
bfe13f54 482 }
4d29515f 483 return false;
bfe13f54
RD
484}
485
364b6055 486bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
487/**
488 * napi_complete - NAPI processing complete
5e82b4b2 489 * @n: NAPI context
bea3348e
SH
490 *
491 * Mark NAPI processing as complete.
3b47d303 492 * Consider using napi_complete_done() instead.
364b6055 493 * Return false if device should avoid rearming interrupts.
bea3348e 494 */
364b6055 495static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
496{
497 return napi_complete_done(n, 0);
498}
bea3348e 499
5fdd2f0e
WW
500int dev_set_threaded(struct net_device *dev, bool threaded);
501
bea3348e
SH
502/**
503 * napi_disable - prevent NAPI from scheduling
5e82b4b2 504 * @n: NAPI context
bea3348e
SH
505 *
506 * Stop NAPI from being scheduled on this context.
507 * Waits till any outstanding processing completes.
508 */
3b47d303 509void napi_disable(struct napi_struct *n);
bea3348e 510
29863d41 511void napi_enable(struct napi_struct *n);
bea3348e 512
c264c3de
SH
513/**
514 * napi_synchronize - wait until NAPI is not running
5e82b4b2 515 * @n: NAPI context
c264c3de
SH
516 *
517 * Wait until NAPI is done being scheduled on this context.
518 * Waits till any outstanding processing completes but
519 * does not disable future activations.
520 */
521static inline void napi_synchronize(const struct napi_struct *n)
522{
facc432f
AB
523 if (IS_ENABLED(CONFIG_SMP))
524 while (test_bit(NAPI_STATE_SCHED, &n->state))
525 msleep(1);
526 else
527 barrier();
c264c3de 528}
c264c3de 529
6c5c9581
MK
530/**
531 * napi_if_scheduled_mark_missed - if napi is running, set the
532 * NAPIF_STATE_MISSED
533 * @n: NAPI context
534 *
535 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
536 * NAPI is scheduled.
537 **/
538static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
539{
540 unsigned long val, new;
541
542 do {
543 val = READ_ONCE(n->state);
544 if (val & NAPIF_STATE_DISABLE)
545 return true;
546
547 if (!(val & NAPIF_STATE_SCHED))
548 return false;
549
550 new = val | NAPIF_STATE_MISSED;
551 } while (cmpxchg(&n->state, val, new) != val);
552
553 return true;
554}
555
d94d9fee 556enum netdev_queue_state_t {
73466498
TH
557 __QUEUE_STATE_DRV_XOFF,
558 __QUEUE_STATE_STACK_XOFF,
c3f26a26 559 __QUEUE_STATE_FROZEN,
79d16385 560};
8e2f1a63
DB
561
562#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
563#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
564#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
565
566#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
567#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
568 QUEUE_STATE_FROZEN)
569#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
570 QUEUE_STATE_FROZEN)
571
73466498
TH
572/*
573 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
574 * netif_tx_* functions below are used to manipulate this flag. The
575 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
576 * queue independently. The netif_xmit_*stopped functions below are called
577 * to check if the queue has been stopped by the driver or stack (either
578 * of the XOFF bits are set in the state). Drivers should not need to call
579 * netif_xmit*stopped functions, they should only be using netif_tx_*.
580 */
79d16385 581
bb949fbd 582struct netdev_queue {
6a321cb3 583/*
5e82b4b2 584 * read-mostly part
6a321cb3 585 */
bb949fbd 586 struct net_device *dev;
46e5da40 587 struct Qdisc __rcu *qdisc;
b0e1e646 588 struct Qdisc *qdisc_sleeping;
ccf5ff69 589#ifdef CONFIG_SYSFS
1d24eb48
TH
590 struct kobject kobj;
591#endif
f2cd2d3e
ED
592#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
593 int numa_node;
594#endif
c0ef079c
FW
595 unsigned long tx_maxrate;
596 /*
597 * Number of TX timeouts for this queue
598 * (/sys/class/net/DEV/Q/trans_timeout)
599 */
600 unsigned long trans_timeout;
ffcfe25b
AD
601
602 /* Subordinate device that the queue has been assigned to */
603 struct net_device *sb_dev;
661b8d1b 604#ifdef CONFIG_XDP_SOCKETS
1742b3d5 605 struct xsk_buff_pool *pool;
661b8d1b 606#endif
6a321cb3 607/*
5e82b4b2 608 * write-mostly part
6a321cb3
ED
609 */
610 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
611 int xmit_lock_owner;
9d21493b 612 /*
9b36627a 613 * Time (in jiffies) of last Tx
9d21493b
ED
614 */
615 unsigned long trans_start;
ccf5ff69 616
114cf580
TH
617 unsigned long state;
618
619#ifdef CONFIG_BQL
620 struct dql dql;
621#endif
e8a0464c 622} ____cacheline_aligned_in_smp;
bb949fbd 623
79134e6c 624extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 625extern int sysctl_devconf_inherit_init_net;
79134e6c 626
316cdaa1
MB
627/*
628 * sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
629 * == 1 : For initns only
630 * == 2 : For none.
631 */
79134e6c
ED
632static inline bool net_has_fallback_tunnels(const struct net *net)
633{
3753d977
MB
634 return !IS_ENABLED(CONFIG_SYSCTL) ||
635 !sysctl_fb_tunnels_only_for_init_net ||
636 (net == &init_net && sysctl_fb_tunnels_only_for_init_net == 1);
79134e6c
ED
637}
638
f2cd2d3e
ED
639static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
640{
641#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
642 return q->numa_node;
643#else
b236da69 644 return NUMA_NO_NODE;
f2cd2d3e
ED
645#endif
646}
647
648static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
649{
650#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
651 q->numa_node = node;
652#endif
653}
654
df334545 655#ifdef CONFIG_RPS
0a9627f2
TH
656/*
657 * This structure holds an RPS map which can be of variable length. The
658 * map is an array of CPUs.
659 */
660struct rps_map {
661 unsigned int len;
662 struct rcu_head rcu;
bb4cf02d 663 u16 cpus[];
0a9627f2 664};
60b778ce 665#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 666
fec5e652 667/*
c445477d
BH
668 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
669 * tail pointer for that CPU's input queue at the time of last enqueue, and
670 * a hardware filter index.
fec5e652
TH
671 */
672struct rps_dev_flow {
673 u16 cpu;
c445477d 674 u16 filter;
fec5e652
TH
675 unsigned int last_qtail;
676};
c445477d 677#define RPS_NO_FILTER 0xffff
fec5e652
TH
678
679/*
680 * The rps_dev_flow_table structure contains a table of flow mappings.
681 */
682struct rps_dev_flow_table {
683 unsigned int mask;
684 struct rcu_head rcu;
bb4cf02d 685 struct rps_dev_flow flows[];
fec5e652
TH
686};
687#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 688 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
689
690/*
691 * The rps_sock_flow_table contains mappings of flows to the last CPU
692 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
693 * Each entry is a 32bit value. Upper part is the high-order bits
694 * of flow hash, lower part is CPU number.
567e4b79 695 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
696 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
697 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 698 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
699 */
700struct rps_sock_flow_table {
567e4b79 701 u32 mask;
93c1af6c 702
bb4cf02d 703 u32 ents[] ____cacheline_aligned_in_smp;
fec5e652 704};
567e4b79 705#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
706
707#define RPS_NO_CPU 0xffff
708
567e4b79
ED
709extern u32 rps_cpu_mask;
710extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
711
fec5e652
TH
712static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
713 u32 hash)
714{
715 if (table && hash) {
567e4b79
ED
716 unsigned int index = hash & table->mask;
717 u32 val = hash & ~rps_cpu_mask;
fec5e652 718
5e82b4b2 719 /* We only give a hint, preemption can change CPU under us */
567e4b79 720 val |= raw_smp_processor_id();
fec5e652 721
567e4b79
ED
722 if (table->ents[index] != val)
723 table->ents[index] = val;
fec5e652
TH
724 }
725}
726
c445477d 727#ifdef CONFIG_RFS_ACCEL
f629d208
JP
728bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
729 u16 filter_id);
c445477d 730#endif
a953be53 731#endif /* CONFIG_RPS */
c445477d 732
0a9627f2
TH
733/* This structure contains an instance of an RX queue. */
734struct netdev_rx_queue {
a953be53 735#ifdef CONFIG_RPS
6e3f7faf
ED
736 struct rps_map __rcu *rps_map;
737 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 738#endif
6e3f7faf 739 struct kobject kobj;
fe822240 740 struct net_device *dev;
e817f856 741 struct xdp_rxq_info xdp_rxq;
661b8d1b 742#ifdef CONFIG_XDP_SOCKETS
1742b3d5 743 struct xsk_buff_pool *pool;
661b8d1b 744#endif
0a9627f2 745} ____cacheline_aligned_in_smp;
a953be53
MD
746
747/*
748 * RX queue sysfs structures and functions.
749 */
750struct rx_queue_attribute {
751 struct attribute attr;
718ad681 752 ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
a953be53 753 ssize_t (*store)(struct netdev_rx_queue *queue,
718ad681 754 const char *buf, size_t len);
a953be53 755};
d314774c 756
044ab86d
AT
757/* XPS map type and offset of the xps map within net_device->xps_maps[]. */
758enum xps_map_type {
759 XPS_CPUS = 0,
760 XPS_RXQS,
761 XPS_MAPS_MAX,
762};
763
bf264145
TH
764#ifdef CONFIG_XPS
765/*
766 * This structure holds an XPS map which can be of variable length. The
767 * map is an array of queues.
768 */
769struct xps_map {
770 unsigned int len;
771 unsigned int alloc_len;
772 struct rcu_head rcu;
bb4cf02d 773 u16 queues[];
bf264145 774};
60b778ce 775#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
776#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
777 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
778
779/*
780 * This structure holds all XPS maps for device. Maps are indexed by CPU.
255c04a8 781 *
5478fcd0
AT
782 * We keep track of the number of cpus/rxqs used when the struct is allocated,
783 * in nr_ids. This will help not accessing out-of-bound memory.
784 *
255c04a8
AT
785 * We keep track of the number of traffic classes used when the struct is
786 * allocated, in num_tc. This will be used to navigate the maps, to ensure we're
787 * not crossing its upper bound, as the original dev->num_tc can be updated in
788 * the meantime.
bf264145
TH
789 */
790struct xps_dev_maps {
791 struct rcu_head rcu;
5478fcd0 792 unsigned int nr_ids;
255c04a8 793 s16 num_tc;
bb4cf02d 794 struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
bf264145 795};
80d19669
AN
796
797#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 798 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
799
800#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
801 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
802
bf264145
TH
803#endif /* CONFIG_XPS */
804
4f57c087
JF
805#define TC_MAX_QUEUE 16
806#define TC_BITMASK 15
807/* HW offloaded queuing disciplines txq count and offset maps */
808struct netdev_tc_txq {
809 u16 count;
810 u16 offset;
811};
812
68bad94e
NP
813#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
814/*
815 * This structure is to hold information about the device
816 * configured to run FCoE protocol stack.
817 */
818struct netdev_fcoe_hbainfo {
819 char manufacturer[64];
820 char serial_number[64];
821 char hardware_version[64];
822 char driver_version[64];
823 char optionrom_version[64];
824 char firmware_version[64];
825 char model[256];
826 char model_description[256];
827};
828#endif
829
02637fce 830#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 831
02637fce
JP
832/* This structure holds a unique identifier to identify some
833 * physical item (port for example) used by a netdevice.
66b52b0d 834 */
02637fce
JP
835struct netdev_phys_item_id {
836 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
837 unsigned char id_len;
838};
839
d754f98b
SF
840static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
841 struct netdev_phys_item_id *b)
842{
843 return a->id_len == b->id_len &&
844 memcmp(a->id, b->id, a->id_len) == 0;
845}
846
99932d4f 847typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
848 struct sk_buff *skb,
849 struct net_device *sb_dev);
99932d4f 850
ddb94eaf
PNA
851enum net_device_path_type {
852 DEV_PATH_ETHERNET = 0,
e4417d69 853 DEV_PATH_VLAN,
ec9d16ba 854 DEV_PATH_BRIDGE,
f6efc675 855 DEV_PATH_PPPOE,
0994d492 856 DEV_PATH_DSA,
ddb94eaf
PNA
857};
858
859struct net_device_path {
860 enum net_device_path_type type;
861 const struct net_device *dev;
e4417d69
PNA
862 union {
863 struct {
864 u16 id;
865 __be16 proto;
f6efc675 866 u8 h_dest[ETH_ALEN];
e4417d69 867 } encap;
bcf2766b
FF
868 struct {
869 enum {
870 DEV_PATH_BR_VLAN_KEEP,
871 DEV_PATH_BR_VLAN_TAG,
872 DEV_PATH_BR_VLAN_UNTAG,
873 } vlan_mode;
874 u16 vlan_id;
875 __be16 vlan_proto;
876 } bridge;
0994d492
FF
877 struct {
878 int port;
879 u16 proto;
880 } dsa;
e4417d69 881 };
ddb94eaf
PNA
882};
883
884#define NET_DEVICE_PATH_STACK_MAX 5
bcf2766b 885#define NET_DEVICE_PATH_VLAN_MAX 2
ddb94eaf
PNA
886
887struct net_device_path_stack {
888 int num_paths;
889 struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
890};
891
892struct net_device_path_ctx {
893 const struct net_device *dev;
894 const u8 *daddr;
bcf2766b
FF
895
896 int num_vlans;
897 struct {
898 u16 id;
899 __be16 proto;
900 } vlan[NET_DEVICE_PATH_VLAN_MAX];
ddb94eaf
PNA
901};
902
2572ac53 903enum tc_setup_type {
575ed7d3 904 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 905 TC_SETUP_CLSU32,
5b33f488 906 TC_SETUP_CLSFLOWER,
ade9b658 907 TC_SETUP_CLSMATCHALL,
332ae8e2 908 TC_SETUP_CLSBPF,
8c4083b3 909 TC_SETUP_BLOCK,
8521db4c 910 TC_SETUP_QDISC_CBS,
602f3baf 911 TC_SETUP_QDISC_RED,
7fdb61b4 912 TC_SETUP_QDISC_PRIO,
f971b132 913 TC_SETUP_QDISC_MQ,
25db26a9 914 TC_SETUP_QDISC_ETF,
98b0e5f6 915 TC_SETUP_ROOT_QDISC,
890d8d23 916 TC_SETUP_QDISC_GRED,
9c66d156 917 TC_SETUP_QDISC_TAPRIO,
c29f74e0 918 TC_SETUP_FT,
d35eb52b 919 TC_SETUP_QDISC_ETS,
ef6aadcc 920 TC_SETUP_QDISC_TBF,
aaca9408 921 TC_SETUP_QDISC_FIFO,
d03b195b 922 TC_SETUP_QDISC_HTB,
16e5cc64
JF
923};
924
f4e63525
JK
925/* These structures hold the attributes of bpf state that are being passed
926 * to the netdevice through the bpf op.
a7862b45 927 */
f4e63525 928enum bpf_netdev_command {
a7862b45
BB
929 /* Set or clear a bpf program used in the earliest stages of packet
930 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
931 * is responsible for calling bpf_prog_put on any old progs that are
932 * stored. In case of error, the callee need not release the new prog
933 * reference, but on success it takes ownership and must bpf_prog_put
934 * when it is no longer used.
935 */
936 XDP_SETUP_PROG,
ee5d032f 937 XDP_SETUP_PROG_HW,
ab3f0063 938 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
939 BPF_OFFLOAD_MAP_ALLOC,
940 BPF_OFFLOAD_MAP_FREE,
1742b3d5 941 XDP_SETUP_XSK_POOL,
a7862b45
BB
942};
943
cae1927c 944struct bpf_prog_offload_ops;
ddf9f970 945struct netlink_ext_ack;
74515c57 946struct xdp_umem;
75ccae62 947struct xdp_dev_bulk_queue;
aa8d3a71 948struct bpf_xdp_link;
ddf9f970 949
7f0a8382
AN
950enum bpf_xdp_mode {
951 XDP_MODE_SKB = 0,
952 XDP_MODE_DRV = 1,
953 XDP_MODE_HW = 2,
954 __MAX_XDP_MODE
955};
956
957struct bpf_xdp_entity {
958 struct bpf_prog *prog;
aa8d3a71 959 struct bpf_xdp_link *link;
7f0a8382 960};
ddf9f970 961
f4e63525
JK
962struct netdev_bpf {
963 enum bpf_netdev_command command;
a7862b45
BB
964 union {
965 /* XDP_SETUP_PROG */
ddf9f970 966 struct {
32d60277 967 u32 flags;
ddf9f970
JK
968 struct bpf_prog *prog;
969 struct netlink_ext_ack *extack;
970 };
a3884572
JK
971 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
972 struct {
973 struct bpf_offloaded_map *offmap;
974 };
1742b3d5 975 /* XDP_SETUP_XSK_POOL */
74515c57 976 struct {
1742b3d5 977 struct xsk_buff_pool *pool;
f8ebfaf6 978 u16 queue_id;
74515c57 979 } xsk;
a7862b45
BB
980 };
981};
16e5cc64 982
9116e5e2
MK
983/* Flags for ndo_xsk_wakeup. */
984#define XDP_WAKEUP_RX (1 << 0)
985#define XDP_WAKEUP_TX (1 << 1)
986
d77e38e6
SK
987#ifdef CONFIG_XFRM_OFFLOAD
988struct xfrmdev_ops {
989 int (*xdo_dev_state_add) (struct xfrm_state *x);
990 void (*xdo_dev_state_delete) (struct xfrm_state *x);
991 void (*xdo_dev_state_free) (struct xfrm_state *x);
992 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
993 struct xfrm_state *x);
50bd870a 994 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
995};
996#endif
997
6c557001
FW
998struct dev_ifalias {
999 struct rcu_head rcuhead;
1000 char ifalias[];
1001};
1002
b473b0d2 1003struct devlink;
da68b4ad 1004struct tlsdev_ops;
b473b0d2 1005
ff927412
JP
1006struct netdev_name_node {
1007 struct hlist_node hlist;
36fbf1e5 1008 struct list_head list;
ff927412
JP
1009 struct net_device *dev;
1010 const char *name;
1011};
1012
36fbf1e5
JP
1013int netdev_name_node_alt_create(struct net_device *dev, const char *name);
1014int netdev_name_node_alt_destroy(struct net_device *dev, const char *name);
1015
93642e14
JP
1016struct netdev_net_notifier {
1017 struct list_head list;
1018 struct notifier_block *nb;
1019};
1020
d314774c
SH
1021/*
1022 * This structure defines the management hooks for network devices.
00829823
SH
1023 * The following hooks can be defined; unless noted otherwise, they are
1024 * optional and can be filled with a null pointer.
d314774c
SH
1025 *
1026 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
1027 * This function is called once when a network device is registered.
1028 * The network device can use this for any late stage initialization
1029 * or semantic validation. It can fail with an error code which will
1030 * be propagated back to register_netdev.
d314774c
SH
1031 *
1032 * void (*ndo_uninit)(struct net_device *dev);
1033 * This function is called when device is unregistered or when registration
1034 * fails. It is not called if init fails.
1035 *
1036 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 1037 * This function is called when a network device transitions to the up
d314774c
SH
1038 * state.
1039 *
1040 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 1041 * This function is called when a network device transitions to the down
d314774c
SH
1042 * state.
1043 *
dc1f8bf6
SH
1044 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1045 * struct net_device *dev);
00829823 1046 * Called when a packet needs to be transmitted.
e79d8429
RR
1047 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
1048 * the queue before that can happen; it's for obsolete devices and weird
1049 * corner cases, but the stack really does a non-trivial amount
1050 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 1051 * Required; cannot be NULL.
00829823 1052 *
1a2a1444
DM
1053 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1054 * struct net_device *dev
1055 * netdev_features_t features);
1056 * Called by core transmit path to determine if device is capable of
1057 * performing offload operations on a given packet. This is to give
1058 * the device an opportunity to implement any restrictions that cannot
1059 * be otherwise expressed by feature flags. The check is called with
1060 * the set of features that the stack has calculated and it returns
1061 * those the driver believes to be appropriate.
cdba756f 1062 *
f663dd9a 1063 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 1064 * struct net_device *sb_dev);
5e82b4b2 1065 * Called to decide which queue to use when device supports multiple
00829823
SH
1066 * transmit queues.
1067 *
d314774c
SH
1068 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1069 * This function is called to allow device receiver to make
5e82b4b2 1070 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
1071 *
1072 * void (*ndo_set_rx_mode)(struct net_device *dev);
1073 * This function is called device changes address list filtering.
01789349 1074 * If driver handles unicast address filtering, it should set
5e82b4b2 1075 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
1076 *
1077 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1078 * This function is called when the Media Access Control address
37b607c5 1079 * needs to be changed. If this interface is not defined, the
5e82b4b2 1080 * MAC address can not be changed.
d314774c
SH
1081 *
1082 * int (*ndo_validate_addr)(struct net_device *dev);
1083 * Test if Media Access Control address is valid for the device.
1084 *
1085 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
1086 * Called when a user requests an ioctl which can't be handled by
1087 * the generic interface code. If not defined ioctls return
d314774c
SH
1088 * not supported error code.
1089 *
1090 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1091 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1092 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1093 * interface (PCI) for low level management.
1094 *
1095 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1096 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1097 * of a device.
d314774c 1098 *
0290bd29 1099 * void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
5e82b4b2 1100 * Callback used when the transmitter has not made any progress
d314774c
SH
1101 * for dev->watchdog ticks.
