Merge tag 'omap-fixes-audio-clock-and-modem-signed' of git://git.kernel.org/pub/scm...
[linux-2.6-block.git] / include / linux / netdevice.h
CommitLineData
2874c5fd 1/* SPDX-License-Identifier: GPL-2.0-or-later */
1da177e4
LT
2/*
3 * INET An implementation of the TCP/IP protocol suite for the LINUX
4 * operating system. INET is implemented using the BSD Socket
5 * interface as the means of communication with the user level.
6 *
7 * Definitions for the Interfaces handler.
8 *
9 * Version: @(#)dev.h 1.0.10 08/12/93
10 *
02c30a84 11 * Authors: Ross Biro
1da177e4
LT
12 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
13 * Corey Minyard <wf-rch!minyard@relay.EU.net>
14 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
113aa838 15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
1da177e4
LT
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
1da177e4
LT
19 * Moved to /usr/include/linux for NET3
20 */
21#ifndef _LINUX_NETDEVICE_H
22#define _LINUX_NETDEVICE_H
23
d7fe0f24 24#include <linux/timer.h>
187f1882 25#include <linux/bug.h>
bea3348e 26#include <linux/delay.h>
60063497 27#include <linux/atomic.h>
53511453 28#include <linux/prefetch.h>
1da177e4
LT
29#include <asm/cache.h>
30#include <asm/byteorder.h>
625788b5 31#include <asm/local.h>
1da177e4 32
1da177e4 33#include <linux/percpu.h>
4d5b78c0 34#include <linux/rculist.h>
bea3348e 35#include <linux/workqueue.h>
114cf580 36#include <linux/dynamic_queue_limits.h>
1da177e4 37
a050c33f 38#include <net/net_namespace.h>
7a6b6f51 39#ifdef CONFIG_DCB
2f90b865
AD
40#include <net/dcbnl.h>
41#endif
5bc1421e 42#include <net/netprio_cgroup.h>
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>
d3d854fd 49#include <uapi/linux/netdev.h>
59cc1f61 50#include <linux/hashtable.h>
406f42fa 51#include <linux/rbtree.h>
9ba74e6c 52#include <net/net_trackers.h>
5b87be9e 53#include <net/net_debug.h>
5b8285cc 54#include <net/dropreason-core.h>
a59e2ecb 55
115c1d6e 56struct netpoll_info;
313162d0 57struct device;
cc69837f 58struct ethtool_ops;
66f72230 59struct kernel_hwtstamp_config;
c1f19b51 60struct phy_device;
2f657a60 61struct dsa_port;
607259a6 62struct ip_tunnel_parm;
30e9bb84
AT
63struct macsec_context;
64struct macsec_ops;
6264f58c
JK
65struct netdev_name_node;
66struct sd_flow_limit;
e679c9c1 67struct sfp_bus;
704232c2
JB
68/* 802.11 specific */
69struct wireless_dev;
98a18b6f
AA
70/* 802.15.4 specific */
71struct wpan_dev;
03c57747 72struct mpls_dev;
7c46a640
AD
73/* UDP Tunnel offloads */
74struct udp_tunnel_info;
cc4e3835
JK
75struct udp_tunnel_nic_info;
76struct udp_tunnel_nic;
a7862b45 77struct bpf_prog;
814abfab 78struct xdp_buff;
680ee045
JK
79struct xdp_frame;
80struct xdp_metadata_ops;
3d76a4d3 81struct xdp_md;
1da177e4 82
680ee045
JK
83typedef u32 xdp_features_t;
84
5198d545 85void synchronize_net(void);
f629d208
JP
86void netdev_set_default_ethtool_ops(struct net_device *dev,
87 const struct ethtool_ops *ops);
d9360708 88void netdev_sw_irq_coalesce_default_on(struct net_device *dev);
d07d7507 89
9a1654ba
JP
90/* Backlog congestion levels */
91#define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
92#define NET_RX_DROP 1 /* packet dropped */
93
7151affe
TY
94#define MAX_NEST_DEV 8
95
572a9d7b
PM
96/*
97 * Transmit return codes: transmit return codes originate from three different
98 * namespaces:
99 *
100 * - qdisc return codes
101 * - driver transmit return codes
102 * - errno values
103 *
104 * Drivers are allowed to return any one of those in their hard_start_xmit()
105 * function. Real network devices commonly used with qdiscs should only return
106 * the driver transmit return codes though - when qdiscs are used, the actual
107 * transmission happens asynchronously, so the value is not propagated to
5e82b4b2
BH
108 * higher layers. Virtual network devices transmit synchronously; in this case
109 * the driver transmit return codes are consumed by dev_queue_xmit(), and all
572a9d7b
PM
110 * others are propagated to higher layers.
111 */
112
113/* qdisc ->enqueue() return codes. */
114#define NET_XMIT_SUCCESS 0x00
9a1654ba
JP
115#define NET_XMIT_DROP 0x01 /* skb dropped */
116#define NET_XMIT_CN 0x02 /* congestion notification */
9a1654ba 117#define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
1da177e4 118
b9df3cb8
GR
119/* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
120 * indicates that the device will soon be dropping packets, or already drops
121 * some packets of the same priority; prompting us to send less aggressively. */
572a9d7b 122#define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
1da177e4
LT
123#define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
124
dc1f8bf6 125/* Driver transmit return codes */
9a1654ba 126#define NETDEV_TX_MASK 0xf0
572a9d7b 127
dc1f8bf6 128enum netdev_tx {
572a9d7b 129 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
9a1654ba
JP
130 NETDEV_TX_OK = 0x00, /* driver took care of packet */
131 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
dc1f8bf6
SH
132};
133typedef enum netdev_tx netdev_tx_t;
134
9a1654ba
JP
135/*
136 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
137 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
138 */
139static inline bool dev_xmit_complete(int rc)
140{
141 /*
142 * Positive cases with an skb consumed by a driver:
143 * - successful transmission (rc == NETDEV_TX_OK)
144 * - error while transmitting (rc < 0)
145 * - error while queueing to a different device (rc & NET_XMIT_MASK)
146 */
147 if (likely(rc < NET_XMIT_MASK))
148 return true;
149
150 return false;
151}
152
1da177e4 153/*
5e82b4b2 154 * Compute the worst-case header length according to the protocols
1da177e4
LT
155 * used.
156 */
fe2918b0 157
c0eb4540
KS
158#if defined(CONFIG_HYPERV_NET)
159# define LL_MAX_HEADER 128
160#elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
8388e3da
DM
161# if defined(CONFIG_MAC80211_MESH)
162# define LL_MAX_HEADER 128
163# else
164# define LL_MAX_HEADER 96
165# endif
1da177e4 166#else
8388e3da 167# define LL_MAX_HEADER 32
1da177e4
LT
168#endif
169
d11ead75
BH
170#if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
171 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
1da177e4
LT
172#define MAX_HEADER LL_MAX_HEADER
173#else
174#define MAX_HEADER (LL_MAX_HEADER + 48)
175#endif
176
177/*
be1f3c2c
BH
178 * Old network device statistics. Fields are native words
179 * (unsigned long) so they can be read and written atomically.
