]> git.proxmox.com Git - mirror_ubuntu-jammy-kernel.git/blame - include/linux/netdevice.h
Merge tag 'kernel.sys.v5.15' of git://git.kernel.org/pub/scm/linux/kernel/git/brauner...
[mirror_ubuntu-jammy-kernel.git] / include / linux / netdevice.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the Interfaces handler.
8 *
9 * Version: @(#)dev.h 1.0.10 08/12/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Corey Minyard <wf-rch!minyard@relay.EU.net>
14 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
1da177e4
LT
19 * Moved to /usr/include/linux for NET3
20 */
21#ifndef _LINUX_NETDEVICE_H
22#define _LINUX_NETDEVICE_H
23
d7fe0f24 24#include <linux/timer.h>
187f1882 25#include <linux/bug.h>
bea3348e 26#include <linux/delay.h>
60063497 27#include <linux/atomic.h>
53511453 28#include <linux/prefetch.h>
1da177e4
LT
29#include <asm/cache.h>
30#include <asm/byteorder.h>
31
1da177e4 32#include <linux/percpu.h>
4d5b78c0 33#include <linux/rculist.h>
bea3348e 34#include <linux/workqueue.h>
114cf580 35#include <linux/dynamic_queue_limits.h>
1da177e4 36
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*/
cb038357 363 NAPI_STATE_SCHED_THREADED, /* Napi is currently scheduled in threaded mode */
217f6974
ED
364};
365
366enum {
7fd3253a
BT
367 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
368 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
369 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
370 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
371 NAPIF_STATE_LISTED = BIT(NAPI_STATE_LISTED),
372 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
373 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
374 NAPIF_STATE_PREFER_BUSY_POLL = BIT(NAPI_STATE_PREFER_BUSY_POLL),
29863d41 375 NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
cb038357 376 NAPIF_STATE_SCHED_THREADED = BIT(NAPI_STATE_SCHED_THREADED),
bea3348e
SH
377};
378
5b252f0c 379enum gro_result {
d1c76af9
HX
380 GRO_MERGED,
381 GRO_MERGED_FREE,
382 GRO_HELD,
383 GRO_NORMAL,
25393d3f 384 GRO_CONSUMED,
d1c76af9 385};
5b252f0c 386typedef enum gro_result gro_result_t;
d1c76af9 387
8a4eb573
JP
388/*
389 * enum rx_handler_result - Possible return values for rx_handlers.
390 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
391 * further.
392 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
393 * case skb->dev was changed by rx_handler.
394 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 395 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
396 *
397 * rx_handlers are functions called from inside __netif_receive_skb(), to do
398 * special processing of the skb, prior to delivery to protocol handlers.
399 *
400 * Currently, a net_device can only have a single rx_handler registered. Trying
401 * to register a second rx_handler will return -EBUSY.
402 *
403 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
404 * To unregister a rx_handler on a net_device, use
405 * netdev_rx_handler_unregister().
406 *
407 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
408 * do with the skb.
409 *
5e82b4b2 410 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 411 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 412 * the skb to be delivered in some other way.
8a4eb573
JP
413 *
414 * If the rx_handler changed skb->dev, to divert the skb to another
415 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
416 * new device will be called if it exists.
417 *
5e82b4b2 418 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 419 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 420 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 421 *
5e82b4b2 422 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
423 * delivered, it should return RX_HANDLER_PASS.
424 *
425 * A device without a registered rx_handler will behave as if rx_handler
426 * returned RX_HANDLER_PASS.
427 */
428
429enum rx_handler_result {
430 RX_HANDLER_CONSUMED,
431 RX_HANDLER_ANOTHER,
432 RX_HANDLER_EXACT,
433 RX_HANDLER_PASS,
434};
435typedef enum rx_handler_result rx_handler_result_t;
436typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 437
f629d208 438void __napi_schedule(struct napi_struct *n);
bc9ad166 439void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 440
4d29515f 441static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
442{
443 return test_bit(NAPI_STATE_DISABLE, &n->state);
444}
445
7fd3253a
BT
446static inline bool napi_prefer_busy_poll(struct napi_struct *n)
447{
448 return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
449}
450
39e6c820 451bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
452
453/**
454 * napi_schedule - schedule NAPI poll
5e82b4b2 455 * @n: NAPI context
bea3348e
SH
456 *
457 * Schedule NAPI poll routine to be called if it is not already
458 * running.
459 */
460static inline void napi_schedule(struct napi_struct *n)
461{
462 if (napi_schedule_prep(n))
463 __napi_schedule(n);
464}
465
bc9ad166
ED
466/**
467 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 468 * @n: NAPI context
bc9ad166
ED
469 *
470 * Variant of napi_schedule(), assuming hard irqs are masked.
471 */
472static inline void napi_schedule_irqoff(struct napi_struct *n)
473{
474 if (napi_schedule_prep(n))
475 __napi_schedule_irqoff(n);
476}
477
bfe13f54 478/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 479static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
480{
481 if (napi_schedule_prep(napi)) {
482 __napi_schedule(napi);
4d29515f 483 return true;
bfe13f54 484 }
4d29515f 485 return false;
bfe13f54
RD
486}
487
364b6055 488bool napi_complete_done(struct napi_struct *n, int work_done);
bea3348e
SH
489/**
490 * napi_complete - NAPI processing complete
5e82b4b2 491 * @n: NAPI context
bea3348e
SH
492 *
493 * Mark NAPI processing as complete.
3b47d303 494 * Consider using napi_complete_done() instead.
364b6055 495 * Return false if device should avoid rearming interrupts.
bea3348e 496 */
364b6055 497static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
498{
499 return napi_complete_done(n, 0);
500}
bea3348e 501
5fdd2f0e
WW
502int dev_set_threaded(struct net_device *dev, bool threaded);
503
bea3348e
SH
504/**
505 * napi_disable - prevent NAPI from scheduling
5e82b4b2 506 * @n: NAPI context
bea3348e
SH
507 *
508 * Stop NAPI from being scheduled on this context.
509 * Waits till any outstanding processing completes.
510 */
3b47d303 511void napi_disable(struct napi_struct *n);
bea3348e 512
29863d41 513void napi_enable(struct napi_struct *n);
bea3348e 514
c264c3de
SH
515/**
516 * napi_synchronize - wait until NAPI is not running
5e82b4b2 517 * @n: NAPI context
c264c3de
SH
518 *
519 * Wait until NAPI is done being scheduled on this context.
520 * Waits till any outstanding processing completes but
521 * does not disable future activations.
522 */
523static inline void napi_synchronize(const struct napi_struct *n)
524{
facc432f
AB
525 if (IS_ENABLED(CONFIG_SMP))
526 while (test_bit(NAPI_STATE_SCHED, &n->state))
527 msleep(1);
528 else
529 barrier();
c264c3de 530}
c264c3de 531
6c5c9581
MK
532/**
533 * napi_if_scheduled_mark_missed - if napi is running, set the
534 * NAPIF_STATE_MISSED
535 * @n: NAPI context
536 *
537 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
538 * NAPI is scheduled.
539 **/
540static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
541{
542 unsigned long val, new;
543
544 do {
545 val = READ_ONCE(n->state);
546 if (val & NAPIF_STATE_DISABLE)
547 return true;
548
549 if (!(val & NAPIF_STATE_SCHED))
550 return false;
551
552 new = val | NAPIF_STATE_MISSED;
553 } while (cmpxchg(&n->state, val, new) != val);
554
555 return true;
556}
557
d94d9fee 558enum netdev_queue_state_t {
73466498
TH
559 __QUEUE_STATE_DRV_XOFF,
560 __QUEUE_STATE_STACK_XOFF,
c3f26a26 561 __QUEUE_STATE_FROZEN,
79d16385 562};
8e2f1a63
DB
563
564#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
565#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
566#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
567
568#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
569#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
570 QUEUE_STATE_FROZEN)
571#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
572 QUEUE_STATE_FROZEN)
573
73466498
TH
574/*
575 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
576 * netif_tx_* functions below are used to manipulate this flag. The
577 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
578 * queue independently. The netif_xmit_*stopped functions below are called
579 * to check if the queue has been stopped by the driver or stack (either
580 * of the XOFF bits are set in the state). Drivers should not need to call
581 * netif_xmit*stopped functions, they should only be using netif_tx_*.
582 */
79d16385 583
bb949fbd 584struct netdev_queue {
6a321cb3 585/*
5e82b4b2 586 * read-mostly part
6a321cb3 587 */
bb949fbd 588 struct net_device *dev;
46e5da40 589 struct Qdisc __rcu *qdisc;
b0e1e646 590 struct Qdisc *qdisc_sleeping;
ccf5ff69 591#ifdef CONFIG_SYSFS
1d24eb48
TH
592 struct kobject kobj;
593#endif
f2cd2d3e
ED
594#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
595 int numa_node;
596#endif
c0ef079c
FW
597 unsigned long tx_maxrate;
598 /*
599 * Number of TX timeouts for this queue
600 * (/sys/class/net/DEV/Q/trans_timeout)
601 */
602 unsigned long trans_timeout;
ffcfe25b
AD
603
604 /* Subordinate device that the queue has been assigned to */
605 struct net_device *sb_dev;
661b8d1b 606#ifdef CONFIG_XDP_SOCKETS
1742b3d5 607 struct xsk_buff_pool *pool;
661b8d1b 608#endif
6a321cb3 609/*
5e82b4b2 610 * write-mostly part
6a321cb3
ED
611 */
612 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
613 int xmit_lock_owner;
9d21493b 614 /*
9b36627a 615 * Time (in jiffies) of last Tx
9d21493b
ED
616 */
617 unsigned long trans_start;
ccf5ff69 618
114cf580
TH
619 unsigned long state;
620
621#ifdef CONFIG_BQL
622 struct dql dql;
623#endif
e8a0464c 624} ____cacheline_aligned_in_smp;
bb949fbd 625
79134e6c 626extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 627extern int sysctl_devconf_inherit_init_net;
79134e6c 628
316cdaa1
MB
629/*
630 * sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
631 * == 1 : For initns only
632 * == 2 : For none.
633 */
79134e6c
ED
634static inline bool net_has_fallback_tunnels(const struct net *net)
635{
3753d977
MB
636 return !IS_ENABLED(CONFIG_SYSCTL) ||
637 !sysctl_fb_tunnels_only_for_init_net ||
638 (net == &init_net && sysctl_fb_tunnels_only_for_init_net == 1);
79134e6c
ED
639}
640
f2cd2d3e
ED
641static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
642{
643#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
644 return q->numa_node;
645#else
b236da69 646 return NUMA_NO_NODE;
f2cd2d3e
ED
647#endif
648}
649
650static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
651{
652#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
653 q->numa_node = node;
654#endif
655}
656
df334545 657#ifdef CONFIG_RPS
0a9627f2
TH
658/*
659 * This structure holds an RPS map which can be of variable length. The
660 * map is an array of CPUs.
661 */
662struct rps_map {
663 unsigned int len;
664 struct rcu_head rcu;
bb4cf02d 665 u16 cpus[];
0a9627f2 666};
60b778ce 667#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 668
fec5e652 669/*
c445477d
BH
670 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
671 * tail pointer for that CPU's input queue at the time of last enqueue, and
672 * a hardware filter index.
fec5e652
TH
673 */
674struct rps_dev_flow {
675 u16 cpu;
c445477d 676 u16 filter;
fec5e652
TH
677 unsigned int last_qtail;
678};
c445477d 679#define RPS_NO_FILTER 0xffff
fec5e652
TH
680
681/*
682 * The rps_dev_flow_table structure contains a table of flow mappings.
683 */
684struct rps_dev_flow_table {
685 unsigned int mask;
686 struct rcu_head rcu;
bb4cf02d 687 struct rps_dev_flow flows[];
fec5e652
TH
688};
689#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 690 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
691
692/*
693 * The rps_sock_flow_table contains mappings of flows to the last CPU
694 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
695 * Each entry is a 32bit value. Upper part is the high-order bits
696 * of flow hash, lower part is CPU number.
567e4b79 697 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
698 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
699 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 700 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
701 */
702struct rps_sock_flow_table {
567e4b79 703 u32 mask;
93c1af6c 704
bb4cf02d 705 u32 ents[] ____cacheline_aligned_in_smp;
fec5e652 706};
567e4b79 707#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
708
709#define RPS_NO_CPU 0xffff
710
567e4b79
ED
711extern u32 rps_cpu_mask;
712extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
713
fec5e652
TH
714static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
715 u32 hash)
716{
717 if (table && hash) {
567e4b79
ED
718 unsigned int index = hash & table->mask;
719 u32 val = hash & ~rps_cpu_mask;
fec5e652 720
5e82b4b2 721 /* We only give a hint, preemption can change CPU under us */
567e4b79 722 val |= raw_smp_processor_id();
fec5e652 723
567e4b79
ED
724 if (table->ents[index] != val)
725 table->ents[index] = val;
fec5e652
TH
726 }
727}
728
c445477d 729#ifdef CONFIG_RFS_ACCEL
f629d208
JP
730bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
731 u16 filter_id);
c445477d 732#endif
a953be53 733#endif /* CONFIG_RPS */
c445477d 734
0a9627f2
TH
735/* This structure contains an instance of an RX queue. */
736struct netdev_rx_queue {
a953be53 737#ifdef CONFIG_RPS
6e3f7faf
ED
738 struct rps_map __rcu *rps_map;
739 struct rps_dev_flow_table __rcu *rps_flow_table;
a953be53 740#endif
6e3f7faf 741 struct kobject kobj;
fe822240 742 struct net_device *dev;
e817f856 743 struct xdp_rxq_info xdp_rxq;
661b8d1b 744#ifdef CONFIG_XDP_SOCKETS
1742b3d5 745 struct xsk_buff_pool *pool;
661b8d1b 746#endif
0a9627f2 747} ____cacheline_aligned_in_smp;
a953be53
MD
748
749/*
750 * RX queue sysfs structures and functions.
751 */
752struct rx_queue_attribute {
753 struct attribute attr;
718ad681 754 ssize_t (*show)(struct netdev_rx_queue *queue, char *buf);
a953be53 755 ssize_t (*store)(struct netdev_rx_queue *queue,
718ad681 756 const char *buf, size_t len);
a953be53 757};
d314774c 758
044ab86d
AT
759/* XPS map type and offset of the xps map within net_device->xps_maps[]. */
760enum xps_map_type {
761 XPS_CPUS = 0,
762 XPS_RXQS,
763 XPS_MAPS_MAX,
764};
765
bf264145
TH
766#ifdef CONFIG_XPS
767/*
768 * This structure holds an XPS map which can be of variable length. The
769 * map is an array of queues.
770 */
771struct xps_map {
772 unsigned int len;
773 unsigned int alloc_len;
774 struct rcu_head rcu;
bb4cf02d 775 u16 queues[];
bf264145 776};
60b778ce 777#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
778#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
779 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
780
781/*
782 * This structure holds all XPS maps for device. Maps are indexed by CPU.
255c04a8 783 *
5478fcd0
AT
784 * We keep track of the number of cpus/rxqs used when the struct is allocated,
785 * in nr_ids. This will help not accessing out-of-bound memory.
786 *
255c04a8
AT
787 * We keep track of the number of traffic classes used when the struct is
788 * allocated, in num_tc. This will be used to navigate the maps, to ensure we're
789 * not crossing its upper bound, as the original dev->num_tc can be updated in
790 * the meantime.
bf264145
TH
791 */
792struct xps_dev_maps {
793 struct rcu_head rcu;
5478fcd0 794 unsigned int nr_ids;
255c04a8 795 s16 num_tc;
bb4cf02d 796 struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
bf264145 797};
80d19669
AN
798
799#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 800 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
801
802#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
803 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
804
bf264145
TH
805#endif /* CONFIG_XPS */
806
4f57c087
JF
807#define TC_MAX_QUEUE 16
808#define TC_BITMASK 15
809/* HW offloaded queuing disciplines txq count and offset maps */
810struct netdev_tc_txq {
811 u16 count;
812 u16 offset;
813};
814
68bad94e
NP
815#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
816/*
817 * This structure is to hold information about the device
818 * configured to run FCoE protocol stack.
819 */
820struct netdev_fcoe_hbainfo {
821 char manufacturer[64];
822 char serial_number[64];
823 char hardware_version[64];
824 char driver_version[64];
825 char optionrom_version[64];
826 char firmware_version[64];
827 char model[256];
828 char model_description[256];
829};
830#endif
831
02637fce 832#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 833
02637fce
JP
834/* This structure holds a unique identifier to identify some
835 * physical item (port for example) used by a netdevice.
66b52b0d 836 */
02637fce
JP
837struct netdev_phys_item_id {
838 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
839 unsigned char id_len;
840};
841
d754f98b
SF
842static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
843 struct netdev_phys_item_id *b)
844{
845 return a->id_len == b->id_len &&
846 memcmp(a->id, b->id, a->id_len) == 0;
847}
848
99932d4f 849typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
850 struct sk_buff *skb,
851 struct net_device *sb_dev);
99932d4f 852
ddb94eaf
PNA
853enum net_device_path_type {
854 DEV_PATH_ETHERNET = 0,
e4417d69 855 DEV_PATH_VLAN,
ec9d16ba 856 DEV_PATH_BRIDGE,
f6efc675 857 DEV_PATH_PPPOE,
0994d492 858 DEV_PATH_DSA,
ddb94eaf
PNA
859};
860
861struct net_device_path {
862 enum net_device_path_type type;
863 const struct net_device *dev;
e4417d69
PNA
864 union {
865 struct {
866 u16 id;
867 __be16 proto;
f6efc675 868 u8 h_dest[ETH_ALEN];
e4417d69 869 } encap;
bcf2766b
FF
870 struct {
871 enum {
872 DEV_PATH_BR_VLAN_KEEP,
873 DEV_PATH_BR_VLAN_TAG,
874 DEV_PATH_BR_VLAN_UNTAG,
26267bf9 875 DEV_PATH_BR_VLAN_UNTAG_HW,
bcf2766b
FF
876 } vlan_mode;
877 u16 vlan_id;
878 __be16 vlan_proto;
879 } bridge;
0994d492
FF
880 struct {
881 int port;
882 u16 proto;
883 } dsa;
e4417d69 884 };
ddb94eaf
PNA
885};
886
887#define NET_DEVICE_PATH_STACK_MAX 5
bcf2766b 888#define NET_DEVICE_PATH_VLAN_MAX 2
ddb94eaf
PNA
889
890struct net_device_path_stack {
891 int num_paths;
892 struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
893};
894
895struct net_device_path_ctx {
896 const struct net_device *dev;
897 const u8 *daddr;
bcf2766b
FF
898
899 int num_vlans;
900 struct {
901 u16 id;
902 __be16 proto;
903 } vlan[NET_DEVICE_PATH_VLAN_MAX];
ddb94eaf
PNA
904};
905
2572ac53 906enum tc_setup_type {
575ed7d3 907 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 908 TC_SETUP_CLSU32,
5b33f488 909 TC_SETUP_CLSFLOWER,
ade9b658 910 TC_SETUP_CLSMATCHALL,
332ae8e2 911 TC_SETUP_CLSBPF,
8c4083b3 912 TC_SETUP_BLOCK,
8521db4c 913 TC_SETUP_QDISC_CBS,
602f3baf 914 TC_SETUP_QDISC_RED,
7fdb61b4 915 TC_SETUP_QDISC_PRIO,
f971b132 916 TC_SETUP_QDISC_MQ,
25db26a9 917 TC_SETUP_QDISC_ETF,
98b0e5f6 918 TC_SETUP_ROOT_QDISC,
890d8d23 919 TC_SETUP_QDISC_GRED,
9c66d156 920 TC_SETUP_QDISC_TAPRIO,
c29f74e0 921 TC_SETUP_FT,
d35eb52b 922 TC_SETUP_QDISC_ETS,
ef6aadcc 923 TC_SETUP_QDISC_TBF,
aaca9408 924 TC_SETUP_QDISC_FIFO,
d03b195b 925 TC_SETUP_QDISC_HTB,
16e5cc64
JF
926};
927
f4e63525
JK
928/* These structures hold the attributes of bpf state that are being passed
929 * to the netdevice through the bpf op.
a7862b45 930 */
f4e63525 931enum bpf_netdev_command {
a7862b45
BB
932 /* Set or clear a bpf program used in the earliest stages of packet
933 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
934 * is responsible for calling bpf_prog_put on any old progs that are
935 * stored. In case of error, the callee need not release the new prog
936 * reference, but on success it takes ownership and must bpf_prog_put
937 * when it is no longer used.
938 */
939 XDP_SETUP_PROG,
ee5d032f 940 XDP_SETUP_PROG_HW,
ab3f0063 941 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
942 BPF_OFFLOAD_MAP_ALLOC,
943 BPF_OFFLOAD_MAP_FREE,
1742b3d5 944 XDP_SETUP_XSK_POOL,
a7862b45
BB
945};
946
cae1927c 947struct bpf_prog_offload_ops;
ddf9f970 948struct netlink_ext_ack;
74515c57 949struct xdp_umem;
75ccae62 950struct xdp_dev_bulk_queue;
aa8d3a71 951struct bpf_xdp_link;
ddf9f970 952
7f0a8382
AN
953enum bpf_xdp_mode {
954 XDP_MODE_SKB = 0,
955 XDP_MODE_DRV = 1,
956 XDP_MODE_HW = 2,
957 __MAX_XDP_MODE
958};
959
960struct bpf_xdp_entity {
961 struct bpf_prog *prog;
aa8d3a71 962 struct bpf_xdp_link *link;
7f0a8382 963};
ddf9f970 964
f4e63525
JK
965struct netdev_bpf {
966 enum bpf_netdev_command command;
a7862b45
BB
967 union {
968 /* XDP_SETUP_PROG */
ddf9f970 969 struct {
32d60277 970 u32 flags;
ddf9f970
JK
971 struct bpf_prog *prog;
972 struct netlink_ext_ack *extack;
973 };
a3884572
JK
974 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
975 struct {
976 struct bpf_offloaded_map *offmap;
977 };
1742b3d5 978 /* XDP_SETUP_XSK_POOL */
74515c57 979 struct {
1742b3d5 980 struct xsk_buff_pool *pool;
f8ebfaf6 981 u16 queue_id;
74515c57 982 } xsk;
a7862b45
BB
983 };
984};
16e5cc64 985
9116e5e2
MK
986/* Flags for ndo_xsk_wakeup. */
987#define XDP_WAKEUP_RX (1 << 0)
988#define XDP_WAKEUP_TX (1 << 1)
989
d77e38e6
SK
990#ifdef CONFIG_XFRM_OFFLOAD
991struct xfrmdev_ops {
992 int (*xdo_dev_state_add) (struct xfrm_state *x);
993 void (*xdo_dev_state_delete) (struct xfrm_state *x);
994 void (*xdo_dev_state_free) (struct xfrm_state *x);
995 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
996 struct xfrm_state *x);
50bd870a 997 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
d77e38e6
SK
998};
999#endif
1000
6c557001
FW
1001struct dev_ifalias {
1002 struct rcu_head rcuhead;
1003 char ifalias[];
1004};
1005
b473b0d2 1006struct devlink;
da68b4ad 1007struct tlsdev_ops;
b473b0d2 1008
ff927412
JP
1009struct netdev_name_node {
1010 struct hlist_node hlist;
36fbf1e5 1011 struct list_head list;
ff927412
JP
1012 struct net_device *dev;
1013 const char *name;
1014};
1015
36fbf1e5
JP
1016int netdev_name_node_alt_create(struct net_device *dev, const char *name);
1017int netdev_name_node_alt_destroy(struct net_device *dev, const char *name);
1018
93642e14
JP
1019struct netdev_net_notifier {
1020 struct list_head list;
1021 struct notifier_block *nb;
1022};
1023
d314774c
SH
1024/*
1025 * This structure defines the management hooks for network devices.