1102 *
bc1f4470 1103 * void (*ndo_get_stats64)(struct net_device *dev,
1104 * struct rtnl_link_stats64 *storage);
d308e38f 1105 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1106 * Called when a user wants to get the network device usage
be1f3c2c 1107 * statistics. Drivers must do one of the following:
3cfde79c
BH
1108 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1109 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1110 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1111 * (which should normally be dev->stats) and return a pointer to
1112 * it. The structure may be changed asynchronously only if each
1113 * field is written atomically.
1114 * 3. Update dev->stats asynchronously and atomically, and define
1115 * neither operation.
d314774c 1116 *
3df5b3c6 1117 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1118 * Return true if this device supports offload stats of this attr_id.
1119 *
1120 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1121 * void *attr_data)
1122 * Get statistics for offload operations by attr_id. Write it into the
1123 * attr_data pointer.
1124 *
5d632cb7 1125 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1126 * If device supports VLAN filtering this function is called when a
80d5c368 1127 * VLAN id is registered.
d314774c 1128 *
5d632cb7 1129 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1130 * If device supports VLAN filtering this function is called when a
80d5c368 1131 * VLAN id is unregistered.
d314774c
SH
1132 *
1133 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1134 *
1135 * SR-IOV management functions.
1136 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1137 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1138 * u8 qos, __be16 proto);
ed616689
SC
1139 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1140 * int max_tx_rate);
5f8444a3 1141 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1142 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1143 * int (*ndo_get_vf_config)(struct net_device *dev,
1144 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1145 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1146 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1147 * struct nlattr *port[]);
01a3d796
VZ
1148 *
1149 * Enable or disable the VF ability to query its RSS Redirection Table and
1150 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1151 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1152 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1153 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1154 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1155 * void *type_data);
6a4bc2b4
FF
1156 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1157 * This is always called from the stack with the rtnl lock held and netif
1158 * tx queues stopped. This allows the netdevice to perform queue
1159 * management safely.
c445477d 1160 *
e9bce845
YZ
1161 * Fiber Channel over Ethernet (FCoE) offload functions.
1162 * int (*ndo_fcoe_enable)(struct net_device *dev);
1163 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1164 * so the underlying device can perform whatever needed configuration or
1165 * initialization to support acceleration of FCoE traffic.
1166 *
1167 * int (*ndo_fcoe_disable)(struct net_device *dev);
1168 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1169 * so the underlying device can perform whatever needed clean-ups to
1170 * stop supporting acceleration of FCoE traffic.
1171 *
1172 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1173 * struct scatterlist *sgl, unsigned int sgc);
1174 * Called when the FCoE Initiator wants to initialize an I/O that
1175 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1176 * perform necessary setup and returns 1 to indicate the device is set up
1177 * successfully to perform DDP on this I/O, otherwise this returns 0.
1178 *
1179 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1180 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1181 * indicated by the FC exchange id 'xid', so the underlying device can
1182 * clean up and reuse resources for later DDP requests.
1183 *
1184 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1185 * struct scatterlist *sgl, unsigned int sgc);
1186 * Called when the FCoE Target wants to initialize an I/O that
1187 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1188 * perform necessary setup and returns 1 to indicate the device is set up
1189 * successfully to perform DDP on this I/O, otherwise this returns 0.
1190 *
68bad94e
NP
1191 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1192 * struct netdev_fcoe_hbainfo *hbainfo);
1193 * Called when the FCoE Protocol stack wants information on the underlying
1194 * device. This information is utilized by the FCoE protocol stack to
1195 * register attributes with Fiber Channel management service as per the
1196 * FC-GS Fabric Device Management Information(FDMI) specification.
1197 *
e9bce845
YZ
1198 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1199 * Called when the underlying device wants to override default World Wide
1200 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1201 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1202 * protocol stack to use.
1203 *
c445477d
BH
1204 * RFS acceleration.
1205 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1206 * u16 rxq_index, u32 flow_id);
1207 * Set hardware filter for RFS. rxq_index is the target queue index;
1208 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1209 * Return the filter ID on success, or a negative error code.
fbaec0ea 1210 *
8b98a70c 1211 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1212 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1213 * Called to make another netdev an underling.
1214 *
1215 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1216 * Called to release previously enslaved netdev.
5455c699 1217 *
cff9f12b
MG
1218 * struct net_device *(*ndo_get_xmit_slave)(struct net_device *dev,
1219 * struct sk_buff *skb,
1220 * bool all_slaves);
1221 * Get the xmit slave of master device. If all_slaves is true, function
1222 * assume all the slaves can transmit.
1223 *
5455c699 1224 * Feature/offload setting functions.
1a2a1444
DM
1225 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1226 * netdev_features_t features);
1227 * Adjusts the requested feature flags according to device-specific
1228 * constraints, and returns the resulting flags. Must not modify
1229 * the device state.
1230 *
c8f44aff 1231 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1232 * Called to update device configuration to new features. Passed
1233 * feature set might be less than what was returned by ndo_fix_features()).
1234 * Must return >0 or -errno if it changed dev->features itself.
1235 *
edc7d573 1236 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1237 * struct net_device *dev,
87b0984e
PM
1238 * const unsigned char *addr, u16 vid, u16 flags,
1239 * struct netlink_ext_ack *extack);
77162022 1240 * Adds an FDB entry to dev for addr.
1690be63
VY
1241 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1242 * struct net_device *dev,
f6f6424b 1243 * const unsigned char *addr, u16 vid)
77162022
JF
1244 * Deletes the FDB entry from dev coresponding to addr.
1245 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1246 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1247 * int *idx)
77162022
JF
1248 * Used to add FDB entries to dump requests. Implementers should add
1249 * entries to skb and update idx with the number of entries.
e5a55a89 1250 *
ad41faa8 1251 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1252 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1253 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1254 * struct net_device *dev, u32 filter_mask,
1255 * int nlflags)
ad41faa8
ND
1256 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1257 * u16 flags);
4bf84c35
JP
1258 *
1259 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1260 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1261 * which do not represent real hardware may define this to allow their
1262 * userspace components to manage their virtual carrier state. Devices
1263 * that determine carrier state from physical hardware properties (eg
1264 * network cables) or protocol-dependent mechanisms (eg
1265 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1266 *
1267 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1268 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1269 * Called to get ID of physical port of this device. If driver does
1270 * not implement this, it is assumed that the hw is not able to have
1271 * multiple net devices on single physical port.
53cf5275 1272 *
d6abc596
FF
1273 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1274 * struct netdev_phys_item_id *ppid)
1275 * Called to get the parent ID of the physical port of this device.
1276 *
a6cc0cfa
JF
1277 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1278 * struct net_device *dev)
1279 * Called by upper layer devices to accelerate switching or other
1280 * station functionality into hardware. 'pdev is the lowerdev
1281 * to use for the offload and 'dev' is the net device that will
1282 * back the offload. Returns a pointer to the private structure
1283 * the upper layer will maintain.
1284 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1285 * Called by upper layer device to delete the station created
1286 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1287 * the station and priv is the structure returned by the add
1288 * operation.
822b3b2e
JF
1289 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1290 * int queue_index, u32 maxrate);
1291 * Called when a user wants to set a max-rate limitation of specific
1292 * TX queue.
a54acb3a
ND
1293 * int (*ndo_get_iflink)(const struct net_device *dev);
1294 * Called to get the iflink value of this device.
d746d707 1295 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1296 * bool proto_down);
d746d707
AK
1297 * This function is used to pass protocol port error state information
1298 * to the switch driver. The switch driver can react to the proto_down
1299 * by doing a phys down on the associated switch port.
fc4099f1
PS
1300 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1301 * This function is used to get egress tunnel information for given skb.
1302 * This is useful for retrieving outer tunnel header parameters while
1303 * sampling packet.
871b642a
PA
1304 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1305 * This function is used to specify the headroom that the skb must
1306 * consider when allocation skb during packet reception. Setting
1307 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1308 * forward. Setting a negative value resets the rx headroom to the
871b642a 1309 * default value.
f4e63525 1310 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1311 * This function is used to set or query state related to XDP on the
f4e63525
JK
1312 * netdevice and manage BPF offload. See definition of
1313 * enum bpf_netdev_command for details.
42b33468
JDB
1314 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1315 * u32 flags);
735fc405
JDB
1316 * This function is used to submit @n XDP packets for transmit on a
1317 * netdevice. Returns number of frames successfully transmitted, frames
1318 * that got dropped are freed/returned via xdp_return_frame().
1319 * Returns negative number, means general error invoking ndo, meaning
1320 * no frames were xmit'ed and core-caller will free all frames.
9116e5e2
MK
1321 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1322 * This function is used to wake up the softirq, ksoftirqd or kthread
1323 * responsible for sending and/or receiving packets on a specific
1324 * queue id bound to an AF_XDP socket. The flags field specifies if
1325 * only RX, only Tx, or both should be woken up using the flags
1326 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
5dc37bb9
JP
1327 * struct devlink_port *(*ndo_get_devlink_port)(struct net_device *dev);
1328 * Get devlink port instance associated with a given netdev.
b473b0d2
JK
1329 * Called with a reference on the netdevice and devlink locks only,
1330 * rtnl_lock is not held.
607259a6
CH
1331 * int (*ndo_tunnel_ctl)(struct net_device *dev, struct ip_tunnel_parm *p,
1332 * int cmd);
1333 * Add, change, delete or get information on an IPv4 tunnel.
9aa1206e
DB
1334 * struct net_device *(*ndo_get_peer_dev)(struct net_device *dev);
1335 * If a device is paired with a peer device, return the peer instance.
1336 * The caller must be under RCU read context.
ddb94eaf
PNA
1337 * int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx, struct net_device_path *path);
1338 * Get the forwarding path to reach the real device from the HW destination address
d314774c
SH
1339 */
1340struct net_device_ops {
1341 int (*ndo_init)(struct net_device *dev);
1342 void (*ndo_uninit)(struct net_device *dev);
1343 int (*ndo_open)(struct net_device *dev);
1344 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1345 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1346 struct net_device *dev);
1347 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1348 struct net_device *dev,
1349 netdev_features_t features);
00829823 1350 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1351 struct sk_buff *skb,
a350ecce 1352 struct net_device *sb_dev);
d314774c
SH
1353 void (*ndo_change_rx_flags)(struct net_device *dev,
1354 int flags);
d314774c 1355 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1356 int (*ndo_set_mac_address)(struct net_device *dev,
1357 void *addr);
d314774c 1358 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1359 int (*ndo_do_ioctl)(struct net_device *dev,
1360 struct ifreq *ifr, int cmd);
d314774c
SH
1361 int (*ndo_set_config)(struct net_device *dev,
1362 struct ifmap *map);
00829823
SH
1363 int (*ndo_change_mtu)(struct net_device *dev,
1364 int new_mtu);
1365 int (*ndo_neigh_setup)(struct net_device *dev,
1366 struct neigh_parms *);
0290bd29
MT
1367 void (*ndo_tx_timeout) (struct net_device *dev,
1368 unsigned int txqueue);
d314774c 1369
bc1f4470 1370 void (*ndo_get_stats64)(struct net_device *dev,
1371 struct rtnl_link_stats64 *storage);
3df5b3c6 1372 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1373 int (*ndo_get_offload_stats)(int attr_id,
1374 const struct net_device *dev,
1375 void *attr_data);
d314774c
SH
1376 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1377
8e586137 1378 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1379 __be16 proto, u16 vid);
8e586137 1380 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1381 __be16 proto, u16 vid);
d314774c 1382#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1383 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1384 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1385 struct netpoll_info *info);
0e34e931 1386 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1387#endif
95c26df8
WM
1388 int (*ndo_set_vf_mac)(struct net_device *dev,
1389 int queue, u8 *mac);
1390 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1391 int queue, u16 vlan,
1392 u8 qos, __be16 proto);
ed616689
SC
1393 int (*ndo_set_vf_rate)(struct net_device *dev,
1394 int vf, int min_tx_rate,
1395 int max_tx_rate);
5f8444a3
GR
1396 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1397 int vf, bool setting);
dd461d6a
HS
1398 int (*ndo_set_vf_trust)(struct net_device *dev,
1399 int vf, bool setting);
95c26df8
WM
1400 int (*ndo_get_vf_config)(struct net_device *dev,
1401 int vf,
1402 struct ifla_vf_info *ivf);
1d8faf48
RE
1403 int (*ndo_set_vf_link_state)(struct net_device *dev,
1404 int vf, int link_state);
3b766cd8
EBE
1405 int (*ndo_get_vf_stats)(struct net_device *dev,
1406 int vf,
1407 struct ifla_vf_stats
1408 *vf_stats);
57b61080
SF
1409 int (*ndo_set_vf_port)(struct net_device *dev,
1410 int vf,
1411 struct nlattr *port[]);
1412 int (*ndo_get_vf_port)(struct net_device *dev,
1413 int vf, struct sk_buff *skb);
30aad417
DG
1414 int (*ndo_get_vf_guid)(struct net_device *dev,
1415 int vf,
1416 struct ifla_vf_guid *node_guid,
1417 struct ifla_vf_guid *port_guid);
cc8e27cc
EC
1418 int (*ndo_set_vf_guid)(struct net_device *dev,
1419 int vf, u64 guid,
1420 int guid_type);
01a3d796
VZ
1421 int (*ndo_set_vf_rss_query_en)(
1422 struct net_device *dev,
1423 int vf, bool setting);
16e5cc64 1424 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1425 enum tc_setup_type type,
de4784ca 1426 void *type_data);
d11ead75 1427#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1428 int (*ndo_fcoe_enable)(struct net_device *dev);
1429 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1430 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1431 u16 xid,
1432 struct scatterlist *sgl,
1433 unsigned int sgc);
1434 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1435 u16 xid);
6247e086
YZ
1436 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1437 u16 xid,
1438 struct scatterlist *sgl,
1439 unsigned int sgc);
68bad94e
NP
1440 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1441 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1442#endif
1443
d11ead75 1444#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1445#define NETDEV_FCOE_WWNN 0
1446#define NETDEV_FCOE_WWPN 1
1447 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1448 u64 *wwn, int type);
4d288d57 1449#endif
3c9c36bc 1450
c445477d
BH
1451#ifdef CONFIG_RFS_ACCEL
1452 int (*ndo_rx_flow_steer)(struct net_device *dev,
1453 const struct sk_buff *skb,
1454 u16 rxq_index,
1455 u32 flow_id);
1456#endif
fbaec0ea 1457 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1458 struct net_device *slave_dev,
1459 struct netlink_ext_ack *extack);
fbaec0ea
JP
1460 int (*ndo_del_slave)(struct net_device *dev,
1461 struct net_device *slave_dev);
cff9f12b
MG
1462 struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
1463 struct sk_buff *skb,
1464 bool all_slaves);
719a402c
TT
1465 struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
1466 struct sock *sk);
c8f44aff
MM
1467 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1468 netdev_features_t features);
5455c699 1469 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1470 netdev_features_t features);
503eebc2
JP
1471 int (*ndo_neigh_construct)(struct net_device *dev,
1472 struct neighbour *n);
1473 void (*ndo_neigh_destroy)(struct net_device *dev,
1474 struct neighbour *n);
77162022
JF
1475
1476 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1477 struct nlattr *tb[],
77162022 1478 struct net_device *dev,
6b6e2725 1479 const unsigned char *addr,
f6f6424b 1480 u16 vid,
87b0984e
PM
1481 u16 flags,
1482 struct netlink_ext_ack *extack);
77162022 1483 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1484 struct nlattr *tb[],
77162022 1485 struct net_device *dev,
f6f6424b
JP
1486 const unsigned char *addr,
1487 u16 vid);
77162022
JF
1488 int (*ndo_fdb_dump)(struct sk_buff *skb,
1489 struct netlink_callback *cb,
1490 struct net_device *dev,
5d5eacb3 1491 struct net_device *filter_dev,
d297653d 1492 int *idx);
5b2f94b2
RP
1493 int (*ndo_fdb_get)(struct sk_buff *skb,
1494 struct nlattr *tb[],
1495 struct net_device *dev,
1496 const unsigned char *addr,
1497 u16 vid, u32 portid, u32 seq,
1498 struct netlink_ext_ack *extack);
e5a55a89 1499 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1500 struct nlmsghdr *nlh,
2fd527b7
PM
1501 u16 flags,
1502 struct netlink_ext_ack *extack);
e5a55a89
JF
1503 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1504 u32 pid, u32 seq,
6cbdceeb 1505 struct net_device *dev,
46c264da
ND
1506 u32 filter_mask,
1507 int nlflags);
407af329 1508 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1509 struct nlmsghdr *nlh,
1510 u16 flags);
4bf84c35
JP
1511 int (*ndo_change_carrier)(struct net_device *dev,
1512 bool new_carrier);
66b52b0d 1513 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1514 struct netdev_phys_item_id *ppid);
d6abc596
FF
1515 int (*ndo_get_port_parent_id)(struct net_device *dev,
1516 struct netdev_phys_item_id *ppid);
db24a904
DA
1517 int (*ndo_get_phys_port_name)(struct net_device *dev,
1518 char *name, size_t len);
a6cc0cfa
JF
1519 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1520 struct net_device *dev);
1521 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1522 void *priv);
1523
822b3b2e
JF
1524 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1525 int queue_index,
1526 u32 maxrate);
a54acb3a 1527 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1528 int (*ndo_change_proto_down)(struct net_device *dev,
1529 bool proto_down);
fc4099f1
PS
1530 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1531 struct sk_buff *skb);
871b642a
PA
1532 void (*ndo_set_rx_headroom)(struct net_device *dev,
1533 int needed_headroom);
f4e63525
JK
1534 int (*ndo_bpf)(struct net_device *dev,
1535 struct netdev_bpf *bpf);
735fc405 1536 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1537 struct xdp_frame **xdp,
1538 u32 flags);
9116e5e2
MK
1539 int (*ndo_xsk_wakeup)(struct net_device *dev,
1540 u32 queue_id, u32 flags);
5dc37bb9 1541 struct devlink_port * (*ndo_get_devlink_port)(struct net_device *dev);
607259a6
CH
1542 int (*ndo_tunnel_ctl)(struct net_device *dev,
1543 struct ip_tunnel_parm *p, int cmd);
9aa1206e 1544 struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
ddb94eaf
PNA
1545 int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
1546 struct net_device_path *path);
d314774c
SH
1547};
1548
7aa98047 1549/**
270f3385 1550 * enum netdev_priv_flags - &struct net_device priv_flags
7aa98047
LR
1551 *
1552 * These are the &struct net_device, they are only set internally
1553 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1554 * userspace; this means that the order of these flags can change
7aa98047
LR
1555 * during any kernel release.
1556 *
1557 * You should have a pretty good reason to be extending these flags.
1558 *
1559 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1560 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1561 * @IFF_BONDING: bonding master or slave
7aa98047 1562 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1563 * @IFF_WAN_HDLC: WAN HDLC device
1564 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1565 * release skb->dst
1566 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1567 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1568 * @IFF_MACVLAN_PORT: device used as macvlan port
1569 * @IFF_BRIDGE_PORT: device used as bridge port
1570 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1571 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1572 * @IFF_UNICAST_FLT: Supports unicast filtering
1573 * @IFF_TEAM_PORT: device used as team port
1574 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1575 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1576 * change when it's running
1577 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1578 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1579 * underlying stacked devices
007979ea 1580 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1581 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1582 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1583 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1584 * @IFF_TEAM: device is a team device
d4ab4286 1585 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1586 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1587 * entity (i.e. the master device for bridged veth)
3c175784 1588 * @IFF_MACSEC: device is a MACsec device
f5426250 1589 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1590 * @IFF_FAILOVER: device is a failover master device
1591 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1592 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8065a779 1593 * @IFF_LIVE_RENAME_OK: rename is allowed while device is up and running
c2ff53d8
XZ
1594 * @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1595 * skb_headlen(skb) == 0 (data starts from frag0)
7aa98047
LR
1596 */
1597enum netdev_priv_flags {
1598 IFF_802_1Q_VLAN = 1<<0,
1599 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1600 IFF_BONDING = 1<<2,
1601 IFF_ISATAP = 1<<3,
1602 IFF_WAN_HDLC = 1<<4,
1603 IFF_XMIT_DST_RELEASE = 1<<5,
1604 IFF_DONT_BRIDGE = 1<<6,
1605 IFF_DISABLE_NETPOLL = 1<<7,
1606 IFF_MACVLAN_PORT = 1<<8,
1607 IFF_BRIDGE_PORT = 1<<9,
1608 IFF_OVS_DATAPATH = 1<<10,
1609 IFF_TX_SKB_SHARING = 1<<11,
1610 IFF_UNICAST_FLT = 1<<12,
1611 IFF_TEAM_PORT = 1<<13,
1612 IFF_SUPP_NOFCS = 1<<14,
1613 IFF_LIVE_ADDR_CHANGE = 1<<15,
1614 IFF_MACVLAN = 1<<16,
1615 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1616 IFF_L3MDEV_MASTER = 1<<18,
1617 IFF_NO_QUEUE = 1<<19,
1618 IFF_OPENVSWITCH = 1<<20,
1619 IFF_L3MDEV_SLAVE = 1<<21,
1620 IFF_TEAM = 1<<22,
1621 IFF_RXFH_CONFIGURED = 1<<23,
1622 IFF_PHONY_HEADROOM = 1<<24,
1623 IFF_MACSEC = 1<<25,
f5426250 1624 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1625 IFF_FAILOVER = 1<<27,
1626 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1627 IFF_L3MDEV_RX_HANDLER = 1<<29,
8065a779 1628 IFF_LIVE_RENAME_OK = 1<<30,
c2ff53d8 1629 IFF_TX_SKB_NO_LINEAR = 1<<31,
7aa98047
LR
1630};
1631
1632#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1633#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1634#define IFF_BONDING IFF_BONDING
7aa98047 1635#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1636#define IFF_WAN_HDLC IFF_WAN_HDLC
1637#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1638#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1639#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1640#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1641#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1642#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1643#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1644#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1645#define IFF_TEAM_PORT IFF_TEAM_PORT
1646#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1647#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1648#define IFF_MACVLAN IFF_MACVLAN
02875878 1649#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1650#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1651#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1652#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1653#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1654#define IFF_TEAM IFF_TEAM
d4ab4286 1655#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
2463e073 1656#define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
3c175784 1657#define IFF_MACSEC IFF_MACSEC
f5426250 1658#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1659#define IFF_FAILOVER IFF_FAILOVER
1660#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
d5256083 1661#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
8065a779 1662#define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
c2ff53d8 1663#define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
7aa98047 1664
4e096a18
OR
1665/* Specifies the type of the struct net_device::ml_priv pointer */
1666enum netdev_ml_priv_type {
1667 ML_PRIV_NONE,
1668 ML_PRIV_CAN,
1669};
1670
536721b1
KK
1671/**
1672 * struct net_device - The DEVICE structure.
d651983d
MCC
1673 *
1674 * Actually, this whole structure is a big mistake. It mixes I/O
1675 * data with strictly "high-level" data, and it has to know about
1676 * almost every data structure used in the INET module.