1da177e4 180 */
fe2918b0 181
6c1c5097
ED
182#define NET_DEV_STAT(FIELD) \
183 union { \
184 unsigned long FIELD; \
185 atomic_long_t __##FIELD; \
186 }
187
d94d9fee 188struct net_device_stats {
6c1c5097
ED
189 NET_DEV_STAT(rx_packets);
190 NET_DEV_STAT(tx_packets);
191 NET_DEV_STAT(rx_bytes);
192 NET_DEV_STAT(tx_bytes);
193 NET_DEV_STAT(rx_errors);
194 NET_DEV_STAT(tx_errors);
195 NET_DEV_STAT(rx_dropped);
196 NET_DEV_STAT(tx_dropped);
197 NET_DEV_STAT(multicast);
198 NET_DEV_STAT(collisions);
199 NET_DEV_STAT(rx_length_errors);
200 NET_DEV_STAT(rx_over_errors);
201 NET_DEV_STAT(rx_crc_errors);
202 NET_DEV_STAT(rx_frame_errors);
203 NET_DEV_STAT(rx_fifo_errors);
204 NET_DEV_STAT(rx_missed_errors);
205 NET_DEV_STAT(tx_aborted_errors);
206 NET_DEV_STAT(tx_carrier_errors);
207 NET_DEV_STAT(tx_fifo_errors);
208 NET_DEV_STAT(tx_heartbeat_errors);
209 NET_DEV_STAT(tx_window_errors);
210 NET_DEV_STAT(rx_compressed);
211 NET_DEV_STAT(tx_compressed);
1da177e4 212};
6c1c5097 213#undef NET_DEV_STAT
1da177e4 214
625788b5
ED
215/* per-cpu stats, allocated on demand.
216 * Try to fit them in a single cache line, for dev_get_stats() sake.
217 */
218struct net_device_core_stats {
6510ea97
SAS
219 unsigned long rx_dropped;
220 unsigned long tx_dropped;
221 unsigned long rx_nohandler;
0e55546b 222 unsigned long rx_otherhost_dropped;
6510ea97 223} __aligned(4 * sizeof(unsigned long));
1da177e4
LT
224
225#include <linux/cache.h>
226#include <linux/skbuff.h>
227
adc9300e 228#ifdef CONFIG_RPS
c5905afb 229#include <linux/static_key.h>
dc05360f
ED
230extern struct static_key_false rps_needed;
231extern struct static_key_false rfs_needed;
adc9300e
ED
232#endif
233
1da177e4
LT
234struct neighbour;
235struct neigh_parms;
236struct sk_buff;
237
f001fde5
JP
238struct netdev_hw_addr {
239 struct list_head list;
406f42fa 240 struct rb_node node;
f001fde5
JP
241 unsigned char addr[MAX_ADDR_LEN];
242 unsigned char type;
ccffad25
JP
243#define NETDEV_HW_ADDR_T_LAN 1
244#define NETDEV_HW_ADDR_T_SAN 2
8e1b3884
TY
245#define NETDEV_HW_ADDR_T_UNICAST 3
246#define NETDEV_HW_ADDR_T_MULTICAST 4
22bedad3 247 bool global_use;
4cd729b0 248 int sync_cnt;
8f8f103d 249 int refcount;
4543fbef 250 int synced;
f001fde5
JP
251 struct rcu_head rcu_head;
252};
253
31278e71
JP
254struct netdev_hw_addr_list {
255 struct list_head list;
256 int count;
406f42fa
GN
257
258 /* Auxiliary tree for faster lookup on addition and deletion */
259 struct rb_root tree;
31278e71
JP
260};
261
22bedad3
JP
262#define netdev_hw_addr_list_count(l) ((l)->count)
263#define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
264#define netdev_hw_addr_list_for_each(ha, l) \
265 list_for_each_entry(ha, &(l)->list, list)
32e7bfc4 266
22bedad3
JP
267#define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
268#define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
269#define netdev_for_each_uc_addr(ha, dev) \
270 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
db01868b
VO
271#define netdev_for_each_synced_uc_addr(_ha, _dev) \
272 netdev_for_each_uc_addr((_ha), (_dev)) \
273 if ((_ha)->sync_cnt)
6683ece3 274
22bedad3
JP
275#define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
276#define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
18e225f2 277#define netdev_for_each_mc_addr(ha, dev) \
22bedad3 278 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
db01868b
VO
279#define netdev_for_each_synced_mc_addr(_ha, _dev) \
280 netdev_for_each_mc_addr((_ha), (_dev)) \
281 if ((_ha)->sync_cnt)
6683ece3 282
d94d9fee 283struct hh_cache {
5b3dc2f3 284 unsigned int hh_len;
3644f0ce 285 seqlock_t hh_lock;
1da177e4
LT
286
287 /* cached hardware header; allow for machine alignment needs. */
288#define HH_DATA_MOD 16
289#define HH_DATA_OFF(__len) \
5ba0eac6 290 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
1da177e4
LT
291#define HH_DATA_ALIGN(__len) \
292 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
293 unsigned long hh_data[HH_DATA_ALIGN(LL_MAX_HEADER) / sizeof(long)];
294};
295
5e82b4b2 296/* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
1da177e4
LT
297 * Alternative is:
298 * dev->hard_header_len ? (dev->hard_header_len +
299 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
300 *
301 * We could use other alignment values, but we must maintain the
302 * relationship HH alignment <= LL alignment.
303 */
304#define LL_RESERVED_SPACE(dev) \
4b397c06
ED
305 ((((dev)->hard_header_len + READ_ONCE((dev)->needed_headroom)) \
306 & ~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 307#define LL_RESERVED_SPACE_EXTRA(dev,extra) \
4b397c06
ED
308 ((((dev)->hard_header_len + READ_ONCE((dev)->needed_headroom) + (extra)) \
309 & ~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
1da177e4 310
3b04ddde
SH
311struct header_ops {
312 int (*create) (struct sk_buff *skb, struct net_device *dev,
313 unsigned short type, const void *daddr,
95c96174 314 const void *saddr, unsigned int len);
3b04ddde 315 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
e69dd336 316 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
3b04ddde
SH
317 void (*cache_update)(struct hh_cache *hh,
318 const struct net_device *dev,
319 const unsigned char *haddr);
2793a23a 320 bool (*validate)(const char *ll_header, unsigned int len);
e78b2915 321 __be16 (*parse_protocol)(const struct sk_buff *skb);
3b04ddde
SH
322};
323
1da177e4 324/* These flag bits are private to the generic network queueing
5e82b4b2 325 * layer; they may not be explicitly referenced by any other
1da177e4
LT
326 * code.