00829823
SH
1026 * The following hooks can be defined; unless noted otherwise, they are
1027 * optional and can be filled with a null pointer.
d314774c
SH
1028 *
1029 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
1030 * This function is called once when a network device is registered.
1031 * The network device can use this for any late stage initialization
1032 * or semantic validation. It can fail with an error code which will
1033 * be propagated back to register_netdev.
d314774c
SH
1034 *
1035 * void (*ndo_uninit)(struct net_device *dev);
1036 * This function is called when device is unregistered or when registration
1037 * fails. It is not called if init fails.
1038 *
1039 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 1040 * This function is called when a network device transitions to the up
d314774c
SH
1041 * state.
1042 *
1043 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 1044 * This function is called when a network device transitions to the down
d314774c
SH
1045 * state.
1046 *
dc1f8bf6
SH
1047 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1048 * struct net_device *dev);
00829823 1049 * Called when a packet needs to be transmitted.
e79d8429
RR
1050 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
1051 * the queue before that can happen; it's for obsolete devices and weird
1052 * corner cases, but the stack really does a non-trivial amount
1053 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 1054 * Required; cannot be NULL.
00829823 1055 *
1a2a1444
DM
1056 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1057 * struct net_device *dev
1058 * netdev_features_t features);
1059 * Called by core transmit path to determine if device is capable of
1060 * performing offload operations on a given packet. This is to give
1061 * the device an opportunity to implement any restrictions that cannot
1062 * be otherwise expressed by feature flags. The check is called with
1063 * the set of features that the stack has calculated and it returns
1064 * those the driver believes to be appropriate.
cdba756f 1065 *
f663dd9a 1066 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 1067 * struct net_device *sb_dev);
5e82b4b2 1068 * Called to decide which queue to use when device supports multiple
00829823
SH
1069 * transmit queues.
1070 *
d314774c
SH
1071 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1072 * This function is called to allow device receiver to make
5e82b4b2 1073 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
1074 *
1075 * void (*ndo_set_rx_mode)(struct net_device *dev);
1076 * This function is called device changes address list filtering.
01789349 1077 * If driver handles unicast address filtering, it should set
5e82b4b2 1078 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
1079 *
1080 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1081 * This function is called when the Media Access Control address
37b607c5 1082 * needs to be changed. If this interface is not defined, the
5e82b4b2 1083 * MAC address can not be changed.
d314774c
SH
1084 *
1085 * int (*ndo_validate_addr)(struct net_device *dev);
1086 * Test if Media Access Control address is valid for the device.
1087 *
1088 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
5e82b4b2
BH
1089 * Called when a user requests an ioctl which can't be handled by
1090 * the generic interface code. If not defined ioctls return
d314774c
SH
1091 * not supported error code.
1092 *
1093 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1094 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1095 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1096 * interface (PCI) for low level management.
1097 *
1098 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1099 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1100 * of a device.
d314774c 1101 *
0290bd29 1102 * void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
5e82b4b2 1103 * Callback used when the transmitter has not made any progress
d314774c
SH
1104 * for dev->watchdog ticks.
1105 *
bc1f4470 1106 * void (*ndo_get_stats64)(struct net_device *dev,
1107 * struct rtnl_link_stats64 *storage);
d308e38f 1108 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1109 * Called when a user wants to get the network device usage
be1f3c2c 1110 * statistics. Drivers must do one of the following:
3cfde79c
BH
1111 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1112 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1113 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1114 * (which should normally be dev->stats) and return a pointer to
1115 * it. The structure may be changed asynchronously only if each
1116 * field is written atomically.
1117 * 3. Update dev->stats asynchronously and atomically, and define
1118 * neither operation.
d314774c 1119 *
3df5b3c6 1120 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1121 * Return true if this device supports offload stats of this attr_id.
1122 *
1123 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1124 * void *attr_data)
1125 * Get statistics for offload operations by attr_id. Write it into the
1126 * attr_data pointer.
1127 *
5d632cb7 1128 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1129 * If device supports VLAN filtering this function is called when a
80d5c368 1130 * VLAN id is registered.
d314774c 1131 *
5d632cb7 1132 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1133 * If device supports VLAN filtering this function is called when a
80d5c368 1134 * VLAN id is unregistered.
d314774c
SH
1135 *
1136 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1137 *
1138 * SR-IOV management functions.
1139 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1140 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1141 * u8 qos, __be16 proto);
ed616689
SC
1142 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1143 * int max_tx_rate);
5f8444a3 1144 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1145 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1146 * int (*ndo_get_vf_config)(struct net_device *dev,
1147 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1148 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1149 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1150 * struct nlattr *port[]);
01a3d796
VZ
1151 *
1152 * Enable or disable the VF ability to query its RSS Redirection Table and
1153 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1154 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1155 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1156 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1157 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1158 * void *type_data);
6a4bc2b4
FF
1159 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1160 * This is always called from the stack with the rtnl lock held and netif
1161 * tx queues stopped. This allows the netdevice to perform queue
1162 * management safely.
c445477d 1163 *
e9bce845
YZ
1164 * Fiber Channel over Ethernet (FCoE) offload functions.
1165 * int (*ndo_fcoe_enable)(struct net_device *dev);
1166 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1167 * so the underlying device can perform whatever needed configuration or
1168 * initialization to support acceleration of FCoE traffic.
1169 *
1170 * int (*ndo_fcoe_disable)(struct net_device *dev);
1171 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1172 * so the underlying device can perform whatever needed clean-ups to
1173 * stop supporting acceleration of FCoE traffic.
1174 *
1175 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1176 * struct scatterlist *sgl, unsigned int sgc);
1177 * Called when the FCoE Initiator wants to initialize an I/O that
1178 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1179 * perform necessary setup and returns 1 to indicate the device is set up
1180 * successfully to perform DDP on this I/O, otherwise this returns 0.
1181 *
1182 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1183 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1184 * indicated by the FC exchange id 'xid', so the underlying device can
1185 * clean up and reuse resources for later DDP requests.
1186 *
1187 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1188 * struct scatterlist *sgl, unsigned int sgc);
1189 * Called when the FCoE Target wants to initialize an I/O that
1190 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1191 * perform necessary setup and returns 1 to indicate the device is set up
1192 * successfully to perform DDP on this I/O, otherwise this returns 0.
1193 *
68bad94e
NP
1194 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1195 * struct netdev_fcoe_hbainfo *hbainfo);
1196 * Called when the FCoE Protocol stack wants information on the underlying
1197 * device. This information is utilized by the FCoE protocol stack to
1198 * register attributes with Fiber Channel management service as per the
1199 * FC-GS Fabric Device Management Information(FDMI) specification.
1200 *
e9bce845
YZ
1201 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1202 * Called when the underlying device wants to override default World Wide
1203 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1204 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1205 * protocol stack to use.
1206 *
c445477d
BH
1207 * RFS acceleration.
1208 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1209 * u16 rxq_index, u32 flow_id);
1210 * Set hardware filter for RFS. rxq_index is the target queue index;
1211 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1212 * Return the filter ID on success, or a negative error code.
fbaec0ea 1213 *
8b98a70c 1214 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1215 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1216 * Called to make another netdev an underling.
1217 *
1218 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1219 * Called to release previously enslaved netdev.
5455c699 1220 *
cff9f12b
MG
1221 * struct net_device *(*ndo_get_xmit_slave)(struct net_device *dev,
1222 * struct sk_buff *skb,
1223 * bool all_slaves);
1224 * Get the xmit slave of master device. If all_slaves is true, function
1225 * assume all the slaves can transmit.
1226 *
5455c699 1227 * Feature/offload setting functions.
1a2a1444
DM
1228 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1229 * netdev_features_t features);
1230 * Adjusts the requested feature flags according to device-specific
1231 * constraints, and returns the resulting flags. Must not modify
1232 * the device state.
1233 *
c8f44aff 1234 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1235 * Called to update device configuration to new features. Passed
1236 * feature set might be less than what was returned by ndo_fix_features()).
1237 * Must return >0 or -errno if it changed dev->features itself.
1238 *
edc7d573 1239 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1240 * struct net_device *dev,
87b0984e
PM
1241 * const unsigned char *addr, u16 vid, u16 flags,
1242 * struct netlink_ext_ack *extack);
77162022 1243 * Adds an FDB entry to dev for addr.
1690be63
VY
1244 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1245 * struct net_device *dev,
f6f6424b 1246 * const unsigned char *addr, u16 vid)
77162022
JF
1247 * Deletes the FDB entry from dev coresponding to addr.
1248 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1249 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1250 * int *idx)
77162022
JF
1251 * Used to add FDB entries to dump requests. Implementers should add
1252 * entries to skb and update idx with the number of entries.
e5a55a89 1253 *
ad41faa8 1254 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1255 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1256 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1257 * struct net_device *dev, u32 filter_mask,
1258 * int nlflags)
ad41faa8
ND
1259 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1260 * u16 flags);
4bf84c35
JP
1261 *
1262 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1263 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1264 * which do not represent real hardware may define this to allow their
1265 * userspace components to manage their virtual carrier state. Devices
1266 * that determine carrier state from physical hardware properties (eg
1267 * network cables) or protocol-dependent mechanisms (eg
1268 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1269 *
1270 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1271 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1272 * Called to get ID of physical port of this device. If driver does
1273 * not implement this, it is assumed that the hw is not able to have
1274 * multiple net devices on single physical port.
53cf5275 1275 *
d6abc596
FF
1276 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1277 * struct netdev_phys_item_id *ppid)
1278 * Called to get the parent ID of the physical port of this device.
1279 *
a6cc0cfa
JF
1280 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1281 * struct net_device *dev)
1282 * Called by upper layer devices to accelerate switching or other
1283 * station functionality into hardware. 'pdev is the lowerdev
1284 * to use for the offload and 'dev' is the net device that will
1285 * back the offload. Returns a pointer to the private structure
1286 * the upper layer will maintain.
1287 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1288 * Called by upper layer device to delete the station created
1289 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1290 * the station and priv is the structure returned by the add
1291 * operation.
822b3b2e
JF
1292 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1293 * int queue_index, u32 maxrate);
1294 * Called when a user wants to set a max-rate limitation of specific
1295 * TX queue.
a54acb3a
ND
1296 * int (*ndo_get_iflink)(const struct net_device *dev);
1297 * Called to get the iflink value of this device.
d746d707 1298 * void (*ndo_change_proto_down)(struct net_device *dev,
5e82b4b2 1299 * bool proto_down);
d746d707
AK
1300 * This function is used to pass protocol port error state information
1301 * to the switch driver. The switch driver can react to the proto_down
1302 * by doing a phys down on the associated switch port.
fc4099f1
PS
1303 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1304 * This function is used to get egress tunnel information for given skb.
1305 * This is useful for retrieving outer tunnel header parameters while
1306 * sampling packet.
871b642a
PA
1307 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1308 * This function is used to specify the headroom that the skb must
1309 * consider when allocation skb during packet reception. Setting
1310 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1311 * forward. Setting a negative value resets the rx headroom to the
871b642a 1312 * default value.
f4e63525 1313 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1314 * This function is used to set or query state related to XDP on the
f4e63525
JK
1315 * netdevice and manage BPF offload. See definition of
1316 * enum bpf_netdev_command for details.
42b33468
JDB
1317 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1318 * u32 flags);
735fc405
JDB
1319 * This function is used to submit @n XDP packets for transmit on a
1320 * netdevice. Returns number of frames successfully transmitted, frames
1321 * that got dropped are freed/returned via xdp_return_frame().
1322 * Returns negative number, means general error invoking ndo, meaning
1323 * no frames were xmit'ed and core-caller will free all frames.
9116e5e2
MK
1324 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1325 * This function is used to wake up the softirq, ksoftirqd or kthread
1326 * responsible for sending and/or receiving packets on a specific
1327 * queue id bound to an AF_XDP socket. The flags field specifies if
1328 * only RX, only Tx, or both should be woken up using the flags
1329 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
5dc37bb9
JP
1330 * struct devlink_port *(*ndo_get_devlink_port)(struct net_device *dev);
1331 * Get devlink port instance associated with a given netdev.
b473b0d2
JK
1332 * Called with a reference on the netdevice and devlink locks only,
1333 * rtnl_lock is not held.
607259a6
CH
1334 * int (*ndo_tunnel_ctl)(struct net_device *dev, struct ip_tunnel_parm *p,
1335 * int cmd);
1336 * Add, change, delete or get information on an IPv4 tunnel.
9aa1206e
DB
1337 * struct net_device *(*ndo_get_peer_dev)(struct net_device *dev);
1338 * If a device is paired with a peer device, return the peer instance.
1339 * The caller must be under RCU read context.
ddb94eaf
PNA
1340 * int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx, struct net_device_path *path);
1341 * Get the forwarding path to reach the real device from the HW destination address
d314774c
SH
1342 */
1343struct net_device_ops {
1344 int (*ndo_init)(struct net_device *dev);
1345 void (*ndo_uninit)(struct net_device *dev);
1346 int (*ndo_open)(struct net_device *dev);
1347 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1348 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1349 struct net_device *dev);
1350 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1351 struct net_device *dev,
1352 netdev_features_t features);
00829823 1353 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1354 struct sk_buff *skb,
a350ecce 1355 struct net_device *sb_dev);
d314774c
SH
1356 void (*ndo_change_rx_flags)(struct net_device *dev,
1357 int flags);
d314774c 1358 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1359 int (*ndo_set_mac_address)(struct net_device *dev,
1360 void *addr);
d314774c 1361 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1362 int (*ndo_do_ioctl)(struct net_device *dev,
1363 struct ifreq *ifr, int cmd);
d314774c
SH
1364 int (*ndo_set_config)(struct net_device *dev,
1365 struct ifmap *map);
00829823
SH
1366 int (*ndo_change_mtu)(struct net_device *dev,
1367 int new_mtu);
1368 int (*ndo_neigh_setup)(struct net_device *dev,
1369 struct neigh_parms *);
0290bd29
MT
1370 void (*ndo_tx_timeout) (struct net_device *dev,
1371 unsigned int txqueue);
d314774c 1372
bc1f4470 1373 void (*ndo_get_stats64)(struct net_device *dev,
1374 struct rtnl_link_stats64 *storage);
3df5b3c6 1375 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1376 int (*ndo_get_offload_stats)(int attr_id,
1377 const struct net_device *dev,
1378 void *attr_data);
d314774c
SH
1379 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1380
8e586137 1381 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1382 __be16 proto, u16 vid);
8e586137 1383 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1384 __be16 proto, u16 vid);
d314774c 1385#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1386 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1387 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1388 struct netpoll_info *info);
0e34e931 1389 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1390#endif
95c26df8
WM
1391 int (*ndo_set_vf_mac)(struct net_device *dev,
1392 int queue, u8 *mac);
1393 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1394 int queue, u16 vlan,
1395 u8 qos, __be16 proto);
ed616689
SC
1396 int (*ndo_set_vf_rate)(struct net_device *dev,
1397 int vf, int min_tx_rate,
1398 int max_tx_rate);
5f8444a3
GR
1399 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1400 int vf, bool setting);
dd461d6a
HS
1401 int (*ndo_set_vf_trust)(struct net_device *dev,
1402 int vf, bool setting);
95c26df8
WM
1403 int (*ndo_get_vf_config)(struct net_device *dev,
1404 int vf,
1405 struct ifla_vf_info *ivf);
1d8faf48
RE
1406 int (*ndo_set_vf_link_state)(struct net_device *dev,
1407 int vf, int link_state);
3b766cd8
EBE
1408 int (*ndo_get_vf_stats)(struct net_device *dev,
1409 int vf,
1410 struct ifla_vf_stats
1411 *vf_stats);
57b61080
SF
1412 int (*ndo_set_vf_port)(struct net_device *dev,
1413 int vf,
1414 struct nlattr *port[]);
1415 int (*ndo_get_vf_port)(struct net_device *dev,
1416 int vf, struct sk_buff *skb);
30aad417
DG
1417 int (*ndo_get_vf_guid)(struct net_device *dev,
1418 int vf,
1419 struct ifla_vf_guid *node_guid,
1420 struct ifla_vf_guid *port_guid);
cc8e27cc
EC
1421 int (*ndo_set_vf_guid)(struct net_device *dev,
1422 int vf, u64 guid,
1423 int guid_type);
01a3d796
VZ
1424 int (*ndo_set_vf_rss_query_en)(
1425 struct net_device *dev,
1426 int vf, bool setting);
16e5cc64 1427 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1428 enum tc_setup_type type,
de4784ca 1429 void *type_data);
d11ead75 1430#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1431 int (*ndo_fcoe_enable)(struct net_device *dev);
1432 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1433 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1434 u16 xid,
1435 struct scatterlist *sgl,
1436 unsigned int sgc);
1437 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1438 u16 xid);
6247e086
YZ
1439 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1440 u16 xid,
1441 struct scatterlist *sgl,
1442 unsigned int sgc);
68bad94e
NP
1443 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1444 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1445#endif
1446
d11ead75 1447#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1448#define NETDEV_FCOE_WWNN 0
1449#define NETDEV_FCOE_WWPN 1
1450 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1451 u64 *wwn, int type);
4d288d57 1452#endif
3c9c36bc 1453
c445477d
BH
1454#ifdef CONFIG_RFS_ACCEL
1455 int (*ndo_rx_flow_steer)(struct net_device *dev,
1456 const struct sk_buff *skb,
1457 u16 rxq_index,
1458 u32 flow_id);
1459#endif
fbaec0ea 1460 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1461 struct net_device *slave_dev,
1462 struct netlink_ext_ack *extack);
fbaec0ea
JP
1463 int (*ndo_del_slave)(struct net_device *dev,
1464 struct net_device *slave_dev);
cff9f12b
MG
1465 struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
1466 struct sk_buff *skb,
1467 bool all_slaves);
719a402c
TT
1468 struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
1469 struct sock *sk);
c8f44aff
MM
1470 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1471 netdev_features_t features);
5455c699 1472 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1473 netdev_features_t features);
503eebc2
JP
1474 int (*ndo_neigh_construct)(struct net_device *dev,
1475 struct neighbour *n);
1476 void (*ndo_neigh_destroy)(struct net_device *dev,
1477 struct neighbour *n);
77162022
JF
1478
1479 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1480 struct nlattr *tb[],
77162022 1481 struct net_device *dev,
6b6e2725 1482 const unsigned char *addr,
f6f6424b 1483 u16 vid,
87b0984e
PM
1484 u16 flags,
1485 struct netlink_ext_ack *extack);
77162022 1486 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1487 struct nlattr *tb[],
77162022 1488 struct net_device *dev,
f6f6424b
JP
1489 const unsigned char *addr,
1490 u16 vid);
77162022
JF
1491 int (*ndo_fdb_dump)(struct sk_buff *skb,
1492 struct netlink_callback *cb,
1493 struct net_device *dev,
5d5eacb3 1494 struct net_device *filter_dev,
d297653d 1495 int *idx);
5b2f94b2
RP
1496 int (*ndo_fdb_get)(struct sk_buff *skb,
1497 struct nlattr *tb[],
1498 struct net_device *dev,
1499 const unsigned char *addr,
1500 u16 vid, u32 portid, u32 seq,
1501 struct netlink_ext_ack *extack);
e5a55a89 1502 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1503 struct nlmsghdr *nlh,
2fd527b7
PM
1504 u16 flags,
1505 struct netlink_ext_ack *extack);
e5a55a89
JF
1506 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1507 u32 pid, u32 seq,
6cbdceeb 1508 struct net_device *dev,
46c264da
ND
1509 u32 filter_mask,
1510 int nlflags);
407af329 1511 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1512 struct nlmsghdr *nlh,
1513 u16 flags);
4bf84c35
JP
1514 int (*ndo_change_carrier)(struct net_device *dev,
1515 bool new_carrier);
66b52b0d 1516 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1517 struct netdev_phys_item_id *ppid);
d6abc596
FF
1518 int (*ndo_get_port_parent_id)(struct net_device *dev,
1519 struct netdev_phys_item_id *ppid);
db24a904
DA
1520 int (*ndo_get_phys_port_name)(struct net_device *dev,
1521 char *name, size_t len);
a6cc0cfa
JF
1522 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1523 struct net_device *dev);
1524 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1525 void *priv);
1526
822b3b2e
JF
1527 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1528 int queue_index,
1529 u32 maxrate);
a54acb3a 1530 int (*ndo_get_iflink)(const struct net_device *dev);
d746d707
AK
1531 int (*ndo_change_proto_down)(struct net_device *dev,
1532 bool proto_down);
fc4099f1
PS
1533 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1534 struct sk_buff *skb);
871b642a
PA
1535 void (*ndo_set_rx_headroom)(struct net_device *dev,
1536 int needed_headroom);
f4e63525
JK
1537 int (*ndo_bpf)(struct net_device *dev,
1538 struct netdev_bpf *bpf);
735fc405 1539 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1540 struct xdp_frame **xdp,
1541 u32 flags);
9116e5e2
MK
1542 int (*ndo_xsk_wakeup)(struct net_device *dev,
1543 u32 queue_id, u32 flags);
5dc37bb9 1544 struct devlink_port * (*ndo_get_devlink_port)(struct net_device *dev);
607259a6
CH
1545 int (*ndo_tunnel_ctl)(struct net_device *dev,
1546 struct ip_tunnel_parm *p, int cmd);
9aa1206e 1547 struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
ddb94eaf
PNA
1548 int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
1549 struct net_device_path *path);
d314774c
SH
1550};
1551
7aa98047 1552/**
270f3385 1553 * enum netdev_priv_flags - &struct net_device priv_flags
7aa98047
LR
1554 *
1555 * These are the &struct net_device, they are only set internally
1556 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1557 * userspace; this means that the order of these flags can change
7aa98047
LR
1558 * during any kernel release.
1559 *
1560 * You should have a pretty good reason to be extending these flags.