536721b1
KK
1677 *
1678 * @name: This is the first field of the "visible" part of this structure
1679 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1680 * of the interface.
536721b1 1681 *
ff927412 1682 * @name_node: Name hashlist node
536721b1
KK
1683 * @ifalias: SNMP alias
1684 * @mem_end: Shared memory end
1685 * @mem_start: Shared memory start
1686 * @base_addr: Device I/O address
1687 * @irq: Device IRQ number
1688 *
1689 * @state: Generic network queuing layer state, see netdev_state_t
1690 * @dev_list: The global list of network devices
5e82b4b2
BH
1691 * @napi_list: List entry used for polling NAPI devices
1692 * @unreg_list: List entry when we are unregistering the
1693 * device; see the function unregister_netdev
1694 * @close_list: List entry used when we are closing the device
62d885fe
BP
1695 * @ptype_all: Device-specific packet handlers for all protocols
1696 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1697 *
1698 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1699 * @features: Currently active device features
1700 * @hw_features: User-changeable features
1701 *
1702 * @wanted_features: User-requested features
1703 * @vlan_features: Mask of features inheritable by VLAN devices
1704 *
1705 * @hw_enc_features: Mask of features inherited by encapsulating devices
1706 * This field indicates what encapsulation
1707 * offloads the hardware is capable of doing,
1708 * and drivers will need to set them appropriately.
1709 *
1710 * @mpls_features: Mask of features inheritable by MPLS
a1fa83bd 1711 * @gso_partial_features: value(s) from NETIF_F_GSO\*
536721b1
KK
1712 *
1713 * @ifindex: interface index
5e82b4b2 1714 * @group: The group the device belongs to
536721b1
KK
1715 *
1716 * @stats: Statistics struct, which was left as a legacy, use
1717 * rtnl_link_stats64 instead
1718 *
1719 * @rx_dropped: Dropped packets by core network,
1720 * do not use this in drivers
1721 * @tx_dropped: Dropped packets by core network,
1722 * do not use this in drivers
6e7333d3
JW
1723 * @rx_nohandler: nohandler dropped packets by core network on
1724 * inactive devices, do not use this in drivers
9e55e5d3
FF
1725 * @carrier_up_count: Number of times the carrier has been up
1726 * @carrier_down_count: Number of times the carrier has been down
536721b1 1727 *
536721b1
KK
1728 * @wireless_handlers: List of functions to handle Wireless Extensions,
1729 * instead of ioctl,
1730 * see <net/iw_handler.h> for details.
1731 * @wireless_data: Instance data managed by the core of wireless extensions
1732 *
1733 * @netdev_ops: Includes several pointers to callbacks,
1734 * if one wants to override the ndo_*() functions
1735 * @ethtool_ops: Management operations
a1fa83bd 1736 * @l3mdev_ops: Layer 3 master device operations
f997c55c
AA
1737 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1738 * discovery handling. Necessary for e.g. 6LoWPAN.
a1fa83bd
RD
1739 * @xfrmdev_ops: Transformation offload operations
1740 * @tlsdev_ops: Transport Layer Security offload operations
d476059e 1741 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1742 * of Layer 2 headers.
1743 *
1744 * @flags: Interface flags (a la BSD)
1745 * @priv_flags: Like 'flags' but invisible to userspace,
1746 * see if.h for the definitions
1747 * @gflags: Global flags ( kept as legacy )
1748 * @padded: How much padding added by alloc_netdev()
1749 * @operstate: RFC2863 operstate
1750 * @link_mode: Mapping policy to operstate
1751 * @if_port: Selectable AUI, TP, ...
1752 * @dma: DMA channel
1753 * @mtu: Interface MTU value
61e84623
JW
1754 * @min_mtu: Interface Minimum MTU value
1755 * @max_mtu: Interface Maximum MTU value
536721b1 1756 * @type: Interface hardware type
2793a23a 1757 * @hard_header_len: Maximum hardware header length.
217e6fa2 1758 * @min_header_len: Minimum hardware header length
536721b1
KK
1759 *
1760 * @needed_headroom: Extra headroom the hardware may need, but not in all
1761 * cases can this be guaranteed
1762 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1763 * cases can this be guaranteed. Some cases also use
1764 * LL_MAX_HEADER instead to allocate the skb
1765 *
1766 * interface address info:
1767 *
1768 * @perm_addr: Permanent hw address
1769 * @addr_assign_type: Hw address assignment type
1770 * @addr_len: Hardware address length
5343da4c
TY
1771 * @upper_level: Maximum depth level of upper devices.
1772 * @lower_level: Maximum depth level of lower devices.
8626a0c8 1773 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1774 * @dev_id: Used to differentiate devices that share
1775 * the same link layer address
1776 * @dev_port: Used to differentiate devices that share
1777 * the same function
1778 * @addr_list_lock: XXX: need comments on this one
a1fa83bd 1779 * @name_assign_type: network interface name assignment type
5e82b4b2 1780 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1781 * has been enabled due to the need to listen to
1782 * additional unicast addresses in a device that
1783 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1784 * @uc: unicast mac addresses
1785 * @mc: multicast mac addresses
1786 * @dev_addrs: list of device hw addresses
1787 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1788 * @promiscuity: Number of times the NIC is told to work in
1789 * promiscuous mode; if it becomes 0 the NIC will
1790 * exit promiscuous mode
536721b1
KK
1791 * @allmulti: Counter, enables or disables allmulticast mode
1792 *
1793 * @vlan_info: VLAN info
1794 * @dsa_ptr: dsa specific data
1795 * @tipc_ptr: TIPC specific data
1796 * @atalk_ptr: AppleTalk link
1797 * @ip_ptr: IPv4 specific data
1798 * @dn_ptr: DECnet specific data
1799 * @ip6_ptr: IPv6 specific data
1800 * @ax25_ptr: AX.25 specific data
1801 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
a1fa83bd
RD
1802 * @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
1803 * device struct
1804 * @mpls_ptr: mpls_dev struct pointer
536721b1 1805 *
536721b1
KK
1806 * @dev_addr: Hw address (before bcast,
1807 * because most packets are unicast)
1808 *
1809 * @_rx: Array of RX queues
1810 * @num_rx_queues: Number of RX queues
1811 * allocated at register_netdev() time
1812 * @real_num_rx_queues: Number of RX queues currently active in device
a1fa83bd
RD
1813 * @xdp_prog: XDP sockets filter program pointer
1814 * @gro_flush_timeout: timeout for GRO layer in NAPI
5c45a918
MCC
1815 * @napi_defer_hard_irqs: If not zero, provides a counter that would
1816 * allow to avoid NIC hard IRQ, on busy queues.
536721b1
KK
1817 *
1818 * @rx_handler: handler for received packets
1819 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1820 * @miniq_ingress: ingress/clsact qdisc specific data for
1821 * ingress processing
536721b1 1822 * @ingress_queue: XXX: need comments on this one
2f5e70c8 1823 * @nf_hooks_ingress: netfilter hooks executed for ingress packets
536721b1
KK
1824 * @broadcast: hw bcast address
1825 *
14ffbbb8
TG
1826 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1827 * indexed by RX queue number. Assigned by driver.
1828 * This must only be set if the ndo_rx_flow_steer
1829 * operation is defined
1830 * @index_hlist: Device index hash chain
1831 *
536721b1
KK
1832 * @_tx: Array of TX queues
1833 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1834 * @real_num_tx_queues: Number of TX queues currently active in device
1835 * @qdisc: Root qdisc from userspace point of view
1836 * @tx_queue_len: Max frames per queue allowed
1837 * @tx_global_lock: XXX: need comments on this one
a1fa83bd 1838 * @xdp_bulkq: XDP device bulk queue
044ab86d 1839 * @xps_maps: all CPUs/RXQs maps for XPS device
536721b1
KK
1840 *
1841 * @xps_maps: XXX: need comments on this one
46209401
JP
1842 * @miniq_egress: clsact qdisc specific data for
1843 * egress processing
a1fa83bd 1844 * @qdisc_hash: qdisc hash table
536721b1 1845 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1846 * the watchdog (see dev_watchdog())
536721b1
KK
1847 * @watchdog_timer: List of timers
1848 *
eb02d39a 1849 * @proto_down_reason: reason a netdev interface is held down
536721b1 1850 * @pcpu_refcnt: Number of references to this device
add2d736 1851 * @dev_refcnt: Number of references to this device
536721b1 1852 * @todo_list: Delayed register/unregister
536721b1
KK
1853 * @link_watch_list: XXX: need comments on this one
1854 *
1855 * @reg_state: Register/unregister state machine
1856 * @dismantle: Device is going to be freed
1857 * @rtnl_link_state: This enum represents the phases of creating
1858 * a new link
1859 *
cf124db5
DM
1860 * @needs_free_netdev: Should unregister perform free_netdev?
1861 * @priv_destructor: Called from unregister
536721b1
KK
1862 * @npinfo: XXX: need comments on this one
1863 * @nd_net: Network namespace this network device is inside
1864 *
1865 * @ml_priv: Mid-layer private
4e096a18 1866 * @ml_priv_type: Mid-layer private type
536721b1
KK
1867 * @lstats: Loopback statistics
1868 * @tstats: Tunnel statistics
1869 * @dstats: Dummy statistics
1870 * @vstats: Virtual ethernet statistics
1871 *
1872 * @garp_port: GARP
1873 * @mrp_port: MRP
1874 *
1875 * @dev: Class/net/name entry
1876 * @sysfs_groups: Space for optional device, statistics and wireless
1877 * sysfs groups
1878 *
1879 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1880 * @rtnl_link_ops: Rtnl_link_ops
1881 *
1882 * @gso_max_size: Maximum size of generic segmentation offload
1883 * @gso_max_segs: Maximum number of segments that can be passed to the
1884 * NIC for GSO
1885 *
1886 * @dcbnl_ops: Data Center Bridging netlink ops
1887 * @num_tc: Number of traffic classes in the net device
1888 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1889 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1890 *
1891 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1892 *
1893 * @priomap: XXX: need comments on this one
1894 * @phydev: Physical device may attach itself
1895 * for hardware timestamping
e679c9c1 1896 * @sfp_bus: attached &struct sfp_bus structure.
1a33e10e 1897 *
1a33e10e
CW
1898 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
1899 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
536721b1 1900 *
d746d707
AK
1901 * @proto_down: protocol port state information can be sent to the
1902 * switch driver and used to set the phys state of the
1903 * switch port.
1904 *
61941143
HK
1905 * @wol_enabled: Wake-on-LAN is enabled
1906 *
29863d41
WW
1907 * @threaded: napi threaded mode is enabled
1908 *
93642e14
JP
1909 * @net_notifier_list: List of per-net netdev notifier block
1910 * that follow this device when it is moved
1911 * to another network namespace.
1912 *
30e9bb84
AT
1913 * @macsec_ops: MACsec offloading ops
1914 *
cc4e3835
JK
1915 * @udp_tunnel_nic_info: static structure describing the UDP tunnel
1916 * offload capabilities of the device
1917 * @udp_tunnel_nic: UDP tunnel offload state
ffa59b0b 1918 * @xdp_state: stores info on attached XDP BPF programs
cc4e3835 1919 *
a93bdcb9
MCC
1920 * @nested_level: Used as as a parameter of spin_lock_nested() of
1921 * dev->addr_list_lock.
1922 * @unlink_list: As netif_addr_lock() can be called recursively,
1923 * keep a list of interfaces to be deleted.
1924 *
1da177e4
LT
1925 * FIXME: cleanup struct net_device such that network protocol info
1926 * moves out.
1927 */
1928
d94d9fee 1929struct net_device {
1da177e4 1930 char name[IFNAMSIZ];
ff927412 1931 struct netdev_name_node *name_node;
6c557001 1932 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1933 /*
1934 * I/O specific fields
1935 * FIXME: Merge these and struct ifmap into one
1936 */
536721b1
KK
1937 unsigned long mem_end;
1938 unsigned long mem_start;
1939 unsigned long base_addr;
1da177e4
LT
1940
1941 /*
536721b1
KK
1942 * Some hardware also needs these fields (state,dev_list,
1943 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1944 * part of the usual set specified in Space.c.
1945 */
1946
1da177e4
LT
1947 unsigned long state;
1948
7562f876 1949 struct list_head dev_list;
bea3348e 1950 struct list_head napi_list;
44a0873d 1951 struct list_head unreg_list;
5cde2829 1952 struct list_head close_list;
7866a621
SN
1953 struct list_head ptype_all;
1954 struct list_head ptype_specific;
2f268f12 1955
2f268f12
VF
1956 struct {
1957 struct list_head upper;
1958 struct list_head lower;
1959 } adj_list;
1960
28af22c6
JDB
1961 /* Read-mostly cache-line for fast-path access */
1962 unsigned int flags;
1963 unsigned int priv_flags;
1964 const struct net_device_ops *netdev_ops;
1965 int ifindex;
1966 unsigned short gflags;
1967 unsigned short hard_header_len;
1968
1969 /* Note : dev->mtu is often read without holding a lock.
1970 * Writers usually hold RTNL.
1971 * It is recommended to use READ_ONCE() to annotate the reads,
1972 * and to use WRITE_ONCE() to annotate the writes.
1973 */
1974 unsigned int mtu;
1975 unsigned short needed_headroom;
1976 unsigned short needed_tailroom;
1977
c8f44aff 1978 netdev_features_t features;
c8f44aff 1979 netdev_features_t hw_features;
c8f44aff 1980 netdev_features_t wanted_features;
c8f44aff 1981 netdev_features_t vlan_features;
6a674e9c 1982 netdev_features_t hw_enc_features;
0d89d203 1983 netdev_features_t mpls_features;
802ab55a 1984 netdev_features_t gso_partial_features;
04ed3e74 1985
28af22c6
JDB
1986 unsigned int min_mtu;
1987 unsigned int max_mtu;
1988 unsigned short type;
1989 unsigned char min_header_len;
1990 unsigned char name_assign_type;
1991
7a66bbc9 1992 int group;
1da177e4 1993
28af22c6 1994 struct net_device_stats stats; /* not used by modern drivers */
015f0688 1995
015f0688
ED
1996 atomic_long_t rx_dropped;
1997 atomic_long_t tx_dropped;
6e7333d3 1998 atomic_long_t rx_nohandler;
1da177e4 1999
b2d3bcfa
DD
2000 /* Stats to monitor link on/off, flapping */
2001 atomic_t carrier_up_count;
2002 atomic_t carrier_down_count;
2003
b86e0280 2004#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
2005 const struct iw_handler_def *wireless_handlers;
2006 struct iw_public_data *wireless_data;
b86e0280 2007#endif
76fd8593 2008 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
2009#ifdef CONFIG_NET_L3_MASTER_DEV
2010 const struct l3mdev_ops *l3mdev_ops;
2011#endif
f997c55c
AA
2012#if IS_ENABLED(CONFIG_IPV6)
2013 const struct ndisc_ops *ndisc_ops;
2014#endif
1da177e4 2015
9cb0d21d 2016#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
2017 const struct xfrmdev_ops *xfrmdev_ops;
2018#endif
2019
a5c37c63
IL
2020#if IS_ENABLED(CONFIG_TLS_DEVICE)
2021 const struct tlsdev_ops *tlsdev_ops;
2022#endif
2023
3b04ddde
SH
2024 const struct header_ops *header_ops;
2025
536721b1
KK
2026 unsigned char operstate;
2027 unsigned char link_mode;
b00055aa 2028
536721b1
KK
2029 unsigned char if_port;
2030 unsigned char dma;
bdc220da 2031
1da177e4 2032 /* Interface address info. */
536721b1
KK
2033 unsigned char perm_addr[MAX_ADDR_LEN];
2034 unsigned char addr_assign_type;
2035 unsigned char addr_len;
5343da4c
TY
2036 unsigned char upper_level;
2037 unsigned char lower_level;
1fc70edb 2038
a0a9663d 2039 unsigned short neigh_priv_len;
536721b1
KK
2040 unsigned short dev_id;
2041 unsigned short dev_port;
28af22c6
JDB
2042 unsigned short padded;
2043
ccffad25 2044 spinlock_t addr_list_lock;
28af22c6 2045 int irq;
1fc70edb 2046
536721b1
KK
2047 struct netdev_hw_addr_list uc;
2048 struct netdev_hw_addr_list mc;
2049 struct netdev_hw_addr_list dev_addrs;
2050
4c3d5e7b
ED
2051#ifdef CONFIG_SYSFS
2052 struct kset *queues_kset;
1fc70edb
TY
2053#endif
2054#ifdef CONFIG_LOCKDEP
2055 struct list_head unlink_list;
4c3d5e7b 2056#endif
9d45abe1
WC
2057 unsigned int promiscuity;
2058 unsigned int allmulti;
1fc70edb
TY
2059 bool uc_promisc;
2060#ifdef CONFIG_LOCKDEP
2061 unsigned char nested_level;
2062#endif
1da177e4 2063
1da177e4 2064
5e82b4b2 2065 /* Protocol-specific pointers */
65ac6a5f 2066
d11ead75 2067#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 2068 struct vlan_info __rcu *vlan_info;
65ac6a5f 2069#endif
34a430d7 2070#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 2071 struct dsa_port *dsa_ptr;
37cb0620
YX
2072#endif
2073#if IS_ENABLED(CONFIG_TIPC)
536721b1 2074 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 2075#endif
89e58148 2076#if IS_ENABLED(CONFIG_IRDA) || IS_ENABLED(CONFIG_ATALK)
536721b1 2077 void *atalk_ptr;
89e58148 2078#endif
536721b1 2079 struct in_device __rcu *ip_ptr;
330c7272 2080#if IS_ENABLED(CONFIG_DECNET)
536721b1 2081 struct dn_dev __rcu *dn_ptr;
330c7272 2082#endif
536721b1 2083 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 2084#if IS_ENABLED(CONFIG_AX25)
536721b1 2085 void *ax25_ptr;
19ff13f2 2086#endif
536721b1 2087 struct wireless_dev *ieee80211_ptr;
98a18b6f 2088 struct wpan_dev *ieee802154_ptr;
03c57747
RS
2089#if IS_ENABLED(CONFIG_MPLS_ROUTING)
2090 struct mpls_dev __rcu *mpls_ptr;
2091#endif
1da177e4 2092
9356b8fc 2093/*
cd13539b 2094 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 2095 */
9356b8fc 2096 /* Interface address info used in eth_type_trans() */
536721b1 2097 unsigned char *dev_addr;
f001fde5 2098
0a9627f2 2099 struct netdev_rx_queue *_rx;
0a9627f2 2100 unsigned int num_rx_queues;
62fe0b40 2101 unsigned int real_num_rx_queues;
0a9627f2 2102
7acedaf5 2103 struct bpf_prog __rcu *xdp_prog;
3b47d303 2104 unsigned long gro_flush_timeout;
6f8b12d6 2105 int napi_defer_hard_irqs;
61391cde 2106 rx_handler_func_t __rcu *rx_handler;
2107 void __rcu *rx_handler_data;
e8a0464c 2108
4cda01e8 2109#ifdef CONFIG_NET_CLS_ACT
46209401 2110 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 2111#endif
24824a09 2112 struct netdev_queue __rcu *ingress_queue;
e687ad60 2113#ifdef CONFIG_NETFILTER_INGRESS
960632ec 2114 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 2115#endif
d2788d34 2116
536721b1 2117 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
2118#ifdef CONFIG_RFS_ACCEL
2119 struct cpu_rmap *rx_cpu_rmap;
2120#endif
2121 struct hlist_node index_hlist;
cd13539b
ED
2122
2123/*
2124 * Cache lines mostly used on transmit path
2125 */
e8a0464c
DM
2126 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
2127 unsigned int num_tx_queues;
fd2ea0a7 2128 unsigned int real_num_tx_queues;
af356afa 2129 struct Qdisc *qdisc;
0cd29503 2130 unsigned int tx_queue_len;
c3f26a26 2131 spinlock_t tx_global_lock;
75ccae62
THJ
2132
2133 struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
cd13539b 2134
bf264145 2135#ifdef CONFIG_XPS
044ab86d 2136 struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
bf264145 2137#endif
1f211a1b 2138#ifdef CONFIG_NET_CLS_ACT
46209401 2139 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 2140#endif
0c4f691f 2141
75ccae62
THJ
2142#ifdef CONFIG_NET_SCHED
2143 DECLARE_HASHTABLE (qdisc_hash, 4);
2144#endif
9356b8fc 2145 /* These may be needed for future network-power-down code. */
9356b8fc 2146 struct timer_list watchdog_timer;
75ccae62 2147 int watchdog_timeo;
9356b8fc 2148
829eb208
RP
2149 u32 proto_down_reason;
2150
1da177e4 2151 struct list_head todo_list;
919067cc
ED
2152
2153#ifdef CONFIG_PCPU_DEV_REFCNT
75ccae62 2154 int __percpu *pcpu_refcnt;
919067cc
ED
2155#else
2156 refcount_t dev_refcnt;