327 */
328
d94d9fee 329enum netdev_state_t {
1da177e4
LT
330 __LINK_STATE_START,
331 __LINK_STATE_PRESENT,
1da177e4 332 __LINK_STATE_NOCARRIER,
b00055aa
SR
333 __LINK_STATE_LINKWATCH_PENDING,
334 __LINK_STATE_DORMANT,
eec517cd 335 __LINK_STATE_TESTING,
1da177e4
LT
336};
337
6312fe77
LR
338struct gro_list {
339 struct list_head list;
340 int count;
341};
342
bea3348e 343/*
d9f37d01
LR
344 * size of gro hash buckets, must less than bit number of
345 * napi_struct::gro_bitmask
bea3348e 346 */
07d78363 347#define GRO_HASH_BUCKETS 8
d9f37d01
LR
348
349/*
350 * Structure for NAPI scheduling similar to tasklet but with weighting
351 */
bea3348e
SH
352struct napi_struct {
353 /* The poll_list must only be managed by the entity which
354 * changes the state of the NAPI_STATE_SCHED bit. This means
355 * whoever atomically sets that bit can add this napi_struct
5e82b4b2 356 * to the per-CPU poll_list, and whoever clears that bit
bea3348e
SH
357 * can remove from the list right before clearing the bit.
358 */
359 struct list_head poll_list;
360
361 unsigned long state;
362 int weight;
6f8b12d6 363 int defer_hard_irqs_count;
d9f37d01 364 unsigned long gro_bitmask;
bea3348e
SH
365 int (*poll)(struct napi_struct *, int);
366#ifdef CONFIG_NETPOLL
8c48eea3 367 /* CPU actively polling if netpoll is configured */
bea3348e 368 int poll_owner;
bea3348e 369#endif
8c48eea3
JK
370 /* CPU on which NAPI has been scheduled for processing */
371 int list_owner;
5d38a079 372 struct net_device *dev;
6312fe77 373 struct gro_list gro_hash[GRO_HASH_BUCKETS];
5d38a079 374 struct sk_buff *skb;
323ebb61
EC
375 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
376 int rx_count; /* length of rx_list */
dd2d6604 377 unsigned int napi_id;
3b47d303 378 struct hrtimer timer;
dd2d6604
JK
379 struct task_struct *thread;
380 /* control-path-only fields follow */
404f7c9e 381 struct list_head dev_list;
af12fa6e 382 struct hlist_node napi_hash_node;
bea3348e
SH
383};
384
d94d9fee 385enum {
7fd3253a
BT
386 NAPI_STATE_SCHED, /* Poll is scheduled */
387 NAPI_STATE_MISSED, /* reschedule a napi */
388 NAPI_STATE_DISABLE, /* Disable pending */
389 NAPI_STATE_NPSVC, /* Netpoll - don't dequeue from poll_list */
390 NAPI_STATE_LISTED, /* NAPI added to system lists */
391 NAPI_STATE_NO_BUSY_POLL, /* Do not add in napi_hash, no busy polling */
392 NAPI_STATE_IN_BUSY_POLL, /* sk_busy_loop() owns this NAPI */
393 NAPI_STATE_PREFER_BUSY_POLL, /* prefer busy-polling over softirq processing*/
29863d41 394 NAPI_STATE_THREADED, /* The poll is performed inside its own thread*/
cb038357 395 NAPI_STATE_SCHED_THREADED, /* Napi is currently scheduled in threaded mode */
217f6974
ED
396};
397
398enum {
7fd3253a
BT
399 NAPIF_STATE_SCHED = BIT(NAPI_STATE_SCHED),
400 NAPIF_STATE_MISSED = BIT(NAPI_STATE_MISSED),
401 NAPIF_STATE_DISABLE = BIT(NAPI_STATE_DISABLE),
402 NAPIF_STATE_NPSVC = BIT(NAPI_STATE_NPSVC),
403 NAPIF_STATE_LISTED = BIT(NAPI_STATE_LISTED),
404 NAPIF_STATE_NO_BUSY_POLL = BIT(NAPI_STATE_NO_BUSY_POLL),
405 NAPIF_STATE_IN_BUSY_POLL = BIT(NAPI_STATE_IN_BUSY_POLL),
406 NAPIF_STATE_PREFER_BUSY_POLL = BIT(NAPI_STATE_PREFER_BUSY_POLL),
29863d41 407 NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
cb038357 408 NAPIF_STATE_SCHED_THREADED = BIT(NAPI_STATE_SCHED_THREADED),
bea3348e
SH
409};
410
5b252f0c 411enum gro_result {
d1c76af9
HX
412 GRO_MERGED,
413 GRO_MERGED_FREE,
414 GRO_HELD,
415 GRO_NORMAL,
25393d3f 416 GRO_CONSUMED,
d1c76af9 417};
5b252f0c 418typedef enum gro_result gro_result_t;
d1c76af9 419
8a4eb573
JP
420/*
421 * enum rx_handler_result - Possible return values for rx_handlers.
422 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
423 * further.
424 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
425 * case skb->dev was changed by rx_handler.
426 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
5e82b4b2 427 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
8a4eb573
JP
428 *
429 * rx_handlers are functions called from inside __netif_receive_skb(), to do
430 * special processing of the skb, prior to delivery to protocol handlers.
431 *
432 * Currently, a net_device can only have a single rx_handler registered. Trying
433 * to register a second rx_handler will return -EBUSY.
434 *
435 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
436 * To unregister a rx_handler on a net_device, use
437 * netdev_rx_handler_unregister().
438 *
439 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
440 * do with the skb.
441 *
5e82b4b2 442 * If the rx_handler consumed the skb in some way, it should return
8a4eb573 443 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
5e82b4b2 444 * the skb to be delivered in some other way.
8a4eb573
JP
445 *
446 * If the rx_handler changed skb->dev, to divert the skb to another
447 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
448 * new device will be called if it exists.
449 *
5e82b4b2 450 * If the rx_handler decides the skb should be ignored, it should return
8a4eb573 451 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
d93cf068 452 * are registered on exact device (ptype->dev == skb->dev).
8a4eb573 453 *
5e82b4b2 454 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
8a4eb573
JP
455 * delivered, it should return RX_HANDLER_PASS.
456 *
457 * A device without a registered rx_handler will behave as if rx_handler
458 * returned RX_HANDLER_PASS.