1561 *
1562 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1563 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1564 * @IFF_BONDING: bonding master or slave
7aa98047 1565 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1566 * @IFF_WAN_HDLC: WAN HDLC device
1567 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1568 * release skb->dst
1569 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1570 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1571 * @IFF_MACVLAN_PORT: device used as macvlan port
1572 * @IFF_BRIDGE_PORT: device used as bridge port
1573 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1574 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1575 * @IFF_UNICAST_FLT: Supports unicast filtering
1576 * @IFF_TEAM_PORT: device used as team port
1577 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1578 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1579 * change when it's running
1580 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1581 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1582 * underlying stacked devices
007979ea 1583 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1584 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1585 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1586 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1587 * @IFF_TEAM: device is a team device
d4ab4286 1588 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1589 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1590 * entity (i.e. the master device for bridged veth)
3c175784 1591 * @IFF_MACSEC: device is a MACsec device
f5426250 1592 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1593 * @IFF_FAILOVER: device is a failover master device
1594 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1595 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8065a779 1596 * @IFF_LIVE_RENAME_OK: rename is allowed while device is up and running
c2ff53d8
XZ
1597 * @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1598 * skb_headlen(skb) == 0 (data starts from frag0)
7aa98047
LR
1599 */
1600enum netdev_priv_flags {
1601 IFF_802_1Q_VLAN = 1<<0,
1602 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1603 IFF_BONDING = 1<<2,
1604 IFF_ISATAP = 1<<3,
1605 IFF_WAN_HDLC = 1<<4,
1606 IFF_XMIT_DST_RELEASE = 1<<5,
1607 IFF_DONT_BRIDGE = 1<<6,
1608 IFF_DISABLE_NETPOLL = 1<<7,
1609 IFF_MACVLAN_PORT = 1<<8,
1610 IFF_BRIDGE_PORT = 1<<9,
1611 IFF_OVS_DATAPATH = 1<<10,
1612 IFF_TX_SKB_SHARING = 1<<11,
1613 IFF_UNICAST_FLT = 1<<12,
1614 IFF_TEAM_PORT = 1<<13,
1615 IFF_SUPP_NOFCS = 1<<14,
1616 IFF_LIVE_ADDR_CHANGE = 1<<15,
1617 IFF_MACVLAN = 1<<16,
1618 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1619 IFF_L3MDEV_MASTER = 1<<18,
1620 IFF_NO_QUEUE = 1<<19,
1621 IFF_OPENVSWITCH = 1<<20,
1622 IFF_L3MDEV_SLAVE = 1<<21,
1623 IFF_TEAM = 1<<22,
1624 IFF_RXFH_CONFIGURED = 1<<23,
1625 IFF_PHONY_HEADROOM = 1<<24,
1626 IFF_MACSEC = 1<<25,
f5426250 1627 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1628 IFF_FAILOVER = 1<<27,
1629 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1630 IFF_L3MDEV_RX_HANDLER = 1<<29,
8065a779 1631 IFF_LIVE_RENAME_OK = 1<<30,
c2ff53d8 1632 IFF_TX_SKB_NO_LINEAR = 1<<31,
7aa98047
LR
1633};
1634
1635#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1636#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1637#define IFF_BONDING IFF_BONDING
7aa98047 1638#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1639#define IFF_WAN_HDLC IFF_WAN_HDLC
1640#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1641#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1642#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1643#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1644#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1645#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1646#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1647#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1648#define IFF_TEAM_PORT IFF_TEAM_PORT
1649#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1650#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1651#define IFF_MACVLAN IFF_MACVLAN
02875878 1652#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1653#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1654#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1655#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1656#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1657#define IFF_TEAM IFF_TEAM
d4ab4286 1658#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
2463e073 1659#define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
3c175784 1660#define IFF_MACSEC IFF_MACSEC
f5426250 1661#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1662#define IFF_FAILOVER IFF_FAILOVER
1663#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
d5256083 1664#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
8065a779 1665#define IFF_LIVE_RENAME_OK IFF_LIVE_RENAME_OK
c2ff53d8 1666#define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
7aa98047 1667
4e096a18
OR
1668/* Specifies the type of the struct net_device::ml_priv pointer */
1669enum netdev_ml_priv_type {
1670 ML_PRIV_NONE,
1671 ML_PRIV_CAN,
1672};
1673
536721b1
KK
1674/**
1675 * struct net_device - The DEVICE structure.
d651983d
MCC
1676 *
1677 * Actually, this whole structure is a big mistake. It mixes I/O
1678 * data with strictly "high-level" data, and it has to know about
1679 * almost every data structure used in the INET module.
536721b1
KK
1680 *
1681 * @name: This is the first field of the "visible" part of this structure
1682 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1683 * of the interface.
536721b1 1684 *
ff927412 1685 * @name_node: Name hashlist node
536721b1
KK
1686 * @ifalias: SNMP alias
1687 * @mem_end: Shared memory end
1688 * @mem_start: Shared memory start
1689 * @base_addr: Device I/O address
1690 * @irq: Device IRQ number
1691 *
1692 * @state: Generic network queuing layer state, see netdev_state_t
1693 * @dev_list: The global list of network devices
5e82b4b2
BH
1694 * @napi_list: List entry used for polling NAPI devices
1695 * @unreg_list: List entry when we are unregistering the
1696 * device; see the function unregister_netdev
1697 * @close_list: List entry used when we are closing the device
62d885fe
BP
1698 * @ptype_all: Device-specific packet handlers for all protocols
1699 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1700 *
1701 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1702 * @features: Currently active device features
1703 * @hw_features: User-changeable features
1704 *
1705 * @wanted_features: User-requested features
1706 * @vlan_features: Mask of features inheritable by VLAN devices
1707 *
1708 * @hw_enc_features: Mask of features inherited by encapsulating devices
1709 * This field indicates what encapsulation
1710 * offloads the hardware is capable of doing,
1711 * and drivers will need to set them appropriately.
1712 *
1713 * @mpls_features: Mask of features inheritable by MPLS
a1fa83bd 1714 * @gso_partial_features: value(s) from NETIF_F_GSO\*
536721b1
KK
1715 *
1716 * @ifindex: interface index
5e82b4b2 1717 * @group: The group the device belongs to
536721b1
KK
1718 *
1719 * @stats: Statistics struct, which was left as a legacy, use
1720 * rtnl_link_stats64 instead
1721 *
1722 * @rx_dropped: Dropped packets by core network,
1723 * do not use this in drivers
1724 * @tx_dropped: Dropped packets by core network,
1725 * do not use this in drivers
6e7333d3
JW
1726 * @rx_nohandler: nohandler dropped packets by core network on
1727 * inactive devices, do not use this in drivers
9e55e5d3
FF
1728 * @carrier_up_count: Number of times the carrier has been up
1729 * @carrier_down_count: Number of times the carrier has been down
536721b1 1730 *
536721b1
KK
1731 * @wireless_handlers: List of functions to handle Wireless Extensions,
1732 * instead of ioctl,
1733 * see <net/iw_handler.h> for details.
1734 * @wireless_data: Instance data managed by the core of wireless extensions
1735 *
1736 * @netdev_ops: Includes several pointers to callbacks,
1737 * if one wants to override the ndo_*() functions
1738 * @ethtool_ops: Management operations
a1fa83bd 1739 * @l3mdev_ops: Layer 3 master device operations
f997c55c
AA
1740 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1741 * discovery handling. Necessary for e.g. 6LoWPAN.
a1fa83bd
RD
1742 * @xfrmdev_ops: Transformation offload operations
1743 * @tlsdev_ops: Transport Layer Security offload operations
d476059e 1744 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1745 * of Layer 2 headers.
1746 *
1747 * @flags: Interface flags (a la BSD)
1748 * @priv_flags: Like 'flags' but invisible to userspace,
1749 * see if.h for the definitions
1750 * @gflags: Global flags ( kept as legacy )
1751 * @padded: How much padding added by alloc_netdev()
1752 * @operstate: RFC2863 operstate
1753 * @link_mode: Mapping policy to operstate
1754 * @if_port: Selectable AUI, TP, ...
1755 * @dma: DMA channel
1756 * @mtu: Interface MTU value
61e84623
JW
1757 * @min_mtu: Interface Minimum MTU value
1758 * @max_mtu: Interface Maximum MTU value
536721b1 1759 * @type: Interface hardware type
2793a23a 1760 * @hard_header_len: Maximum hardware header length.
217e6fa2 1761 * @min_header_len: Minimum hardware header length
536721b1
KK
1762 *
1763 * @needed_headroom: Extra headroom the hardware may need, but not in all
1764 * cases can this be guaranteed
1765 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1766 * cases can this be guaranteed. Some cases also use
1767 * LL_MAX_HEADER instead to allocate the skb
1768 *
1769 * interface address info:
1770 *
1771 * @perm_addr: Permanent hw address
1772 * @addr_assign_type: Hw address assignment type
1773 * @addr_len: Hardware address length
5343da4c
TY
1774 * @upper_level: Maximum depth level of upper devices.
1775 * @lower_level: Maximum depth level of lower devices.
8626a0c8 1776 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1777 * @dev_id: Used to differentiate devices that share
1778 * the same link layer address
1779 * @dev_port: Used to differentiate devices that share
1780 * the same function
1781 * @addr_list_lock: XXX: need comments on this one
a1fa83bd 1782 * @name_assign_type: network interface name assignment type
5e82b4b2 1783 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1784 * has been enabled due to the need to listen to
1785 * additional unicast addresses in a device that
1786 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1787 * @uc: unicast mac addresses
1788 * @mc: multicast mac addresses
1789 * @dev_addrs: list of device hw addresses
1790 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1791 * @promiscuity: Number of times the NIC is told to work in
1792 * promiscuous mode; if it becomes 0 the NIC will
1793 * exit promiscuous mode
536721b1
KK
1794 * @allmulti: Counter, enables or disables allmulticast mode
1795 *
1796 * @vlan_info: VLAN info
1797 * @dsa_ptr: dsa specific data
1798 * @tipc_ptr: TIPC specific data
1799 * @atalk_ptr: AppleTalk link
1800 * @ip_ptr: IPv4 specific data
1801 * @dn_ptr: DECnet specific data
1802 * @ip6_ptr: IPv6 specific data
1803 * @ax25_ptr: AX.25 specific data
1804 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
a1fa83bd
RD
1805 * @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
1806 * device struct
1807 * @mpls_ptr: mpls_dev struct pointer
536721b1 1808 *
536721b1
KK
1809 * @dev_addr: Hw address (before bcast,
1810 * because most packets are unicast)
1811 *
1812 * @_rx: Array of RX queues
1813 * @num_rx_queues: Number of RX queues
1814 * allocated at register_netdev() time
1815 * @real_num_rx_queues: Number of RX queues currently active in device
a1fa83bd
RD
1816 * @xdp_prog: XDP sockets filter program pointer
1817 * @gro_flush_timeout: timeout for GRO layer in NAPI
5c45a918
MCC
1818 * @napi_defer_hard_irqs: If not zero, provides a counter that would
1819 * allow to avoid NIC hard IRQ, on busy queues.
536721b1
KK
1820 *
1821 * @rx_handler: handler for received packets
1822 * @rx_handler_data: XXX: need comments on this one
46209401
JP
1823 * @miniq_ingress: ingress/clsact qdisc specific data for
1824 * ingress processing
536721b1 1825 * @ingress_queue: XXX: need comments on this one
2f5e70c8 1826 * @nf_hooks_ingress: netfilter hooks executed for ingress packets
536721b1
KK
1827 * @broadcast: hw bcast address
1828 *
14ffbbb8
TG
1829 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1830 * indexed by RX queue number. Assigned by driver.
1831 * This must only be set if the ndo_rx_flow_steer
1832 * operation is defined
1833 * @index_hlist: Device index hash chain
1834 *
536721b1
KK
1835 * @_tx: Array of TX queues
1836 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1837 * @real_num_tx_queues: Number of TX queues currently active in device
1838 * @qdisc: Root qdisc from userspace point of view
1839 * @tx_queue_len: Max frames per queue allowed
1840 * @tx_global_lock: XXX: need comments on this one
a1fa83bd 1841 * @xdp_bulkq: XDP device bulk queue
044ab86d 1842 * @xps_maps: all CPUs/RXQs maps for XPS device
536721b1
KK
1843 *
1844 * @xps_maps: XXX: need comments on this one
46209401
JP
1845 * @miniq_egress: clsact qdisc specific data for
1846 * egress processing
a1fa83bd 1847 * @qdisc_hash: qdisc hash table
536721b1 1848 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1849 * the watchdog (see dev_watchdog())
536721b1
KK
1850 * @watchdog_timer: List of timers
1851 *
eb02d39a 1852 * @proto_down_reason: reason a netdev interface is held down
536721b1 1853 * @pcpu_refcnt: Number of references to this device
add2d736 1854 * @dev_refcnt: Number of references to this device
536721b1 1855 * @todo_list: Delayed register/unregister
536721b1
KK
1856 * @link_watch_list: XXX: need comments on this one
1857 *
1858 * @reg_state: Register/unregister state machine
1859 * @dismantle: Device is going to be freed
1860 * @rtnl_link_state: This enum represents the phases of creating
1861 * a new link
1862 *
cf124db5
DM
1863 * @needs_free_netdev: Should unregister perform free_netdev?
1864 * @priv_destructor: Called from unregister
536721b1
KK
1865 * @npinfo: XXX: need comments on this one
1866 * @nd_net: Network namespace this network device is inside
1867 *
1868 * @ml_priv: Mid-layer private
4e096a18 1869 * @ml_priv_type: Mid-layer private type
536721b1
KK
1870 * @lstats: Loopback statistics
1871 * @tstats: Tunnel statistics
1872 * @dstats: Dummy statistics
1873 * @vstats: Virtual ethernet statistics
1874 *
1875 * @garp_port: GARP
1876 * @mrp_port: MRP
1877 *
1878 * @dev: Class/net/name entry
1879 * @sysfs_groups: Space for optional device, statistics and wireless
1880 * sysfs groups
1881 *
1882 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1883 * @rtnl_link_ops: Rtnl_link_ops
1884 *
1885 * @gso_max_size: Maximum size of generic segmentation offload
1886 * @gso_max_segs: Maximum number of segments that can be passed to the
1887 * NIC for GSO
1888 *
1889 * @dcbnl_ops: Data Center Bridging netlink ops
1890 * @num_tc: Number of traffic classes in the net device
1891 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1892 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
1893 *
1894 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1895 *
1896 * @priomap: XXX: need comments on this one
1897 * @phydev: Physical device may attach itself
1898 * for hardware timestamping
e679c9c1 1899 * @sfp_bus: attached &struct sfp_bus structure.
1a33e10e 1900 *
1a33e10e
CW
1901 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
1902 * @qdisc_running_key: lockdep class annotating Qdisc->running seqcount
536721b1 1903 *
d746d707
AK
1904 * @proto_down: protocol port state information can be sent to the
1905 * switch driver and used to set the phys state of the
1906 * switch port.
1907 *
61941143
HK
1908 * @wol_enabled: Wake-on-LAN is enabled
1909 *
29863d41
WW
1910 * @threaded: napi threaded mode is enabled
1911 *
93642e14
JP
1912 * @net_notifier_list: List of per-net netdev notifier block
1913 * that follow this device when it is moved
1914 * to another network namespace.
1915 *
30e9bb84
AT
1916 * @macsec_ops: MACsec offloading ops
1917 *
cc4e3835
JK
1918 * @udp_tunnel_nic_info: static structure describing the UDP tunnel
1919 * offload capabilities of the device
1920 * @udp_tunnel_nic: UDP tunnel offload state
ffa59b0b 1921 * @xdp_state: stores info on attached XDP BPF programs
cc4e3835 1922 *
a93bdcb9
MCC
1923 * @nested_level: Used as as a parameter of spin_lock_nested() of
1924 * dev->addr_list_lock.
1925 * @unlink_list: As netif_addr_lock() can be called recursively,
1926 * keep a list of interfaces to be deleted.
1927 *
1da177e4
LT
1928 * FIXME: cleanup struct net_device such that network protocol info
1929 * moves out.
1930 */
1931
d94d9fee 1932struct net_device {
1da177e4 1933 char name[IFNAMSIZ];
ff927412 1934 struct netdev_name_node *name_node;
6c557001 1935 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
1936 /*
1937 * I/O specific fields
1938 * FIXME: Merge these and struct ifmap into one
1939 */
536721b1
KK
1940 unsigned long mem_end;
1941 unsigned long mem_start;
1942 unsigned long base_addr;
1da177e4
LT
1943
1944 /*
536721b1
KK
1945 * Some hardware also needs these fields (state,dev_list,
1946 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
1947 * part of the usual set specified in Space.c.
1948 */
1949
1da177e4
LT
1950 unsigned long state;
1951
7562f876 1952 struct list_head dev_list;
bea3348e 1953 struct list_head napi_list;
44a0873d 1954 struct list_head unreg_list;
5cde2829 1955 struct list_head close_list;
7866a621
SN
1956 struct list_head ptype_all;
1957 struct list_head ptype_specific;
2f268f12 1958
2f268f12
VF
1959 struct {
1960 struct list_head upper;
1961 struct list_head lower;
1962 } adj_list;
1963
28af22c6
JDB
1964 /* Read-mostly cache-line for fast-path access */
1965 unsigned int flags;
1966 unsigned int priv_flags;
1967 const struct net_device_ops *netdev_ops;
1968 int ifindex;
1969 unsigned short gflags;
1970 unsigned short hard_header_len;
1971
1972 /* Note : dev->mtu is often read without holding a lock.
1973 * Writers usually hold RTNL.
1974 * It is recommended to use READ_ONCE() to annotate the reads,
1975 * and to use WRITE_ONCE() to annotate the writes.