2157#endif
1da177e4 2158
e014debe 2159 struct list_head link_watch_list;
572a103d 2160
1da177e4 2161 enum { NETREG_UNINITIALIZED=0,
b17a7c17 2162 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
2163 NETREG_UNREGISTERING, /* called unregister_netdevice */
2164 NETREG_UNREGISTERED, /* completed unregister todo */
2165 NETREG_RELEASED, /* called free_netdev */
937f1ba5 2166 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
2167 } reg_state:8;
2168
536721b1 2169 bool dismantle;
a2835763
PM
2170
2171 enum {
2172 RTNL_LINK_INITIALIZED,
2173 RTNL_LINK_INITIALIZING,
2174 } rtnl_link_state:16;
1da177e4 2175
cf124db5
DM
2176 bool needs_free_netdev;
2177 void (*priv_destructor)(struct net_device *dev);
1da177e4 2178
1da177e4 2179#ifdef CONFIG_NETPOLL
5fbee843 2180 struct netpoll_info __rcu *npinfo;
1da177e4 2181#endif
eae792b7 2182
0c5c9fb5 2183 possible_net_t nd_net;
4a1c5371 2184
4951704b 2185 /* mid-layer private */
4e096a18
OR
2186 void *ml_priv;
2187 enum netdev_ml_priv_type ml_priv_type;
2188
a7855c78 2189 union {
536721b1 2190 struct pcpu_lstats __percpu *lstats;
8f84985f 2191 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2192 struct pcpu_dstats __percpu *dstats;
a7855c78 2193 };
536721b1 2194
fb585b44 2195#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2196 struct garp_port __rcu *garp_port;
fb585b44
TK
2197#endif
2198#if IS_ENABLED(CONFIG_MRP)
febf018d 2199 struct mrp_port __rcu *mrp_port;
fb585b44 2200#endif
1da177e4 2201
5e82b4b2 2202 struct device dev;
0c509a6c 2203 const struct attribute_group *sysfs_groups[4];
a953be53 2204 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2205
38f7b870 2206 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2207
82cc1a7a
PWJ
2208 /* for setting kernel sock attribute on TCP connection setup */
2209#define GSO_MAX_SIZE 65536
2210 unsigned int gso_max_size;
30b678d8
BH
2211#define GSO_MAX_SEGS 65535
2212 u16 gso_max_segs;
743b03a8 2213
7a6b6f51 2214#ifdef CONFIG_DCB
32953543 2215 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2216#endif
ffcfe25b 2217 s16 num_tc;
5e82b4b2
BH
2218 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2219 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2220
d11ead75 2221#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2222 unsigned int fcoe_ddp_xid;
5bc1421e 2223#endif
86f8515f 2224#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2225 struct netprio_map __rcu *priomap;
4d288d57 2226#endif
5e82b4b2 2227 struct phy_device *phydev;
e679c9c1 2228 struct sfp_bus *sfp_bus;
1a33e10e
CW
2229 struct lock_class_key *qdisc_tx_busylock;
2230 struct lock_class_key *qdisc_running_key;
5e82b4b2 2231 bool proto_down;
61941143 2232 unsigned wol_enabled:1;
29863d41 2233 unsigned threaded:1;
93642e14
JP
2234
2235 struct list_head net_notifier_list;
30e9bb84
AT
2236
2237#if IS_ENABLED(CONFIG_MACSEC)
2238 /* MACsec management functions */
2239 const struct macsec_ops *macsec_ops;
2240#endif
cc4e3835
JK
2241 const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
2242 struct udp_tunnel_nic *udp_tunnel_nic;
7f0a8382
AN
2243
2244 /* protected by rtnl_lock */
2245 struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
1da177e4 2246};
43cb76d9 2247#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2248
b5cdae32
DM
2249static inline bool netif_elide_gro(const struct net_device *dev)
2250{
2251 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2252 return true;
2253 return false;
2254}
2255
1da177e4 2256#define NETDEV_ALIGN 32
1da177e4 2257
4f57c087
JF
2258static inline
2259int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2260{
2261 return dev->prio_tc_map[prio & TC_BITMASK];
2262}
2263
2264static inline
2265int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2266{
2267 if (tc >= dev->num_tc)
2268 return -EINVAL;
2269
2270 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2271 return 0;
2272}
2273
8d059b0f 2274int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2275void netdev_reset_tc(struct net_device *dev);
2276int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2277int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2278
2279static inline
2280int netdev_get_num_tc(struct net_device *dev)
2281{
2282 return dev->num_tc;
2283}
2284
f468f21b
TT
2285static inline void net_prefetch(void *p)
2286{
2287 prefetch(p);
2288#if L1_CACHE_BYTES < 128
2289 prefetch((u8 *)p + L1_CACHE_BYTES);
2290#endif
2291}
2292
2293static inline void net_prefetchw(void *p)
2294{
2295 prefetchw(p);
2296#if L1_CACHE_BYTES < 128
2297 prefetchw((u8 *)p + L1_CACHE_BYTES);
2298#endif
2299}
2300
ffcfe25b
AD
2301void netdev_unbind_sb_channel(struct net_device *dev,
2302 struct net_device *sb_dev);
2303int netdev_bind_sb_channel_queue(struct net_device *dev,
2304 struct net_device *sb_dev,
2305 u8 tc, u16 count, u16 offset);
2306int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2307static inline int netdev_get_sb_channel(struct net_device *dev)
2308{
2309 return max_t(int, -dev->num_tc, 0);
2310}
2311
e8a0464c
DM
2312static inline
2313struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2314 unsigned int index)
2315{
2316 return &dev->_tx[index];
2317}
2318
10c51b56
DB
2319static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2320 const struct sk_buff *skb)
2321{
2322 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2323}
2324
e8a0464c
DM
2325static inline void netdev_for_each_tx_queue(struct net_device *dev,
2326 void (*f)(struct net_device *,
2327 struct netdev_queue *,
2328 void *),
2329 void *arg)
2330{
2331 unsigned int i;
2332
2333 for (i = 0; i < dev->num_tx_queues; i++)
2334 f(dev, &dev->_tx[i], arg);
2335}
2336
1a33e10e
CW
2337#define netdev_lockdep_set_classes(dev) \
2338{ \
2339 static struct lock_class_key qdisc_tx_busylock_key; \
2340 static struct lock_class_key qdisc_running_key; \
2341 static struct lock_class_key qdisc_xmit_lock_key; \
845e0ebb 2342 static struct lock_class_key dev_addr_list_lock_key; \
1a33e10e
CW
2343 unsigned int i; \
2344 \
2345 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
2346 (dev)->qdisc_running_key = &qdisc_running_key; \
845e0ebb
CW
2347 lockdep_set_class(&(dev)->addr_list_lock, \
2348 &dev_addr_list_lock_key); \
1a33e10e
CW
2349 for (i = 0; i < (dev)->num_tx_queues; i++) \
2350 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2351 &qdisc_xmit_lock_key); \
2352}
2353
b71b5837
PA
2354u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2355 struct net_device *sb_dev);
4bd97d51
PA
2356struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2357 struct sk_buff *skb,
2358 struct net_device *sb_dev);
8c4c49df 2359
871b642a
PA
2360/* returns the headroom that the master device needs to take in account
2361 * when forwarding to this dev
2362 */
2363static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2364{
2365 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2366}
2367
2368static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2369{
2370 if (dev->netdev_ops->ndo_set_rx_headroom)
2371 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2372}
2373
2374/* set the device rx headroom to the dev's default */
2375static inline void netdev_reset_rx_headroom(struct net_device *dev)
2376{
2377 netdev_set_rx_headroom(dev, -1);
2378}
2379
4e096a18
OR
2380static inline void *netdev_get_ml_priv(struct net_device *dev,
2381 enum netdev_ml_priv_type type)
2382{
2383 if (dev->ml_priv_type != type)
2384 return NULL;
2385
2386 return dev->ml_priv;
2387}
2388
2389static inline void netdev_set_ml_priv(struct net_device *dev,
2390 void *ml_priv,
2391 enum netdev_ml_priv_type type)
2392{
2393 WARN(dev->ml_priv_type && dev->ml_priv_type != type,
2394 "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
2395 dev->ml_priv_type, type);
2396 WARN(!dev->ml_priv_type && dev->ml_priv,
2397 "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
2398
2399 dev->ml_priv = ml_priv;
2400 dev->ml_priv_type = type;
2401}
2402
c346dca1
YH
2403/*
2404 * Net namespace inlines
2405 */
2406static inline
2407struct net *dev_net(const struct net_device *dev)
2408{
c2d9ba9b 2409 return read_pnet(&dev->nd_net);
c346dca1
YH
2410}
2411
2412static inline
f5aa23fd 2413void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2414{
0c5c9fb5 2415 write_pnet(&dev->nd_net, net);
c346dca1
YH
2416}
2417
bea3348e
SH
2418/**
2419 * netdev_priv - access network device private data
2420 * @dev: network device
2421 *
2422 * Get network device private data
2423 */
6472ce60 2424static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2425{
1ce8e7b5 2426 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2427}
2428
1da177e4
LT
2429/* Set the sysfs physical device reference for the network logical device
2430 * if set prior to registration will cause a symlink during initialization.
2431 */
43cb76d9 2432#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2433
384912ed 2434/* Set the sysfs device type for the network logical device to allow
3f79410c 2435 * fine-grained identification of different network device types. For
5e82b4b2 2436 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2437 */
2438#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2439
82dc3c63
ED
2440/* Default NAPI poll() weight
2441 * Device drivers are strongly advised to not use bigger value
2442 */
2443#define NAPI_POLL_WEIGHT 64
2444
3b582cc1 2445/**
5e82b4b2 2446 * netif_napi_add - initialize a NAPI context
3b582cc1 2447 * @dev: network device
5e82b4b2 2448 * @napi: NAPI context
3b582cc1
SH
2449 * @poll: polling function
2450 * @weight: default weight
2451 *
5e82b4b2
BH
2452 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2453 * *any* of the other NAPI-related functions.
3b582cc1 2454 */
d565b0a1
HX
2455void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2456 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2457
d64b5e85 2458/**
5e82b4b2 2459 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2460 * @dev: network device
5e82b4b2 2461 * @napi: NAPI context
d64b5e85
ED
2462 * @poll: polling function
2463 * @weight: default weight
2464 *
2465 * This variant of netif_napi_add() should be used from drivers using NAPI
2466 * to exclusively poll a TX queue.
2467 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2468 */
2469static inline void netif_tx_napi_add(struct net_device *dev,
2470 struct napi_struct *napi,
2471 int (*poll)(struct napi_struct *, int),
2472 int weight)
2473{
2474 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2475 netif_napi_add(dev, napi, poll, weight);
2476}
2477
5198d545
JK
2478/**
2479 * __netif_napi_del - remove a NAPI context
2480 * @napi: NAPI context
2481 *
2482 * Warning: caller must observe RCU grace period before freeing memory
2483 * containing @napi. Drivers might want to call this helper to combine
2484 * all the needed RCU grace periods into a single one.
2485 */
2486void __netif_napi_del(struct napi_struct *napi);
2487
d8156534 2488/**
5e82b4b2
BH
2489 * netif_napi_del - remove a NAPI context
2490 * @napi: NAPI context
d8156534 2491 *
5e82b4b2 2492 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2493 */
5198d545
JK
2494static inline void netif_napi_del(struct napi_struct *napi)
2495{
2496 __netif_napi_del(napi);
2497 synchronize_net();
2498}
d565b0a1
HX
2499
2500struct napi_gro_cb {
78a478d0 2501 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2502 void *frag0;
78a478d0 2503
7489594c
HX
2504 /* Length of frag0. */
2505 unsigned int frag0_len;
2506
86911732 2507 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2508 int data_offset;
86911732 2509
d565b0a1 2510 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2511 u16 flush;
2512
2513 /* Save the IP ID here and check when we get to the transport layer */
2514 u16 flush_id;
d565b0a1
HX
2515
2516 /* Number of segments aggregated. */
2e71a6f8
ED
2517 u16 count;
2518
15e2396d
TH
2519 /* Start offset for remote checksum offload */
2520 u16 gro_remcsum_start;
2521
2e71a6f8
ED
2522 /* jiffies when first packet was created/queued */
2523 unsigned long age;
86347245 2524
afe93325 2525 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2526 u16 proto;
2527
baa32ff4
TH
2528 /* This is non-zero if the packet may be of the same flow. */
2529 u8 same_flow:1;
2530
fac8e0f5
JG
2531 /* Used in tunnel GRO receive */
2532 u8 encap_mark:1;
573e8fca
TH
2533
2534 /* GRO checksum is valid */
2535 u8 csum_valid:1;
2536
662880f4
TH
2537 /* Number of checksums via CHECKSUM_UNNECESSARY */
2538 u8 csum_cnt:3;
c3c7c254 2539
baa32ff4
TH
2540 /* Free the skb? */
2541 u8 free:2;
2542#define NAPI_GRO_FREE 1
2543#define NAPI_GRO_FREE_STOLEN_HEAD 2
2544
efc98d08
TH
2545 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2546 u8 is_ipv6:1;
2547
a0ca153f
AD
2548 /* Used in GRE, set in fou/gue_gro_receive */
2549 u8 is_fou:1;
2550
1530545e
AD
2551 /* Used to determine if flush_id can be ignored */
2552 u8 is_atomic:1;
2553
fcd91dd4
SD
2554 /* Number of gro_receive callbacks this packet already went through */
2555 u8 recursion_counter:4;
2556
3a1296a3
SK
2557 /* GRO is done by frag_list pointer chaining. */
2558 u8 is_flist:1;
baa32ff4 2559
bf5a755f
JC
2560 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2561 __wsum csum;
2562
c3c7c254
ED
2563 /* used in skb_gro_receive() slow path */
2564 struct sk_buff *last;
d565b0a1
HX
2565};
2566
2567#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2568
fcd91dd4
SD
2569#define GRO_RECURSION_LIMIT 15
2570static inline int gro_recursion_inc_test(struct sk_buff *skb)
2571{
2572 return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
2573}
2574
d4546c25
DM
2575typedef struct sk_buff *(*gro_receive_t)(struct list_head *, struct sk_buff *);
2576static inline struct sk_buff *call_gro_receive(gro_receive_t cb,
2577 struct list_head *head,
2578 struct sk_buff *skb)
fcd91dd4
SD
2579{
2580 if (unlikely(gro_recursion_inc_test(skb))) {
2581 NAPI_GRO_CB(skb)->flush |= 1;
2582 return NULL;
2583 }
2584
2585 return cb(head, skb);
2586}
2587
d4546c25
DM
2588typedef struct sk_buff *(*gro_receive_sk_t)(struct sock *, struct list_head *,
2589 struct sk_buff *);
2590static inline struct sk_buff *call_gro_receive_sk(gro_receive_sk_t cb,
2591 struct sock *sk,
2592 struct list_head *head,
2593 struct sk_buff *skb)
fcd91dd4
SD
2594{
2595 if (unlikely(gro_recursion_inc_test(skb))) {
2596 NAPI_GRO_CB(skb)->flush |= 1;
2597 return NULL;
2598 }
2599
2600 return cb(sk, head, skb);
2601}
2602
1da177e4 2603struct packet_type {
f2ccd8fa 2604 __be16 type; /* This is really htons(ether_type). */
fa788d98 2605 bool ignore_outgoing;
f2ccd8fa
DM
2606 struct net_device *dev; /* NULL is wildcarded here */
2607 int (*func) (struct sk_buff *,
2608 struct net_device *,
2609 struct packet_type *,
2610 struct net_device *);
17266ee9
EC
2611 void (*list_func) (struct list_head *,
2612 struct packet_type *,
2613 struct net_device *);
c0de08d0
EL
2614 bool (*id_match)(struct packet_type *ptype,
2615 struct sock *sk);
1da177e4
LT
2616 void *af_packet_priv;
2617 struct list_head list;
2618};
2619
f191a1d1 2620struct offload_callbacks {
576a30eb 2621 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2622 netdev_features_t features);
d4546c25
DM
2623 struct sk_buff *(*gro_receive)(struct list_head *head,
2624 struct sk_buff *skb);
299603e8 2625 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2626};
2627
2628struct packet_offload {
2629 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2630 u16 priority;
f191a1d1
VY
2631 struct offload_callbacks callbacks;
2632 struct list_head list;
1da177e4
LT
2633};
2634
5e82b4b2 2635/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2636struct pcpu_sw_netstats {
2637 u64 rx_packets;
2638 u64 rx_bytes;
2639 u64 tx_packets;
2640 u64 tx_bytes;
2641 struct u64_stats_sync syncp;
9a5ee462 2642} __aligned(4 * sizeof(u64));
52bb6677
LR
2643
2644struct pcpu_lstats {
fd2f4737
ED
2645 u64_stats_t packets;
2646 u64_stats_t bytes;
52bb6677 2647 struct u64_stats_sync syncp;
9a5ee462 2648} __aligned(2 * sizeof(u64));
8f84985f 2649
de7d5084
ED
2650void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2651
451b05f4
FF
2652static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
2653{
2654 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2655
2656 u64_stats_update_begin(&tstats->syncp);
2657 tstats->rx_bytes += len;
2658 tstats->rx_packets++;
2659 u64_stats_update_end(&tstats->syncp);
2660}
2661
d3fd6548
HK
2662static inline void dev_sw_netstats_tx_add(struct net_device *dev,
2663 unsigned int packets,
2664 unsigned int len)
2665{
2666 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2667
2668 u64_stats_update_begin(&tstats->syncp);
2669 tstats->tx_bytes += len;
2670 tstats->tx_packets += packets;
2671 u64_stats_update_end(&tstats->syncp);
2672}
2673
dd5382a0
ED
2674static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
2675{
2676 struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
2677
2678 u64_stats_update_begin(&lstats->syncp);
fd2f4737
ED
2679 u64_stats_add(&lstats->bytes, len);
2680 u64_stats_inc(&lstats->packets);
dd5382a0
ED
2681 u64_stats_update_end(&lstats->syncp);
2682}
2683
aabc92bb
PNA
2684#define __netdev_alloc_pcpu_stats(type, gfp) \
2685({ \
2686 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2687 if (pcpu_stats) { \
2688 int __cpu; \
2689 for_each_possible_cpu(__cpu) { \
2690 typeof(type) *stat; \
2691 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2692 u64_stats_init(&stat->syncp); \
2693 } \
2694 } \
2695 pcpu_stats; \
1c213bd2
WC
2696})
2697
aabc92bb 2698#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2699 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2700
81b01894
HK
2701#define devm_netdev_alloc_pcpu_stats(dev, type) \
2702({ \
2703 typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
2704 if (pcpu_stats) { \
2705 int __cpu; \
2706 for_each_possible_cpu(__cpu) { \
2707 typeof(type) *stat; \
2708 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2709 u64_stats_init(&stat->syncp); \
2710 } \
2711 } \
2712 pcpu_stats; \
2713})
2714
764f5e54
JP
2715enum netdev_lag_tx_type {
2716 NETDEV_LAG_TX_TYPE_UNKNOWN,
2717 NETDEV_LAG_TX_TYPE_RANDOM,
2718 NETDEV_LAG_TX_TYPE_BROADCAST,
2719 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2720 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2721 NETDEV_LAG_TX_TYPE_HASH,
2722};
2723
f44aa9ef
JH
2724enum netdev_lag_hash {
2725 NETDEV_LAG_HASH_NONE,
2726 NETDEV_LAG_HASH_L2,
2727 NETDEV_LAG_HASH_L34,
2728 NETDEV_LAG_HASH_L23,
2729 NETDEV_LAG_HASH_E23,
2730 NETDEV_LAG_HASH_E34,
7b8fc010 2731 NETDEV_LAG_HASH_VLAN_SRCMAC,
f44aa9ef
JH
2732 NETDEV_LAG_HASH_UNKNOWN,
2733};
2734
764f5e54
JP
2735struct netdev_lag_upper_info {
2736 enum netdev_lag_tx_type tx_type;
f44aa9ef 2737 enum netdev_lag_hash hash_type;
764f5e54
JP
2738};
2739
fb1b2e3c
JP
2740struct netdev_lag_lower_state_info {
2741 u8 link_up : 1,
2742 tx_enabled : 1;
2743};
2744
1da177e4
LT
2745#include <linux/notifier.h>
2746
ede2762d
KT
2747/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2748 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2749 * adding new types.