459 */
460
461enum rx_handler_result {
462 RX_HANDLER_CONSUMED,
463 RX_HANDLER_ANOTHER,
464 RX_HANDLER_EXACT,
465 RX_HANDLER_PASS,
466};
467typedef enum rx_handler_result rx_handler_result_t;
468typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
ab95bfe0 469
f629d208 470void __napi_schedule(struct napi_struct *n);
bc9ad166 471void __napi_schedule_irqoff(struct napi_struct *n);
bea3348e 472
4d29515f 473static inline bool napi_disable_pending(struct napi_struct *n)
a0a46196
DM
474{
475 return test_bit(NAPI_STATE_DISABLE, &n->state);
476}
477
7fd3253a
BT
478static inline bool napi_prefer_busy_poll(struct napi_struct *n)
479{
480 return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
481}
482
39e6c820 483bool napi_schedule_prep(struct napi_struct *n);
bea3348e
SH
484
485/**
486 * napi_schedule - schedule NAPI poll
5e82b4b2 487 * @n: NAPI context
bea3348e
SH
488 *
489 * Schedule NAPI poll routine to be called if it is not already
490 * running.
491 */
492static inline void napi_schedule(struct napi_struct *n)
493{
494 if (napi_schedule_prep(n))
495 __napi_schedule(n);
496}
497
bc9ad166
ED
498/**
499 * napi_schedule_irqoff - schedule NAPI poll
5e82b4b2 500 * @n: NAPI context
bc9ad166
ED
501 *
502 * Variant of napi_schedule(), assuming hard irqs are masked.
503 */
504static inline void napi_schedule_irqoff(struct napi_struct *n)
505{
506 if (napi_schedule_prep(n))
507 __napi_schedule_irqoff(n);
508}
509
bfe13f54 510/* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
4d29515f 511static inline bool napi_reschedule(struct napi_struct *napi)
bfe13f54
RD
512{
513 if (napi_schedule_prep(napi)) {
514 __napi_schedule(napi);
4d29515f 515 return true;
bfe13f54 516 }
4d29515f 517 return false;
bfe13f54
RD
518}
519
bea3348e 520/**
3eb8eea2
JK
521 * napi_complete_done - NAPI processing complete
522 * @n: NAPI context
523 * @work_done: number of packets processed
bea3348e 524 *
3eb8eea2
JK
525 * Mark NAPI processing as complete. Should only be called if poll budget
526 * has not been completely consumed.
527 * Prefer over napi_complete().
364b6055 528 * Return false if device should avoid rearming interrupts.
bea3348e 529 */
3eb8eea2
JK
530bool napi_complete_done(struct napi_struct *n, int work_done);
531
364b6055 532static inline bool napi_complete(struct napi_struct *n)
3b47d303
ED
533{
534 return napi_complete_done(n, 0);
535}
bea3348e 536
5fdd2f0e
WW
537int dev_set_threaded(struct net_device *dev, bool threaded);
538
bea3348e
SH
539/**
540 * napi_disable - prevent NAPI from scheduling
5e82b4b2 541 * @n: NAPI context
bea3348e
SH
542 *
543 * Stop NAPI from being scheduled on this context.
544 * Waits till any outstanding processing completes.
545 */
3b47d303 546void napi_disable(struct napi_struct *n);
bea3348e 547
29863d41 548void napi_enable(struct napi_struct *n);
bea3348e 549
c264c3de
SH
550/**
551 * napi_synchronize - wait until NAPI is not running
5e82b4b2 552 * @n: NAPI context
c264c3de
SH
553 *
554 * Wait until NAPI is done being scheduled on this context.
555 * Waits till any outstanding processing completes but
556 * does not disable future activations.
557 */
558static inline void napi_synchronize(const struct napi_struct *n)
559{
facc432f
AB
560 if (IS_ENABLED(CONFIG_SMP))
561 while (test_bit(NAPI_STATE_SCHED, &n->state))
562 msleep(1);
563 else
564 barrier();
c264c3de 565}
c264c3de 566
6c5c9581
MK
567/**
568 * napi_if_scheduled_mark_missed - if napi is running, set the
569 * NAPIF_STATE_MISSED
570 * @n: NAPI context
571 *
572 * If napi is running, set the NAPIF_STATE_MISSED, and return true if
573 * NAPI is scheduled.
574 **/
575static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
576{
577 unsigned long val, new;
578
b9030780 579 val = READ_ONCE(n->state);
6c5c9581 580 do {
6c5c9581
MK
581 if (val & NAPIF_STATE_DISABLE)
582 return true;
583
584 if (!(val & NAPIF_STATE_SCHED))
585 return false;
586
587 new = val | NAPIF_STATE_MISSED;
b9030780 588 } while (!try_cmpxchg(&n->state, &val, new));
6c5c9581
MK
589
590 return true;
591}
592
d94d9fee 593enum netdev_queue_state_t {
73466498
TH
594 __QUEUE_STATE_DRV_XOFF,
595 __QUEUE_STATE_STACK_XOFF,
c3f26a26 596 __QUEUE_STATE_FROZEN,
79d16385 597};
8e2f1a63
DB
598
599#define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
600#define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
601#define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
602
603#define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
604#define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
605 QUEUE_STATE_FROZEN)
606#define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
607 QUEUE_STATE_FROZEN)
608
73466498
TH
609/*
610 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
611 * netif_tx_* functions below are used to manipulate this flag. The
612 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
613 * queue independently. The netif_xmit_*stopped functions below are called
614 * to check if the queue has been stopped by the driver or stack (either
615 * of the XOFF bits are set in the state). Drivers should not need to call
616 * netif_xmit*stopped functions, they should only be using netif_tx_*.
617 */
79d16385 618
bb949fbd 619struct netdev_queue {
6a321cb3 620/*
5e82b4b2 621 * read-mostly part
6a321cb3 622 */
bb949fbd 623 struct net_device *dev;
0b688f24
ED
624 netdevice_tracker dev_tracker;
625
46e5da40 626 struct Qdisc __rcu *qdisc;
d636fc5d 627 struct Qdisc __rcu *qdisc_sleeping;
ccf5ff69 628#ifdef CONFIG_SYSFS
1d24eb48
TH
629 struct kobject kobj;
630#endif
f2cd2d3e
ED
631#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
632 int numa_node;
633#endif
c0ef079c
FW
634 unsigned long tx_maxrate;
635 /*
636 * Number of TX timeouts for this queue
637 * (/sys/class/net/DEV/Q/trans_timeout)
638 */
8160fb43 639 atomic_long_t trans_timeout;
ffcfe25b
AD
640
641 /* Subordinate device that the queue has been assigned to */
642 struct net_device *sb_dev;
661b8d1b 643#ifdef CONFIG_XDP_SOCKETS
1742b3d5 644 struct xsk_buff_pool *pool;
661b8d1b 645#endif
6a321cb3 646/*
5e82b4b2 647 * write-mostly part
6a321cb3
ED
648 */
649 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
650 int xmit_lock_owner;
9d21493b 651 /*
9b36627a 652 * Time (in jiffies) of last Tx
9d21493b
ED
653 */
654 unsigned long trans_start;
ccf5ff69 655
114cf580
TH
656 unsigned long state;
657
658#ifdef CONFIG_BQL
659 struct dql dql;
660#endif
e8a0464c 661} ____cacheline_aligned_in_smp;
bb949fbd 662
79134e6c 663extern int sysctl_fb_tunnels_only_for_init_net;
856c395c 664extern int sysctl_devconf_inherit_init_net;
79134e6c 665
316cdaa1
MB
666/*
667 * sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
668 * == 1 : For initns only
669 * == 2 : For none.