1976 */
1977 unsigned int mtu;
1978 unsigned short needed_headroom;
1979 unsigned short needed_tailroom;
1980
c8f44aff 1981 netdev_features_t features;
c8f44aff 1982 netdev_features_t hw_features;
c8f44aff 1983 netdev_features_t wanted_features;
c8f44aff 1984 netdev_features_t vlan_features;
6a674e9c 1985 netdev_features_t hw_enc_features;
0d89d203 1986 netdev_features_t mpls_features;
802ab55a 1987 netdev_features_t gso_partial_features;
04ed3e74 1988
28af22c6
JDB
1989 unsigned int min_mtu;
1990 unsigned int max_mtu;
1991 unsigned short type;
1992 unsigned char min_header_len;
1993 unsigned char name_assign_type;
1994
7a66bbc9 1995 int group;
1da177e4 1996
28af22c6 1997 struct net_device_stats stats; /* not used by modern drivers */
015f0688 1998
015f0688
ED
1999 atomic_long_t rx_dropped;
2000 atomic_long_t tx_dropped;
6e7333d3 2001 atomic_long_t rx_nohandler;
1da177e4 2002
b2d3bcfa
DD
2003 /* Stats to monitor link on/off, flapping */
2004 atomic_t carrier_up_count;
2005 atomic_t carrier_down_count;
2006
b86e0280 2007#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
2008 const struct iw_handler_def *wireless_handlers;
2009 struct iw_public_data *wireless_data;
b86e0280 2010#endif
76fd8593 2011 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
2012#ifdef CONFIG_NET_L3_MASTER_DEV
2013 const struct l3mdev_ops *l3mdev_ops;
2014#endif
f997c55c
AA
2015#if IS_ENABLED(CONFIG_IPV6)
2016 const struct ndisc_ops *ndisc_ops;
2017#endif
1da177e4 2018
9cb0d21d 2019#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
2020 const struct xfrmdev_ops *xfrmdev_ops;
2021#endif
2022
a5c37c63
IL
2023#if IS_ENABLED(CONFIG_TLS_DEVICE)
2024 const struct tlsdev_ops *tlsdev_ops;
2025#endif
2026
3b04ddde
SH
2027 const struct header_ops *header_ops;
2028
536721b1
KK
2029 unsigned char operstate;
2030 unsigned char link_mode;
b00055aa 2031
536721b1
KK
2032 unsigned char if_port;
2033 unsigned char dma;
bdc220da 2034
1da177e4 2035 /* Interface address info. */
536721b1
KK
2036 unsigned char perm_addr[MAX_ADDR_LEN];
2037 unsigned char addr_assign_type;
2038 unsigned char addr_len;
5343da4c
TY
2039 unsigned char upper_level;
2040 unsigned char lower_level;
1fc70edb 2041
a0a9663d 2042 unsigned short neigh_priv_len;
536721b1
KK
2043 unsigned short dev_id;
2044 unsigned short dev_port;
28af22c6
JDB
2045 unsigned short padded;
2046
ccffad25 2047 spinlock_t addr_list_lock;
28af22c6 2048 int irq;
1fc70edb 2049
536721b1
KK
2050 struct netdev_hw_addr_list uc;
2051 struct netdev_hw_addr_list mc;
2052 struct netdev_hw_addr_list dev_addrs;
2053
4c3d5e7b
ED
2054#ifdef CONFIG_SYSFS
2055 struct kset *queues_kset;
1fc70edb
TY
2056#endif
2057#ifdef CONFIG_LOCKDEP
2058 struct list_head unlink_list;
4c3d5e7b 2059#endif
9d45abe1
WC
2060 unsigned int promiscuity;
2061 unsigned int allmulti;
1fc70edb
TY
2062 bool uc_promisc;
2063#ifdef CONFIG_LOCKDEP
2064 unsigned char nested_level;
2065#endif
1da177e4 2066
1da177e4 2067
5e82b4b2 2068 /* Protocol-specific pointers */
65ac6a5f 2069
d11ead75 2070#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 2071 struct vlan_info __rcu *vlan_info;
65ac6a5f 2072#endif
34a430d7 2073#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 2074 struct dsa_port *dsa_ptr;
37cb0620
YX
2075#endif
2076#if IS_ENABLED(CONFIG_TIPC)
536721b1 2077 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 2078#endif
89e58148 2079#if IS_ENABLED(CONFIG_IRDA) || IS_ENABLED(CONFIG_ATALK)
536721b1 2080 void *atalk_ptr;
89e58148 2081#endif
536721b1 2082 struct in_device __rcu *ip_ptr;
330c7272 2083#if IS_ENABLED(CONFIG_DECNET)
536721b1 2084 struct dn_dev __rcu *dn_ptr;
330c7272 2085#endif
536721b1 2086 struct inet6_dev __rcu *ip6_ptr;
19ff13f2 2087#if IS_ENABLED(CONFIG_AX25)
536721b1 2088 void *ax25_ptr;
19ff13f2 2089#endif
536721b1 2090 struct wireless_dev *ieee80211_ptr;
98a18b6f 2091 struct wpan_dev *ieee802154_ptr;
03c57747
RS
2092#if IS_ENABLED(CONFIG_MPLS_ROUTING)
2093 struct mpls_dev __rcu *mpls_ptr;
2094#endif
1da177e4 2095
9356b8fc 2096/*
cd13539b 2097 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 2098 */
9356b8fc 2099 /* Interface address info used in eth_type_trans() */
536721b1 2100 unsigned char *dev_addr;
f001fde5 2101
0a9627f2 2102 struct netdev_rx_queue *_rx;
0a9627f2 2103 unsigned int num_rx_queues;
62fe0b40 2104 unsigned int real_num_rx_queues;
0a9627f2 2105
7acedaf5 2106 struct bpf_prog __rcu *xdp_prog;
3b47d303 2107 unsigned long gro_flush_timeout;
6f8b12d6 2108 int napi_defer_hard_irqs;
61391cde 2109 rx_handler_func_t __rcu *rx_handler;
2110 void __rcu *rx_handler_data;
e8a0464c 2111
4cda01e8 2112#ifdef CONFIG_NET_CLS_ACT
46209401 2113 struct mini_Qdisc __rcu *miniq_ingress;
d2788d34 2114#endif
24824a09 2115 struct netdev_queue __rcu *ingress_queue;
e687ad60 2116#ifdef CONFIG_NETFILTER_INGRESS
960632ec 2117 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 2118#endif
d2788d34 2119
536721b1 2120 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
2121#ifdef CONFIG_RFS_ACCEL
2122 struct cpu_rmap *rx_cpu_rmap;
2123#endif
2124 struct hlist_node index_hlist;
cd13539b
ED
2125
2126/*
2127 * Cache lines mostly used on transmit path
2128 */
e8a0464c
DM
2129 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
2130 unsigned int num_tx_queues;
fd2ea0a7 2131 unsigned int real_num_tx_queues;
af356afa 2132 struct Qdisc *qdisc;
0cd29503 2133 unsigned int tx_queue_len;
c3f26a26 2134 spinlock_t tx_global_lock;
75ccae62
THJ
2135
2136 struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
cd13539b 2137
bf264145 2138#ifdef CONFIG_XPS
044ab86d 2139 struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
bf264145 2140#endif
1f211a1b 2141#ifdef CONFIG_NET_CLS_ACT
46209401 2142 struct mini_Qdisc __rcu *miniq_egress;
1f211a1b 2143#endif
0c4f691f 2144
75ccae62
THJ
2145#ifdef CONFIG_NET_SCHED
2146 DECLARE_HASHTABLE (qdisc_hash, 4);
2147#endif
9356b8fc 2148 /* These may be needed for future network-power-down code. */
9356b8fc 2149 struct timer_list watchdog_timer;
75ccae62 2150 int watchdog_timeo;
9356b8fc 2151
829eb208
RP
2152 u32 proto_down_reason;
2153
1da177e4 2154 struct list_head todo_list;
919067cc
ED
2155
2156#ifdef CONFIG_PCPU_DEV_REFCNT
75ccae62 2157 int __percpu *pcpu_refcnt;
919067cc
ED
2158#else
2159 refcount_t dev_refcnt;
2160#endif
1da177e4 2161
e014debe 2162 struct list_head link_watch_list;
572a103d 2163
1da177e4 2164 enum { NETREG_UNINITIALIZED=0,
b17a7c17 2165 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
2166 NETREG_UNREGISTERING, /* called unregister_netdevice */
2167 NETREG_UNREGISTERED, /* completed unregister todo */
2168 NETREG_RELEASED, /* called free_netdev */
937f1ba5 2169 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
2170 } reg_state:8;
2171
536721b1 2172 bool dismantle;
a2835763
PM
2173
2174 enum {
2175 RTNL_LINK_INITIALIZED,
2176 RTNL_LINK_INITIALIZING,
2177 } rtnl_link_state:16;
1da177e4 2178
cf124db5
DM
2179 bool needs_free_netdev;
2180 void (*priv_destructor)(struct net_device *dev);
1da177e4 2181
1da177e4 2182#ifdef CONFIG_NETPOLL
5fbee843 2183 struct netpoll_info __rcu *npinfo;
1da177e4 2184#endif
eae792b7 2185
0c5c9fb5 2186 possible_net_t nd_net;
4a1c5371 2187
4951704b 2188 /* mid-layer private */
4e096a18
OR
2189 void *ml_priv;
2190 enum netdev_ml_priv_type ml_priv_type;
2191
a7855c78 2192 union {
536721b1 2193 struct pcpu_lstats __percpu *lstats;
8f84985f 2194 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2195 struct pcpu_dstats __percpu *dstats;
a7855c78 2196 };
536721b1 2197
fb585b44 2198#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2199 struct garp_port __rcu *garp_port;
fb585b44
TK
2200#endif
2201#if IS_ENABLED(CONFIG_MRP)
febf018d 2202 struct mrp_port __rcu *mrp_port;
fb585b44 2203#endif
1da177e4 2204
5e82b4b2 2205 struct device dev;
0c509a6c 2206 const struct attribute_group *sysfs_groups[4];
a953be53 2207 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2208
38f7b870 2209 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2210
82cc1a7a
PWJ
2211 /* for setting kernel sock attribute on TCP connection setup */
2212#define GSO_MAX_SIZE 65536
2213 unsigned int gso_max_size;
30b678d8
BH
2214#define GSO_MAX_SEGS 65535
2215 u16 gso_max_segs;
743b03a8 2216
7a6b6f51 2217#ifdef CONFIG_DCB
32953543 2218 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2219#endif
ffcfe25b 2220 s16 num_tc;
5e82b4b2
BH
2221 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2222 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2223
d11ead75 2224#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2225 unsigned int fcoe_ddp_xid;
5bc1421e 2226#endif
86f8515f 2227#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2228 struct netprio_map __rcu *priomap;
4d288d57 2229#endif
5e82b4b2 2230 struct phy_device *phydev;
e679c9c1 2231 struct sfp_bus *sfp_bus;
1a33e10e
CW
2232 struct lock_class_key *qdisc_tx_busylock;
2233 struct lock_class_key *qdisc_running_key;
5e82b4b2 2234 bool proto_down;
61941143 2235 unsigned wol_enabled:1;
29863d41 2236 unsigned threaded:1;
93642e14
JP
2237
2238 struct list_head net_notifier_list;
30e9bb84
AT
2239
2240#if IS_ENABLED(CONFIG_MACSEC)
2241 /* MACsec management functions */
2242 const struct macsec_ops *macsec_ops;
2243#endif
cc4e3835
JK
2244 const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
2245 struct udp_tunnel_nic *udp_tunnel_nic;
7f0a8382
AN
2246
2247 /* protected by rtnl_lock */
2248 struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
1da177e4 2249};
43cb76d9 2250#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2251
b5cdae32
DM
2252static inline bool netif_elide_gro(const struct net_device *dev)
2253{
2254 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2255 return true;
2256 return false;
2257}
2258
1da177e4 2259#define NETDEV_ALIGN 32
1da177e4 2260
4f57c087
JF
2261static inline
2262int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2263{
2264 return dev->prio_tc_map[prio & TC_BITMASK];
2265}
2266
2267static inline
2268int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2269{
2270 if (tc >= dev->num_tc)
2271 return -EINVAL;
2272
2273 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2274 return 0;
2275}
2276
8d059b0f 2277int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2278void netdev_reset_tc(struct net_device *dev);
2279int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2280int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2281
2282static inline
2283int netdev_get_num_tc(struct net_device *dev)
2284{
2285 return dev->num_tc;
2286}
2287
f468f21b
TT
2288static inline void net_prefetch(void *p)
2289{
2290 prefetch(p);
2291#if L1_CACHE_BYTES < 128
2292 prefetch((u8 *)p + L1_CACHE_BYTES);
2293#endif
2294}
2295
2296static inline void net_prefetchw(void *p)
2297{
2298 prefetchw(p);
2299#if L1_CACHE_BYTES < 128
2300 prefetchw((u8 *)p + L1_CACHE_BYTES);
2301#endif
2302}
2303
ffcfe25b
AD
2304void netdev_unbind_sb_channel(struct net_device *dev,
2305 struct net_device *sb_dev);
2306int netdev_bind_sb_channel_queue(struct net_device *dev,
2307 struct net_device *sb_dev,
2308 u8 tc, u16 count, u16 offset);
2309int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2310static inline int netdev_get_sb_channel(struct net_device *dev)
2311{
2312 return max_t(int, -dev->num_tc, 0);
2313}
2314
e8a0464c
DM
2315static inline
2316struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2317 unsigned int index)
2318{
2319 return &dev->_tx[index];
2320}
2321
10c51b56
DB
2322static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2323 const struct sk_buff *skb)
2324{
2325 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2326}
2327
e8a0464c
DM
2328static inline void netdev_for_each_tx_queue(struct net_device *dev,
2329 void (*f)(struct net_device *,
2330 struct netdev_queue *,
2331 void *),
2332 void *arg)
2333{
2334 unsigned int i;
2335
2336 for (i = 0; i < dev->num_tx_queues; i++)
2337 f(dev, &dev->_tx[i], arg);
2338}
2339
1a33e10e
CW
2340#define netdev_lockdep_set_classes(dev) \
2341{ \
2342 static struct lock_class_key qdisc_tx_busylock_key; \
2343 static struct lock_class_key qdisc_running_key; \
2344 static struct lock_class_key qdisc_xmit_lock_key; \
845e0ebb 2345 static struct lock_class_key dev_addr_list_lock_key; \
1a33e10e
CW
2346 unsigned int i; \
2347 \
2348 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
2349 (dev)->qdisc_running_key = &qdisc_running_key; \
845e0ebb
CW
2350 lockdep_set_class(&(dev)->addr_list_lock, \
2351 &dev_addr_list_lock_key); \
1a33e10e
CW
2352 for (i = 0; i < (dev)->num_tx_queues; i++) \
2353 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2354 &qdisc_xmit_lock_key); \
2355}
2356
b71b5837
PA
2357u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2358 struct net_device *sb_dev);
4bd97d51
PA
2359struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2360 struct sk_buff *skb,
2361 struct net_device *sb_dev);
8c4c49df 2362
871b642a
PA
2363/* returns the headroom that the master device needs to take in account
2364 * when forwarding to this dev
2365 */
2366static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2367{
2368 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2369}
2370
2371static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2372{
2373 if (dev->netdev_ops->ndo_set_rx_headroom)
2374 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2375}
2376
2377/* set the device rx headroom to the dev's default */
2378static inline void netdev_reset_rx_headroom(struct net_device *dev)
2379{
2380 netdev_set_rx_headroom(dev, -1);
2381}
2382
4e096a18
OR
2383static inline void *netdev_get_ml_priv(struct net_device *dev,
2384 enum netdev_ml_priv_type type)
2385{
2386 if (dev->ml_priv_type != type)
2387 return NULL;
2388
2389 return dev->ml_priv;
2390}
2391
2392static inline void netdev_set_ml_priv(struct net_device *dev,
2393 void *ml_priv,
2394 enum netdev_ml_priv_type type)
2395{
2396 WARN(dev->ml_priv_type && dev->ml_priv_type != type,
2397 "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
2398 dev->ml_priv_type, type);
2399 WARN(!dev->ml_priv_type && dev->ml_priv,
2400 "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
2401
2402 dev->ml_priv = ml_priv;
2403 dev->ml_priv_type = type;
2404}
2405
c346dca1
YH
2406/*
2407 * Net namespace inlines
2408 */
2409static inline
2410struct net *dev_net(const struct net_device *dev)
2411{
c2d9ba9b 2412 return read_pnet(&dev->nd_net);
c346dca1
YH
2413}
2414
2415static inline
f5aa23fd 2416void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2417{
0c5c9fb5 2418 write_pnet(&dev->nd_net, net);
c346dca1
YH
2419}
2420
bea3348e
SH
2421/**
2422 * netdev_priv - access network device private data
2423 * @dev: network device
2424 *
2425 * Get network device private data
2426 */
6472ce60 2427static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2428{
1ce8e7b5 2429 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2430}
2431
1da177e4
LT
2432/* Set the sysfs physical device reference for the network logical device
2433 * if set prior to registration will cause a symlink during initialization.
2434 */
43cb76d9 2435#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2436
384912ed 2437/* Set the sysfs device type for the network logical device to allow
3f79410c 2438 * fine-grained identification of different network device types. For
5e82b4b2 2439 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2440 */
2441#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2442
82dc3c63
ED
2443/* Default NAPI poll() weight
2444 * Device drivers are strongly advised to not use bigger value
2445 */
2446#define NAPI_POLL_WEIGHT 64
2447
3b582cc1 2448/**
5e82b4b2 2449 * netif_napi_add - initialize a NAPI context
3b582cc1 2450 * @dev: network device
5e82b4b2 2451 * @napi: NAPI context
3b582cc1
SH
2452 * @poll: polling function
2453 * @weight: default weight
2454 *
5e82b4b2
BH
2455 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2456 * *any* of the other NAPI-related functions.
3b582cc1 2457 */
d565b0a1
HX
2458void netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2459 int (*poll)(struct napi_struct *, int), int weight);
bea3348e 2460
d64b5e85 2461/**
5e82b4b2 2462 * netif_tx_napi_add - initialize a NAPI context
d64b5e85 2463 * @dev: network device
5e82b4b2 2464 * @napi: NAPI context
d64b5e85
ED
2465 * @poll: polling function
2466 * @weight: default weight
2467 *
2468 * This variant of netif_napi_add() should be used from drivers using NAPI
2469 * to exclusively poll a TX queue.
2470 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2471 */
2472static inline void netif_tx_napi_add(struct net_device *dev,
2473 struct napi_struct *napi,
2474 int (*poll)(struct napi_struct *, int),
2475 int weight)
2476{
2477 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2478 netif_napi_add(dev, napi, poll, weight);
2479}
2480
5198d545
JK
2481/**
2482 * __netif_napi_del - remove a NAPI context
2483 * @napi: NAPI context
2484 *
2485 * Warning: caller must observe RCU grace period before freeing memory
2486 * containing @napi. Drivers might want to call this helper to combine
2487 * all the needed RCU grace periods into a single one.
2488 */
2489void __netif_napi_del(struct napi_struct *napi);
2490
d8156534 2491/**
5e82b4b2
BH
2492 * netif_napi_del - remove a NAPI context
2493 * @napi: NAPI context
d8156534 2494 *
5e82b4b2 2495 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2496 */
5198d545
JK
2497static inline void netif_napi_del(struct napi_struct *napi)
2498{
2499 __netif_napi_del(napi);
2500 synchronize_net();
2501}
d565b0a1
HX
2502
2503struct napi_gro_cb {
78a478d0 2504 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
5e82b4b2 2505 void *frag0;
78a478d0 2506
7489594c
HX
2507 /* Length of frag0. */
2508 unsigned int frag0_len;
2509
86911732 2510 /* This indicates where we are processing relative to skb->data. */
5e82b4b2 2511 int data_offset;
86911732 2512
d565b0a1 2513 /* This is non-zero if the packet cannot be merged with the new skb. */
bf5a755f
JC
2514 u16 flush;
2515
2516 /* Save the IP ID here and check when we get to the transport layer */
2517 u16 flush_id;
d565b0a1
HX
2518
2519 /* Number of segments aggregated. */
2e71a6f8
ED
2520 u16 count;
2521
15e2396d
TH
2522 /* Start offset for remote checksum offload */
2523 u16 gro_remcsum_start;
2524
2e71a6f8
ED
2525 /* jiffies when first packet was created/queued */
2526 unsigned long age;
86347245 2527
afe93325 2528 /* Used in ipv6_gro_receive() and foo-over-udp */
b582ef09
OG
2529 u16 proto;
2530
baa32ff4
TH
2531 /* This is non-zero if the packet may be of the same flow. */
2532 u8 same_flow:1;
2533
fac8e0f5
JG
2534 /* Used in tunnel GRO receive */
2535 u8 encap_mark:1;
573e8fca
TH
2536
2537 /* GRO checksum is valid */
2538 u8 csum_valid:1;
2539
662880f4
TH
2540 /* Number of checksums via CHECKSUM_UNNECESSARY */
2541 u8 csum_cnt:3;
c3c7c254 2542
baa32ff4
TH
2543 /* Free the skb? */
2544 u8 free:2;
2545#define NAPI_GRO_FREE 1
2546#define NAPI_GRO_FREE_STOLEN_HEAD 2
2547
efc98d08
TH
2548 /* Used in foo-over-udp, set in udp[46]_gro_receive */
2549 u8 is_ipv6:1;
2550
a0ca153f
AD
2551 /* Used in GRE, set in fou/gue_gro_receive */
2552 u8 is_fou:1;
2553
1530545e
AD
2554 /* Used to determine if flush_id can be ignored */
2555 u8 is_atomic:1;
2556
fcd91dd4
SD
2557 /* Number of gro_receive callbacks this packet already went through */
2558 u8 recursion_counter:4;
2559
3a1296a3
SK
2560 /* GRO is done by frag_list pointer chaining. */
2561 u8 is_flist:1;
baa32ff4 2562
bf5a755f
JC
2563 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
2564 __wsum csum;
2565
c3c7c254
ED
2566 /* used in skb_gro_receive() slow path */
2567 struct sk_buff *last;
d565b0a1
HX
2568};
2569
2570#define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
d8156534 2571
fcd91dd4
SD
2572#define GRO_RECURSION_LIMIT 15
2573static inline int gro_recursion_inc_test(struct sk_buff *skb)
2574{
2575 return ++NAPI_GRO_CB(skb)->recursion_counter == GRO_RECURSION_LIMIT;
2576}
2577
d4546c25
DM
2578typedef struct sk_buff *(*gro_receive_t)(struct list_head *, struct sk_buff *);
2579static inline struct sk_buff *call_gro_receive(gro_receive_t cb,
2580 struct list_head *head,
2581 struct sk_buff *skb)
fcd91dd4
SD
2582{
2583 if (unlikely(gro_recursion_inc_test(skb))) {
2584 NAPI_GRO_CB(skb)->flush |= 1;
2585 return NULL;
2586 }
2587
2588 return cb(head, skb);
2589}
2590
d4546c25
DM
2591typedef struct sk_buff *(*gro_receive_sk_t)(struct sock *, struct list_head *,
2592 struct sk_buff *);
2593static inline struct sk_buff *call_gro_receive_sk(gro_receive_sk_t cb,
2594 struct sock *sk,
2595 struct list_head *head,
2596 struct sk_buff *skb)
fcd91dd4
SD
2597{
2598 if (unlikely(gro_recursion_inc_test(skb))) {
2599 NAPI_GRO_CB(skb)->flush |= 1;
2600 return NULL;
2601 }
2602
2603 return cb(sk, head, skb);
2604}
2605
1da177e4 2606struct packet_type {
f2ccd8fa 2607 __be16 type; /* This is really htons(ether_type). */
fa788d98 2608 bool ignore_outgoing;
f2ccd8fa
DM
2609 struct net_device *dev; /* NULL is wildcarded here */
2610 int (*func) (struct sk_buff *,
2611 struct net_device *,
2612 struct packet_type *,
2613 struct net_device *);
17266ee9
EC
2614 void (*list_func) (struct list_head *,
2615 struct packet_type *,
2616 struct net_device *);
c0de08d0
EL
2617 bool (*id_match)(struct packet_type *ptype,
2618 struct sock *sk);
1da177e4
LT
2619 void *af_packet_priv;
2620 struct list_head list;
2621};
2622
f191a1d1 2623struct offload_callbacks {
576a30eb 2624 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2625 netdev_features_t features);
d4546c25
DM
2626 struct sk_buff *(*gro_receive)(struct list_head *head,
2627 struct sk_buff *skb);
299603e8 2628 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2629};
2630
2631struct packet_offload {
2632 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2633 u16 priority;
f191a1d1
VY
2634 struct offload_callbacks callbacks;
2635 struct list_head list;
1da177e4
LT
2636};
2637
5e82b4b2 2638/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f
LR
2639struct pcpu_sw_netstats {
2640 u64 rx_packets;
2641 u64 rx_bytes;
2642 u64 tx_packets;
2643 u64 tx_bytes;
2644 struct u64_stats_sync syncp;
9a5ee462 2645} __aligned(4 * sizeof(u64));
52bb6677
LR
2646
2647struct pcpu_lstats {
fd2f4737
ED
2648 u64_stats_t packets;
2649 u64_stats_t bytes;
52bb6677 2650 struct u64_stats_sync syncp;
9a5ee462 2651} __aligned(2 * sizeof(u64));
8f84985f 2652
de7d5084
ED
2653void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2654
451b05f4
FF
2655static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
2656{
2657 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2658
2659 u64_stats_update_begin(&tstats->syncp);
2660 tstats->rx_bytes += len;
2661 tstats->rx_packets++;
2662 u64_stats_update_end(&tstats->syncp);
2663}
2664
d3fd6548
HK
2665static inline void dev_sw_netstats_tx_add(struct net_device *dev,
2666 unsigned int packets,
2667 unsigned int len)
2668{
2669 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2670
2671 u64_stats_update_begin(&tstats->syncp);
2672 tstats->tx_bytes += len;
2673 tstats->tx_packets += packets;
2674 u64_stats_update_end(&tstats->syncp);
2675}
2676
dd5382a0
ED
2677static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
2678{
2679 struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
2680
2681 u64_stats_update_begin(&lstats->syncp);
fd2f4737
ED
2682 u64_stats_add(&lstats->bytes, len);
2683 u64_stats_inc(&lstats->packets);
dd5382a0
ED
2684 u64_stats_update_end(&lstats->syncp);
2685}
2686
aabc92bb
PNA
2687#define __netdev_alloc_pcpu_stats(type, gfp) \
2688({ \
2689 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2690 if (pcpu_stats) { \
2691 int __cpu; \
2692 for_each_possible_cpu(__cpu) { \
2693 typeof(type) *stat; \
2694 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2695 u64_stats_init(&stat->syncp); \
2696 } \
2697 } \
2698 pcpu_stats; \
1c213bd2
WC
2699})
2700
aabc92bb 2701#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2702 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2703
81b01894
HK
2704#define devm_netdev_alloc_pcpu_stats(dev, type) \
2705({ \
2706 typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
2707 if (pcpu_stats) { \
2708 int __cpu; \
2709 for_each_possible_cpu(__cpu) { \
2710 typeof(type) *stat; \
2711 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2712 u64_stats_init(&stat->syncp); \
2713 } \
2714 } \
2715 pcpu_stats; \
2716})
2717
764f5e54
JP
2718enum netdev_lag_tx_type {
2719 NETDEV_LAG_TX_TYPE_UNKNOWN,
2720 NETDEV_LAG_TX_TYPE_RANDOM,
2721 NETDEV_LAG_TX_TYPE_BROADCAST,
2722 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2723 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2724 NETDEV_LAG_TX_TYPE_HASH,
2725};
2726
f44aa9ef
JH
2727enum netdev_lag_hash {
2728 NETDEV_LAG_HASH_NONE,
2729 NETDEV_LAG_HASH_L2,
2730 NETDEV_LAG_HASH_L34,
2731 NETDEV_LAG_HASH_L23,
2732 NETDEV_LAG_HASH_E23,
2733 NETDEV_LAG_HASH_E34,
7b8fc010 2734 NETDEV_LAG_HASH_VLAN_SRCMAC,
f44aa9ef
JH
2735 NETDEV_LAG_HASH_UNKNOWN,
2736};
2737
764f5e54
JP
2738struct netdev_lag_upper_info {
2739 enum netdev_lag_tx_type tx_type;
f44aa9ef 2740 enum netdev_lag_hash hash_type;
764f5e54
JP
2741};
2742
fb1b2e3c
JP
2743struct netdev_lag_lower_state_info {
2744 u8 link_up : 1,
2745 tx_enabled : 1;
2746};
2747
1da177e4
LT
2748#include <linux/notifier.h>
2749
ede2762d
KT
2750/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2751 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2752 * adding new types.