dcfe1421 2750 */
ede2762d
KT
2751enum netdev_cmd {
2752 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2753 NETDEV_DOWN,
2754 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2755 detected a hardware crash and restarted
2756 - we can use this eg to kick tcp sessions
2757 once done */
ede2762d
KT
2758 NETDEV_CHANGE, /* Notify device state change */
2759 NETDEV_REGISTER,
2760 NETDEV_UNREGISTER,
2761 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2762 NETDEV_CHANGEADDR, /* notify after the address change */
2763 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2764 NETDEV_GOING_DOWN,
2765 NETDEV_CHANGENAME,
2766 NETDEV_FEAT_CHANGE,
2767 NETDEV_BONDING_FAILOVER,
2768 NETDEV_PRE_UP,
2769 NETDEV_PRE_TYPE_CHANGE,
2770 NETDEV_POST_TYPE_CHANGE,
2771 NETDEV_POST_INIT,
ede2762d
KT
2772 NETDEV_RELEASE,
2773 NETDEV_NOTIFY_PEERS,
2774 NETDEV_JOIN,
2775 NETDEV_CHANGEUPPER,
2776 NETDEV_RESEND_IGMP,
2777 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2778 NETDEV_CHANGEINFODATA,
2779 NETDEV_BONDING_INFO,
2780 NETDEV_PRECHANGEUPPER,
2781 NETDEV_CHANGELOWERSTATE,
2782 NETDEV_UDP_TUNNEL_PUSH_INFO,
2783 NETDEV_UDP_TUNNEL_DROP_INFO,
2784 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2785 NETDEV_CVLAN_FILTER_PUSH_INFO,
2786 NETDEV_CVLAN_FILTER_DROP_INFO,
2787 NETDEV_SVLAN_FILTER_PUSH_INFO,
2788 NETDEV_SVLAN_FILTER_DROP_INFO,
ede2762d
KT
2789};
2790const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2791
f629d208
JP
2792int register_netdevice_notifier(struct notifier_block *nb);
2793int unregister_netdevice_notifier(struct notifier_block *nb);
a30c7b42
JP
2794int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2795int unregister_netdevice_notifier_net(struct net *net,
2796 struct notifier_block *nb);
93642e14
JP
2797int register_netdevice_notifier_dev_net(struct net_device *dev,
2798 struct notifier_block *nb,
2799 struct netdev_net_notifier *nn);
2800int unregister_netdevice_notifier_dev_net(struct net_device *dev,
2801 struct notifier_block *nb,
2802 struct netdev_net_notifier *nn);
351638e7
JP
2803
2804struct netdev_notifier_info {
51d0c047
DA
2805 struct net_device *dev;
2806 struct netlink_ext_ack *extack;
351638e7
JP
2807};
2808
af7d6cce
SD
2809struct netdev_notifier_info_ext {
2810 struct netdev_notifier_info info; /* must be first */
2811 union {
2812 u32 mtu;
2813 } ext;
2814};
2815
be9efd36
JP
2816struct netdev_notifier_change_info {
2817 struct netdev_notifier_info info; /* must be first */
2818 unsigned int flags_changed;
2819};
2820
0e4ead9d
JP
2821struct netdev_notifier_changeupper_info {
2822 struct netdev_notifier_info info; /* must be first */
2823 struct net_device *upper_dev; /* new upper dev */
2824 bool master; /* is upper dev master */
5e82b4b2 2825 bool linking; /* is the notification for link or unlink */
29bf24af 2826 void *upper_info; /* upper dev info */
0e4ead9d
JP
2827};
2828
04d48266
JP
2829struct netdev_notifier_changelowerstate_info {
2830 struct netdev_notifier_info info; /* must be first */
2831 void *lower_state_info; /* is lower dev state */
2832};
2833
1570415f
PM
2834struct netdev_notifier_pre_changeaddr_info {
2835 struct netdev_notifier_info info; /* must be first */
2836 const unsigned char *dev_addr;
2837};
2838
75538c2b
CW
2839static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2840 struct net_device *dev)
2841{
2842 info->dev = dev;
51d0c047 2843 info->extack = NULL;
75538c2b
CW
2844}
2845
351638e7
JP
2846static inline struct net_device *
2847netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2848{
2849 return info->dev;
2850}
2851
51d0c047
DA
2852static inline struct netlink_ext_ack *
2853netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2854{
2855 return info->extack;
2856}
2857
f629d208 2858int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2859
2860
1da177e4
LT
2861extern rwlock_t dev_base_lock; /* Device list lock */
2862
881d966b
EB
2863#define for_each_netdev(net, d) \
2864 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2865#define for_each_netdev_reverse(net, d) \
2866 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2867#define for_each_netdev_rcu(net, d) \
2868 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2869#define for_each_netdev_safe(net, d, n) \
2870 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2871#define for_each_netdev_continue(net, d) \
2872 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
afa0df59
JP
2873#define for_each_netdev_continue_reverse(net, d) \
2874 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
2875 dev_list)
254245d2 2876#define for_each_netdev_continue_rcu(net, d) \
2877 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2878#define for_each_netdev_in_bond_rcu(bond, slave) \
2879 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2880 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2881#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2882
a050c33f
DL
2883static inline struct net_device *next_net_device(struct net_device *dev)
2884{
2885 struct list_head *lh;
2886 struct net *net;
2887
c346dca1 2888 net = dev_net(dev);
a050c33f
DL
2889 lh = dev->dev_list.next;
2890 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2891}
2892
ce81b76a
ED
2893static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2894{
2895 struct list_head *lh;
2896 struct net *net;
2897
2898 net = dev_net(dev);
ccf43438 2899 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2900 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2901}
2902
a050c33f
DL
2903static inline struct net_device *first_net_device(struct net *net)
2904{
2905 return list_empty(&net->dev_base_head) ? NULL :
2906 net_device_entry(net->dev_base_head.next);
2907}
7562f876 2908
ccf43438
ED
2909static inline struct net_device *first_net_device_rcu(struct net *net)
2910{
2911 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2912
2913 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2914}
2915
f629d208
JP
2916int netdev_boot_setup_check(struct net_device *dev);
2917unsigned long netdev_boot_base(const char *prefix, int unit);
2918struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2919 const char *hwaddr);
2920struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
f629d208
JP
2921void dev_add_pack(struct packet_type *pt);
2922void dev_remove_pack(struct packet_type *pt);
2923void __dev_remove_pack(struct packet_type *pt);
2924void dev_add_offload(struct packet_offload *po);
2925void dev_remove_offload(struct packet_offload *po);
f629d208 2926
a54acb3a 2927int dev_get_iflink(const struct net_device *dev);
fc4099f1 2928int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
ddb94eaf
PNA
2929int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
2930 struct net_device_path_stack *stack);
6c555490
WC
2931struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2932 unsigned short mask);
f629d208
JP
2933struct net_device *dev_get_by_name(struct net *net, const char *name);
2934struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2935struct net_device *__dev_get_by_name(struct net *net, const char *name);
2936int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 2937int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 2938void dev_close(struct net_device *dev);
2939void dev_close_many(struct list_head *head, bool unlink);
f629d208 2940void dev_disable_lro(struct net_device *dev);
0c4b51f0 2941int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2942u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 2943 struct net_device *sb_dev);
a4ea8a3d 2944u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 2945 struct net_device *sb_dev);
36ccdf85 2946
2b4aa3ce 2947int dev_queue_xmit(struct sk_buff *skb);
eadec877 2948int dev_queue_xmit_accel(struct sk_buff *skb, struct net_device *sb_dev);
36ccdf85
BT
2949int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
2950
2951static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
2952{
2953 int ret;
2954
2955 ret = __dev_direct_xmit(skb, queue_id);
2956 if (!dev_xmit_complete(ret))
2957 kfree_skb(skb);
2958 return ret;
2959}
2960
f629d208
JP
2961int register_netdevice(struct net_device *dev);
2962void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2963void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2964static inline void unregister_netdevice(struct net_device *dev)
2965{
2966 unregister_netdevice_queue(dev, NULL);
2967}
2968
f629d208
JP
2969int netdev_refcnt_read(const struct net_device *dev);
2970void free_netdev(struct net_device *dev);
74d332c1 2971void netdev_freemem(struct net_device *dev);
f629d208 2972int init_dummy_netdev(struct net_device *dev);
937f1ba5 2973
cff9f12b
MG
2974struct net_device *netdev_get_xmit_slave(struct net_device *dev,
2975 struct sk_buff *skb,
2976 bool all_slaves);
719a402c
TT
2977struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
2978 struct sock *sk);
f629d208
JP
2979struct net_device *dev_get_by_index(struct net *net, int ifindex);
2980struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2981struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2982struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208
JP
2983int netdev_get_name(struct net *net, char *name, int ifindex);
2984int dev_restart(struct net_device *dev);
d4546c25 2985int skb_gro_receive(struct sk_buff *p, struct sk_buff *skb);
3a1296a3 2986int skb_gro_receive_list(struct sk_buff *p, struct sk_buff *skb);
86911732
HX
2987
2988static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2989{
2990 return NAPI_GRO_CB(skb)->data_offset;
2991}
2992
2993static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2994{
2995 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2996}
2997
2998static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
2999{
3000 NAPI_GRO_CB(skb)->data_offset += len;
3001}
3002
a5b1cf28
HX
3003static inline void *skb_gro_header_fast(struct sk_buff *skb,
3004 unsigned int offset)
86911732 3005{
a5b1cf28
HX
3006 return NAPI_GRO_CB(skb)->frag0 + offset;
3007}
78a478d0 3008
a5b1cf28
HX
3009static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
3010{
3011 return NAPI_GRO_CB(skb)->frag0_len < hlen;
3012}
78a478d0 3013
57ea52a8
HX
3014static inline void skb_gro_frag0_invalidate(struct sk_buff *skb)
3015{
3016 NAPI_GRO_CB(skb)->frag0 = NULL;
3017 NAPI_GRO_CB(skb)->frag0_len = 0;
3018}
3019
a5b1cf28
HX
3020static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
3021 unsigned int offset)
3022{
17dd759c
HX
3023 if (!pskb_may_pull(skb, hlen))
3024 return NULL;
3025
57ea52a8 3026 skb_gro_frag0_invalidate(skb);
17dd759c 3027 return skb->data + offset;
86911732 3028}
1da177e4 3029
36e7b1b8
HX
3030static inline void *skb_gro_network_header(struct sk_buff *skb)
3031{
78d3fd0b
HX
3032 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
3033 skb_network_offset(skb);
36e7b1b8
HX
3034}
3035
bf5a755f
JC
3036static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
3037 const void *start, unsigned int len)
3038{
573e8fca 3039 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
3040 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
3041 csum_partial(start, len, 0));
3042}
3043
573e8fca
TH
3044/* GRO checksum functions. These are logical equivalents of the normal
3045 * checksum functions (in skbuff.h) except that they operate on the GRO
3046 * offsets and fields in sk_buff.
3047 */
3048
3049__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
3050
15e2396d
TH
3051static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
3052{
b7fe10e5 3053 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
3054}
3055
573e8fca
TH
3056static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
3057 bool zero_okay,
3058 __sum16 check)
3059{
6edec0e6
TH
3060 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
3061 skb_checksum_start_offset(skb) <
3062 skb_gro_offset(skb)) &&
15e2396d 3063 !skb_at_gro_remcsum_start(skb) &&
662880f4 3064 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
3065 (!zero_okay || check));
3066}
3067
3068static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
3069 __wsum psum)
3070{
3071 if (NAPI_GRO_CB(skb)->csum_valid &&
3072 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
3073 return 0;
3074
3075 NAPI_GRO_CB(skb)->csum = psum;
3076
3077 return __skb_gro_checksum_complete(skb);
3078}
3079
573e8fca
TH
3080static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
3081{
662880f4
TH
3082 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
3083 /* Consume a checksum from CHECKSUM_UNNECESSARY */
3084 NAPI_GRO_CB(skb)->csum_cnt--;
3085 } else {
3086 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
3087 * verified a new top level checksum or an encapsulated one
3088 * during GRO. This saves work if we fallback to normal path.
3089 */
3090 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
3091 }
3092}
3093
3094#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
3095 compute_pseudo) \
3096({ \
3097 __sum16 __ret = 0; \
3098 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
3099 __ret = __skb_gro_checksum_validate_complete(skb, \
3100 compute_pseudo(skb, proto)); \
219f1d79 3101 if (!__ret) \
573e8fca
TH
3102 skb_gro_incr_csum_unnecessary(skb); \
3103 __ret; \
3104})
3105
3106#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
3107 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
3108
3109#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
3110 compute_pseudo) \
3111 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
3112
3113#define skb_gro_checksum_simple_validate(skb) \
3114 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
3115
d96535a1
TH
3116static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
3117{
3118 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
3119 !NAPI_GRO_CB(skb)->csum_valid);
3120}
3121
3122static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
b39c78b2 3123 __wsum pseudo)
d96535a1
TH
3124{
3125 NAPI_GRO_CB(skb)->csum = ~pseudo;
3126 NAPI_GRO_CB(skb)->csum_valid = 1;
3127}
3128
b39c78b2 3129#define skb_gro_checksum_try_convert(skb, proto, compute_pseudo) \
d96535a1
TH
3130do { \
3131 if (__skb_gro_checksum_convert_check(skb)) \
b39c78b2 3132 __skb_gro_checksum_convert(skb, \
d96535a1
TH
3133 compute_pseudo(skb, proto)); \
3134} while (0)
3135
26c4f7da
TH
3136struct gro_remcsum {
3137 int offset;
3138 __wsum delta;
3139};
3140
3141static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
3142{
846cd667 3143 grc->offset = 0;
26c4f7da
TH
3144 grc->delta = 0;
3145}
3146
b7fe10e5
TH
3147static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
3148 unsigned int off, size_t hdrlen,
3149 int start, int offset,
3150 struct gro_remcsum *grc,
3151 bool nopartial)
dcdc8994
TH
3152{
3153 __wsum delta;
b7fe10e5 3154 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
3155
3156 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
3157
15e2396d 3158 if (!nopartial) {
b7fe10e5
TH
3159 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
3160 return ptr;
3161 }
3162
3163 ptr = skb_gro_header_fast(skb, off);
3164 if (skb_gro_header_hard(skb, off + plen)) {
3165 ptr = skb_gro_header_slow(skb, off + plen, off);
3166 if (!ptr)
3167 return NULL;
15e2396d
TH
3168 }
3169
b7fe10e5
TH
3170 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
3171 start, offset);
dcdc8994
TH
3172
3173 /* Adjust skb->csum since we changed the packet */
dcdc8994 3174 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 3175
b7fe10e5 3176 grc->offset = off + hdrlen + offset;
26c4f7da 3177 grc->delta = delta;
b7fe10e5
TH
3178
3179 return ptr;
dcdc8994
TH
3180}
3181
26c4f7da
TH
3182static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
3183 struct gro_remcsum *grc)
3184{
b7fe10e5
TH
3185 void *ptr;
3186 size_t plen = grc->offset + sizeof(u16);
3187
26c4f7da
TH
3188 if (!grc->delta)
3189 return;
3190
b7fe10e5
TH
3191 ptr = skb_gro_header_fast(skb, grc->offset);
3192 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
3193 ptr = skb_gro_header_slow(skb, plen, grc->offset);
3194 if (!ptr)
3195 return;
3196 }
3197
3198 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 3199}
dcdc8994 3200
25393d3f 3201#ifdef CONFIG_XFRM_OFFLOAD
d4546c25 3202static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
25393d3f
SK
3203{
3204 if (PTR_ERR(pp) != -EINPROGRESS)
3205 NAPI_GRO_CB(skb)->flush |= flush;
3206}
603d4cf8 3207static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 3208 struct sk_buff *pp,
603d4cf8
SD
3209 int flush,
3210 struct gro_remcsum *grc)
3211{
3212 if (PTR_ERR(pp) != -EINPROGRESS) {
3213 NAPI_GRO_CB(skb)->flush |= flush;
3214 skb_gro_remcsum_cleanup(skb, grc);
3215 skb->remcsum_offload = 0;
3216 }
3217}
25393d3f 3218#else
d4546c25 3219static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
5f114163
SK
3220{
3221 NAPI_GRO_CB(skb)->flush |= flush;
3222}
603d4cf8 3223static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 3224 struct sk_buff *pp,
603d4cf8
SD
3225 int flush,
3226 struct gro_remcsum *grc)
3227{
3228 NAPI_GRO_CB(skb)->flush |= flush;
3229 skb_gro_remcsum_cleanup(skb, grc);
3230 skb->remcsum_offload = 0;
3231}
25393d3f 3232#endif
5f114163 3233
0c4e8581
SH
3234static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
3235 unsigned short type,
3b04ddde 3236 const void *daddr, const void *saddr,
95c96174 3237 unsigned int len)
0c4e8581 3238{
f1ecfd5d 3239 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 3240 return 0;
3b04ddde
SH
3241
3242 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
3243}
3244
b95cce35
SH
3245static inline int dev_parse_header(const struct sk_buff *skb,
3246 unsigned char *haddr)
3247{
3248 const struct net_device *dev = skb->dev;
3249
1b83336b 3250 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 3251 return 0;
3b04ddde 3252 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
3253}
3254
e78b2915
MM
3255static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
3256{
3257 const struct net_device *dev = skb->dev;
3258
3259 if (!dev->header_ops || !dev->header_ops->parse_protocol)
3260 return 0;
3261 return dev->header_ops->parse_protocol(skb);
3262}
3263
2793a23a
WB
3264/* ll_header must have at least hard_header_len allocated */
3265static inline bool dev_validate_header(const struct net_device *dev,
3266 char *ll_header, int len)
3267{
3268 if (likely(len >= dev->hard_header_len))
3269 return true;
217e6fa2
WB
3270 if (len < dev->min_header_len)
3271 return false;
2793a23a
WB
3272
3273 if (capable(CAP_SYS_RAWIO)) {
3274 memset(ll_header + len, 0, dev->hard_header_len - len);
3275 return true;
3276 }
3277
3278 if (dev->header_ops && dev->header_ops->validate)
3279 return dev->header_ops->validate(ll_header, len);
3280
3281 return false;
3282}
3283
d5496990
EB
3284static inline bool dev_has_header(const struct net_device *dev)
3285{
3286 return dev->header_ops && dev->header_ops->create;
3287}
3288
36fd633e
AV
3289typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
3290 int len, int size);
f629d208 3291int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
3292static inline int unregister_gifconf(unsigned int family)
3293{
3294 return register_gifconf(family, NULL);
3295}
3296
99bbc707 3297#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 3298#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
3299struct sd_flow_limit {
3300 u64 count;
3301 unsigned int num_buckets;
3302 unsigned int history_head;
3303 u16 history[FLOW_LIMIT_HISTORY];
3304 u8 buckets[];
3305};
3306
3307extern int netdev_flow_limit_table_len;
3308#endif /* CONFIG_NET_FLOW_LIMIT */
3309
1da177e4 3310/*
5e82b4b2 3311 * Incoming packets are placed on per-CPU queues
1da177e4 3312 */
d94d9fee 3313struct softnet_data {
1da177e4 3314 struct list_head poll_list;
6e7676c1 3315 struct sk_buff_head process_queue;
1da177e4 3316
dee42870 3317 /* stats */
cd7b5396
DM
3318 unsigned int processed;
3319 unsigned int time_squeeze;
cd7b5396 3320 unsigned int received_rps;
fd793d89 3321#ifdef CONFIG_RPS
88751275 3322 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
3323#endif
3324#ifdef CONFIG_NET_FLOW_LIMIT
3325 struct sd_flow_limit __rcu *flow_limit;
3326#endif
3327 struct Qdisc *output_queue;
3328 struct Qdisc **output_queue_tailp;
3329 struct sk_buff *completion_queue;
f53c7239
SK
3330#ifdef CONFIG_XFRM_OFFLOAD
3331 struct sk_buff_head xfrm_backlog;
3332#endif
97cdcf37
FW
3333 /* written and read only by owning cpu: */
3334 struct {
3335 u16 recursion;
3336 u8 more;
3337 } xmit;
4cdb1e2e 3338#ifdef CONFIG_RPS
501e7ef5
ED
3339 /* input_queue_head should be written by cpu owning this struct,
3340 * and only read by other cpus. Worth using a cache line.
3341 */
3342 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3343
3344 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3345 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3346 struct softnet_data *rps_ipi_next;
3347 unsigned int cpu;
76cc8b13 3348 unsigned int input_queue_tail;
1e94d72f 3349#endif
95c96174 3350 unsigned int dropped;
0a9627f2 3351 struct sk_buff_head input_pkt_queue;
bea3348e 3352 struct napi_struct backlog;
99bbc707 3353
1da177e4
LT
3354};
3355
76cc8b13 3356static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3357{
3358#ifdef CONFIG_RPS
76cc8b13
TH
3359 sd->input_queue_head++;
3360#endif
3361}
3362
3363static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3364 unsigned int *qtail)
3365{
3366#ifdef CONFIG_RPS
3367 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3368#endif
3369}
3370
0a9627f2 3371DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3372
97cdcf37
FW
3373static inline int dev_recursion_level(void)
3374{
28b05b92 3375 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3376}
3377
fb7861d1 3378#define XMIT_RECURSION_LIMIT 8
97cdcf37
FW
3379static inline bool dev_xmit_recursion(void)
3380{
3381 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3382 XMIT_RECURSION_LIMIT);
3383}
3384
3385static inline void dev_xmit_recursion_inc(void)
3386{
3387 __this_cpu_inc(softnet_data.xmit.recursion);
3388}
3389
3390static inline void dev_xmit_recursion_dec(void)
3391{
3392 __this_cpu_dec(softnet_data.xmit.recursion);
3393}
3394
f629d208 3395void __netif_schedule(struct Qdisc *q);
46e5da40 3396void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3397
fd2ea0a7
DM
3398static inline void netif_tx_schedule_all(struct net_device *dev)
3399{
3400 unsigned int i;
3401
3402 for (i = 0; i < dev->num_tx_queues; i++)
3403 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3404}
3405
f9a7cbbf 3406static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3407{
73466498 3408 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3409}
3410
bea3348e
SH
3411/**
3412 * netif_start_queue - allow transmit
3413 * @dev: network device
3414 *
3415 * Allow upper layers to call the device hard_start_xmit routine.