670 */
79134e6c
ED
671static inline bool net_has_fallback_tunnels(const struct net *net)
672{
af67508e
KI
673#if IS_ENABLED(CONFIG_SYSCTL)
674 int fb_tunnels_only_for_init_net = READ_ONCE(sysctl_fb_tunnels_only_for_init_net);
675
676 return !fb_tunnels_only_for_init_net ||
677 (net_eq(net, &init_net) && fb_tunnels_only_for_init_net == 1);
678#else
679 return true;
680#endif
79134e6c
ED
681}
682
a5612ca1
KI
683static inline int net_inherit_devconf(void)
684{
685#if IS_ENABLED(CONFIG_SYSCTL)
686 return READ_ONCE(sysctl_devconf_inherit_init_net);
687#else
688 return 0;
689#endif
79134e6c
ED
690}
691
f2cd2d3e
ED
692static inline int netdev_queue_numa_node_read(const struct netdev_queue *q)
693{
694#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
695 return q->numa_node;
696#else
b236da69 697 return NUMA_NO_NODE;
f2cd2d3e
ED
698#endif
699}
700
701static inline void netdev_queue_numa_node_write(struct netdev_queue *q, int node)
702{
703#if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
704 q->numa_node = node;
705#endif
706}
707
df334545 708#ifdef CONFIG_RPS
0a9627f2
TH
709/*
710 * This structure holds an RPS map which can be of variable length. The
711 * map is an array of CPUs.
712 */
713struct rps_map {
714 unsigned int len;
715 struct rcu_head rcu;
bb4cf02d 716 u16 cpus[];
0a9627f2 717};
60b778ce 718#define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
0a9627f2 719
fec5e652 720/*
c445477d
BH
721 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
722 * tail pointer for that CPU's input queue at the time of last enqueue, and
723 * a hardware filter index.
fec5e652
TH
724 */
725struct rps_dev_flow {
726 u16 cpu;
c445477d 727 u16 filter;
fec5e652
TH
728 unsigned int last_qtail;
729};
c445477d 730#define RPS_NO_FILTER 0xffff
fec5e652
TH
731
732/*
733 * The rps_dev_flow_table structure contains a table of flow mappings.
734 */
735struct rps_dev_flow_table {
736 unsigned int mask;
737 struct rcu_head rcu;
bb4cf02d 738 struct rps_dev_flow flows[];
fec5e652
TH
739};
740#define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
60b778ce 741 ((_num) * sizeof(struct rps_dev_flow)))
fec5e652
TH
742
743/*
744 * The rps_sock_flow_table contains mappings of flows to the last CPU
745 * on which they were processed by the application (set in recvmsg).
5e82b4b2
BH
746 * Each entry is a 32bit value. Upper part is the high-order bits
747 * of flow hash, lower part is CPU number.
567e4b79 748 * rps_cpu_mask is used to partition the space, depending on number of
5e82b4b2
BH
749 * possible CPUs : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
750 * For example, if 64 CPUs are possible, rps_cpu_mask = 0x3f,
567e4b79 751 * meaning we use 32-6=26 bits for the hash.
fec5e652
TH
752 */
753struct rps_sock_flow_table {
567e4b79 754 u32 mask;
93c1af6c 755
bb4cf02d 756 u32 ents[] ____cacheline_aligned_in_smp;
fec5e652 757};
567e4b79 758#define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
fec5e652
TH
759
760#define RPS_NO_CPU 0xffff
761
567e4b79
ED
762extern u32 rps_cpu_mask;
763extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
764
fec5e652
TH
765static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
766 u32 hash)
767{
768 if (table && hash) {
567e4b79
ED
769 unsigned int index = hash & table->mask;
770 u32 val = hash & ~rps_cpu_mask;
fec5e652 771
5e82b4b2 772 /* We only give a hint, preemption can change CPU under us */
567e4b79 773 val |= raw_smp_processor_id();
fec5e652 774
5c3b74a9
ED
775 /* The following WRITE_ONCE() is paired with the READ_ONCE()
776 * here, and another one in get_rps_cpu().
777 */
778 if (READ_ONCE(table->ents[index]) != val)
779 WRITE_ONCE(table->ents[index], val);
fec5e652
TH
780 }
781}
782
c445477d 783#ifdef CONFIG_RFS_ACCEL
f629d208
JP
784bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
785 u16 filter_id);
c445477d 786#endif
a953be53 787#endif /* CONFIG_RPS */
c445477d 788
044ab86d
AT
789/* XPS map type and offset of the xps map within net_device->xps_maps[]. */
790enum xps_map_type {
791 XPS_CPUS = 0,
792 XPS_RXQS,
793 XPS_MAPS_MAX,
794};
795
bf264145
TH
796#ifdef CONFIG_XPS
797/*
798 * This structure holds an XPS map which can be of variable length. The
799 * map is an array of queues.
800 */
801struct xps_map {
802 unsigned int len;
803 unsigned int alloc_len;
804 struct rcu_head rcu;
bb4cf02d 805 u16 queues[];
bf264145 806};
60b778ce 807#define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
c59f419b
HD
808#define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
809 - sizeof(struct xps_map)) / sizeof(u16))
bf264145
TH
810
811/*
812 * This structure holds all XPS maps for device. Maps are indexed by CPU.
255c04a8 813 *
5478fcd0
AT
814 * We keep track of the number of cpus/rxqs used when the struct is allocated,
815 * in nr_ids. This will help not accessing out-of-bound memory.
816 *
255c04a8
AT
817 * We keep track of the number of traffic classes used when the struct is
818 * allocated, in num_tc. This will be used to navigate the maps, to ensure we're
819 * not crossing its upper bound, as the original dev->num_tc can be updated in
820 * the meantime.
bf264145
TH
821 */
822struct xps_dev_maps {
823 struct rcu_head rcu;
5478fcd0 824 unsigned int nr_ids;
255c04a8 825 s16 num_tc;
bb4cf02d 826 struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
bf264145 827};
80d19669
AN
828
829#define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
184c449f 830 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
80d19669
AN
831
832#define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
833 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
834
bf264145
TH
835#endif /* CONFIG_XPS */
836
4f57c087
JF
837#define TC_MAX_QUEUE 16
838#define TC_BITMASK 15
839/* HW offloaded queuing disciplines txq count and offset maps */
840struct netdev_tc_txq {
841 u16 count;
842 u16 offset;
843};
844
68bad94e
NP
845#if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
846/*
847 * This structure is to hold information about the device
848 * configured to run FCoE protocol stack.