dcfe1421 2753 */
ede2762d
KT
2754enum netdev_cmd {
2755 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2756 NETDEV_DOWN,
2757 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2758 detected a hardware crash and restarted
2759 - we can use this eg to kick tcp sessions
2760 once done */
ede2762d
KT
2761 NETDEV_CHANGE, /* Notify device state change */
2762 NETDEV_REGISTER,
2763 NETDEV_UNREGISTER,
2764 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2765 NETDEV_CHANGEADDR, /* notify after the address change */
2766 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2767 NETDEV_GOING_DOWN,
2768 NETDEV_CHANGENAME,
2769 NETDEV_FEAT_CHANGE,
2770 NETDEV_BONDING_FAILOVER,
2771 NETDEV_PRE_UP,
2772 NETDEV_PRE_TYPE_CHANGE,
2773 NETDEV_POST_TYPE_CHANGE,
2774 NETDEV_POST_INIT,
ede2762d
KT
2775 NETDEV_RELEASE,
2776 NETDEV_NOTIFY_PEERS,
2777 NETDEV_JOIN,
2778 NETDEV_CHANGEUPPER,
2779 NETDEV_RESEND_IGMP,
2780 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2781 NETDEV_CHANGEINFODATA,
2782 NETDEV_BONDING_INFO,
2783 NETDEV_PRECHANGEUPPER,
2784 NETDEV_CHANGELOWERSTATE,
2785 NETDEV_UDP_TUNNEL_PUSH_INFO,
2786 NETDEV_UDP_TUNNEL_DROP_INFO,
2787 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2788 NETDEV_CVLAN_FILTER_PUSH_INFO,
2789 NETDEV_CVLAN_FILTER_DROP_INFO,
2790 NETDEV_SVLAN_FILTER_PUSH_INFO,
2791 NETDEV_SVLAN_FILTER_DROP_INFO,
ede2762d
KT
2792};
2793const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2794
f629d208
JP
2795int register_netdevice_notifier(struct notifier_block *nb);
2796int unregister_netdevice_notifier(struct notifier_block *nb);
a30c7b42
JP
2797int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2798int unregister_netdevice_notifier_net(struct net *net,
2799 struct notifier_block *nb);
93642e14
JP
2800int register_netdevice_notifier_dev_net(struct net_device *dev,
2801 struct notifier_block *nb,
2802 struct netdev_net_notifier *nn);
2803int unregister_netdevice_notifier_dev_net(struct net_device *dev,
2804 struct notifier_block *nb,
2805 struct netdev_net_notifier *nn);
351638e7
JP
2806
2807struct netdev_notifier_info {
51d0c047
DA
2808 struct net_device *dev;
2809 struct netlink_ext_ack *extack;
351638e7
JP
2810};
2811
af7d6cce
SD
2812struct netdev_notifier_info_ext {
2813 struct netdev_notifier_info info; /* must be first */
2814 union {
2815 u32 mtu;
2816 } ext;
2817};
2818
be9efd36
JP
2819struct netdev_notifier_change_info {
2820 struct netdev_notifier_info info; /* must be first */
2821 unsigned int flags_changed;
2822};
2823
0e4ead9d
JP
2824struct netdev_notifier_changeupper_info {
2825 struct netdev_notifier_info info; /* must be first */
2826 struct net_device *upper_dev; /* new upper dev */
2827 bool master; /* is upper dev master */
5e82b4b2 2828 bool linking; /* is the notification for link or unlink */
29bf24af 2829 void *upper_info; /* upper dev info */
0e4ead9d
JP
2830};
2831
04d48266
JP
2832struct netdev_notifier_changelowerstate_info {
2833 struct netdev_notifier_info info; /* must be first */
2834 void *lower_state_info; /* is lower dev state */
2835};
2836
1570415f
PM
2837struct netdev_notifier_pre_changeaddr_info {
2838 struct netdev_notifier_info info; /* must be first */
2839 const unsigned char *dev_addr;
2840};
2841
75538c2b
CW
2842static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2843 struct net_device *dev)
2844{
2845 info->dev = dev;
51d0c047 2846 info->extack = NULL;
75538c2b
CW
2847}
2848
351638e7
JP
2849static inline struct net_device *
2850netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2851{
2852 return info->dev;
2853}
2854
51d0c047
DA
2855static inline struct netlink_ext_ack *
2856netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2857{
2858 return info->extack;
2859}
2860
f629d208 2861int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
dcfe1421
AW
2862
2863
1da177e4
LT
2864extern rwlock_t dev_base_lock; /* Device list lock */
2865
881d966b
EB
2866#define for_each_netdev(net, d) \
2867 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2868#define for_each_netdev_reverse(net, d) \
2869 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2870#define for_each_netdev_rcu(net, d) \
2871 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2872#define for_each_netdev_safe(net, d, n) \
2873 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2874#define for_each_netdev_continue(net, d) \
2875 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
afa0df59
JP
2876#define for_each_netdev_continue_reverse(net, d) \
2877 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
2878 dev_list)
254245d2 2879#define for_each_netdev_continue_rcu(net, d) \
2880 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 2881#define for_each_netdev_in_bond_rcu(bond, slave) \
2882 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 2883 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 2884#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 2885
a050c33f
DL
2886static inline struct net_device *next_net_device(struct net_device *dev)
2887{
2888 struct list_head *lh;
2889 struct net *net;
2890
c346dca1 2891 net = dev_net(dev);
a050c33f
DL
2892 lh = dev->dev_list.next;
2893 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2894}
2895
ce81b76a
ED
2896static inline struct net_device *next_net_device_rcu(struct net_device *dev)
2897{
2898 struct list_head *lh;
2899 struct net *net;
2900
2901 net = dev_net(dev);
ccf43438 2902 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
2903 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2904}
2905
a050c33f
DL
2906static inline struct net_device *first_net_device(struct net *net)
2907{
2908 return list_empty(&net->dev_base_head) ? NULL :
2909 net_device_entry(net->dev_base_head.next);
2910}
7562f876 2911
ccf43438
ED
2912static inline struct net_device *first_net_device_rcu(struct net *net)
2913{
2914 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
2915
2916 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
2917}
2918
f629d208
JP
2919int netdev_boot_setup_check(struct net_device *dev);
2920unsigned long netdev_boot_base(const char *prefix, int unit);
2921struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
2922 const char *hwaddr);
2923struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
f629d208
JP
2924void dev_add_pack(struct packet_type *pt);
2925void dev_remove_pack(struct packet_type *pt);
2926void __dev_remove_pack(struct packet_type *pt);
2927void dev_add_offload(struct packet_offload *po);
2928void dev_remove_offload(struct packet_offload *po);
f629d208 2929
a54acb3a 2930int dev_get_iflink(const struct net_device *dev);
fc4099f1 2931int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
ddb94eaf
PNA
2932int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
2933 struct net_device_path_stack *stack);
6c555490
WC
2934struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
2935 unsigned short mask);
f629d208
JP
2936struct net_device *dev_get_by_name(struct net *net, const char *name);
2937struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
2938struct net_device *__dev_get_by_name(struct net *net, const char *name);
2939int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 2940int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 2941void dev_close(struct net_device *dev);
2942void dev_close_many(struct list_head *head, bool unlink);
f629d208 2943void dev_disable_lro(struct net_device *dev);
0c4b51f0 2944int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 2945u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 2946 struct net_device *sb_dev);
a4ea8a3d 2947u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 2948 struct net_device *sb_dev);
36ccdf85 2949
2b4aa3ce 2950int dev_queue_xmit(struct sk_buff *skb);
eadec877 2951int dev_queue_xmit_accel(struct sk_buff *skb, struct net_device *sb_dev);
36ccdf85
BT
2952int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
2953
2954static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
2955{
2956 int ret;
2957
2958 ret = __dev_direct_xmit(skb, queue_id);
2959 if (!dev_xmit_complete(ret))
2960 kfree_skb(skb);
2961 return ret;
2962}
2963
f629d208
JP
2964int register_netdevice(struct net_device *dev);
2965void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
2966void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
2967static inline void unregister_netdevice(struct net_device *dev)
2968{
2969 unregister_netdevice_queue(dev, NULL);
2970}
2971
f629d208
JP
2972int netdev_refcnt_read(const struct net_device *dev);
2973void free_netdev(struct net_device *dev);
74d332c1 2974void netdev_freemem(struct net_device *dev);
f629d208 2975int init_dummy_netdev(struct net_device *dev);
937f1ba5 2976
cff9f12b
MG
2977struct net_device *netdev_get_xmit_slave(struct net_device *dev,
2978 struct sk_buff *skb,
2979 bool all_slaves);
719a402c
TT
2980struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
2981 struct sock *sk);
f629d208
JP
2982struct net_device *dev_get_by_index(struct net *net, int ifindex);
2983struct net_device *__dev_get_by_index(struct net *net, int ifindex);
2984struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 2985struct net_device *dev_get_by_napi_id(unsigned int napi_id);
f629d208
JP
2986int netdev_get_name(struct net *net, char *name, int ifindex);
2987int dev_restart(struct net_device *dev);
d4546c25 2988int skb_gro_receive(struct sk_buff *p, struct sk_buff *skb);
3a1296a3 2989int skb_gro_receive_list(struct sk_buff *p, struct sk_buff *skb);
86911732
HX
2990
2991static inline unsigned int skb_gro_offset(const struct sk_buff *skb)
2992{
2993 return NAPI_GRO_CB(skb)->data_offset;
2994}
2995
2996static inline unsigned int skb_gro_len(const struct sk_buff *skb)
2997{
2998 return skb->len - NAPI_GRO_CB(skb)->data_offset;
2999}
3000
3001static inline void skb_gro_pull(struct sk_buff *skb, unsigned int len)
3002{
3003 NAPI_GRO_CB(skb)->data_offset += len;
3004}
3005
a5b1cf28
HX
3006static inline void *skb_gro_header_fast(struct sk_buff *skb,
3007 unsigned int offset)
86911732 3008{
a5b1cf28
HX
3009 return NAPI_GRO_CB(skb)->frag0 + offset;
3010}
78a478d0 3011
a5b1cf28
HX
3012static inline int skb_gro_header_hard(struct sk_buff *skb, unsigned int hlen)
3013{
3014 return NAPI_GRO_CB(skb)->frag0_len < hlen;
3015}
78a478d0 3016
57ea52a8
HX
3017static inline void skb_gro_frag0_invalidate(struct sk_buff *skb)
3018{
3019 NAPI_GRO_CB(skb)->frag0 = NULL;
3020 NAPI_GRO_CB(skb)->frag0_len = 0;
3021}
3022
a5b1cf28
HX
3023static inline void *skb_gro_header_slow(struct sk_buff *skb, unsigned int hlen,
3024 unsigned int offset)
3025{
17dd759c
HX
3026 if (!pskb_may_pull(skb, hlen))
3027 return NULL;
3028
57ea52a8 3029 skb_gro_frag0_invalidate(skb);
17dd759c 3030 return skb->data + offset;
86911732 3031}
1da177e4 3032
36e7b1b8
HX
3033static inline void *skb_gro_network_header(struct sk_buff *skb)
3034{
78d3fd0b
HX
3035 return (NAPI_GRO_CB(skb)->frag0 ?: skb->data) +
3036 skb_network_offset(skb);
36e7b1b8
HX
3037}
3038
bf5a755f
JC
3039static inline void skb_gro_postpull_rcsum(struct sk_buff *skb,
3040 const void *start, unsigned int len)
3041{
573e8fca 3042 if (NAPI_GRO_CB(skb)->csum_valid)
bf5a755f
JC
3043 NAPI_GRO_CB(skb)->csum = csum_sub(NAPI_GRO_CB(skb)->csum,
3044 csum_partial(start, len, 0));
3045}
3046
573e8fca
TH
3047/* GRO checksum functions. These are logical equivalents of the normal
3048 * checksum functions (in skbuff.h) except that they operate on the GRO
3049 * offsets and fields in sk_buff.
3050 */
3051
3052__sum16 __skb_gro_checksum_complete(struct sk_buff *skb);
3053
15e2396d
TH
3054static inline bool skb_at_gro_remcsum_start(struct sk_buff *skb)
3055{
b7fe10e5 3056 return (NAPI_GRO_CB(skb)->gro_remcsum_start == skb_gro_offset(skb));
15e2396d
TH
3057}
3058
573e8fca
TH
3059static inline bool __skb_gro_checksum_validate_needed(struct sk_buff *skb,
3060 bool zero_okay,
3061 __sum16 check)
3062{
6edec0e6
TH
3063 return ((skb->ip_summed != CHECKSUM_PARTIAL ||
3064 skb_checksum_start_offset(skb) <
3065 skb_gro_offset(skb)) &&
15e2396d 3066 !skb_at_gro_remcsum_start(skb) &&
662880f4 3067 NAPI_GRO_CB(skb)->csum_cnt == 0 &&
573e8fca
TH
3068 (!zero_okay || check));
3069}
3070
3071static inline __sum16 __skb_gro_checksum_validate_complete(struct sk_buff *skb,
3072 __wsum psum)
3073{
3074 if (NAPI_GRO_CB(skb)->csum_valid &&
3075 !csum_fold(csum_add(psum, NAPI_GRO_CB(skb)->csum)))
3076 return 0;
3077
3078 NAPI_GRO_CB(skb)->csum = psum;
3079
3080 return __skb_gro_checksum_complete(skb);
3081}
3082
573e8fca
TH
3083static inline void skb_gro_incr_csum_unnecessary(struct sk_buff *skb)
3084{
662880f4
TH
3085 if (NAPI_GRO_CB(skb)->csum_cnt > 0) {
3086 /* Consume a checksum from CHECKSUM_UNNECESSARY */
3087 NAPI_GRO_CB(skb)->csum_cnt--;
3088 } else {
3089 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
3090 * verified a new top level checksum or an encapsulated one
3091 * during GRO. This saves work if we fallback to normal path.
3092 */
3093 __skb_incr_checksum_unnecessary(skb);
573e8fca
TH
3094 }
3095}
3096
3097#define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
3098 compute_pseudo) \
3099({ \
3100 __sum16 __ret = 0; \
3101 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
3102 __ret = __skb_gro_checksum_validate_complete(skb, \
3103 compute_pseudo(skb, proto)); \
219f1d79 3104 if (!__ret) \
573e8fca
TH
3105 skb_gro_incr_csum_unnecessary(skb); \
3106 __ret; \
3107})
3108
3109#define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
3110 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
3111
3112#define skb_gro_checksum_validate_zero_check(skb, proto, check, \
3113 compute_pseudo) \
3114 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
3115
3116#define skb_gro_checksum_simple_validate(skb) \
3117 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
3118
d96535a1
TH
3119static inline bool __skb_gro_checksum_convert_check(struct sk_buff *skb)
3120{
3121 return (NAPI_GRO_CB(skb)->csum_cnt == 0 &&
3122 !NAPI_GRO_CB(skb)->csum_valid);
3123}
3124
3125static inline void __skb_gro_checksum_convert(struct sk_buff *skb,
b39c78b2 3126 __wsum pseudo)
d96535a1
TH
3127{
3128 NAPI_GRO_CB(skb)->csum = ~pseudo;
3129 NAPI_GRO_CB(skb)->csum_valid = 1;
3130}
3131
b39c78b2 3132#define skb_gro_checksum_try_convert(skb, proto, compute_pseudo) \
d96535a1
TH
3133do { \
3134 if (__skb_gro_checksum_convert_check(skb)) \
b39c78b2 3135 __skb_gro_checksum_convert(skb, \
d96535a1
TH
3136 compute_pseudo(skb, proto)); \
3137} while (0)
3138
26c4f7da
TH
3139struct gro_remcsum {
3140 int offset;
3141 __wsum delta;
3142};
3143
3144static inline void skb_gro_remcsum_init(struct gro_remcsum *grc)
3145{
846cd667 3146 grc->offset = 0;
26c4f7da
TH
3147 grc->delta = 0;
3148}
3149
b7fe10e5
TH
3150static inline void *skb_gro_remcsum_process(struct sk_buff *skb, void *ptr,
3151 unsigned int off, size_t hdrlen,
3152 int start, int offset,
3153 struct gro_remcsum *grc,
3154 bool nopartial)
dcdc8994
TH
3155{
3156 __wsum delta;
b7fe10e5 3157 size_t plen = hdrlen + max_t(size_t, offset + sizeof(u16), start);
dcdc8994
TH
3158
3159 BUG_ON(!NAPI_GRO_CB(skb)->csum_valid);
3160
15e2396d 3161 if (!nopartial) {
b7fe10e5
TH
3162 NAPI_GRO_CB(skb)->gro_remcsum_start = off + hdrlen + start;
3163 return ptr;
3164 }
3165
3166 ptr = skb_gro_header_fast(skb, off);
3167 if (skb_gro_header_hard(skb, off + plen)) {
3168 ptr = skb_gro_header_slow(skb, off + plen, off);
3169 if (!ptr)
3170 return NULL;
15e2396d
TH
3171 }
3172
b7fe10e5
TH
3173 delta = remcsum_adjust(ptr + hdrlen, NAPI_GRO_CB(skb)->csum,
3174 start, offset);
dcdc8994
TH
3175
3176 /* Adjust skb->csum since we changed the packet */
dcdc8994 3177 NAPI_GRO_CB(skb)->csum = csum_add(NAPI_GRO_CB(skb)->csum, delta);
26c4f7da 3178
b7fe10e5 3179 grc->offset = off + hdrlen + offset;
26c4f7da 3180 grc->delta = delta;
b7fe10e5
TH
3181
3182 return ptr;
dcdc8994
TH
3183}
3184
26c4f7da
TH
3185static inline void skb_gro_remcsum_cleanup(struct sk_buff *skb,
3186 struct gro_remcsum *grc)
3187{
b7fe10e5
TH
3188 void *ptr;
3189 size_t plen = grc->offset + sizeof(u16);
3190
26c4f7da
TH
3191 if (!grc->delta)
3192 return;
3193
b7fe10e5
TH
3194 ptr = skb_gro_header_fast(skb, grc->offset);
3195 if (skb_gro_header_hard(skb, grc->offset + sizeof(u16))) {
3196 ptr = skb_gro_header_slow(skb, plen, grc->offset);
3197 if (!ptr)
3198 return;
3199 }
3200
3201 remcsum_unadjust((__sum16 *)ptr, grc->delta);
26c4f7da 3202}
dcdc8994 3203
25393d3f 3204#ifdef CONFIG_XFRM_OFFLOAD
d4546c25 3205static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
25393d3f
SK
3206{
3207 if (PTR_ERR(pp) != -EINPROGRESS)
3208 NAPI_GRO_CB(skb)->flush |= flush;
3209}
603d4cf8 3210static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 3211 struct sk_buff *pp,
603d4cf8
SD
3212 int flush,
3213 struct gro_remcsum *grc)
3214{
3215 if (PTR_ERR(pp) != -EINPROGRESS) {
3216 NAPI_GRO_CB(skb)->flush |= flush;
3217 skb_gro_remcsum_cleanup(skb, grc);
3218 skb->remcsum_offload = 0;
3219 }
3220}
25393d3f 3221#else
d4546c25 3222static inline void skb_gro_flush_final(struct sk_buff *skb, struct sk_buff *pp, int flush)
5f114163
SK
3223{
3224 NAPI_GRO_CB(skb)->flush |= flush;
3225}
603d4cf8 3226static inline void skb_gro_flush_final_remcsum(struct sk_buff *skb,
5cd3da4b 3227 struct sk_buff *pp,
603d4cf8
SD
3228 int flush,
3229 struct gro_remcsum *grc)
3230{
3231 NAPI_GRO_CB(skb)->flush |= flush;
3232 skb_gro_remcsum_cleanup(skb, grc);
3233 skb->remcsum_offload = 0;
3234}
25393d3f 3235#endif
5f114163 3236
0c4e8581
SH
3237static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
3238 unsigned short type,
3b04ddde 3239 const void *daddr, const void *saddr,
95c96174 3240 unsigned int len)
0c4e8581 3241{
f1ecfd5d 3242 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 3243 return 0;
3b04ddde
SH
3244
3245 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
3246}
3247
b95cce35
SH
3248static inline int dev_parse_header(const struct sk_buff *skb,
3249 unsigned char *haddr)
3250{
3251 const struct net_device *dev = skb->dev;
3252
1b83336b 3253 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 3254 return 0;
3b04ddde 3255 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
3256}
3257
e78b2915
MM
3258static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
3259{
3260 const struct net_device *dev = skb->dev;
3261
3262 if (!dev->header_ops || !dev->header_ops->parse_protocol)
3263 return 0;
3264 return dev->header_ops->parse_protocol(skb);
3265}
3266
2793a23a
WB
3267/* ll_header must have at least hard_header_len allocated */
3268static inline bool dev_validate_header(const struct net_device *dev,
3269 char *ll_header, int len)
3270{
3271 if (likely(len >= dev->hard_header_len))
3272 return true;
217e6fa2
WB
3273 if (len < dev->min_header_len)
3274 return false;
2793a23a
WB
3275
3276 if (capable(CAP_SYS_RAWIO)) {
3277 memset(ll_header + len, 0, dev->hard_header_len - len);
3278 return true;
3279 }
3280
3281 if (dev->header_ops && dev->header_ops->validate)
3282 return dev->header_ops->validate(ll_header, len);
3283
3284 return false;
3285}
3286
d5496990
EB
3287static inline bool dev_has_header(const struct net_device *dev)
3288{
3289 return dev->header_ops && dev->header_ops->create;
3290}
3291
36fd633e
AV
3292typedef int gifconf_func_t(struct net_device * dev, char __user * bufptr,
3293 int len, int size);
f629d208 3294int register_gifconf(unsigned int family, gifconf_func_t *gifconf);
1da177e4
LT
3295static inline int unregister_gifconf(unsigned int family)
3296{
3297 return register_gifconf(family, NULL);
3298}
3299
99bbc707 3300#ifdef CONFIG_NET_FLOW_LIMIT
5f121b9a 3301#define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
99bbc707
WB
3302struct sd_flow_limit {
3303 u64 count;
3304 unsigned int num_buckets;
3305 unsigned int history_head;
3306 u16 history[FLOW_LIMIT_HISTORY];
3307 u8 buckets[];
3308};
3309
3310extern int netdev_flow_limit_table_len;
3311#endif /* CONFIG_NET_FLOW_LIMIT */
3312
1da177e4 3313/*
5e82b4b2 3314 * Incoming packets are placed on per-CPU queues
1da177e4 3315 */
d94d9fee 3316struct softnet_data {
1da177e4 3317 struct list_head poll_list;
6e7676c1 3318 struct sk_buff_head process_queue;
1da177e4 3319
dee42870 3320 /* stats */
cd7b5396
DM
3321 unsigned int processed;
3322 unsigned int time_squeeze;
cd7b5396 3323 unsigned int received_rps;
fd793d89 3324#ifdef CONFIG_RPS
88751275 3325 struct softnet_data *rps_ipi_list;
4cdb1e2e
ED
3326#endif
3327#ifdef CONFIG_NET_FLOW_LIMIT
3328 struct sd_flow_limit __rcu *flow_limit;
3329#endif
3330 struct Qdisc *output_queue;
3331 struct Qdisc **output_queue_tailp;
3332 struct sk_buff *completion_queue;
f53c7239
SK
3333#ifdef CONFIG_XFRM_OFFLOAD
3334 struct sk_buff_head xfrm_backlog;
3335#endif
97cdcf37
FW
3336 /* written and read only by owning cpu: */
3337 struct {
3338 u16 recursion;
3339 u8 more;
3340 } xmit;
4cdb1e2e 3341#ifdef CONFIG_RPS
501e7ef5
ED
3342 /* input_queue_head should be written by cpu owning this struct,
3343 * and only read by other cpus. Worth using a cache line.