3416 */
1da177e4
LT
3417static inline void netif_start_queue(struct net_device *dev)
3418{
e8a0464c 3419 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3420}
3421
fd2ea0a7
DM
3422static inline void netif_tx_start_all_queues(struct net_device *dev)
3423{
3424 unsigned int i;
3425
3426 for (i = 0; i < dev->num_tx_queues; i++) {
3427 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3428 netif_tx_start_queue(txq);
3429 }
3430}
3431
46e5da40 3432void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3433
d29f749e
DJ
3434/**
3435 * netif_wake_queue - restart transmit
3436 * @dev: network device
3437 *
3438 * Allow upper layers to call the device hard_start_xmit routine.
3439 * Used for flow control when transmit resources are available.
3440 */
79d16385
DM
3441static inline void netif_wake_queue(struct net_device *dev)
3442{
e8a0464c 3443 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3444}
3445
fd2ea0a7
DM
3446static inline void netif_tx_wake_all_queues(struct net_device *dev)
3447{
3448 unsigned int i;
3449
3450 for (i = 0; i < dev->num_tx_queues; i++) {
3451 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3452 netif_tx_wake_queue(txq);
3453 }
3454}
3455
f9a7cbbf 3456static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3457{
73466498 3458 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3459}
3460
bea3348e
SH
3461/**
3462 * netif_stop_queue - stop transmitted packets
3463 * @dev: network device
3464 *
3465 * Stop upper layers calling the device hard_start_xmit routine.
3466 * Used for flow control when transmit resources are unavailable.
3467 */
1da177e4
LT
3468static inline void netif_stop_queue(struct net_device *dev)
3469{
e8a0464c 3470 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3471}
3472
a2029240 3473void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3474
4d29515f 3475static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3476{
73466498 3477 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3478}
3479
bea3348e
SH
3480/**
3481 * netif_queue_stopped - test if transmit queue is flowblocked
3482 * @dev: network device
3483 *
3484 * Test if transmit queue on device is currently unable to send.
3485 */
4d29515f 3486static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3487{
e8a0464c 3488 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3489}
3490
4d29515f 3491static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3492{
73466498
TH
3493 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3494}
3495
8e2f1a63
DB
3496static inline bool
3497netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3498{
3499 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3500}
3501
8e2f1a63
DB
3502static inline bool
3503netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3504{
3505 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3506}
3507
f57bac3c
VM
3508/**
3509 * netdev_queue_set_dql_min_limit - set dql minimum limit
3510 * @dev_queue: pointer to transmit queue
3511 * @min_limit: dql minimum limit
3512 *
3513 * Forces xmit_more() to return true until the minimum threshold
3514 * defined by @min_limit is reached (or until the tx queue is
3515 * empty). Warning: to be use with care, misuse will impact the
3516 * latency.
3517 */
3518static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
3519 unsigned int min_limit)
3520{
3521#ifdef CONFIG_BQL
3522 dev_queue->dql.min_limit = min_limit;
3523#endif
3524}
3525
53511453
ED
3526/**
3527 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3528 * @dev_queue: pointer to transmit queue
3529 *
3530 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3531 * to give appropriate hint to the CPU.
53511453
ED
3532 */
3533static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3534{
3535#ifdef CONFIG_BQL
3536 prefetchw(&dev_queue->dql.num_queued);
3537#endif
3538}
3539
3540/**
3541 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3542 * @dev_queue: pointer to transmit queue
3543 *
3544 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3545 * to give appropriate hint to the CPU.
53511453
ED
3546 */
3547static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3548{
3549#ifdef CONFIG_BQL
3550 prefetchw(&dev_queue->dql.limit);
3551#endif
3552}
3553
c5d67bd7
TH
3554static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3555 unsigned int bytes)
3556{
114cf580
TH
3557#ifdef CONFIG_BQL
3558 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3559
3560 if (likely(dql_avail(&dev_queue->dql) >= 0))
3561 return;
3562
3563 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3564
3565 /*
3566 * The XOFF flag must be set before checking the dql_avail below,
3567 * because in netdev_tx_completed_queue we update the dql_completed
3568 * before checking the XOFF flag.
3569 */
3570 smp_mb();
3571
3572 /* check again in case another CPU has just made room avail */
3573 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3574 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3575#endif
c5d67bd7
TH
3576}
3577
3e59020a
ED
3578/* Variant of netdev_tx_sent_queue() for drivers that are aware
3579 * that they should not test BQL status themselves.
3580 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3581 * skb of a batch.
3582 * Returns true if the doorbell must be used to kick the NIC.
3583 */
3584static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3585 unsigned int bytes,
3586 bool xmit_more)
3587{
3588 if (xmit_more) {
3589#ifdef CONFIG_BQL
3590 dql_queued(&dev_queue->dql, bytes);
3591#endif
3592 return netif_tx_queue_stopped(dev_queue);
3593 }
3594 netdev_tx_sent_queue(dev_queue, bytes);
3595 return true;
3596}
3597
0042d0c8
FF
3598/**
3599 * netdev_sent_queue - report the number of bytes queued to hardware
3600 * @dev: network device
3601 * @bytes: number of bytes queued to the hardware device queue
3602 *
3603 * Report the number of bytes queued for sending/completion to the network
3604 * device hardware queue. @bytes should be a good approximation and should
3605 * exactly match netdev_completed_queue() @bytes
3606 */
c5d67bd7
TH
3607static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3608{
3609 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3610}
3611
620344c4
HK
3612static inline bool __netdev_sent_queue(struct net_device *dev,
3613 unsigned int bytes,
3614 bool xmit_more)
3615{
3616 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3617 xmit_more);
3618}
3619
c5d67bd7 3620static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3621 unsigned int pkts, unsigned int bytes)
c5d67bd7 3622{
114cf580 3623#ifdef CONFIG_BQL
b37c0fbe
AD
3624 if (unlikely(!bytes))
3625 return;
3626
3627 dql_completed(&dev_queue->dql, bytes);
3628
3629 /*
3630 * Without the memory barrier there is a small possiblity that
3631 * netdev_tx_sent_queue will miss the update and cause the queue to
3632 * be stopped forever
3633 */
3634 smp_mb();
3635
f3acd33d 3636 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3637 return;
3638
3639 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3640 netif_schedule_queue(dev_queue);
114cf580 3641#endif
c5d67bd7
TH
3642}
3643
0042d0c8
FF
3644/**
3645 * netdev_completed_queue - report bytes and packets completed by device
3646 * @dev: network device
3647 * @pkts: actual number of packets sent over the medium
3648 * @bytes: actual number of bytes sent over the medium
3649 *
3650 * Report the number of bytes and packets transmitted by the network device
3651 * hardware queue over the physical medium, @bytes must exactly match the
3652 * @bytes amount passed to netdev_sent_queue()
3653 */
c5d67bd7 3654static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3655 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3656{
3657 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3658}
3659
3660static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3661{
114cf580 3662#ifdef CONFIG_BQL
5c490354 3663 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3664 dql_reset(&q->dql);
3665#endif
c5d67bd7
TH
3666}
3667
0042d0c8
FF
3668/**
3669 * netdev_reset_queue - reset the packets and bytes count of a network device
3670 * @dev_queue: network device
3671 *
3672 * Reset the bytes and packet count of a network device and clear the
3673 * software flow control OFF bit for this network device
3674 */
c5d67bd7
TH
3675static inline void netdev_reset_queue(struct net_device *dev_queue)
3676{
3677 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3678}
3679
b9507bda
DB
3680/**
3681 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3682 * @dev: network device
3683 * @queue_index: given tx queue index
3684 *
3685 * Returns 0 if given tx queue index >= number of device tx queues,
3686 * otherwise returns the originally passed tx queue index.
3687 */
3688static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3689{
3690 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3691 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3692 dev->name, queue_index,
3693 dev->real_num_tx_queues);
3694 return 0;
3695 }
3696
3697 return queue_index;
3698}
3699
bea3348e
SH
3700/**
3701 * netif_running - test if up
3702 * @dev: network device
3703 *
3704 * Test if the device has been brought up.
3705 */
4d29515f 3706static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3707{
3708 return test_bit(__LINK_STATE_START, &dev->state);
3709}
3710
f25f4e44 3711/*
5e82b4b2 3712 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3713 * stop, and a check if it's stopped. All other device management is
3714 * done at the overall netdevice level.
3715 * Also test the device if we're multiqueue.
3716 */
bea3348e
SH
3717
3718/**
3719 * netif_start_subqueue - allow sending packets on subqueue
3720 * @dev: network device
3721 * @queue_index: sub queue index
3722 *
3723 * Start individual transmit queue of a device with multiple transmit queues.
3724 */
f25f4e44
PWJ
3725static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3726{
fd2ea0a7 3727 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3728
3729 netif_tx_start_queue(txq);
f25f4e44
PWJ
3730}
3731
bea3348e
SH
3732/**
3733 * netif_stop_subqueue - stop sending packets on subqueue
3734 * @dev: network device
3735 * @queue_index: sub queue index
3736 *
3737 * Stop individual transmit queue of a device with multiple transmit queues.
3738 */
f25f4e44
PWJ
3739static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3740{
fd2ea0a7 3741 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3742 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3743}
3744
bea3348e 3745/**
270f3385 3746 * __netif_subqueue_stopped - test status of subqueue
bea3348e
SH
3747 * @dev: network device
3748 * @queue_index: sub queue index
3749 *
3750 * Check individual transmit queue of a device with multiple transmit queues.
3751 */
4d29515f
DM
3752static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3753 u16 queue_index)
f25f4e44 3754{
fd2ea0a7 3755 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3756
3757 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3758}
3759
270f3385
MCC
3760/**
3761 * netif_subqueue_stopped - test status of subqueue
3762 * @dev: network device
3763 * @skb: sub queue buffer pointer
3764 *
3765 * Check individual transmit queue of a device with multiple transmit queues.
3766 */
4d29515f
DM
3767static inline bool netif_subqueue_stopped(const struct net_device *dev,
3768 struct sk_buff *skb)
668f895a
PE
3769{
3770 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3771}
bea3348e 3772
738b35cc
FF
3773/**
3774 * netif_wake_subqueue - allow sending packets on subqueue
3775 * @dev: network device
3776 * @queue_index: sub queue index
3777 *
3778 * Resume individual transmit queue of a device with multiple transmit queues.
3779 */
3780static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3781{
3782 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3783
3784 netif_tx_wake_queue(txq);
3785}
f25f4e44 3786
537c00de 3787#ifdef CONFIG_XPS
53af53ae 3788int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3789 u16 index);
80d19669 3790int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
044ab86d 3791 u16 index, enum xps_map_type type);
80d19669
AN
3792
3793/**
3794 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3795 * @j: CPU/Rx queue index
3796 * @mask: bitmask of all cpus/rx queues
3797 * @nr_bits: number of bits in the bitmask
3798 *
3799 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3800 */
3801static inline bool netif_attr_test_mask(unsigned long j,
3802 const unsigned long *mask,
3803 unsigned int nr_bits)
3804{
3805 cpu_max_bits_warn(j, nr_bits);
3806 return test_bit(j, mask);
3807}
3808
3809/**
3810 * netif_attr_test_online - Test for online CPU/Rx queue
3811 * @j: CPU/Rx queue index
3812 * @online_mask: bitmask for CPUs/Rx queues that are online
3813 * @nr_bits: number of bits in the bitmask
3814 *
3815 * Returns true if a CPU/Rx queue is online.
3816 */
3817static inline bool netif_attr_test_online(unsigned long j,
3818 const unsigned long *online_mask,
3819 unsigned int nr_bits)
3820{
3821 cpu_max_bits_warn(j, nr_bits);
3822
3823 if (online_mask)
3824 return test_bit(j, online_mask);
3825
3826 return (j < nr_bits);
3827}
3828
3829/**
3830 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3831 * @n: CPU/Rx queue index
3832 * @srcp: the cpumask/Rx queue mask pointer
3833 * @nr_bits: number of bits in the bitmask
3834 *
3835 * Returns >= nr_bits if no further CPUs/Rx queues set.
3836 */
3837static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3838 unsigned int nr_bits)
3839{
3840 /* -1 is a legal arg here. */
3841 if (n != -1)
3842 cpu_max_bits_warn(n, nr_bits);
3843
3844 if (srcp)
3845 return find_next_bit(srcp, nr_bits, n + 1);
3846
3847 return n + 1;
3848}
3849
3850/**
a1fa83bd 3851 * netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
80d19669
AN
3852 * @n: CPU/Rx queue index
3853 * @src1p: the first CPUs/Rx queues mask pointer
3854 * @src2p: the second CPUs/Rx queues mask pointer
3855 * @nr_bits: number of bits in the bitmask
3856 *
3857 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3858 */
3859static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3860 const unsigned long *src2p,
3861 unsigned int nr_bits)
3862{
3863 /* -1 is a legal arg here. */
3864 if (n != -1)
3865 cpu_max_bits_warn(n, nr_bits);
3866
3867 if (src1p && src2p)
3868 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3869 else if (src1p)
3870 return find_next_bit(src1p, nr_bits, n + 1);
3871 else if (src2p)
3872 return find_next_bit(src2p, nr_bits, n + 1);
3873
3874 return n + 1;
3875}
537c00de
AD
3876#else
3877static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3878 const struct cpumask *mask,
537c00de
AD
3879 u16 index)
3880{
3881 return 0;
3882}
c9fbb2d2
KK
3883
3884static inline int __netif_set_xps_queue(struct net_device *dev,
3885 const unsigned long *mask,
044ab86d 3886 u16 index, enum xps_map_type type)
c9fbb2d2
KK
3887{
3888 return 0;
3889}
537c00de
AD
3890#endif
3891
bea3348e
SH
3892/**
3893 * netif_is_multiqueue - test if device has multiple transmit queues
3894 * @dev: network device
3895 *
3896 * Check if device has multiple transmit queues
bea3348e 3897 */
4d29515f 3898static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3899{
a02cec21 3900 return dev->num_tx_queues > 1;
f25f4e44 3901}
1da177e4 3902
f629d208 3903int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3904
a953be53 3905#ifdef CONFIG_SYSFS
f629d208 3906int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3907#else
3908static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3909 unsigned int rxqs)
62fe0b40 3910{
c29c2ebd 3911 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3912 return 0;
3913}
3914#endif
3915
65073a67
DB
3916static inline struct netdev_rx_queue *
3917__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3918{
3919 return dev->_rx + rxq;
3920}
3921
a953be53
MD
3922#ifdef CONFIG_SYSFS
3923static inline unsigned int get_netdev_rx_queue_index(
3924 struct netdev_rx_queue *queue)
3925{
3926 struct net_device *dev = queue->dev;
3927 int index = queue - dev->_rx;
3928
3929 BUG_ON(index >= dev->num_rx_queues);
3930 return index;
3931}
3932#endif
3933
16917b87 3934#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3935int netif_get_num_default_rss_queues(void);
16917b87 3936
e6247027
ED
3937enum skb_free_reason {
3938 SKB_REASON_CONSUMED,
3939 SKB_REASON_DROPPED,
3940};
3941
3942void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3943void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3944
e6247027
ED
3945/*
3946 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3947 * interrupt context or with hardware interrupts being disabled.
3948 * (in_irq() || irqs_disabled())
3949 *
3950 * We provide four helpers that can be used in following contexts :
3951 *
3952 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3953 * replacing kfree_skb(skb)
3954 *
3955 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3956 * Typically used in place of consume_skb(skb) in TX completion path
3957 *
3958 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3959 * replacing kfree_skb(skb)
3960 *
3961 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3962 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3963 */
e6247027
ED
3964static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3965{
3966 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3967}
3968
3969static inline void dev_consume_skb_irq(struct sk_buff *skb)
3970{
3971 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3972}
3973
3974static inline void dev_kfree_skb_any(struct sk_buff *skb)
3975{
3976 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3977}
3978
3979static inline void dev_consume_skb_any(struct sk_buff *skb)
3980{
3981 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3982}
1da177e4 3983
7c497478
JW
3984void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3985int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208
JP
3986int netif_rx(struct sk_buff *skb);
3987int netif_rx_ni(struct sk_buff *skb);
c11171a4 3988int netif_rx_any_context(struct sk_buff *skb);
04eb4489 3989int netif_receive_skb(struct sk_buff *skb);
1c601d82 3990int netif_receive_skb_core(struct sk_buff *skb);
f6ad8c1b 3991void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3992gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3993void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3994struct sk_buff *napi_get_frags(struct napi_struct *napi);
3995gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3996struct packet_offload *gro_find_receive_by_type(__be16 type);
3997struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3998
3999static inline void napi_free_frags(struct napi_struct *napi)
4000{
4001 kfree_skb(napi->skb);
4002 napi->skb = NULL;
4003}
4004
24b27fc4 4005bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
4006int netdev_rx_handler_register(struct net_device *dev,
4007 rx_handler_func_t *rx_handler,
4008 void *rx_handler_data);
4009void netdev_rx_handler_unregister(struct net_device *dev);
4010
4011bool dev_valid_name(const char *name);
44c02a2c
AV
4012int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
4013 bool *need_copyout);
36fd633e 4014int dev_ifconf(struct net *net, struct ifconf *, int);
f629d208
JP
4015int dev_ethtool(struct net *net, struct ifreq *);
4016unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
4017int __dev_change_flags(struct net_device *dev, unsigned int flags,
4018 struct netlink_ext_ack *extack);
567c5e13
PM
4019int dev_change_flags(struct net_device *dev, unsigned int flags,
4020 struct netlink_ext_ack *extack);
cb178190
DM
4021void __dev_notify_flags(struct net_device *, unsigned int old_flags,
4022 unsigned int gchanges);
f629d208
JP
4023int dev_change_name(struct net_device *, const char *);
4024int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 4025int dev_get_alias(const struct net_device *, char *, size_t);
f629d208 4026int dev_change_net_namespace(struct net_device *, struct net *, const char *);
f51048c3 4027int __dev_set_mtu(struct net_device *, int);
d836f5c6
ED
4028int dev_validate_mtu(struct net_device *dev, int mtu,
4029 struct netlink_ext_ack *extack);
7a4c53be
SH
4030int dev_set_mtu_ext(struct net_device *dev, int mtu,
4031 struct netlink_ext_ack *extack);
f629d208 4032int dev_set_mtu(struct net_device *, int);
6a643ddb 4033int dev_change_tx_queue_len(struct net_device *, unsigned long);
f629d208 4034void dev_set_group(struct net_device *, int);
d59cdf94
PM
4035int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
4036 struct netlink_ext_ack *extack);
3a37a963
PM
4037int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
4038 struct netlink_ext_ack *extack);
3b23a32a
CW
4039int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
4040 struct netlink_ext_ack *extack);
4041int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
f629d208
JP
4042int dev_change_carrier(struct net_device *, bool new_carrier);
4043int dev_get_phys_port_id(struct net_device *dev,
02637fce 4044 struct netdev_phys_item_id *ppid);
db24a904
DA
4045int dev_get_phys_port_name(struct net_device *dev,
4046 char *name, size_t len);
d6abc596
FF
4047int dev_get_port_parent_id(struct net_device *dev,
4048 struct netdev_phys_item_id *ppid, bool recurse);
4049bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
d746d707 4050int dev_change_proto_down(struct net_device *dev, bool proto_down);
b5899679 4051int dev_change_proto_down_generic(struct net_device *dev, bool proto_down);
829eb208
RP
4052void dev_change_proto_down_reason(struct net_device *dev, unsigned long mask,
4053 u32 value);
f53c7239 4054struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
4055struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
4056 struct netdev_queue *txq, int *ret);
d67b9cd2 4057
f4e63525 4058typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf);
d67b9cd2 4059int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
92234c8f 4060 int fd, int expected_fd, u32 flags);
aa8d3a71 4061int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
7f0a8382 4062u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
aa8d3a71 4063
a0265d28 4064int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 4065int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
5f7d5728 4066int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
4067bool is_skb_forwardable(const struct net_device *dev,
4068 const struct sk_buff *skb);
1da177e4 4069
5f7d5728
JDB
4070static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
4071 const struct sk_buff *skb,
4072 const bool check_mtu)
4073{
4074 const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
4075 unsigned int len;
4076
4077 if (!(dev->flags & IFF_UP))
4078 return false;
4079
4080 if (!check_mtu)
4081 return true;
4082
4083 len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
4084 if (skb->len <= len)
4085 return true;
4086
4087 /* if TSO is enabled, we don't care about the length as the packet
4088 * could be forwarded without being segmented before
4089 */
4090 if (skb_is_gso(skb))
4091 return true;
4092
4093 return false;
4094}
4095
4e3264d2 4096static __always_inline int ____dev_forward_skb(struct net_device *dev,
5f7d5728
JDB
4097 struct sk_buff *skb,
4098 const bool check_mtu)
4e3264d2
MKL
4099{
4100 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
5f7d5728 4101 unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
4e3264d2
MKL
4102 atomic_long_inc(&dev->rx_dropped);
4103 kfree_skb(skb);
4104 return NET_RX_DROP;
4105 }
4106
4107 skb_scrub_packet(skb, true);
4108 skb->priority = 0;
4109 return 0;
4110}
4111
9f9a742d 4112bool dev_nit_active(struct net_device *dev);
74b20582
DA
4113void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
4114
20380731 4115extern int netdev_budget;
7acf8a1e 4116extern unsigned int netdev_budget_usecs;
1da177e4
LT
4117
4118/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 4119void netdev_run_todo(void);
1da177e4 4120
bea3348e
SH
4121/**
4122 * dev_put - release reference to device
4123 * @dev: network device
4124 *
9ef4429b 4125 * Release reference to device to allow it to be freed.
bea3348e 4126 */
1da177e4
LT
4127static inline void dev_put(struct net_device *dev)
4128{
919067cc 4129#ifdef CONFIG_PCPU_DEV_REFCNT
933393f5 4130 this_cpu_dec(*dev->pcpu_refcnt);
919067cc
ED
4131#else
4132 refcount_dec(&dev->dev_refcnt);
4133#endif
1da177e4
LT
4134}
4135
bea3348e
SH
4136/**
4137 * dev_hold - get reference to device
4138 * @dev: network device
4139 *
9ef4429b 4140 * Hold reference to device to keep it from being freed.
bea3348e 4141 */
15333061
SH
4142static inline void dev_hold(struct net_device *dev)
4143{
919067cc 4144#ifdef CONFIG_PCPU_DEV_REFCNT
933393f5 4145 this_cpu_inc(*dev->pcpu_refcnt);
919067cc
ED
4146#else
4147 refcount_inc(&dev->dev_refcnt);
4148#endif
15333061 4149}
1da177e4
LT
4150
4151/* Carrier loss detection, dial on demand. The functions netif_carrier_on
4152 * and _off may be called from IRQ context, but it is caller
4153 * who is responsible for serialization of these calls.
b00055aa
SR
4154 *
4155 * The name carrier is inappropriate, these functions should really be
4156 * called netif_lowerlayer_*() because they represent the state of any
4157 * kind of lower layer not just hardware media.