849 */
850struct netdev_fcoe_hbainfo {
851 char manufacturer[64];
852 char serial_number[64];
853 char hardware_version[64];
854 char driver_version[64];
855 char optionrom_version[64];
856 char firmware_version[64];
857 char model[256];
858 char model_description[256];
859};
860#endif
861
02637fce 862#define MAX_PHYS_ITEM_ID_LEN 32
66b52b0d 863
02637fce
JP
864/* This structure holds a unique identifier to identify some
865 * physical item (port for example) used by a netdevice.
66b52b0d 866 */
02637fce
JP
867struct netdev_phys_item_id {
868 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
66b52b0d
JP
869 unsigned char id_len;
870};
871
d754f98b
SF
872static inline bool netdev_phys_item_id_same(struct netdev_phys_item_id *a,
873 struct netdev_phys_item_id *b)
874{
875 return a->id_len == b->id_len &&
876 memcmp(a->id, b->id, a->id_len) == 0;
877}
878
99932d4f 879typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
8ec56fc3
AD
880 struct sk_buff *skb,
881 struct net_device *sb_dev);
99932d4f 882
ddb94eaf
PNA
883enum net_device_path_type {
884 DEV_PATH_ETHERNET = 0,
e4417d69 885 DEV_PATH_VLAN,
ec9d16ba 886 DEV_PATH_BRIDGE,
f6efc675 887 DEV_PATH_PPPOE,
0994d492 888 DEV_PATH_DSA,
a333215e 889 DEV_PATH_MTK_WDMA,
ddb94eaf
PNA
890};
891
892struct net_device_path {
893 enum net_device_path_type type;
894 const struct net_device *dev;
e4417d69
PNA
895 union {
896 struct {
897 u16 id;
898 __be16 proto;
f6efc675 899 u8 h_dest[ETH_ALEN];
e4417d69 900 } encap;
bcf2766b
FF
901 struct {
902 enum {
903 DEV_PATH_BR_VLAN_KEEP,
904 DEV_PATH_BR_VLAN_TAG,
905 DEV_PATH_BR_VLAN_UNTAG,
26267bf9 906 DEV_PATH_BR_VLAN_UNTAG_HW,
bcf2766b
FF
907 } vlan_mode;
908 u16 vlan_id;
909 __be16 vlan_proto;
910 } bridge;
0994d492
FF
911 struct {
912 int port;
913 u16 proto;
914 } dsa;
a333215e
FF
915 struct {
916 u8 wdma_idx;
917 u8 queue;
918 u16 wcid;
919 u8 bss;
920 } mtk_wdma;
e4417d69 921 };
ddb94eaf
PNA
922};
923
924#define NET_DEVICE_PATH_STACK_MAX 5
bcf2766b 925#define NET_DEVICE_PATH_VLAN_MAX 2
ddb94eaf
PNA
926
927struct net_device_path_stack {
928 int num_paths;
929 struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
930};
931
932struct net_device_path_ctx {
933 const struct net_device *dev;
cf2df74e 934 u8 daddr[ETH_ALEN];
bcf2766b
FF
935
936 int num_vlans;
937 struct {
938 u16 id;
939 __be16 proto;
940 } vlan[NET_DEVICE_PATH_VLAN_MAX];
ddb94eaf
PNA
941};
942
2572ac53 943enum tc_setup_type {
aac4daa8 944 TC_QUERY_CAPS,
575ed7d3 945 TC_SETUP_QDISC_MQPRIO,
a1b7c5fd 946 TC_SETUP_CLSU32,
5b33f488 947 TC_SETUP_CLSFLOWER,
ade9b658 948 TC_SETUP_CLSMATCHALL,
332ae8e2 949 TC_SETUP_CLSBPF,
8c4083b3 950 TC_SETUP_BLOCK,
8521db4c 951 TC_SETUP_QDISC_CBS,
602f3baf 952 TC_SETUP_QDISC_RED,
7fdb61b4 953 TC_SETUP_QDISC_PRIO,
f971b132 954 TC_SETUP_QDISC_MQ,
25db26a9 955 TC_SETUP_QDISC_ETF,
98b0e5f6 956 TC_SETUP_ROOT_QDISC,
890d8d23 957 TC_SETUP_QDISC_GRED,
9c66d156 958 TC_SETUP_QDISC_TAPRIO,
c29f74e0 959 TC_SETUP_FT,
d35eb52b 960 TC_SETUP_QDISC_ETS,
ef6aadcc 961 TC_SETUP_QDISC_TBF,
aaca9408 962 TC_SETUP_QDISC_FIFO,
d03b195b 963 TC_SETUP_QDISC_HTB,
8cbfe939 964 TC_SETUP_ACT,
16e5cc64
JF
965};
966
f4e63525
JK
967/* These structures hold the attributes of bpf state that are being passed
968 * to the netdevice through the bpf op.
a7862b45 969 */
f4e63525 970enum bpf_netdev_command {
a7862b45
BB
971 /* Set or clear a bpf program used in the earliest stages of packet
972 * rx. The prog will have been loaded as BPF_PROG_TYPE_XDP. The callee
973 * is responsible for calling bpf_prog_put on any old progs that are
974 * stored. In case of error, the callee need not release the new prog
975 * reference, but on success it takes ownership and must bpf_prog_put
976 * when it is no longer used.