3344 */
3345 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3346
3347 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3348 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3349 struct softnet_data *rps_ipi_next;
3350 unsigned int cpu;
76cc8b13 3351 unsigned int input_queue_tail;
1e94d72f 3352#endif
95c96174 3353 unsigned int dropped;
0a9627f2 3354 struct sk_buff_head input_pkt_queue;
bea3348e 3355 struct napi_struct backlog;
99bbc707 3356
1da177e4
LT
3357};
3358
76cc8b13 3359static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3360{
3361#ifdef CONFIG_RPS
76cc8b13
TH
3362 sd->input_queue_head++;
3363#endif
3364}
3365
3366static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3367 unsigned int *qtail)
3368{
3369#ifdef CONFIG_RPS
3370 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3371#endif
3372}
3373
0a9627f2 3374DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3375
97cdcf37
FW
3376static inline int dev_recursion_level(void)
3377{
28b05b92 3378 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3379}
3380
fb7861d1 3381#define XMIT_RECURSION_LIMIT 8
97cdcf37
FW
3382static inline bool dev_xmit_recursion(void)
3383{
3384 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3385 XMIT_RECURSION_LIMIT);
3386}
3387
3388static inline void dev_xmit_recursion_inc(void)
3389{
3390 __this_cpu_inc(softnet_data.xmit.recursion);
3391}
3392
3393static inline void dev_xmit_recursion_dec(void)
3394{
3395 __this_cpu_dec(softnet_data.xmit.recursion);
3396}
3397
f629d208 3398void __netif_schedule(struct Qdisc *q);
46e5da40 3399void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3400
fd2ea0a7
DM
3401static inline void netif_tx_schedule_all(struct net_device *dev)
3402{
3403 unsigned int i;
3404
3405 for (i = 0; i < dev->num_tx_queues; i++)
3406 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3407}
3408
f9a7cbbf 3409static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3410{
73466498 3411 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3412}
3413
bea3348e
SH
3414/**
3415 * netif_start_queue - allow transmit
3416 * @dev: network device
3417 *
3418 * Allow upper layers to call the device hard_start_xmit routine.
3419 */
1da177e4
LT
3420static inline void netif_start_queue(struct net_device *dev)
3421{
e8a0464c 3422 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3423}
3424
fd2ea0a7
DM
3425static inline void netif_tx_start_all_queues(struct net_device *dev)
3426{
3427 unsigned int i;
3428
3429 for (i = 0; i < dev->num_tx_queues; i++) {
3430 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3431 netif_tx_start_queue(txq);
3432 }
3433}
3434
46e5da40 3435void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3436
d29f749e
DJ
3437/**
3438 * netif_wake_queue - restart transmit
3439 * @dev: network device
3440 *
3441 * Allow upper layers to call the device hard_start_xmit routine.
3442 * Used for flow control when transmit resources are available.
3443 */
79d16385
DM
3444static inline void netif_wake_queue(struct net_device *dev)
3445{
e8a0464c 3446 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3447}
3448
fd2ea0a7
DM
3449static inline void netif_tx_wake_all_queues(struct net_device *dev)
3450{
3451 unsigned int i;
3452
3453 for (i = 0; i < dev->num_tx_queues; i++) {
3454 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3455 netif_tx_wake_queue(txq);
3456 }
3457}
3458
f9a7cbbf 3459static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3460{
73466498 3461 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3462}
3463
bea3348e
SH
3464/**
3465 * netif_stop_queue - stop transmitted packets
3466 * @dev: network device
3467 *
3468 * Stop upper layers calling the device hard_start_xmit routine.
3469 * Used for flow control when transmit resources are unavailable.
3470 */
1da177e4
LT
3471static inline void netif_stop_queue(struct net_device *dev)
3472{
e8a0464c 3473 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3474}
3475
a2029240 3476void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3477
4d29515f 3478static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3479{
73466498 3480 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3481}
3482
bea3348e
SH
3483/**
3484 * netif_queue_stopped - test if transmit queue is flowblocked
3485 * @dev: network device
3486 *
3487 * Test if transmit queue on device is currently unable to send.
3488 */
4d29515f 3489static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3490{
e8a0464c 3491 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3492}
3493
4d29515f 3494static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3495{
73466498
TH
3496 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3497}
3498
8e2f1a63
DB
3499static inline bool
3500netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3501{
3502 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3503}
3504
8e2f1a63
DB
3505static inline bool
3506netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3507{
3508 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3509}
3510
f57bac3c
VM
3511/**
3512 * netdev_queue_set_dql_min_limit - set dql minimum limit
3513 * @dev_queue: pointer to transmit queue
3514 * @min_limit: dql minimum limit
3515 *
3516 * Forces xmit_more() to return true until the minimum threshold
3517 * defined by @min_limit is reached (or until the tx queue is
3518 * empty). Warning: to be use with care, misuse will impact the
3519 * latency.
3520 */
3521static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
3522 unsigned int min_limit)
3523{
3524#ifdef CONFIG_BQL
3525 dev_queue->dql.min_limit = min_limit;
3526#endif
3527}
3528
53511453
ED
3529/**
3530 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3531 * @dev_queue: pointer to transmit queue
3532 *
3533 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3534 * to give appropriate hint to the CPU.
53511453
ED
3535 */
3536static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3537{
3538#ifdef CONFIG_BQL
3539 prefetchw(&dev_queue->dql.num_queued);
3540#endif
3541}
3542
3543/**
3544 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3545 * @dev_queue: pointer to transmit queue
3546 *
3547 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3548 * to give appropriate hint to the CPU.
53511453
ED
3549 */
3550static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3551{
3552#ifdef CONFIG_BQL
3553 prefetchw(&dev_queue->dql.limit);
3554#endif
3555}
3556
c5d67bd7
TH
3557static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3558 unsigned int bytes)
3559{
114cf580
TH
3560#ifdef CONFIG_BQL
3561 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3562
3563 if (likely(dql_avail(&dev_queue->dql) >= 0))
3564 return;
3565
3566 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3567
3568 /*
3569 * The XOFF flag must be set before checking the dql_avail below,
3570 * because in netdev_tx_completed_queue we update the dql_completed
3571 * before checking the XOFF flag.
3572 */
3573 smp_mb();
3574
3575 /* check again in case another CPU has just made room avail */
3576 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3577 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3578#endif
c5d67bd7
TH
3579}
3580
3e59020a
ED
3581/* Variant of netdev_tx_sent_queue() for drivers that are aware
3582 * that they should not test BQL status themselves.
3583 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3584 * skb of a batch.
3585 * Returns true if the doorbell must be used to kick the NIC.
3586 */
3587static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3588 unsigned int bytes,
3589 bool xmit_more)
3590{
3591 if (xmit_more) {
3592#ifdef CONFIG_BQL
3593 dql_queued(&dev_queue->dql, bytes);
3594#endif
3595 return netif_tx_queue_stopped(dev_queue);
3596 }
3597 netdev_tx_sent_queue(dev_queue, bytes);
3598 return true;
3599}
3600
0042d0c8
FF
3601/**
3602 * netdev_sent_queue - report the number of bytes queued to hardware
3603 * @dev: network device
3604 * @bytes: number of bytes queued to the hardware device queue
3605 *
3606 * Report the number of bytes queued for sending/completion to the network
3607 * device hardware queue. @bytes should be a good approximation and should
3608 * exactly match netdev_completed_queue() @bytes
3609 */
c5d67bd7
TH
3610static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3611{
3612 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3613}
3614
620344c4
HK
3615static inline bool __netdev_sent_queue(struct net_device *dev,
3616 unsigned int bytes,
3617 bool xmit_more)
3618{
3619 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3620 xmit_more);
3621}
3622
c5d67bd7 3623static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3624 unsigned int pkts, unsigned int bytes)
c5d67bd7 3625{
114cf580 3626#ifdef CONFIG_BQL
b37c0fbe
AD
3627 if (unlikely(!bytes))
3628 return;
3629
3630 dql_completed(&dev_queue->dql, bytes);
3631
3632 /*
3633 * Without the memory barrier there is a small possiblity that
3634 * netdev_tx_sent_queue will miss the update and cause the queue to
3635 * be stopped forever
3636 */
3637 smp_mb();
3638
f3acd33d 3639 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3640 return;
3641
3642 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3643 netif_schedule_queue(dev_queue);
114cf580 3644#endif
c5d67bd7
TH
3645}
3646
0042d0c8
FF
3647/**
3648 * netdev_completed_queue - report bytes and packets completed by device
3649 * @dev: network device
3650 * @pkts: actual number of packets sent over the medium
3651 * @bytes: actual number of bytes sent over the medium
3652 *
3653 * Report the number of bytes and packets transmitted by the network device
3654 * hardware queue over the physical medium, @bytes must exactly match the
3655 * @bytes amount passed to netdev_sent_queue()
3656 */
c5d67bd7 3657static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3658 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3659{
3660 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3661}
3662
3663static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3664{
114cf580 3665#ifdef CONFIG_BQL
5c490354 3666 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3667 dql_reset(&q->dql);
3668#endif
c5d67bd7
TH
3669}
3670
0042d0c8
FF
3671/**
3672 * netdev_reset_queue - reset the packets and bytes count of a network device
3673 * @dev_queue: network device
3674 *
3675 * Reset the bytes and packet count of a network device and clear the
3676 * software flow control OFF bit for this network device
3677 */
c5d67bd7
TH
3678static inline void netdev_reset_queue(struct net_device *dev_queue)
3679{
3680 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3681}
3682
b9507bda
DB
3683/**
3684 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3685 * @dev: network device
3686 * @queue_index: given tx queue index
3687 *
3688 * Returns 0 if given tx queue index >= number of device tx queues,
3689 * otherwise returns the originally passed tx queue index.
3690 */
3691static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3692{
3693 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3694 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3695 dev->name, queue_index,
3696 dev->real_num_tx_queues);
3697 return 0;
3698 }
3699
3700 return queue_index;
3701}
3702
bea3348e
SH
3703/**
3704 * netif_running - test if up
3705 * @dev: network device
3706 *
3707 * Test if the device has been brought up.
3708 */
4d29515f 3709static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3710{
3711 return test_bit(__LINK_STATE_START, &dev->state);
3712}
3713
f25f4e44 3714/*
5e82b4b2 3715 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3716 * stop, and a check if it's stopped. All other device management is
3717 * done at the overall netdevice level.
3718 * Also test the device if we're multiqueue.
3719 */
bea3348e
SH
3720
3721/**
3722 * netif_start_subqueue - allow sending packets on subqueue
3723 * @dev: network device
3724 * @queue_index: sub queue index
3725 *
3726 * Start individual transmit queue of a device with multiple transmit queues.
3727 */
f25f4e44
PWJ
3728static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3729{
fd2ea0a7 3730 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3731
3732 netif_tx_start_queue(txq);
f25f4e44
PWJ
3733}
3734
bea3348e
SH
3735/**
3736 * netif_stop_subqueue - stop sending packets on subqueue
3737 * @dev: network device
3738 * @queue_index: sub queue index
3739 *
3740 * Stop individual transmit queue of a device with multiple transmit queues.
3741 */
f25f4e44
PWJ
3742static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3743{
fd2ea0a7 3744 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3745 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3746}
3747
bea3348e 3748/**
270f3385 3749 * __netif_subqueue_stopped - test status of subqueue
bea3348e
SH
3750 * @dev: network device
3751 * @queue_index: sub queue index
3752 *
3753 * Check individual transmit queue of a device with multiple transmit queues.
3754 */
4d29515f
DM
3755static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3756 u16 queue_index)
f25f4e44 3757{
fd2ea0a7 3758 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3759
3760 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3761}
3762
270f3385
MCC
3763/**
3764 * netif_subqueue_stopped - test status of subqueue
3765 * @dev: network device
3766 * @skb: sub queue buffer pointer
3767 *
3768 * Check individual transmit queue of a device with multiple transmit queues.
3769 */
4d29515f
DM
3770static inline bool netif_subqueue_stopped(const struct net_device *dev,
3771 struct sk_buff *skb)
668f895a
PE
3772{
3773 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3774}
bea3348e 3775
738b35cc
FF
3776/**
3777 * netif_wake_subqueue - allow sending packets on subqueue
3778 * @dev: network device
3779 * @queue_index: sub queue index
3780 *
3781 * Resume individual transmit queue of a device with multiple transmit queues.
3782 */
3783static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3784{
3785 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3786
3787 netif_tx_wake_queue(txq);
3788}
f25f4e44 3789
537c00de 3790#ifdef CONFIG_XPS
53af53ae 3791int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3792 u16 index);
80d19669 3793int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
044ab86d 3794 u16 index, enum xps_map_type type);
80d19669
AN
3795
3796/**
3797 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3798 * @j: CPU/Rx queue index
3799 * @mask: bitmask of all cpus/rx queues
3800 * @nr_bits: number of bits in the bitmask
3801 *
3802 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3803 */
3804static inline bool netif_attr_test_mask(unsigned long j,
3805 const unsigned long *mask,
3806 unsigned int nr_bits)
3807{
3808 cpu_max_bits_warn(j, nr_bits);
3809 return test_bit(j, mask);
3810}
3811
3812/**
3813 * netif_attr_test_online - Test for online CPU/Rx queue
3814 * @j: CPU/Rx queue index
3815 * @online_mask: bitmask for CPUs/Rx queues that are online
3816 * @nr_bits: number of bits in the bitmask
3817 *
3818 * Returns true if a CPU/Rx queue is online.
3819 */
3820static inline bool netif_attr_test_online(unsigned long j,
3821 const unsigned long *online_mask,
3822 unsigned int nr_bits)
3823{
3824 cpu_max_bits_warn(j, nr_bits);
3825
3826 if (online_mask)
3827 return test_bit(j, online_mask);
3828
3829 return (j < nr_bits);
3830}
3831
3832/**
3833 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3834 * @n: CPU/Rx queue index
3835 * @srcp: the cpumask/Rx queue mask pointer
3836 * @nr_bits: number of bits in the bitmask
3837 *
3838 * Returns >= nr_bits if no further CPUs/Rx queues set.
3839 */
3840static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3841 unsigned int nr_bits)
3842{
3843 /* -1 is a legal arg here. */
3844 if (n != -1)
3845 cpu_max_bits_warn(n, nr_bits);
3846
3847 if (srcp)
3848 return find_next_bit(srcp, nr_bits, n + 1);
3849
3850 return n + 1;
3851}
3852
3853/**
a1fa83bd 3854 * netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
80d19669
AN
3855 * @n: CPU/Rx queue index
3856 * @src1p: the first CPUs/Rx queues mask pointer
3857 * @src2p: the second CPUs/Rx queues mask pointer
3858 * @nr_bits: number of bits in the bitmask
3859 *
3860 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3861 */
3862static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3863 const unsigned long *src2p,
3864 unsigned int nr_bits)
3865{
3866 /* -1 is a legal arg here. */
3867 if (n != -1)
3868 cpu_max_bits_warn(n, nr_bits);
3869
3870 if (src1p && src2p)
3871 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3872 else if (src1p)
3873 return find_next_bit(src1p, nr_bits, n + 1);
3874 else if (src2p)
3875 return find_next_bit(src2p, nr_bits, n + 1);
3876
3877 return n + 1;
3878}
537c00de
AD
3879#else
3880static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3881 const struct cpumask *mask,
537c00de
AD
3882 u16 index)
3883{
3884 return 0;
3885}
c9fbb2d2
KK
3886
3887static inline int __netif_set_xps_queue(struct net_device *dev,
3888 const unsigned long *mask,
044ab86d 3889 u16 index, enum xps_map_type type)
c9fbb2d2
KK
3890{
3891 return 0;
3892}
537c00de
AD
3893#endif
3894
bea3348e
SH
3895/**
3896 * netif_is_multiqueue - test if device has multiple transmit queues
3897 * @dev: network device
3898 *
3899 * Check if device has multiple transmit queues
bea3348e 3900 */
4d29515f 3901static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3902{
a02cec21 3903 return dev->num_tx_queues > 1;
f25f4e44 3904}
1da177e4 3905
f629d208 3906int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3907
a953be53 3908#ifdef CONFIG_SYSFS
f629d208 3909int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3910#else
3911static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3912 unsigned int rxqs)
62fe0b40 3913{
c29c2ebd 3914 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3915 return 0;
3916}
3917#endif
3918
65073a67
DB
3919static inline struct netdev_rx_queue *
3920__netif_get_rx_queue(struct net_device *dev, unsigned int rxq)
3921{
3922 return dev->_rx + rxq;
3923}
3924
a953be53
MD
3925#ifdef CONFIG_SYSFS
3926static inline unsigned int get_netdev_rx_queue_index(
3927 struct netdev_rx_queue *queue)
3928{
3929 struct net_device *dev = queue->dev;
3930 int index = queue - dev->_rx;
3931
3932 BUG_ON(index >= dev->num_rx_queues);
3933 return index;
3934}
3935#endif
3936
16917b87 3937#define DEFAULT_MAX_NUM_RSS_QUEUES (8)
f629d208 3938int netif_get_num_default_rss_queues(void);
16917b87 3939
e6247027
ED
3940enum skb_free_reason {
3941 SKB_REASON_CONSUMED,
3942 SKB_REASON_DROPPED,
3943};
3944
3945void __dev_kfree_skb_irq(struct sk_buff *skb, enum skb_free_reason reason);
3946void __dev_kfree_skb_any(struct sk_buff *skb, enum skb_free_reason reason);
1da177e4 3947
e6247027
ED
3948/*
3949 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3950 * interrupt context or with hardware interrupts being disabled.
3951 * (in_irq() || irqs_disabled())
3952 *
3953 * We provide four helpers that can be used in following contexts :
3954 *
3955 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3956 * replacing kfree_skb(skb)
3957 *
3958 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3959 * Typically used in place of consume_skb(skb) in TX completion path
3960 *
3961 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3962 * replacing kfree_skb(skb)
3963 *
3964 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3965 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3966 */
e6247027
ED
3967static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3968{
3969 __dev_kfree_skb_irq(skb, SKB_REASON_DROPPED);
3970}
3971
3972static inline void dev_consume_skb_irq(struct sk_buff *skb)
3973{
3974 __dev_kfree_skb_irq(skb, SKB_REASON_CONSUMED);
3975}
3976
3977static inline void dev_kfree_skb_any(struct sk_buff *skb)
3978{
3979 __dev_kfree_skb_any(skb, SKB_REASON_DROPPED);
3980}
3981
3982static inline void dev_consume_skb_any(struct sk_buff *skb)
3983{
3984 __dev_kfree_skb_any(skb, SKB_REASON_CONSUMED);
3985}
1da177e4 3986
7c497478
JW
3987void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3988int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208
JP
3989int netif_rx(struct sk_buff *skb);
3990int netif_rx_ni(struct sk_buff *skb);
c11171a4 3991int netif_rx_any_context(struct sk_buff *skb);
04eb4489 3992int netif_receive_skb(struct sk_buff *skb);
1c601d82 3993int netif_receive_skb_core(struct sk_buff *skb);
f6ad8c1b 3994void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3995gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3996void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3997struct sk_buff *napi_get_frags(struct napi_struct *napi);
3998gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3999struct packet_offload *gro_find_receive_by_type(__be16 type);
4000struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
4001
4002static inline void napi_free_frags(struct napi_struct *napi)
4003{
4004 kfree_skb(napi->skb);
4005 napi->skb = NULL;
4006}
4007
24b27fc4 4008bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
4009int netdev_rx_handler_register(struct net_device *dev,
4010 rx_handler_func_t *rx_handler,
4011 void *rx_handler_data);
4012void netdev_rx_handler_unregister(struct net_device *dev);
4013
4014bool dev_valid_name(const char *name);
44c02a2c
AV
4015int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
4016 bool *need_copyout);
36fd633e 4017int dev_ifconf(struct net *net, struct ifconf *, int);
f629d208
JP
4018int dev_ethtool(struct net *net, struct ifreq *);
4019unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
4020int __dev_change_flags(struct net_device *dev, unsigned int flags,
4021 struct netlink_ext_ack *extack);
567c5e13
PM
4022int dev_change_flags(struct net_device *dev, unsigned int flags,
4023 struct netlink_ext_ack *extack);
cb178190
DM
4024void __dev_notify_flags(struct net_device *, unsigned int old_flags,
4025 unsigned int gchanges);
f629d208
JP
4026int dev_change_name(struct net_device *, const char *);
4027int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 4028int dev_get_alias(const struct net_device *, char *, size_t);
0854fa82
AV
4029int __dev_change_net_namespace(struct net_device *dev, struct net *net,
4030 const char *pat, int new_ifindex);
4031static inline
eeb85a14 4032int dev_change_net_namespace(struct net_device *dev, struct net *net,
0854fa82
AV
4033 const char *pat)
4034{
4035 return __dev_change_net_namespace(dev, net, pat, 0);
4036}
f51048c3 4037int __dev_set_mtu(struct net_device *, int);
d836f5c6
ED
4038int dev_validate_mtu(struct net_device *dev, int mtu,
4039 struct netlink_ext_ack *extack);
7a4c53be
SH
4040int dev_set_mtu_ext(struct net_device *dev, int mtu,
4041 struct netlink_ext_ack *extack);
f629d208 4042int dev_set_mtu(struct net_device *, int);
6a643ddb 4043int dev_change_tx_queue_len(struct net_device *, unsigned long);
f629d208 4044void dev_set_group(struct net_device *, int);
d59cdf94
PM
4045int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
4046 struct netlink_ext_ack *extack);
3a37a963
PM
4047int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
4048 struct netlink_ext_ack *extack);
3b23a32a
CW
4049int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
4050 struct netlink_ext_ack *extack);
4051int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
f629d208
JP
4052int dev_change_carrier(struct net_device *, bool new_carrier);
4053int dev_get_phys_port_id(struct net_device *dev,
02637fce 4054 struct netdev_phys_item_id *ppid);
db24a904
DA
4055int dev_get_phys_port_name(struct net_device *dev,
4056 char *name, size_t len);
d6abc596
FF
4057int dev_get_port_parent_id(struct net_device *dev,
4058 struct netdev_phys_item_id *ppid, bool recurse);
4059bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
d746d707 4060int dev_change_proto_down(struct net_device *dev, bool proto_down);
b5899679 4061int dev_change_proto_down_generic(struct net_device *dev, bool proto_down);
829eb208
RP
4062void dev_change_proto_down_reason(struct net_device *dev, unsigned long mask,
4063 u32 value);
f53c7239 4064struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
4065struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
4066 struct netdev_queue *txq, int *ret);
d67b9cd2 4067
f4e63525 4068typedef int (*bpf_op_t)(struct net_device *dev, struct netdev_bpf *bpf);
d67b9cd2 4069int dev_change_xdp_fd(struct net_device *dev, struct netlink_ext_ack *extack,
92234c8f 4070 int fd, int expected_fd, u32 flags);
aa8d3a71 4071int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
7f0a8382 4072u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
aa8d3a71 4073
a0265d28 4074int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 4075int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
5f7d5728 4076int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
4077bool is_skb_forwardable(const struct net_device *dev,
4078 const struct sk_buff *skb);
1da177e4 4079
5f7d5728
JDB
4080static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
4081 const struct sk_buff *skb,
4082 const bool check_mtu)
4083{
4084 const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
4085 unsigned int len;
4086
4087 if (!(dev->flags & IFF_UP))
4088 return false;
4089
4090 if (!check_mtu)
4091 return true;
4092
4093 len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
4094 if (skb->len <= len)
4095 return true;
4096
4097 /* if TSO is enabled, we don't care about the length as the packet
4098 * could be forwarded without being segmented before
4099 */
4100 if (skb_is_gso(skb))
4101 return true;
4102
4103 return false;
4104}
4105
4e3264d2 4106static __always_inline int ____dev_forward_skb(struct net_device *dev,
5f7d5728
JDB
4107 struct sk_buff *skb,
4108 const bool check_mtu)
4e3264d2
MKL
4109{
4110 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
5f7d5728 4111 unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
4e3264d2
MKL
4112 atomic_long_inc(&dev->rx_dropped);
4113 kfree_skb(skb);
4114 return NET_RX_DROP;
4115 }
4116
ff70202b 4117 skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
4e3264d2
MKL
4118 skb->priority = 0;
4119 return 0;
4120}
4121
9f9a742d 4122bool dev_nit_active(struct net_device *dev);
74b20582
DA
4123void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
4124
20380731 4125extern int netdev_budget;
7acf8a1e 4126extern unsigned int netdev_budget_usecs;
1da177e4
LT
4127
4128/* Called by rtnetlink.c:rtnl_unlock() */
f629d208 4129void netdev_run_todo(void);
1da177e4 4130
bea3348e
SH
4131/**
4132 * dev_put - release reference to device
4133 * @dev: network device
4134 *
9ef4429b 4135 * Release reference to device to allow it to be freed.
bea3348e 4136 */
1da177e4
LT
4137static inline void dev_put(struct net_device *dev)
4138{
919067cc 4139#ifdef CONFIG_PCPU_DEV_REFCNT
933393f5 4140 this_cpu_dec(*dev->pcpu_refcnt);
919067cc
ED
4141#else
4142 refcount_dec(&dev->dev_refcnt);
4143#endif
1da177e4
LT
4144}
4145
bea3348e
SH
4146/**
4147 * dev_hold - get reference to device
4148 * @dev: network device
4149 *
9ef4429b 4150 * Hold reference to device to keep it from being freed.
bea3348e 4151 */
15333061
SH
4152static inline void dev_hold(struct net_device *dev)
4153{
919067cc 4154#ifdef CONFIG_PCPU_DEV_REFCNT
933393f5 4155 this_cpu_inc(*dev->pcpu_refcnt);
919067cc
ED
4156#else
4157 refcount_inc(&dev->dev_refcnt);
4158#endif
15333061 4159}
1da177e4
LT
4160
4161/* Carrier loss detection, dial on demand. The functions netif_carrier_on
4162 * and _off may be called from IRQ context, but it is caller
4163 * who is responsible for serialization of these calls.
b00055aa
SR
4164 *
4165 * The name carrier is inappropriate, these functions should really be
4166 * called netif_lowerlayer_*() because they represent the state of any
4167 * kind of lower layer not just hardware media.