1da177e4
LT
4158 */
4159
f629d208
JP
4160void linkwatch_init_dev(struct net_device *dev);
4161void linkwatch_fire_event(struct net_device *dev);
4162void linkwatch_forget_dev(struct net_device *dev);
1da177e4 4163
bea3348e
SH
4164/**
4165 * netif_carrier_ok - test if carrier present
4166 * @dev: network device
4167 *
4168 * Check if carrier is present on device
4169 */
4d29515f 4170static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
4171{
4172 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
4173}
4174
f629d208 4175unsigned long dev_trans_start(struct net_device *dev);
9d21493b 4176
f629d208 4177void __netdev_watchdog_up(struct net_device *dev);
1da177e4 4178
f629d208 4179void netif_carrier_on(struct net_device *dev);
1da177e4 4180
f629d208 4181void netif_carrier_off(struct net_device *dev);
1da177e4 4182
bea3348e
SH
4183/**
4184 * netif_dormant_on - mark device as dormant.
4185 * @dev: network device
4186 *
4187 * Mark device as dormant (as per RFC2863).
4188 *
4189 * The dormant state indicates that the relevant interface is not
4190 * actually in a condition to pass packets (i.e., it is not 'up') but is
4191 * in a "pending" state, waiting for some external event. For "on-
4192 * demand" interfaces, this new state identifies the situation where the
4193 * interface is waiting for events to place it in the up state.
bea3348e 4194 */
b00055aa
SR
4195static inline void netif_dormant_on(struct net_device *dev)
4196{
4197 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
4198 linkwatch_fire_event(dev);
4199}
4200
bea3348e
SH
4201/**
4202 * netif_dormant_off - set device as not dormant.
4203 * @dev: network device
4204 *
4205 * Device is not in dormant state.
4206 */
b00055aa
SR
4207static inline void netif_dormant_off(struct net_device *dev)
4208{
4209 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
4210 linkwatch_fire_event(dev);
4211}
4212
bea3348e 4213/**
8ecbc40a 4214 * netif_dormant - test if device is dormant
bea3348e
SH
4215 * @dev: network device
4216 *
8ecbc40a 4217 * Check if device is dormant.
bea3348e 4218 */
4d29515f 4219static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
4220{
4221 return test_bit(__LINK_STATE_DORMANT, &dev->state);
4222}
4223
4224
eec517cd
AL
4225/**
4226 * netif_testing_on - mark device as under test.
4227 * @dev: network device
4228 *
4229 * Mark device as under test (as per RFC2863).
4230 *
4231 * The testing state indicates that some test(s) must be performed on
4232 * the interface. After completion, of the test, the interface state
4233 * will change to up, dormant, or down, as appropriate.
4234 */
4235static inline void netif_testing_on(struct net_device *dev)
4236{
4237 if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
4238 linkwatch_fire_event(dev);
4239}
4240
4241/**
4242 * netif_testing_off - set device as not under test.
4243 * @dev: network device
4244 *
4245 * Device is not in testing state.
4246 */
4247static inline void netif_testing_off(struct net_device *dev)
4248{
4249 if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
4250 linkwatch_fire_event(dev);
4251}
4252
4253/**
4254 * netif_testing - test if device is under test
4255 * @dev: network device
4256 *
4257 * Check if device is under test
4258 */
4259static inline bool netif_testing(const struct net_device *dev)
4260{
4261 return test_bit(__LINK_STATE_TESTING, &dev->state);
4262}
4263
4264
bea3348e
SH
4265/**
4266 * netif_oper_up - test if device is operational
4267 * @dev: network device
4268 *
4269 * Check if carrier is operational
4270 */
4d29515f 4271static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 4272{
b00055aa
SR
4273 return (dev->operstate == IF_OPER_UP ||
4274 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
4275}
4276
bea3348e
SH
4277/**
4278 * netif_device_present - is device available or removed
4279 * @dev: network device
4280 *
4281 * Check if device has not been removed from system.
4282 */
7a126a43 4283static inline bool netif_device_present(const struct net_device *dev)
1da177e4
LT
4284{
4285 return test_bit(__LINK_STATE_PRESENT, &dev->state);
4286}
4287
f629d208 4288void netif_device_detach(struct net_device *dev);
1da177e4 4289
f629d208 4290void netif_device_attach(struct net_device *dev);
1da177e4
LT
4291
4292/*
4293 * Network interface message level settings
4294 */
1da177e4
LT
4295
4296enum {
6a94b8cc
MK
4297 NETIF_MSG_DRV_BIT,
4298 NETIF_MSG_PROBE_BIT,
4299 NETIF_MSG_LINK_BIT,
4300 NETIF_MSG_TIMER_BIT,
4301 NETIF_MSG_IFDOWN_BIT,
4302 NETIF_MSG_IFUP_BIT,
4303 NETIF_MSG_RX_ERR_BIT,
4304 NETIF_MSG_TX_ERR_BIT,
4305 NETIF_MSG_TX_QUEUED_BIT,
4306 NETIF_MSG_INTR_BIT,
4307 NETIF_MSG_TX_DONE_BIT,
4308 NETIF_MSG_RX_STATUS_BIT,
4309 NETIF_MSG_PKTDATA_BIT,
4310 NETIF_MSG_HW_BIT,
4311 NETIF_MSG_WOL_BIT,
4312
4313 /* When you add a new bit above, update netif_msg_class_names array
4314 * in net/ethtool/common.c
4315 */
4316 NETIF_MSG_CLASS_COUNT,
1da177e4 4317};
6a94b8cc
MK
4318/* Both ethtool_ops interface and internal driver implementation use u32 */
4319static_assert(NETIF_MSG_CLASS_COUNT <= 32);
4320
4321#define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
4322#define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
4323
4324#define NETIF_MSG_DRV __NETIF_MSG(DRV)
4325#define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
4326#define NETIF_MSG_LINK __NETIF_MSG(LINK)
4327#define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
4328#define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
4329#define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
4330#define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
4331#define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
4332#define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
4333#define NETIF_MSG_INTR __NETIF_MSG(INTR)
4334#define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
4335#define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
4336#define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
4337#define NETIF_MSG_HW __NETIF_MSG(HW)
4338#define NETIF_MSG_WOL __NETIF_MSG(WOL)
1da177e4
LT
4339
4340#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
4341#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
4342#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
4343#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
4344#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
4345#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
4346#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
4347#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
4348#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
4349#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
4350#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
4351#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
4352#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
4353#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
4354#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
4355
4356static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
4357{
4358 /* use default */
4359 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
4360 return default_msg_enable_bits;
4361 if (debug_value == 0) /* no output */
4362 return 0;
4363 /* set low N bits */
f4d7b3e2 4364 return (1U << debug_value) - 1;
1da177e4
LT
4365}
4366
c773e847 4367static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 4368{
c773e847
DM
4369 spin_lock(&txq->_xmit_lock);
4370 txq->xmit_lock_owner = cpu;
22dd7495
JHS
4371}
4372
5a717f4f
MT
4373static inline bool __netif_tx_acquire(struct netdev_queue *txq)
4374{
4375 __acquire(&txq->_xmit_lock);
4376 return true;
4377}
4378
4379static inline void __netif_tx_release(struct netdev_queue *txq)
4380{
4381 __release(&txq->_xmit_lock);
4382}
4383
fd2ea0a7
DM
4384static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
4385{
4386 spin_lock_bh(&txq->_xmit_lock);
4387 txq->xmit_lock_owner = smp_processor_id();
4388}
4389
4d29515f 4390static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 4391{
4d29515f 4392 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
4393 if (likely(ok))
4394 txq->xmit_lock_owner = smp_processor_id();
4395 return ok;
4396}
4397
4398static inline void __netif_tx_unlock(struct netdev_queue *txq)
4399{
4400 txq->xmit_lock_owner = -1;
4401 spin_unlock(&txq->_xmit_lock);
4402}
4403
4404static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
4405{
4406 txq->xmit_lock_owner = -1;
4407 spin_unlock_bh(&txq->_xmit_lock);
4408}
4409
08baf561
ED
4410static inline void txq_trans_update(struct netdev_queue *txq)
4411{
4412 if (txq->xmit_lock_owner != -1)
4413 txq->trans_start = jiffies;
4414}
4415
ba162f8e
FW
4416/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
4417static inline void netif_trans_update(struct net_device *dev)
4418{
9b36627a
FW
4419 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
4420
4421 if (txq->trans_start != jiffies)
4422 txq->trans_start = jiffies;
ba162f8e
FW
4423}
4424
d29f749e
DJ
4425/**
4426 * netif_tx_lock - grab network device transmit lock
4427 * @dev: network device
d29f749e
DJ
4428 *
4429 * Get network device transmit lock
4430 */
22dd7495
JHS
4431static inline void netif_tx_lock(struct net_device *dev)
4432{
e8a0464c 4433 unsigned int i;
c3f26a26 4434 int cpu;
c773e847 4435
c3f26a26
DM
4436 spin_lock(&dev->tx_global_lock);
4437 cpu = smp_processor_id();
e8a0464c
DM
4438 for (i = 0; i < dev->num_tx_queues; i++) {
4439 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4440
4441 /* We are the only thread of execution doing a
4442 * freeze, but we have to grab the _xmit_lock in
4443 * order to synchronize with threads which are in
4444 * the ->hard_start_xmit() handler and already
4445 * checked the frozen bit.
4446 */
e8a0464c 4447 __netif_tx_lock(txq, cpu);
c3f26a26
DM
4448 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
4449 __netif_tx_unlock(txq);
e8a0464c 4450 }
932ff279
HX
4451}
4452
4453static inline void netif_tx_lock_bh(struct net_device *dev)
4454{
e8a0464c
DM
4455 local_bh_disable();
4456 netif_tx_lock(dev);
932ff279
HX
4457}
4458
932ff279
HX
4459static inline void netif_tx_unlock(struct net_device *dev)
4460{
e8a0464c
DM
4461 unsigned int i;
4462
4463 for (i = 0; i < dev->num_tx_queues; i++) {
4464 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 4465
c3f26a26
DM
4466 /* No need to grab the _xmit_lock here. If the
4467 * queue is not stopped for another reason, we
4468 * force a schedule.
4469 */
4470 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 4471 netif_schedule_queue(txq);
c3f26a26
DM
4472 }
4473 spin_unlock(&dev->tx_global_lock);
932ff279
HX
4474}
4475
4476static inline void netif_tx_unlock_bh(struct net_device *dev)
4477{
e8a0464c
DM
4478 netif_tx_unlock(dev);
4479 local_bh_enable();
932ff279
HX
4480}
4481
c773e847 4482#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4483 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4484 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4485 } else { \
4486 __netif_tx_acquire(txq); \
22dd7495
JHS
4487 } \
4488}
4489
5efeac44
EB
4490#define HARD_TX_TRYLOCK(dev, txq) \
4491 (((dev->features & NETIF_F_LLTX) == 0) ? \
4492 __netif_tx_trylock(txq) : \
5a717f4f 4493 __netif_tx_acquire(txq))
5efeac44 4494
c773e847 4495#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4496 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4497 __netif_tx_unlock(txq); \
5a717f4f
MT
4498 } else { \
4499 __netif_tx_release(txq); \
22dd7495
JHS
4500 } \
4501}
4502
1da177e4
LT
4503static inline void netif_tx_disable(struct net_device *dev)
4504{
fd2ea0a7 4505 unsigned int i;
c3f26a26 4506 int cpu;
fd2ea0a7 4507
c3f26a26
DM
4508 local_bh_disable();
4509 cpu = smp_processor_id();
3aa6bce9 4510 spin_lock(&dev->tx_global_lock);
fd2ea0a7
DM
4511 for (i = 0; i < dev->num_tx_queues; i++) {
4512 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4513
4514 __netif_tx_lock(txq, cpu);
fd2ea0a7 4515 netif_tx_stop_queue(txq);
c3f26a26 4516 __netif_tx_unlock(txq);
fd2ea0a7 4517 }
3aa6bce9 4518 spin_unlock(&dev->tx_global_lock);
c3f26a26 4519 local_bh_enable();
1da177e4
LT
4520}
4521
e308a5d8
DM
4522static inline void netif_addr_lock(struct net_device *dev)
4523{
1fc70edb 4524 unsigned char nest_level = 0;
e308a5d8 4525
1fc70edb
TY
4526#ifdef CONFIG_LOCKDEP
4527 nest_level = dev->nested_level;
4528#endif
4529 spin_lock_nested(&dev->addr_list_lock, nest_level);
845e0ebb
CW
4530}
4531
e308a5d8
DM
4532static inline void netif_addr_lock_bh(struct net_device *dev)
4533{
1fc70edb
TY
4534 unsigned char nest_level = 0;
4535
4536#ifdef CONFIG_LOCKDEP
4537 nest_level = dev->nested_level;
4538#endif
4539 local_bh_disable();
4540 spin_lock_nested(&dev->addr_list_lock, nest_level);
e308a5d8
DM
4541}
4542
4543static inline void netif_addr_unlock(struct net_device *dev)
4544{
4545 spin_unlock(&dev->addr_list_lock);
4546}
4547
4548static inline void netif_addr_unlock_bh(struct net_device *dev)
4549{
4550 spin_unlock_bh(&dev->addr_list_lock);
4551}
4552
f001fde5 4553/*
31278e71 4554 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4555 * rcu_read_lock held.
4556 */
4557#define for_each_dev_addr(dev, ha) \
31278e71 4558 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4559
1da177e4
LT
4560/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4561
f629d208 4562void ether_setup(struct net_device *dev);
1da177e4
LT
4563
4564/* Support for loadable net-drivers */
f629d208 4565struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4566 unsigned char name_assign_type,
f629d208
JP
4567 void (*setup)(struct net_device *),
4568 unsigned int txqs, unsigned int rxqs);
c835a677
TG
4569#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4570 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4571
c835a677
TG
4572#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4573 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4574 count)
36909ea4 4575
f629d208
JP
4576int register_netdev(struct net_device *dev);
4577void unregister_netdev(struct net_device *dev);
f001fde5 4578
cd16627f
BG
4579int devm_register_netdev(struct device *dev, struct net_device *ndev);
4580
22bedad3 4581/* General hardware address lists handling functions */
f629d208
JP
4582int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4583 struct netdev_hw_addr_list *from_list, int addr_len);
4584void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4585 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4586int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4587 struct net_device *dev,
4588 int (*sync)(struct net_device *, const unsigned char *),
4589 int (*unsync)(struct net_device *,
4590 const unsigned char *));
e7946760
IK
4591int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4592 struct net_device *dev,
4593 int (*sync)(struct net_device *,
4594 const unsigned char *, int),
4595 int (*unsync)(struct net_device *,
4596 const unsigned char *, int));
4597void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4598 struct net_device *dev,
4599 int (*unsync)(struct net_device *,
4600 const unsigned char *, int));
670e5b8e
AD
4601void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4602 struct net_device *dev,
4603 int (*unsync)(struct net_device *,
4604 const unsigned char *));
f629d208 4605void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4606
f001fde5 4607/* Functions used for device addresses handling */
f629d208
JP
4608int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4609 unsigned char addr_type);
4610int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4611 unsigned char addr_type);
f629d208
JP
4612void dev_addr_flush(struct net_device *dev);
4613int dev_addr_init(struct net_device *dev);
a748ee24
JP
4614
4615/* Functions used for unicast addresses handling */
f629d208
JP
4616int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4617int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4618int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4619int dev_uc_sync(struct net_device *to, struct net_device *from);
4620int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4621void dev_uc_unsync(struct net_device *to, struct net_device *from);
4622void dev_uc_flush(struct net_device *dev);
4623void dev_uc_init(struct net_device *dev);
f001fde5 4624
670e5b8e
AD
4625/**
4626 * __dev_uc_sync - Synchonize device's unicast list
4627 * @dev: device to sync
4628 * @sync: function to call if address should be added
4629 * @unsync: function to call if address should be removed
4630 *
4631 * Add newly added addresses to the interface, and release
4632 * addresses that have been deleted.
5e82b4b2 4633 */
670e5b8e
AD
4634static inline int __dev_uc_sync(struct net_device *dev,
4635 int (*sync)(struct net_device *,
4636 const unsigned char *),
4637 int (*unsync)(struct net_device *,
4638 const unsigned char *))
4639{
4640 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4641}
4642
4643/**
e793c0f7 4644 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4645 * @dev: device to sync
4646 * @unsync: function to call if address should be removed
4647 *
4648 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4649 */
670e5b8e
AD
4650static inline void __dev_uc_unsync(struct net_device *dev,
4651 int (*unsync)(struct net_device *,
4652 const unsigned char *))
4653{
4654 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4655}
4656
22bedad3 4657/* Functions used for multicast addresses handling */
f629d208
JP
4658int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4659int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4660int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4661int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4662int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4663int dev_mc_sync(struct net_device *to, struct net_device *from);
4664int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4665void dev_mc_unsync(struct net_device *to, struct net_device *from);
4666void dev_mc_flush(struct net_device *dev);
4667void dev_mc_init(struct net_device *dev);
f001fde5 4668
670e5b8e
AD
4669/**
4670 * __dev_mc_sync - Synchonize device's multicast list
4671 * @dev: device to sync
4672 * @sync: function to call if address should be added
4673 * @unsync: function to call if address should be removed
4674 *
4675 * Add newly added addresses to the interface, and release
4676 * addresses that have been deleted.