977 */
978 XDP_SETUP_PROG,
ee5d032f 979 XDP_SETUP_PROG_HW,
ab3f0063 980 /* BPF program for offload callbacks, invoked at program load time. */
a3884572
JK
981 BPF_OFFLOAD_MAP_ALLOC,
982 BPF_OFFLOAD_MAP_FREE,
1742b3d5 983 XDP_SETUP_XSK_POOL,
a7862b45
BB
984};
985
cae1927c 986struct bpf_prog_offload_ops;
ddf9f970 987struct netlink_ext_ack;
74515c57 988struct xdp_umem;
75ccae62 989struct xdp_dev_bulk_queue;
aa8d3a71 990struct bpf_xdp_link;
ddf9f970 991
7f0a8382
AN
992enum bpf_xdp_mode {
993 XDP_MODE_SKB = 0,
994 XDP_MODE_DRV = 1,
995 XDP_MODE_HW = 2,
996 __MAX_XDP_MODE
997};
998
999struct bpf_xdp_entity {
1000 struct bpf_prog *prog;
aa8d3a71 1001 struct bpf_xdp_link *link;
7f0a8382 1002};
ddf9f970 1003
f4e63525
JK
1004struct netdev_bpf {
1005 enum bpf_netdev_command command;
a7862b45
BB
1006 union {
1007 /* XDP_SETUP_PROG */
ddf9f970 1008 struct {
32d60277 1009 u32 flags;
ddf9f970
JK
1010 struct bpf_prog *prog;
1011 struct netlink_ext_ack *extack;
1012 };
a3884572
JK
1013 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
1014 struct {
1015 struct bpf_offloaded_map *offmap;
1016 };
1742b3d5 1017 /* XDP_SETUP_XSK_POOL */
74515c57 1018 struct {
1742b3d5 1019 struct xsk_buff_pool *pool;
f8ebfaf6 1020 u16 queue_id;
74515c57 1021 } xsk;
a7862b45
BB
1022 };
1023};
16e5cc64 1024
9116e5e2
MK
1025/* Flags for ndo_xsk_wakeup. */
1026#define XDP_WAKEUP_RX (1 << 0)
1027#define XDP_WAKEUP_TX (1 << 1)
1028
d77e38e6
SK
1029#ifdef CONFIG_XFRM_OFFLOAD
1030struct xfrmdev_ops {
7681a4f5 1031 int (*xdo_dev_state_add) (struct xfrm_state *x, struct netlink_ext_ack *extack);
d77e38e6
SK
1032 void (*xdo_dev_state_delete) (struct xfrm_state *x);
1033 void (*xdo_dev_state_free) (struct xfrm_state *x);
1034 bool (*xdo_dev_offload_ok) (struct sk_buff *skb,
1035 struct xfrm_state *x);
50bd870a 1036 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
f3da86dc 1037 void (*xdo_dev_state_update_curlft) (struct xfrm_state *x);
3089386d 1038 int (*xdo_dev_policy_add) (struct xfrm_policy *x, struct netlink_ext_ack *extack);
919e43fa
LR
1039 void (*xdo_dev_policy_delete) (struct xfrm_policy *x);
1040 void (*xdo_dev_policy_free) (struct xfrm_policy *x);
d77e38e6
SK
1041};
1042#endif
1043
6c557001
FW
1044struct dev_ifalias {
1045 struct rcu_head rcuhead;
1046 char ifalias[];
1047};
1048
b473b0d2 1049struct devlink;
da68b4ad 1050struct tlsdev_ops;
b473b0d2 1051
93642e14
JP
1052struct netdev_net_notifier {
1053 struct list_head list;
1054 struct notifier_block *nb;
1055};
1056
d314774c
SH
1057/*
1058 * This structure defines the management hooks for network devices.
00829823
SH
1059 * The following hooks can be defined; unless noted otherwise, they are
1060 * optional and can be filled with a null pointer.
d314774c
SH
1061 *
1062 * int (*ndo_init)(struct net_device *dev);
5e82b4b2
BH
1063 * This function is called once when a network device is registered.
1064 * The network device can use this for any late stage initialization
1065 * or semantic validation. It can fail with an error code which will
1066 * be propagated back to register_netdev.
d314774c
SH
1067 *
1068 * void (*ndo_uninit)(struct net_device *dev);
1069 * This function is called when device is unregistered or when registration
1070 * fails. It is not called if init fails.
1071 *
1072 * int (*ndo_open)(struct net_device *dev);
5e82b4b2 1073 * This function is called when a network device transitions to the up
d314774c
SH
1074 * state.
1075 *
1076 * int (*ndo_stop)(struct net_device *dev);
5e82b4b2 1077 * This function is called when a network device transitions to the down
d314774c
SH
1078 * state.
1079 *
dc1f8bf6
SH
1080 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1081 * struct net_device *dev);
00829823 1082 * Called when a packet needs to be transmitted.
e79d8429
RR
1083 * Returns NETDEV_TX_OK. Can return NETDEV_TX_BUSY, but you should stop
1084 * the queue before that can happen; it's for obsolete devices and weird
1085 * corner cases, but the stack really does a non-trivial amount
1086 * of useless work if you return NETDEV_TX_BUSY.
5e82b4b2 1087 * Required; cannot be NULL.
00829823 1088 *
1a2a1444
DM
1089 * netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1090 * struct net_device *dev
1091 * netdev_features_t features);
1092 * Called by core transmit path to determine if device is capable of
1093 * performing offload operations on a given packet. This is to give
1094 * the device an opportunity to implement any restrictions that cannot
1095 * be otherwise expressed by feature flags. The check is called with
1096 * the set of features that the stack has calculated and it returns
1097 * those the driver believes to be appropriate.
cdba756f 1098 *
f663dd9a 1099 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
a350ecce 1100 * struct net_device *sb_dev);
5e82b4b2 1101 * Called to decide which queue to use when device supports multiple
00829823
SH
1102 * transmit queues.
1103 *
d314774c
SH
1104 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1105 * This function is called to allow device receiver to make
5e82b4b2 1106 * changes to configuration when multicast or promiscuous is enabled.
d314774c
SH
1107 *
1108 * void (*ndo_set_rx_mode)(struct net_device *dev);
1109 * This function is called device changes address list filtering.
01789349 1110 * If driver handles unicast address filtering, it should set
5e82b4b2 1111 * IFF_UNICAST_FLT in its priv_flags.
d314774c
SH
1112 *
1113 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1114 * This function is called when the Media Access Control address
37b607c5 1115 * needs to be changed. If this interface is not defined, the
5e82b4b2 1116 * MAC address can not be changed.
d314774c
SH
1117 *
1118 * int (*ndo_validate_addr)(struct net_device *dev);
1119 * Test if Media Access Control address is valid for the device.
1120 *
1121 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
3d9d00bd
AB
1122 * Old-style ioctl entry point. This is used internally by the
1123 * appletalk and ieee802154 subsystems but is no longer called by
1124 * the device ioctl handler.
1125 *
1126 * int (*ndo_siocbond)(struct net_device *dev, struct ifreq *ifr, int cmd);
1127 * Used by the bonding driver for its device specific ioctls:
1128 * SIOCBONDENSLAVE, SIOCBONDRELEASE, SIOCBONDSETHWADDR, SIOCBONDCHANGEACTIVE,
1129 * SIOCBONDSLAVEINFOQUERY, and SIOCBONDINFOQUERY
d314774c 1130 *
a7605370
AB
1131 * * int (*ndo_eth_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
1132 * Called for ethernet specific ioctls: SIOCGMIIPHY, SIOCGMIIREG,
1133 * SIOCSMIIREG, SIOCSHWTSTAMP and SIOCGHWTSTAMP.
1134 *
d314774c
SH
1135 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1136 * Used to set network devices bus interface parameters. This interface
5e82b4b2 1137 * is retained for legacy reasons; new devices should use the bus
d314774c
SH
1138 * interface (PCI) for low level management.
1139 *
1140 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
1141 * Called when a user wants to change the Maximum Transfer Unit
db46a0e1 1142 * of a device.
d314774c 1143 *
0290bd29 1144 * void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
5e82b4b2 1145 * Callback used when the transmitter has not made any progress
d314774c
SH
1146 * for dev->watchdog ticks.
1147 *
bc1f4470 1148 * void (*ndo_get_stats64)(struct net_device *dev,
1149 * struct rtnl_link_stats64 *storage);
d308e38f 1150 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
d314774c 1151 * Called when a user wants to get the network device usage
be1f3c2c 1152 * statistics. Drivers must do one of the following:
3cfde79c
BH
1153 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1154 * rtnl_link_stats64 structure passed by the caller.