1da177e4
LT
4168 */
4169
f629d208
JP
4170void linkwatch_init_dev(struct net_device *dev);
4171void linkwatch_fire_event(struct net_device *dev);
4172void linkwatch_forget_dev(struct net_device *dev);
1da177e4 4173
bea3348e
SH
4174/**
4175 * netif_carrier_ok - test if carrier present
4176 * @dev: network device
4177 *
4178 * Check if carrier is present on device
4179 */
4d29515f 4180static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
4181{
4182 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
4183}
4184
f629d208 4185unsigned long dev_trans_start(struct net_device *dev);
9d21493b 4186
f629d208 4187void __netdev_watchdog_up(struct net_device *dev);
1da177e4 4188
f629d208 4189void netif_carrier_on(struct net_device *dev);
f629d208 4190void netif_carrier_off(struct net_device *dev);
490dceca 4191void netif_carrier_event(struct net_device *dev);
1da177e4 4192
bea3348e
SH
4193/**
4194 * netif_dormant_on - mark device as dormant.
4195 * @dev: network device
4196 *
4197 * Mark device as dormant (as per RFC2863).
4198 *
4199 * The dormant state indicates that the relevant interface is not
4200 * actually in a condition to pass packets (i.e., it is not 'up') but is
4201 * in a "pending" state, waiting for some external event. For "on-
4202 * demand" interfaces, this new state identifies the situation where the
4203 * interface is waiting for events to place it in the up state.
bea3348e 4204 */
b00055aa
SR
4205static inline void netif_dormant_on(struct net_device *dev)
4206{
4207 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
4208 linkwatch_fire_event(dev);
4209}
4210
bea3348e
SH
4211/**
4212 * netif_dormant_off - set device as not dormant.
4213 * @dev: network device
4214 *
4215 * Device is not in dormant state.
4216 */
b00055aa
SR
4217static inline void netif_dormant_off(struct net_device *dev)
4218{
4219 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
4220 linkwatch_fire_event(dev);
4221}
4222
bea3348e 4223/**
8ecbc40a 4224 * netif_dormant - test if device is dormant
bea3348e
SH
4225 * @dev: network device
4226 *
8ecbc40a 4227 * Check if device is dormant.
bea3348e 4228 */
4d29515f 4229static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
4230{
4231 return test_bit(__LINK_STATE_DORMANT, &dev->state);
4232}
4233
4234
eec517cd
AL
4235/**
4236 * netif_testing_on - mark device as under test.
4237 * @dev: network device
4238 *
4239 * Mark device as under test (as per RFC2863).
4240 *
4241 * The testing state indicates that some test(s) must be performed on
4242 * the interface. After completion, of the test, the interface state
4243 * will change to up, dormant, or down, as appropriate.
4244 */
4245static inline void netif_testing_on(struct net_device *dev)
4246{
4247 if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
4248 linkwatch_fire_event(dev);
4249}
4250
4251/**
4252 * netif_testing_off - set device as not under test.
4253 * @dev: network device
4254 *
4255 * Device is not in testing state.
4256 */
4257static inline void netif_testing_off(struct net_device *dev)
4258{
4259 if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
4260 linkwatch_fire_event(dev);
4261}
4262
4263/**
4264 * netif_testing - test if device is under test
4265 * @dev: network device
4266 *
4267 * Check if device is under test
4268 */
4269static inline bool netif_testing(const struct net_device *dev)
4270{
4271 return test_bit(__LINK_STATE_TESTING, &dev->state);
4272}
4273
4274
bea3348e
SH
4275/**
4276 * netif_oper_up - test if device is operational
4277 * @dev: network device
4278 *
4279 * Check if carrier is operational
4280 */
4d29515f 4281static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 4282{
b00055aa
SR
4283 return (dev->operstate == IF_OPER_UP ||
4284 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
4285}
4286
bea3348e
SH
4287/**
4288 * netif_device_present - is device available or removed
4289 * @dev: network device
4290 *
4291 * Check if device has not been removed from system.
4292 */
7a126a43 4293static inline bool netif_device_present(const struct net_device *dev)
1da177e4
LT
4294{
4295 return test_bit(__LINK_STATE_PRESENT, &dev->state);
4296}
4297
f629d208 4298void netif_device_detach(struct net_device *dev);
1da177e4 4299
f629d208 4300void netif_device_attach(struct net_device *dev);
1da177e4
LT
4301
4302/*
4303 * Network interface message level settings
4304 */
1da177e4
LT
4305
4306enum {
6a94b8cc
MK
4307 NETIF_MSG_DRV_BIT,
4308 NETIF_MSG_PROBE_BIT,
4309 NETIF_MSG_LINK_BIT,
4310 NETIF_MSG_TIMER_BIT,
4311 NETIF_MSG_IFDOWN_BIT,
4312 NETIF_MSG_IFUP_BIT,
4313 NETIF_MSG_RX_ERR_BIT,
4314 NETIF_MSG_TX_ERR_BIT,
4315 NETIF_MSG_TX_QUEUED_BIT,
4316 NETIF_MSG_INTR_BIT,
4317 NETIF_MSG_TX_DONE_BIT,
4318 NETIF_MSG_RX_STATUS_BIT,
4319 NETIF_MSG_PKTDATA_BIT,
4320 NETIF_MSG_HW_BIT,
4321 NETIF_MSG_WOL_BIT,
4322
4323 /* When you add a new bit above, update netif_msg_class_names array
4324 * in net/ethtool/common.c
4325 */
4326 NETIF_MSG_CLASS_COUNT,
1da177e4 4327};
6a94b8cc
MK
4328/* Both ethtool_ops interface and internal driver implementation use u32 */
4329static_assert(NETIF_MSG_CLASS_COUNT <= 32);
4330
4331#define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
4332#define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
4333
4334#define NETIF_MSG_DRV __NETIF_MSG(DRV)
4335#define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
4336#define NETIF_MSG_LINK __NETIF_MSG(LINK)
4337#define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
4338#define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
4339#define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
4340#define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
4341#define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
4342#define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
4343#define NETIF_MSG_INTR __NETIF_MSG(INTR)
4344#define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
4345#define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
4346#define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
4347#define NETIF_MSG_HW __NETIF_MSG(HW)
4348#define NETIF_MSG_WOL __NETIF_MSG(WOL)
1da177e4
LT
4349
4350#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
4351#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
4352#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
4353#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
4354#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
4355#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
4356#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
4357#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
4358#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
4359#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
4360#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
4361#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
4362#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
4363#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
4364#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
4365
4366static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
4367{
4368 /* use default */
4369 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
4370 return default_msg_enable_bits;
4371 if (debug_value == 0) /* no output */
4372 return 0;
4373 /* set low N bits */
f4d7b3e2 4374 return (1U << debug_value) - 1;
1da177e4
LT
4375}
4376
c773e847 4377static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 4378{
c773e847
DM
4379 spin_lock(&txq->_xmit_lock);
4380 txq->xmit_lock_owner = cpu;
22dd7495
JHS
4381}
4382
5a717f4f
MT
4383static inline bool __netif_tx_acquire(struct netdev_queue *txq)
4384{
4385 __acquire(&txq->_xmit_lock);
4386 return true;
4387}
4388
4389static inline void __netif_tx_release(struct netdev_queue *txq)
4390{
4391 __release(&txq->_xmit_lock);
4392}
4393
fd2ea0a7
DM
4394static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
4395{
4396 spin_lock_bh(&txq->_xmit_lock);
4397 txq->xmit_lock_owner = smp_processor_id();
4398}
4399
4d29515f 4400static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 4401{
4d29515f 4402 bool ok = spin_trylock(&txq->_xmit_lock);
c3f26a26
DM
4403 if (likely(ok))
4404 txq->xmit_lock_owner = smp_processor_id();
4405 return ok;
4406}
4407
4408static inline void __netif_tx_unlock(struct netdev_queue *txq)
4409{
4410 txq->xmit_lock_owner = -1;
4411 spin_unlock(&txq->_xmit_lock);
4412}
4413
4414static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
4415{
4416 txq->xmit_lock_owner = -1;
4417 spin_unlock_bh(&txq->_xmit_lock);
4418}
4419
08baf561
ED
4420static inline void txq_trans_update(struct netdev_queue *txq)
4421{
4422 if (txq->xmit_lock_owner != -1)
4423 txq->trans_start = jiffies;
4424}
4425
ba162f8e
FW
4426/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
4427static inline void netif_trans_update(struct net_device *dev)
4428{
9b36627a
FW
4429 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
4430
4431 if (txq->trans_start != jiffies)
4432 txq->trans_start = jiffies;
ba162f8e
FW
4433}
4434
d29f749e
DJ
4435/**
4436 * netif_tx_lock - grab network device transmit lock
4437 * @dev: network device
d29f749e
DJ
4438 *
4439 * Get network device transmit lock
4440 */
22dd7495
JHS
4441static inline void netif_tx_lock(struct net_device *dev)
4442{
e8a0464c 4443 unsigned int i;
c3f26a26 4444 int cpu;
c773e847 4445
c3f26a26
DM
4446 spin_lock(&dev->tx_global_lock);
4447 cpu = smp_processor_id();
e8a0464c
DM
4448 for (i = 0; i < dev->num_tx_queues; i++) {
4449 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4450
4451 /* We are the only thread of execution doing a
4452 * freeze, but we have to grab the _xmit_lock in
4453 * order to synchronize with threads which are in
4454 * the ->hard_start_xmit() handler and already
4455 * checked the frozen bit.
4456 */
e8a0464c 4457 __netif_tx_lock(txq, cpu);
c3f26a26
DM
4458 set_bit(__QUEUE_STATE_FROZEN, &txq->state);
4459 __netif_tx_unlock(txq);
e8a0464c 4460 }
932ff279
HX
4461}
4462
4463static inline void netif_tx_lock_bh(struct net_device *dev)
4464{
e8a0464c
DM
4465 local_bh_disable();
4466 netif_tx_lock(dev);
932ff279
HX
4467}
4468
932ff279
HX
4469static inline void netif_tx_unlock(struct net_device *dev)
4470{
e8a0464c
DM
4471 unsigned int i;
4472
4473 for (i = 0; i < dev->num_tx_queues; i++) {
4474 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c773e847 4475
c3f26a26
DM
4476 /* No need to grab the _xmit_lock here. If the
4477 * queue is not stopped for another reason, we
4478 * force a schedule.
4479 */
4480 clear_bit(__QUEUE_STATE_FROZEN, &txq->state);
7b3d3e4f 4481 netif_schedule_queue(txq);
c3f26a26
DM
4482 }
4483 spin_unlock(&dev->tx_global_lock);
932ff279
HX
4484}
4485
4486static inline void netif_tx_unlock_bh(struct net_device *dev)
4487{
e8a0464c
DM
4488 netif_tx_unlock(dev);
4489 local_bh_enable();
932ff279
HX
4490}
4491
c773e847 4492#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4493 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4494 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4495 } else { \
4496 __netif_tx_acquire(txq); \
22dd7495
JHS
4497 } \
4498}
4499
5efeac44
EB
4500#define HARD_TX_TRYLOCK(dev, txq) \
4501 (((dev->features & NETIF_F_LLTX) == 0) ? \
4502 __netif_tx_trylock(txq) : \
5a717f4f 4503 __netif_tx_acquire(txq))
5efeac44 4504
c773e847 4505#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4506 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4507 __netif_tx_unlock(txq); \
5a717f4f
MT
4508 } else { \
4509 __netif_tx_release(txq); \
22dd7495
JHS
4510 } \
4511}
4512
1da177e4
LT
4513static inline void netif_tx_disable(struct net_device *dev)
4514{
fd2ea0a7 4515 unsigned int i;
c3f26a26 4516 int cpu;
fd2ea0a7 4517
c3f26a26
DM
4518 local_bh_disable();
4519 cpu = smp_processor_id();
3aa6bce9 4520 spin_lock(&dev->tx_global_lock);
fd2ea0a7
DM
4521 for (i = 0; i < dev->num_tx_queues; i++) {
4522 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4523
4524 __netif_tx_lock(txq, cpu);
fd2ea0a7 4525 netif_tx_stop_queue(txq);
c3f26a26 4526 __netif_tx_unlock(txq);
fd2ea0a7 4527 }
3aa6bce9 4528 spin_unlock(&dev->tx_global_lock);
c3f26a26 4529 local_bh_enable();
1da177e4
LT
4530}
4531
e308a5d8
DM
4532static inline void netif_addr_lock(struct net_device *dev)
4533{
1fc70edb 4534 unsigned char nest_level = 0;
e308a5d8 4535
1fc70edb
TY
4536#ifdef CONFIG_LOCKDEP
4537 nest_level = dev->nested_level;
4538#endif
4539 spin_lock_nested(&dev->addr_list_lock, nest_level);
845e0ebb
CW
4540}
4541
e308a5d8
DM
4542static inline void netif_addr_lock_bh(struct net_device *dev)
4543{
1fc70edb
TY
4544 unsigned char nest_level = 0;
4545
4546#ifdef CONFIG_LOCKDEP
4547 nest_level = dev->nested_level;
4548#endif
4549 local_bh_disable();
4550 spin_lock_nested(&dev->addr_list_lock, nest_level);
e308a5d8
DM
4551}
4552
4553static inline void netif_addr_unlock(struct net_device *dev)
4554{
4555 spin_unlock(&dev->addr_list_lock);
4556}
4557
4558static inline void netif_addr_unlock_bh(struct net_device *dev)
4559{
4560 spin_unlock_bh(&dev->addr_list_lock);
4561}
4562
f001fde5 4563/*
31278e71 4564 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4565 * rcu_read_lock held.
4566 */
4567#define for_each_dev_addr(dev, ha) \
31278e71 4568 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4569
1da177e4
LT
4570/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4571
f629d208 4572void ether_setup(struct net_device *dev);
1da177e4
LT
4573
4574/* Support for loadable net-drivers */
f629d208 4575struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4576 unsigned char name_assign_type,
f629d208
JP
4577 void (*setup)(struct net_device *),
4578 unsigned int txqs, unsigned int rxqs);
c835a677
TG
4579#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4580 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4581
c835a677
TG
4582#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4583 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4584 count)
36909ea4 4585
f629d208
JP
4586int register_netdev(struct net_device *dev);
4587void unregister_netdev(struct net_device *dev);
f001fde5 4588
cd16627f
BG
4589int devm_register_netdev(struct device *dev, struct net_device *ndev);
4590
22bedad3 4591/* General hardware address lists handling functions */
f629d208
JP
4592int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4593 struct netdev_hw_addr_list *from_list, int addr_len);
4594void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4595 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4596int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4597 struct net_device *dev,
4598 int (*sync)(struct net_device *, const unsigned char *),
4599 int (*unsync)(struct net_device *,
4600 const unsigned char *));
e7946760
IK
4601int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4602 struct net_device *dev,
4603 int (*sync)(struct net_device *,
4604 const unsigned char *, int),
4605 int (*unsync)(struct net_device *,
4606 const unsigned char *, int));
4607void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4608 struct net_device *dev,
4609 int (*unsync)(struct net_device *,
4610 const unsigned char *, int));
670e5b8e
AD
4611void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4612 struct net_device *dev,
4613 int (*unsync)(struct net_device *,
4614 const unsigned char *));
f629d208 4615void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4616
f001fde5 4617/* Functions used for device addresses handling */
f629d208
JP
4618int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4619 unsigned char addr_type);
4620int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4621 unsigned char addr_type);
f629d208
JP
4622void dev_addr_flush(struct net_device *dev);
4623int dev_addr_init(struct net_device *dev);
a748ee24
JP
4624
4625/* Functions used for unicast addresses handling */
f629d208
JP
4626int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4627int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4628int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4629int dev_uc_sync(struct net_device *to, struct net_device *from);
4630int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4631void dev_uc_unsync(struct net_device *to, struct net_device *from);
4632void dev_uc_flush(struct net_device *dev);
4633void dev_uc_init(struct net_device *dev);
f001fde5 4634
670e5b8e
AD
4635/**
4636 * __dev_uc_sync - Synchonize device's unicast list
4637 * @dev: device to sync
4638 * @sync: function to call if address should be added
4639 * @unsync: function to call if address should be removed
4640 *
4641 * Add newly added addresses to the interface, and release
4642 * addresses that have been deleted.
5e82b4b2 4643 */
670e5b8e
AD
4644static inline int __dev_uc_sync(struct net_device *dev,
4645 int (*sync)(struct net_device *,
4646 const unsigned char *),
4647 int (*unsync)(struct net_device *,
4648 const unsigned char *))
4649{
4650 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4651}
4652
4653/**
e793c0f7 4654 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4655 * @dev: device to sync
4656 * @unsync: function to call if address should be removed
4657 *
4658 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4659 */
670e5b8e
AD
4660static inline void __dev_uc_unsync(struct net_device *dev,
4661 int (*unsync)(struct net_device *,
4662 const unsigned char *))
4663{
4664 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4665}
4666
22bedad3 4667/* Functions used for multicast addresses handling */
f629d208
JP
4668int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4669int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4670int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4671int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4672int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4673int dev_mc_sync(struct net_device *to, struct net_device *from);
4674int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4675void dev_mc_unsync(struct net_device *to, struct net_device *from);
4676void dev_mc_flush(struct net_device *dev);
4677void dev_mc_init(struct net_device *dev);
f001fde5 4678
670e5b8e
AD
4679/**
4680 * __dev_mc_sync - Synchonize device's multicast list
4681 * @dev: device to sync
4682 * @sync: function to call if address should be added
4683 * @unsync: function to call if address should be removed
4684 *
4685 * Add newly added addresses to the interface, and release
4686 * addresses that have been deleted.