5e82b4b2 4677 */
670e5b8e
AD
4678static inline int __dev_mc_sync(struct net_device *dev,
4679 int (*sync)(struct net_device *,
4680 const unsigned char *),
4681 int (*unsync)(struct net_device *,
4682 const unsigned char *))
4683{
4684 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4685}
4686
4687/**
e793c0f7 4688 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4689 * @dev: device to sync
4690 * @unsync: function to call if address should be removed
4691 *
4692 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4693 */
670e5b8e
AD
4694static inline void __dev_mc_unsync(struct net_device *dev,
4695 int (*unsync)(struct net_device *,
4696 const unsigned char *))
4697{
4698 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4699}
4700
4417da66 4701/* Functions used for secondary unicast and multicast support */
f629d208
JP
4702void dev_set_rx_mode(struct net_device *dev);
4703void __dev_set_rx_mode(struct net_device *dev);
4704int dev_set_promiscuity(struct net_device *dev, int inc);
4705int dev_set_allmulti(struct net_device *dev, int inc);
4706void netdev_state_change(struct net_device *dev);
7061eb8c 4707void __netdev_notify_peers(struct net_device *dev);
f629d208
JP
4708void netdev_notify_peers(struct net_device *dev);
4709void netdev_features_change(struct net_device *dev);
1da177e4 4710/* Load a device via the kmod */
f629d208
JP
4711void dev_load(struct net *net, const char *name);
4712struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4713 struct rtnl_link_stats64 *storage);
4714void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4715 const struct net_device_stats *netdev_stats);
44fa32f0
HK
4716void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
4717 const struct pcpu_sw_netstats __percpu *netstats);
a1839426 4718void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
eeda3fd6 4719
1da177e4 4720extern int netdev_max_backlog;
3b098e2d 4721extern int netdev_tstamp_prequeue;
5aa3afe1 4722extern int netdev_unregister_timeout_secs;
1da177e4 4723extern int weight_p;
3d48b53f
MT
4724extern int dev_weight_rx_bias;
4725extern int dev_weight_tx_bias;
4726extern int dev_rx_weight;
4727extern int dev_tx_weight;
323ebb61 4728extern int gro_normal_batch;
9ff162a8 4729
1fc70edb
TY
4730enum {
4731 NESTED_SYNC_IMM_BIT,
4732 NESTED_SYNC_TODO_BIT,
4733};
4734
4735#define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
4736#define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
4737
4738#define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
4739#define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
4740
eff74233 4741struct netdev_nested_priv {
1fc70edb 4742 unsigned char flags;
eff74233
TY
4743 void *data;
4744};
4745
f629d208 4746bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4747struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4748 struct list_head **iter);
f629d208
JP
4749struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
4750 struct list_head **iter);
8b5be856 4751
1fc70edb
TY
4752#ifdef CONFIG_LOCKDEP
4753static LIST_HEAD(net_unlink_list);
4754
4755static inline void net_unlink_todo(struct net_device *dev)
4756{
4757 if (list_empty(&dev->unlink_list))
4758 list_add_tail(&dev->unlink_list, &net_unlink_list);
4759}
4760#endif
4761
44a40855
VY
4762/* iterate through upper list, must be called under RCU read lock */
4763#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4764 for (iter = &(dev)->adj_list.upper, \
4765 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4766 updev; \
4767 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4768
1a3f060c
DA
4769int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4770 int (*fn)(struct net_device *upper_dev,
eff74233
TY
4771 struct netdev_nested_priv *priv),
4772 struct netdev_nested_priv *priv);
1a3f060c
DA
4773
4774bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4775 struct net_device *upper_dev);
4776
25cc72a3
IS
4777bool netdev_has_any_upper_dev(struct net_device *dev);
4778
f629d208
JP
4779void *netdev_lower_get_next_private(struct net_device *dev,
4780 struct list_head **iter);
4781void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4782 struct list_head **iter);
31088a11
VF
4783
4784#define netdev_for_each_lower_private(dev, priv, iter) \
4785 for (iter = (dev)->adj_list.lower.next, \
4786 priv = netdev_lower_get_next_private(dev, &(iter)); \
4787 priv; \
4788 priv = netdev_lower_get_next_private(dev, &(iter)))
4789
4790#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4791 for (iter = &(dev)->adj_list.lower, \
4792 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4793 priv; \
4794 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4795
4085ebe8
VY
4796void *netdev_lower_get_next(struct net_device *dev,
4797 struct list_head **iter);
7ce856aa 4798
4085ebe8 4799#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4800 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4801 ldev = netdev_lower_get_next(dev, &(iter)); \
4802 ldev; \
4803 ldev = netdev_lower_get_next(dev, &(iter)))
4804
7151affe 4805struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
7ce856aa 4806 struct list_head **iter);
1a3f060c
DA
4807int netdev_walk_all_lower_dev(struct net_device *dev,
4808 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4809 struct netdev_nested_priv *priv),
4810 struct netdev_nested_priv *priv);
1a3f060c
DA
4811int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4812 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4813 struct netdev_nested_priv *priv),
4814 struct netdev_nested_priv *priv);
1a3f060c 4815
f629d208 4816void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4817void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4818struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4819struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4820int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4821 struct netlink_ext_ack *extack);
f629d208 4822int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4823 struct net_device *upper_dev,
42ab19ee
DA
4824 void *upper_priv, void *upper_info,
4825 struct netlink_ext_ack *extack);
f629d208
JP
4826void netdev_upper_dev_unlink(struct net_device *dev,
4827 struct net_device *upper_dev);
32b6d34f
TY
4828int netdev_adjacent_change_prepare(struct net_device *old_dev,
4829 struct net_device *new_dev,
4830 struct net_device *dev,
4831 struct netlink_ext_ack *extack);
4832void netdev_adjacent_change_commit(struct net_device *old_dev,
4833 struct net_device *new_dev,
4834 struct net_device *dev);
4835void netdev_adjacent_change_abort(struct net_device *old_dev,
4836 struct net_device *new_dev,
4837 struct net_device *dev);
5bb025fa 4838void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4839void *netdev_lower_dev_get_private(struct net_device *dev,
4840 struct net_device *lower_dev);
04d48266
JP
4841void netdev_lower_state_changed(struct net_device *lower_dev,
4842 void *lower_state_info);
960fb622
ED
4843
4844/* RSS keys are 40 or 52 bytes long */
4845#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4846extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4847void netdev_rss_key_fill(void *buffer, size_t len);
4848
f629d208 4849int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4850int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4851int skb_csum_hwoffload_help(struct sk_buff *skb,
4852 const netdev_features_t features);
4853
f629d208
JP
4854struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4855 netdev_features_t features, bool tx_path);
4856struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4857 netdev_features_t features);
12b0004d 4858
61bd3857
MS
4859struct netdev_bonding_info {
4860 ifslave slave;
4861 ifbond master;
4862};
4863
4864struct netdev_notifier_bonding_info {
4865 struct netdev_notifier_info info; /* must be first */
4866 struct netdev_bonding_info bonding_info;
4867};
4868
4869void netdev_bonding_info_change(struct net_device *dev,
4870 struct netdev_bonding_info *bonding_info);
4871
6b08d6c1
MK
4872#if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
4873void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
4874#else
4875static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
4876 const void *data)
4877{
4878}
4879#endif
4880
12b0004d
CW
4881static inline
4882struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4883{
4884 return __skb_gso_segment(skb, features, true);
4885}
53d6471c 4886__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4887
4888static inline bool can_checksum_protocol(netdev_features_t features,
4889 __be16 protocol)
4890{
c8cd0989
TH
4891 if (protocol == htons(ETH_P_FCOE))
4892 return !!(features & NETIF_F_FCOE_CRC);
4893
4894 /* Assume this is an IP checksum (not SCTP CRC) */
4895
4896 if (features & NETIF_F_HW_CSUM) {
4897 /* Can checksum everything */
4898 return true;
4899 }
4900
4901 switch (protocol) {
4902 case htons(ETH_P_IP):
4903 return !!(features & NETIF_F_IP_CSUM);
4904 case htons(ETH_P_IPV6):
4905 return !!(features & NETIF_F_IPV6_CSUM);
4906 default:
4907 return false;
4908 }
ec5f0615 4909}
12b0004d 4910
fb286bb2 4911#ifdef CONFIG_BUG
7fe50ac8 4912void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4913#else
7fe50ac8
CW
4914static inline void netdev_rx_csum_fault(struct net_device *dev,
4915 struct sk_buff *skb)
fb286bb2
HX
4916{
4917}
4918#endif
1da177e4 4919/* rx skb timestamps */
f629d208
JP
4920void net_enable_timestamp(void);
4921void net_disable_timestamp(void);
1da177e4 4922
20380731 4923#ifdef CONFIG_PROC_FS
f629d208 4924int __init dev_proc_init(void);
900ff8c6
CW
4925#else
4926#define dev_proc_init() 0
20380731
ACM
4927#endif
4928
4798248e 4929static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4930 struct sk_buff *skb, struct net_device *dev,
4931 bool more)
4798248e 4932{
6b16f9ee 4933 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4934 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4935}
4936
97cdcf37
FW
4937static inline bool netdev_xmit_more(void)
4938{
4939 return __this_cpu_read(softnet_data.xmit.more);
4940}
4941
10b3ad8c 4942static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4943 struct netdev_queue *txq, bool more)
4798248e
DM
4944{
4945 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4946 netdev_tx_t rc;
4798248e 4947
fa2dbdc2 4948 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4949 if (rc == NETDEV_TX_OK)
4950 txq_trans_update(txq);
4951
4952 return rc;
4798248e
DM
4953}
4954
b793dc5c 4955int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4956 const void *ns);
b793dc5c 4957void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4958 const void *ns);
58292cbe 4959
737aec57 4960extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4961
f629d208 4962const char *netdev_drivername(const struct net_device *dev);
6579e57b 4963
f629d208 4964void linkwatch_run_queue(void);
20380731 4965
da08143b
MK
4966static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4967 netdev_features_t f2)
4968{
c8cd0989
TH
4969 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4970 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4971 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4972 else
b6a0e72a 4973 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4974 }
da08143b 4975
c8cd0989 4976 return f1 & f2;
da08143b
MK
4977}
4978
c8f44aff
MM
4979static inline netdev_features_t netdev_get_wanted_features(
4980 struct net_device *dev)
5455c699
MM
4981{
4982 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4983}
c8f44aff
MM
4984netdev_features_t netdev_increment_features(netdev_features_t all,
4985 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4986
4987/* Allow TSO being used on stacked device :
4988 * Performing the GSO segmentation before last device
4989 * is a performance improvement.
4990 */
4991static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4992 netdev_features_t mask)
4993{
4994 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4995}
4996
6cb6a27c 4997int __netdev_update_features(struct net_device *dev);
5455c699 4998void netdev_update_features(struct net_device *dev);
afe12cc8 4999void netdev_change_features(struct net_device *dev);
7f353bf2 5000
fc4a7489
PM
5001void netif_stacked_transfer_operstate(const struct net_device *rootdev,
5002 struct net_device *dev);
5003
e38f3025
TM
5004netdev_features_t passthru_features_check(struct sk_buff *skb,
5005 struct net_device *dev,
5006 netdev_features_t features);
c1e756bf 5007netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 5008
4d29515f 5009static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 5010{
7b748340 5011 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
5012
5013 /* check flags correspondence */
5014 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
5015 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
5016 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 5017 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
5018 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
5019 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
5020 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
5021 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
5022 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
5023 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
5024 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
5025 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 5026 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 5027 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 5028 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 5029 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 5030 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 5031 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
3b335832 5032 BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
0345e186 5033
d6b4991a 5034 return (features & feature) == feature;
576a30eb
HX
5035}
5036
4d29515f 5037static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 5038{
278b2513 5039 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 5040 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
5041}
5042
8b86a61d 5043static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 5044 netdev_features_t features)
7967168c 5045{
fc741216 5046 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
5047 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
5048 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
5049}
5050
82cc1a7a
PWJ
5051static inline void netif_set_gso_max_size(struct net_device *dev,
5052 unsigned int size)
5053{
5054 dev->gso_max_size = size;
5055}
5056
7a7ffbab
WCC
5057static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
5058 int pulled_hlen, u16 mac_offset,
5059 int mac_len)
5060{
5061 skb->protocol = protocol;
5062 skb->encapsulation = 1;
5063 skb_push(skb, pulled_hlen);
5064 skb_reset_transport_header(skb);
5065 skb->mac_header = mac_offset;
5066 skb->network_header = skb->mac_header + mac_len;
5067 skb->mac_len = mac_len;
5068}
5069
3c175784
SD
5070static inline bool netif_is_macsec(const struct net_device *dev)
5071{
5072 return dev->priv_flags & IFF_MACSEC;
5073}
5074
b618aaa9 5075static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
5076{
5077 return dev->priv_flags & IFF_MACVLAN;
5078}
5079
b618aaa9 5080static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
5081{
5082 return dev->priv_flags & IFF_MACVLAN_PORT;
5083}
5084
b618aaa9 5085static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 5086{
5087 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
5088}
5089
b618aaa9 5090static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
5091{
5092 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
5093}
5094
3bdc0eba
BG
5095static inline bool netif_supports_nofcs(struct net_device *dev)
5096{
5097 return dev->priv_flags & IFF_SUPP_NOFCS;
5098}
5099
d5256083
DB
5100static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
5101{
5102 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
5103}
5104
007979ea 5105static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 5106{
007979ea 5107 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
5108}
5109
fee6d4c7
DA
5110static inline bool netif_is_l3_slave(const struct net_device *dev)
5111{
5112 return dev->priv_flags & IFF_L3MDEV_SLAVE;
5113}
5114
0894ae3f
JP
5115static inline bool netif_is_bridge_master(const struct net_device *dev)
5116{
5117 return dev->priv_flags & IFF_EBRIDGE;
5118}
5119
28f9ee22
VY
5120static inline bool netif_is_bridge_port(const struct net_device *dev)
5121{
5122 return dev->priv_flags & IFF_BRIDGE_PORT;
5123}
5124
35d4e172
JP
5125static inline bool netif_is_ovs_master(const struct net_device *dev)
5126{
5127 return dev->priv_flags & IFF_OPENVSWITCH;
5128}
5129
5be66141
JP
5130static inline bool netif_is_ovs_port(const struct net_device *dev)
5131{
5132 return dev->priv_flags & IFF_OVS_DATAPATH;
5133}
5134
df23bb18
SB
5135static inline bool netif_is_any_bridge_port(const struct net_device *dev)
5136{
5137 return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
5138}
5139
b618aaa9 5140static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
5141{
5142 return dev->priv_flags & IFF_TEAM;
5143}
5144
b618aaa9 5145static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
5146{
5147 return dev->priv_flags & IFF_TEAM_PORT;
5148}
5149
b618aaa9 5150static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
5151{
5152 return netif_is_bond_master(dev) || netif_is_team_master(dev);
5153}
5154
b618aaa9 5155static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
5156{
5157 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
5158}
5159
d4ab4286
KJ
5160static inline bool netif_is_rxfh_configured(const struct net_device *dev)
5161{
5162 return dev->priv_flags & IFF_RXFH_CONFIGURED;
5163}
5164
30c8bd5a
SS
5165static inline bool netif_is_failover(const struct net_device *dev)
5166{
5167 return dev->priv_flags & IFF_FAILOVER;
5168}
5169
5170static inline bool netif_is_failover_slave(const struct net_device *dev)
5171{
5172 return dev->priv_flags & IFF_FAILOVER_SLAVE;
5173}
5174
02875878
ED
5175/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
5176static inline void netif_keep_dst(struct net_device *dev)
5177{
5178 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
5179}
5180
18d3df3e
PA
5181/* return true if dev can't cope with mtu frames that need vlan tag insertion */
5182static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
5183{
5184 /* TODO: reserve and use an additional IFF bit, if we get more users */
5185 return dev->priv_flags & IFF_MACSEC;
5186}
5187
505d4f73 5188extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 5189
571ba423
JP
5190/* Logging, debugging and troubleshooting/diagnostic helpers. */
5191
5192/* netdev_printk helpers, similar to dev_printk */
5193
5194static inline const char *netdev_name(const struct net_device *dev)
5195{
c6f854d5
VF
5196 if (!dev->name[0] || strchr(dev->name, '%'))
5197 return "(unnamed net_device)";
571ba423
JP
5198 return dev->name;
5199}
5200
8397ed36
DA
5201static inline bool netdev_unregistering(const struct net_device *dev)
5202{
5203 return dev->reg_state == NETREG_UNREGISTERING;
5204}
5205
ccc7f496
VF
5206static inline const char *netdev_reg_state(const struct net_device *dev)
5207{
5208 switch (dev->reg_state) {
5209 case NETREG_UNINITIALIZED: return " (uninitialized)";
5210 case NETREG_REGISTERED: return "";
5211 case NETREG_UNREGISTERING: return " (unregistering)";
5212 case NETREG_UNREGISTERED: return " (unregistered)";
5213 case NETREG_RELEASED: return " (released)";
5214 case NETREG_DUMMY: return " (dummy)";
5215 }
5216
5217 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
5218 return " (unknown)";
5219}
5220
ce3fdb69 5221__printf(3, 4) __cold
6ea754eb
JP
5222void netdev_printk(const char *level, const struct net_device *dev,
5223 const char *format, ...);
ce3fdb69 5224__printf(2, 3) __cold
6ea754eb 5225void netdev_emerg(const struct net_device *dev, const char *format, ...);
ce3fdb69 5226__printf(2, 3) __cold
6ea754eb 5227void netdev_alert(const struct net_device *dev, const char *format, ...);
ce3fdb69 5228__printf(2, 3) __cold
6ea754eb 5229void netdev_crit(const struct net_device *dev, const char *format, ...);
ce3fdb69 5230__printf(2, 3) __cold
6ea754eb 5231void netdev_err(const struct net_device *dev, const char *format, ...);
ce3fdb69 5232__printf(2, 3) __cold
6ea754eb 5233void netdev_warn(const struct net_device *dev, const char *format, ...);
ce3fdb69 5234__printf(2, 3) __cold
6ea754eb 5235void netdev_notice(const struct net_device *dev, const char *format, ...);
ce3fdb69 5236__printf(2, 3) __cold
6ea754eb 5237void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 5238
375ef2b1
GP
5239#define netdev_level_once(level, dev, fmt, ...) \
5240do { \
5241 static bool __print_once __read_mostly; \
5242 \
5243 if (!__print_once) { \
5244 __print_once = true; \
5245 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
5246 } \
5247} while (0)
5248
5249#define netdev_emerg_once(dev, fmt, ...) \
5250 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
5251#define netdev_alert_once(dev, fmt, ...) \
5252 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
5253#define netdev_crit_once(dev, fmt, ...) \
5254 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
5255#define netdev_err_once(dev, fmt, ...) \
5256 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
5257#define netdev_warn_once(dev, fmt, ...) \
5258 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
5259#define netdev_notice_once(dev, fmt, ...) \
5260 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
5261#define netdev_info_once(dev, fmt, ...) \
5262 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
5263
8909c9ad
VK
5264#define MODULE_ALIAS_NETDEV(device) \
5265 MODULE_ALIAS("netdev-" device)
5266
ceabef7d
OZ
5267#if defined(CONFIG_DYNAMIC_DEBUG) || \
5268 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
571ba423
JP
5269#define netdev_dbg(__dev, format, args...) \
5270do { \
ffa10cb4 5271 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 5272} while (0)
b558c96f
JC
5273#elif defined(DEBUG)
5274#define netdev_dbg(__dev, format, args...) \
5275 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
5276#else
5277#define netdev_dbg(__dev, format, args...) \
5278({ \
5279 if (0) \
5280 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
5281})
5282#endif
5283
5284#if defined(VERBOSE_DEBUG)
5285#define netdev_vdbg netdev_dbg
5286#else
5287
5288#define netdev_vdbg(dev, format, args...) \
5289({ \
5290 if (0) \
5291 netdev_printk(KERN_DEBUG, dev, format, ##args); \
5292 0; \
5293})
5294#endif
5295
5296/*
5297 * netdev_WARN() acts like dev_printk(), but with the key difference
5298 * of using a WARN/WARN_ON to get the message out, including the
5299 * file/line information and a backtrace.
5300 */
5301#define netdev_WARN(dev, format, args...) \
e1cfe3d0 5302 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 5303 netdev_reg_state(dev), ##args)
571ba423 5304
72dd831e 5305#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 5306 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
5307 netdev_reg_state(dev), ##args)
5308
b3d95c5c
JP
5309/* netif printk helpers, similar to netdev_printk */
5310
5311#define netif_printk(priv, type, level, dev, fmt, args...) \
5312do { \
5313 if (netif_msg_##type(priv)) \
5314 netdev_printk(level, (dev), fmt, ##args); \
5315} while (0)
5316
f45f4321
JP
5317#define netif_level(level, priv, type, dev, fmt, args...) \
5318do { \
5319 if (netif_msg_##type(priv)) \
5320 netdev_##level(dev, fmt, ##args); \
5321} while (0)
5322
b3d95c5c 5323#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 5324 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 5325#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 5326 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 5327#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 5328 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 5329#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 5330 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 5331#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 5332 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 5333#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 5334 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 5335#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 5336 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 5337
ceabef7d
OZ
5338#if defined(CONFIG_DYNAMIC_DEBUG) || \
5339 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
b3d95c5c
JP
5340#define netif_dbg(priv, type, netdev, format, args...) \
5341do { \
5342 if (netif_msg_##type(priv)) \
b5fb0a03 5343 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 5344} while (0)
0053ea9c
JP
5345#elif defined(DEBUG)
5346#define netif_dbg(priv, type, dev, format, args...) \
5347 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
5348#else
5349#define netif_dbg(priv, type, dev, format, args...) \
5350({ \
5351 if (0) \
5352 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
5353 0; \
5354})
5355#endif
5356
f617f276
EC
5357/* if @cond then downgrade to debug, else print at @level */
5358#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
5359 do { \
5360 if (cond) \
5361 netif_dbg(priv, type, netdev, fmt, ##args); \
5362 else \
5363 netif_ ## level(priv, type, netdev, fmt, ##args); \
5364 } while (0)
5365
b3d95c5c 5366#if defined(VERBOSE_DEBUG)
bcfcc450 5367#define netif_vdbg netif_dbg
b3d95c5c
JP
5368#else
5369#define netif_vdbg(priv, type, dev, format, args...) \
5370({ \
5371 if (0) \
a4ed89cb 5372 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
5373 0; \
5374})
5375#endif
571ba423 5376
900ff8c6
CW
5377/*
5378 * The list of packet types we will receive (as opposed to discard)
5379 * and the routines to invoke.
5380 *
5381 * Why 16. Because with 16 the only overlap we get on a hash of the
5382 * low nibble of the protocol value is RARP/SNAP/X.25.
5383 *
900ff8c6 5384 * 0800 IP
900ff8c6
CW
5385 * 0001 802.3
5386 * 0002 AX.25
5387 * 0004 802.2
5388 * 8035 RARP
5389 * 0005 SNAP
5390 * 0805 X.25
5391 * 0806 ARP
5392 * 8137 IPX
5393 * 0009 Localtalk
5394 * 86DD IPv6
5395 */
5396#define PTYPE_HASH_SIZE (16)
5397#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
5398
744b8376
VO
5399extern struct list_head ptype_all __read_mostly;
5400extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
5401
4de83b88
MB
5402extern struct net_device *blackhole_netdev;
5403
385a154c 5404#endif /* _LINUX_NETDEVICE_H */