5e82b4b2 4687 */
670e5b8e
AD
4688static inline int __dev_mc_sync(struct net_device *dev,
4689 int (*sync)(struct net_device *,
4690 const unsigned char *),
4691 int (*unsync)(struct net_device *,
4692 const unsigned char *))
4693{
4694 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4695}
4696
4697/**
e793c0f7 4698 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4699 * @dev: device to sync
4700 * @unsync: function to call if address should be removed
4701 *
4702 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4703 */
670e5b8e
AD
4704static inline void __dev_mc_unsync(struct net_device *dev,
4705 int (*unsync)(struct net_device *,
4706 const unsigned char *))
4707{
4708 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4709}
4710
4417da66 4711/* Functions used for secondary unicast and multicast support */
f629d208
JP
4712void dev_set_rx_mode(struct net_device *dev);
4713void __dev_set_rx_mode(struct net_device *dev);
4714int dev_set_promiscuity(struct net_device *dev, int inc);
4715int dev_set_allmulti(struct net_device *dev, int inc);
4716void netdev_state_change(struct net_device *dev);
7061eb8c 4717void __netdev_notify_peers(struct net_device *dev);
f629d208
JP
4718void netdev_notify_peers(struct net_device *dev);
4719void netdev_features_change(struct net_device *dev);
1da177e4 4720/* Load a device via the kmod */
f629d208
JP
4721void dev_load(struct net *net, const char *name);
4722struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4723 struct rtnl_link_stats64 *storage);
4724void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4725 const struct net_device_stats *netdev_stats);
44fa32f0
HK
4726void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
4727 const struct pcpu_sw_netstats __percpu *netstats);
a1839426 4728void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
eeda3fd6 4729
1da177e4 4730extern int netdev_max_backlog;
3b098e2d 4731extern int netdev_tstamp_prequeue;
5aa3afe1 4732extern int netdev_unregister_timeout_secs;
1da177e4 4733extern int weight_p;
3d48b53f
MT
4734extern int dev_weight_rx_bias;
4735extern int dev_weight_tx_bias;
4736extern int dev_rx_weight;
4737extern int dev_tx_weight;
323ebb61 4738extern int gro_normal_batch;
9ff162a8 4739
1fc70edb
TY
4740enum {
4741 NESTED_SYNC_IMM_BIT,
4742 NESTED_SYNC_TODO_BIT,
4743};
4744
4745#define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
4746#define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
4747
4748#define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
4749#define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
4750
eff74233 4751struct netdev_nested_priv {
1fc70edb 4752 unsigned char flags;
eff74233
TY
4753 void *data;
4754};
4755
f629d208 4756bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4757struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4758 struct list_head **iter);
f629d208
JP
4759struct net_device *netdev_all_upper_get_next_dev_rcu(struct net_device *dev,
4760 struct list_head **iter);
8b5be856 4761
1fc70edb
TY
4762#ifdef CONFIG_LOCKDEP
4763static LIST_HEAD(net_unlink_list);
4764
4765static inline void net_unlink_todo(struct net_device *dev)
4766{
4767 if (list_empty(&dev->unlink_list))
4768 list_add_tail(&dev->unlink_list, &net_unlink_list);
4769}
4770#endif
4771
44a40855
VY
4772/* iterate through upper list, must be called under RCU read lock */
4773#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4774 for (iter = &(dev)->adj_list.upper, \
4775 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4776 updev; \
4777 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4778
1a3f060c
DA
4779int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4780 int (*fn)(struct net_device *upper_dev,
eff74233
TY
4781 struct netdev_nested_priv *priv),
4782 struct netdev_nested_priv *priv);
1a3f060c
DA
4783
4784bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4785 struct net_device *upper_dev);
4786
25cc72a3
IS
4787bool netdev_has_any_upper_dev(struct net_device *dev);
4788
f629d208
JP
4789void *netdev_lower_get_next_private(struct net_device *dev,
4790 struct list_head **iter);
4791void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4792 struct list_head **iter);
31088a11
VF
4793
4794#define netdev_for_each_lower_private(dev, priv, iter) \
4795 for (iter = (dev)->adj_list.lower.next, \
4796 priv = netdev_lower_get_next_private(dev, &(iter)); \
4797 priv; \
4798 priv = netdev_lower_get_next_private(dev, &(iter)))
4799
4800#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4801 for (iter = &(dev)->adj_list.lower, \
4802 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4803 priv; \
4804 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4805
4085ebe8
VY
4806void *netdev_lower_get_next(struct net_device *dev,
4807 struct list_head **iter);
7ce856aa 4808
4085ebe8 4809#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4810 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4811 ldev = netdev_lower_get_next(dev, &(iter)); \
4812 ldev; \
4813 ldev = netdev_lower_get_next(dev, &(iter)))
4814
7151affe 4815struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
7ce856aa 4816 struct list_head **iter);
1a3f060c
DA
4817int netdev_walk_all_lower_dev(struct net_device *dev,
4818 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4819 struct netdev_nested_priv *priv),
4820 struct netdev_nested_priv *priv);
1a3f060c
DA
4821int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4822 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4823 struct netdev_nested_priv *priv),
4824 struct netdev_nested_priv *priv);
1a3f060c 4825
f629d208 4826void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4827void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4828struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4829struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4830int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4831 struct netlink_ext_ack *extack);
f629d208 4832int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4833 struct net_device *upper_dev,
42ab19ee
DA
4834 void *upper_priv, void *upper_info,
4835 struct netlink_ext_ack *extack);
f629d208
JP
4836void netdev_upper_dev_unlink(struct net_device *dev,
4837 struct net_device *upper_dev);
32b6d34f
TY
4838int netdev_adjacent_change_prepare(struct net_device *old_dev,
4839 struct net_device *new_dev,
4840 struct net_device *dev,
4841 struct netlink_ext_ack *extack);
4842void netdev_adjacent_change_commit(struct net_device *old_dev,
4843 struct net_device *new_dev,
4844 struct net_device *dev);
4845void netdev_adjacent_change_abort(struct net_device *old_dev,
4846 struct net_device *new_dev,
4847 struct net_device *dev);
5bb025fa 4848void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4849void *netdev_lower_dev_get_private(struct net_device *dev,
4850 struct net_device *lower_dev);
04d48266
JP
4851void netdev_lower_state_changed(struct net_device *lower_dev,
4852 void *lower_state_info);
960fb622
ED
4853
4854/* RSS keys are 40 or 52 bytes long */
4855#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4856extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4857void netdev_rss_key_fill(void *buffer, size_t len);
4858
f629d208 4859int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4860int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4861int skb_csum_hwoffload_help(struct sk_buff *skb,
4862 const netdev_features_t features);
4863
f629d208
JP
4864struct sk_buff *__skb_gso_segment(struct sk_buff *skb,
4865 netdev_features_t features, bool tx_path);
4866struct sk_buff *skb_mac_gso_segment(struct sk_buff *skb,
4867 netdev_features_t features);
12b0004d 4868
61bd3857
MS
4869struct netdev_bonding_info {
4870 ifslave slave;
4871 ifbond master;
4872};
4873
4874struct netdev_notifier_bonding_info {
4875 struct netdev_notifier_info info; /* must be first */
4876 struct netdev_bonding_info bonding_info;
4877};
4878
4879void netdev_bonding_info_change(struct net_device *dev,
4880 struct netdev_bonding_info *bonding_info);
4881
6b08d6c1
MK
4882#if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
4883void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
4884#else
4885static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
4886 const void *data)
4887{
4888}
4889#endif
4890
12b0004d
CW
4891static inline
4892struct sk_buff *skb_gso_segment(struct sk_buff *skb, netdev_features_t features)
4893{
4894 return __skb_gso_segment(skb, features, true);
4895}
53d6471c 4896__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4897
4898static inline bool can_checksum_protocol(netdev_features_t features,
4899 __be16 protocol)
4900{
c8cd0989
TH
4901 if (protocol == htons(ETH_P_FCOE))
4902 return !!(features & NETIF_F_FCOE_CRC);
4903
4904 /* Assume this is an IP checksum (not SCTP CRC) */
4905
4906 if (features & NETIF_F_HW_CSUM) {
4907 /* Can checksum everything */
4908 return true;
4909 }
4910
4911 switch (protocol) {
4912 case htons(ETH_P_IP):
4913 return !!(features & NETIF_F_IP_CSUM);
4914 case htons(ETH_P_IPV6):
4915 return !!(features & NETIF_F_IPV6_CSUM);
4916 default:
4917 return false;
4918 }
ec5f0615 4919}
12b0004d 4920
fb286bb2 4921#ifdef CONFIG_BUG
7fe50ac8 4922void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4923#else
7fe50ac8
CW
4924static inline void netdev_rx_csum_fault(struct net_device *dev,
4925 struct sk_buff *skb)
fb286bb2
HX
4926{
4927}
4928#endif
1da177e4 4929/* rx skb timestamps */
f629d208
JP
4930void net_enable_timestamp(void);
4931void net_disable_timestamp(void);
1da177e4 4932
20380731 4933#ifdef CONFIG_PROC_FS
f629d208 4934int __init dev_proc_init(void);
900ff8c6
CW
4935#else
4936#define dev_proc_init() 0
20380731
ACM
4937#endif
4938
4798248e 4939static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4940 struct sk_buff *skb, struct net_device *dev,
4941 bool more)
4798248e 4942{
6b16f9ee 4943 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4944 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4945}
4946
97cdcf37
FW
4947static inline bool netdev_xmit_more(void)
4948{
4949 return __this_cpu_read(softnet_data.xmit.more);
4950}
4951
10b3ad8c 4952static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4953 struct netdev_queue *txq, bool more)
4798248e
DM
4954{
4955 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4956 netdev_tx_t rc;
4798248e 4957
fa2dbdc2 4958 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4959 if (rc == NETDEV_TX_OK)
4960 txq_trans_update(txq);
4961
4962 return rc;
4798248e
DM
4963}
4964
b793dc5c 4965int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4966 const void *ns);
b793dc5c 4967void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4968 const void *ns);
58292cbe 4969
737aec57 4970extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4971
f629d208 4972const char *netdev_drivername(const struct net_device *dev);
6579e57b 4973
f629d208 4974void linkwatch_run_queue(void);
20380731 4975
da08143b
MK
4976static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4977 netdev_features_t f2)
4978{
c8cd0989
TH
4979 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4980 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4981 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4982 else
b6a0e72a 4983 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4984 }
da08143b 4985
c8cd0989 4986 return f1 & f2;
da08143b
MK
4987}
4988
c8f44aff
MM
4989static inline netdev_features_t netdev_get_wanted_features(
4990 struct net_device *dev)
5455c699
MM
4991{
4992 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4993}
c8f44aff
MM
4994netdev_features_t netdev_increment_features(netdev_features_t all,
4995 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4996
4997/* Allow TSO being used on stacked device :
4998 * Performing the GSO segmentation before last device
4999 * is a performance improvement.
5000 */
5001static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
5002 netdev_features_t mask)
5003{
5004 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
5005}
5006
6cb6a27c 5007int __netdev_update_features(struct net_device *dev);
5455c699 5008void netdev_update_features(struct net_device *dev);
afe12cc8 5009void netdev_change_features(struct net_device *dev);
7f353bf2 5010
fc4a7489
PM
5011void netif_stacked_transfer_operstate(const struct net_device *rootdev,
5012 struct net_device *dev);
5013
e38f3025
TM
5014netdev_features_t passthru_features_check(struct sk_buff *skb,
5015 struct net_device *dev,
5016 netdev_features_t features);
c1e756bf 5017netdev_features_t netif_skb_features(struct sk_buff *skb);
58e998c6 5018
4d29515f 5019static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 5020{
7b748340 5021 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
5022
5023 /* check flags correspondence */
5024 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
5025 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
5026 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 5027 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
5028 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
5029 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
5030 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
5031 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
5032 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
5033 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
5034 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
5035 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 5036 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 5037 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 5038 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 5039 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 5040 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 5041 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
3b335832 5042 BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
0345e186 5043
d6b4991a 5044 return (features & feature) == feature;
576a30eb
HX
5045}
5046
4d29515f 5047static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 5048{
278b2513 5049 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 5050 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
5051}
5052
8b86a61d 5053static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 5054 netdev_features_t features)
7967168c 5055{
fc741216 5056 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
5057 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
5058 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
5059}
5060
82cc1a7a
PWJ
5061static inline void netif_set_gso_max_size(struct net_device *dev,
5062 unsigned int size)
5063{
5064 dev->gso_max_size = size;
5065}
5066
7a7ffbab
WCC
5067static inline void skb_gso_error_unwind(struct sk_buff *skb, __be16 protocol,
5068 int pulled_hlen, u16 mac_offset,
5069 int mac_len)
5070{
5071 skb->protocol = protocol;
5072 skb->encapsulation = 1;
5073 skb_push(skb, pulled_hlen);
5074 skb_reset_transport_header(skb);
5075 skb->mac_header = mac_offset;
5076 skb->network_header = skb->mac_header + mac_len;
5077 skb->mac_len = mac_len;
5078}
5079
3c175784
SD
5080static inline bool netif_is_macsec(const struct net_device *dev)
5081{
5082 return dev->priv_flags & IFF_MACSEC;
5083}
5084
b618aaa9 5085static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
5086{
5087 return dev->priv_flags & IFF_MACVLAN;
5088}
5089
b618aaa9 5090static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
5091{
5092 return dev->priv_flags & IFF_MACVLAN_PORT;
5093}
5094
b618aaa9 5095static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 5096{
5097 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
5098}
5099
b618aaa9 5100static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
5101{
5102 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
5103}
5104
3bdc0eba
BG
5105static inline bool netif_supports_nofcs(struct net_device *dev)
5106{
5107 return dev->priv_flags & IFF_SUPP_NOFCS;
5108}
5109
d5256083
DB
5110static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
5111{
5112 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
5113}
5114
007979ea 5115static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 5116{
007979ea 5117 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
5118}
5119
fee6d4c7
DA
5120static inline bool netif_is_l3_slave(const struct net_device *dev)
5121{
5122 return dev->priv_flags & IFF_L3MDEV_SLAVE;
5123}
5124
0894ae3f
JP
5125static inline bool netif_is_bridge_master(const struct net_device *dev)
5126{
5127 return dev->priv_flags & IFF_EBRIDGE;
5128}
5129
28f9ee22
VY
5130static inline bool netif_is_bridge_port(const struct net_device *dev)
5131{
5132 return dev->priv_flags & IFF_BRIDGE_PORT;
5133}
5134
35d4e172
JP
5135static inline bool netif_is_ovs_master(const struct net_device *dev)
5136{
5137 return dev->priv_flags & IFF_OPENVSWITCH;
5138}
5139
5be66141
JP
5140static inline bool netif_is_ovs_port(const struct net_device *dev)
5141{
5142 return dev->priv_flags & IFF_OVS_DATAPATH;
5143}
5144
df23bb18
SB
5145static inline bool netif_is_any_bridge_port(const struct net_device *dev)
5146{
5147 return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
5148}
5149
b618aaa9 5150static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
5151{
5152 return dev->priv_flags & IFF_TEAM;
5153}
5154
b618aaa9 5155static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
5156{
5157 return dev->priv_flags & IFF_TEAM_PORT;
5158}
5159
b618aaa9 5160static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
5161{
5162 return netif_is_bond_master(dev) || netif_is_team_master(dev);
5163}
5164
b618aaa9 5165static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
5166{
5167 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
5168}
5169
d4ab4286
KJ
5170static inline bool netif_is_rxfh_configured(const struct net_device *dev)
5171{
5172 return dev->priv_flags & IFF_RXFH_CONFIGURED;
5173}
5174
30c8bd5a
SS
5175static inline bool netif_is_failover(const struct net_device *dev)
5176{
5177 return dev->priv_flags & IFF_FAILOVER;
5178}
5179
5180static inline bool netif_is_failover_slave(const struct net_device *dev)
5181{
5182 return dev->priv_flags & IFF_FAILOVER_SLAVE;
5183}
5184
02875878
ED
5185/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
5186static inline void netif_keep_dst(struct net_device *dev)
5187{
5188 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
5189}
5190
18d3df3e
PA
5191/* return true if dev can't cope with mtu frames that need vlan tag insertion */
5192static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
5193{
5194 /* TODO: reserve and use an additional IFF bit, if we get more users */
5195 return dev->priv_flags & IFF_MACSEC;
5196}
5197
505d4f73 5198extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 5199
571ba423
JP
5200/* Logging, debugging and troubleshooting/diagnostic helpers. */
5201
5202/* netdev_printk helpers, similar to dev_printk */
5203
5204static inline const char *netdev_name(const struct net_device *dev)
5205{
c6f854d5
VF
5206 if (!dev->name[0] || strchr(dev->name, '%'))
5207 return "(unnamed net_device)";
571ba423
JP
5208 return dev->name;
5209}
5210
8397ed36
DA
5211static inline bool netdev_unregistering(const struct net_device *dev)
5212{
5213 return dev->reg_state == NETREG_UNREGISTERING;
5214}
5215
ccc7f496
VF
5216static inline const char *netdev_reg_state(const struct net_device *dev)
5217{
5218 switch (dev->reg_state) {
5219 case NETREG_UNINITIALIZED: return " (uninitialized)";
5220 case NETREG_REGISTERED: return "";
5221 case NETREG_UNREGISTERING: return " (unregistering)";
5222 case NETREG_UNREGISTERED: return " (unregistered)";
5223 case NETREG_RELEASED: return " (released)";
5224 case NETREG_DUMMY: return " (dummy)";
5225 }
5226
5227 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
5228 return " (unknown)";
5229}
5230
ce3fdb69 5231__printf(3, 4) __cold
6ea754eb
JP
5232void netdev_printk(const char *level, const struct net_device *dev,
5233 const char *format, ...);
ce3fdb69 5234__printf(2, 3) __cold
6ea754eb 5235void netdev_emerg(const struct net_device *dev, const char *format, ...);
ce3fdb69 5236__printf(2, 3) __cold
6ea754eb 5237void netdev_alert(const struct net_device *dev, const char *format, ...);
ce3fdb69 5238__printf(2, 3) __cold
6ea754eb 5239void netdev_crit(const struct net_device *dev, const char *format, ...);
ce3fdb69 5240__printf(2, 3) __cold
6ea754eb 5241void netdev_err(const struct net_device *dev, const char *format, ...);
ce3fdb69 5242__printf(2, 3) __cold
6ea754eb 5243void netdev_warn(const struct net_device *dev, const char *format, ...);
ce3fdb69 5244__printf(2, 3) __cold
6ea754eb 5245void netdev_notice(const struct net_device *dev, const char *format, ...);
ce3fdb69 5246__printf(2, 3) __cold
6ea754eb 5247void netdev_info(const struct net_device *dev, const char *format, ...);
571ba423 5248
375ef2b1
GP
5249#define netdev_level_once(level, dev, fmt, ...) \
5250do { \
5251 static bool __print_once __read_mostly; \
5252 \
5253 if (!__print_once) { \
5254 __print_once = true; \
5255 netdev_printk(level, dev, fmt, ##__VA_ARGS__); \
5256 } \
5257} while (0)
5258
5259#define netdev_emerg_once(dev, fmt, ...) \
5260 netdev_level_once(KERN_EMERG, dev, fmt, ##__VA_ARGS__)
5261#define netdev_alert_once(dev, fmt, ...) \
5262 netdev_level_once(KERN_ALERT, dev, fmt, ##__VA_ARGS__)
5263#define netdev_crit_once(dev, fmt, ...) \
5264 netdev_level_once(KERN_CRIT, dev, fmt, ##__VA_ARGS__)
5265#define netdev_err_once(dev, fmt, ...) \
5266 netdev_level_once(KERN_ERR, dev, fmt, ##__VA_ARGS__)
5267#define netdev_warn_once(dev, fmt, ...) \
5268 netdev_level_once(KERN_WARNING, dev, fmt, ##__VA_ARGS__)
5269#define netdev_notice_once(dev, fmt, ...) \
5270 netdev_level_once(KERN_NOTICE, dev, fmt, ##__VA_ARGS__)
5271#define netdev_info_once(dev, fmt, ...) \
5272 netdev_level_once(KERN_INFO, dev, fmt, ##__VA_ARGS__)
5273
8909c9ad
VK
5274#define MODULE_ALIAS_NETDEV(device) \
5275 MODULE_ALIAS("netdev-" device)
5276
ceabef7d
OZ
5277#if defined(CONFIG_DYNAMIC_DEBUG) || \
5278 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
571ba423
JP
5279#define netdev_dbg(__dev, format, args...) \
5280do { \
ffa10cb4 5281 dynamic_netdev_dbg(__dev, format, ##args); \
571ba423 5282} while (0)
b558c96f
JC
5283#elif defined(DEBUG)
5284#define netdev_dbg(__dev, format, args...) \
5285 netdev_printk(KERN_DEBUG, __dev, format, ##args)
571ba423
JP
5286#else
5287#define netdev_dbg(__dev, format, args...) \
5288({ \
5289 if (0) \
5290 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
571ba423
JP
5291})
5292#endif
5293
5294#if defined(VERBOSE_DEBUG)
5295#define netdev_vdbg netdev_dbg
5296#else
5297
5298#define netdev_vdbg(dev, format, args...) \
5299({ \
5300 if (0) \
5301 netdev_printk(KERN_DEBUG, dev, format, ##args); \
5302 0; \
5303})
5304#endif
5305
5306/*
5307 * netdev_WARN() acts like dev_printk(), but with the key difference
5308 * of using a WARN/WARN_ON to get the message out, including the
5309 * file/line information and a backtrace.
5310 */
5311#define netdev_WARN(dev, format, args...) \
e1cfe3d0 5312 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 5313 netdev_reg_state(dev), ##args)
571ba423 5314
72dd831e 5315#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 5316 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
5317 netdev_reg_state(dev), ##args)
5318
b3d95c5c
JP
5319/* netif printk helpers, similar to netdev_printk */
5320
5321#define netif_printk(priv, type, level, dev, fmt, args...) \
5322do { \
5323 if (netif_msg_##type(priv)) \
5324 netdev_printk(level, (dev), fmt, ##args); \
5325} while (0)
5326
f45f4321
JP
5327#define netif_level(level, priv, type, dev, fmt, args...) \
5328do { \
5329 if (netif_msg_##type(priv)) \
5330 netdev_##level(dev, fmt, ##args); \
5331} while (0)
5332
b3d95c5c 5333#define netif_emerg(priv, type, dev, fmt, args...) \
f45f4321 5334 netif_level(emerg, priv, type, dev, fmt, ##args)
b3d95c5c 5335#define netif_alert(priv, type, dev, fmt, args...) \
f45f4321 5336 netif_level(alert, priv, type, dev, fmt, ##args)
b3d95c5c 5337#define netif_crit(priv, type, dev, fmt, args...) \
f45f4321 5338 netif_level(crit, priv, type, dev, fmt, ##args)
b3d95c5c 5339#define netif_err(priv, type, dev, fmt, args...) \
f45f4321 5340 netif_level(err, priv, type, dev, fmt, ##args)
b3d95c5c 5341#define netif_warn(priv, type, dev, fmt, args...) \
f45f4321 5342 netif_level(warn, priv, type, dev, fmt, ##args)
b3d95c5c 5343#define netif_notice(priv, type, dev, fmt, args...) \
f45f4321 5344 netif_level(notice, priv, type, dev, fmt, ##args)
b3d95c5c 5345#define netif_info(priv, type, dev, fmt, args...) \
f45f4321 5346 netif_level(info, priv, type, dev, fmt, ##args)
b3d95c5c 5347
ceabef7d
OZ
5348#if defined(CONFIG_DYNAMIC_DEBUG) || \
5349 (defined(CONFIG_DYNAMIC_DEBUG_CORE) && defined(DYNAMIC_DEBUG_MODULE))
b3d95c5c
JP
5350#define netif_dbg(priv, type, netdev, format, args...) \
5351do { \
5352 if (netif_msg_##type(priv)) \
b5fb0a03 5353 dynamic_netdev_dbg(netdev, format, ##args); \
b3d95c5c 5354} while (0)
0053ea9c
JP
5355#elif defined(DEBUG)
5356#define netif_dbg(priv, type, dev, format, args...) \
5357 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
b3d95c5c
JP
5358#else
5359#define netif_dbg(priv, type, dev, format, args...) \
5360({ \
5361 if (0) \
5362 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
5363 0; \
5364})
5365#endif
5366
f617f276
EC
5367/* if @cond then downgrade to debug, else print at @level */
5368#define netif_cond_dbg(priv, type, netdev, cond, level, fmt, args...) \
5369 do { \
5370 if (cond) \
5371 netif_dbg(priv, type, netdev, fmt, ##args); \
5372 else \
5373 netif_ ## level(priv, type, netdev, fmt, ##args); \
5374 } while (0)
5375
b3d95c5c 5376#if defined(VERBOSE_DEBUG)
bcfcc450 5377#define netif_vdbg netif_dbg
b3d95c5c
JP
5378#else
5379#define netif_vdbg(priv, type, dev, format, args...) \
5380({ \
5381 if (0) \
a4ed89cb 5382 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
b3d95c5c
JP
5383 0; \
5384})
5385#endif
571ba423 5386
900ff8c6
CW
5387/*
5388 * The list of packet types we will receive (as opposed to discard)
5389 * and the routines to invoke.
5390 *
5391 * Why 16. Because with 16 the only overlap we get on a hash of the
5392 * low nibble of the protocol value is RARP/SNAP/X.25.
5393 *
900ff8c6 5394 * 0800 IP
900ff8c6
CW
5395 * 0001 802.3
5396 * 0002 AX.25
5397 * 0004 802.2
5398 * 8035 RARP
5399 * 0005 SNAP
5400 * 0805 X.25
5401 * 0806 ARP
5402 * 8137 IPX
5403 * 0009 Localtalk
5404 * 86DD IPv6
5405 */
5406#define PTYPE_HASH_SIZE (16)
5407#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
5408
744b8376
VO
5409extern struct list_head ptype_all __read_mostly;
5410extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
5411
4de83b88
MB
5412extern struct net_device *blackhole_netdev;
5413
385a154c 5414#endif /* _LINUX_NETDEVICE_H */