82695d9b 1155 * 2. Define @ndo_get_stats to update a net_device_stats structure
be1f3c2c
BH
1156 * (which should normally be dev->stats) and return a pointer to
1157 * it. The structure may be changed asynchronously only if each
1158 * field is written atomically.
1159 * 3. Update dev->stats asynchronously and atomically, and define
1160 * neither operation.
d314774c 1161 *
3df5b3c6 1162 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
2c9d85d4
NF
1163 * Return true if this device supports offload stats of this attr_id.
1164 *
1165 * int (*ndo_get_offload_stats)(int attr_id, const struct net_device *dev,
1166 * void *attr_data)
1167 * Get statistics for offload operations by attr_id. Write it into the
1168 * attr_data pointer.
1169 *
5d632cb7 1170 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1171 * If device supports VLAN filtering this function is called when a
80d5c368 1172 * VLAN id is registered.
d314774c 1173 *
5d632cb7 1174 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
5e82b4b2 1175 * If device supports VLAN filtering this function is called when a
80d5c368 1176 * VLAN id is unregistered.
d314774c
SH
1177 *
1178 * void (*ndo_poll_controller)(struct net_device *dev);
95c26df8
WM
1179 *
1180 * SR-IOV management functions.
1181 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
79aab093
MS
1182 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1183 * u8 qos, __be16 proto);
ed616689
SC
1184 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1185 * int max_tx_rate);
5f8444a3 1186 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
dd461d6a 1187 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
95c26df8
WM
1188 * int (*ndo_get_vf_config)(struct net_device *dev,
1189 * int vf, struct ifla_vf_info *ivf);
1d8faf48 1190 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
57b61080
SF
1191 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1192 * struct nlattr *port[]);
01a3d796
VZ
1193 *
1194 * Enable or disable the VF ability to query its RSS Redirection Table and
1195 * Hash Key. This is needed since on some devices VF share this information
5e82b4b2 1196 * with PF and querying it may introduce a theoretical security risk.
01a3d796 1197 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
57b61080 1198 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
2572ac53 1199 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
de4784ca 1200 * void *type_data);
6a4bc2b4
FF
1201 * Called to setup any 'tc' scheduler, classifier or action on @dev.
1202 * This is always called from the stack with the rtnl lock held and netif
1203 * tx queues stopped. This allows the netdevice to perform queue
1204 * management safely.
c445477d 1205 *
e9bce845
YZ
1206 * Fiber Channel over Ethernet (FCoE) offload functions.
1207 * int (*ndo_fcoe_enable)(struct net_device *dev);
1208 * Called when the FCoE protocol stack wants to start using LLD for FCoE
1209 * so the underlying device can perform whatever needed configuration or
1210 * initialization to support acceleration of FCoE traffic.
1211 *
1212 * int (*ndo_fcoe_disable)(struct net_device *dev);
1213 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
1214 * so the underlying device can perform whatever needed clean-ups to
1215 * stop supporting acceleration of FCoE traffic.
1216 *
1217 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
1218 * struct scatterlist *sgl, unsigned int sgc);
1219 * Called when the FCoE Initiator wants to initialize an I/O that
1220 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1221 * perform necessary setup and returns 1 to indicate the device is set up
1222 * successfully to perform DDP on this I/O, otherwise this returns 0.
1223 *
1224 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
1225 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
1226 * indicated by the FC exchange id 'xid', so the underlying device can
1227 * clean up and reuse resources for later DDP requests.
1228 *
1229 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
1230 * struct scatterlist *sgl, unsigned int sgc);
1231 * Called when the FCoE Target wants to initialize an I/O that
1232 * is a possible candidate for Direct Data Placement (DDP). The LLD can
1233 * perform necessary setup and returns 1 to indicate the device is set up
1234 * successfully to perform DDP on this I/O, otherwise this returns 0.
1235 *
68bad94e
NP
1236 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1237 * struct netdev_fcoe_hbainfo *hbainfo);
1238 * Called when the FCoE Protocol stack wants information on the underlying
1239 * device. This information is utilized by the FCoE protocol stack to
1240 * register attributes with Fiber Channel management service as per the
1241 * FC-GS Fabric Device Management Information(FDMI) specification.
1242 *
e9bce845
YZ
1243 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
1244 * Called when the underlying device wants to override default World Wide
1245 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
1246 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
1247 * protocol stack to use.
1248 *
c445477d
BH
1249 * RFS acceleration.
1250 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
1251 * u16 rxq_index, u32 flow_id);
1252 * Set hardware filter for RFS. rxq_index is the target queue index;
1253 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
1254 * Return the filter ID on success, or a negative error code.
fbaec0ea 1255 *
8b98a70c 1256 * Slave management functions (for bridge, bonding, etc).
fbaec0ea
JP
1257 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
1258 * Called to make another netdev an underling.
1259 *
1260 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
1261 * Called to release previously enslaved netdev.
5455c699 1262 *
cff9f12b
MG
1263 * struct net_device *(*ndo_get_xmit_slave)(struct net_device *dev,
1264 * struct sk_buff *skb,
1265 * bool all_slaves);
1266 * Get the xmit slave of master device. If all_slaves is true, function
1267 * assume all the slaves can transmit.
1268 *
5455c699 1269 * Feature/offload setting functions.
1a2a1444
DM
1270 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1271 * netdev_features_t features);
1272 * Adjusts the requested feature flags according to device-specific
1273 * constraints, and returns the resulting flags. Must not modify
1274 * the device state.
1275 *
c8f44aff 1276 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
5455c699
MM
1277 * Called to update device configuration to new features. Passed
1278 * feature set might be less than what was returned by ndo_fix_features()).
1279 * Must return >0 or -errno if it changed dev->features itself.
1280 *
edc7d573 1281 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1282 * struct net_device *dev,
87b0984e
PM
1283 * const unsigned char *addr, u16 vid, u16 flags,
1284 * struct netlink_ext_ack *extack);
77162022 1285 * Adds an FDB entry to dev for addr.
1690be63
VY
1286 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1287 * struct net_device *dev,
f6f6424b 1288 * const unsigned char *addr, u16 vid)
77162022 1289 * Deletes the FDB entry from dev coresponding to addr.
1306d536
NA
1290 * int (*ndo_fdb_del_bulk)(struct ndmsg *ndm, struct nlattr *tb[],
1291 * struct net_device *dev,
1292 * u16 vid,
1293 * struct netlink_ext_ack *extack);
77162022 1294 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
5d5eacb3 1295 * struct net_device *dev, struct net_device *filter_dev,
d297653d 1296 * int *idx)
77162022
JF
1297 * Used to add FDB entries to dump requests. Implementers should add
1298 * entries to skb and update idx with the number of entries.
e5a55a89 1299 *
8c44fa12
IS
1300 * int (*ndo_mdb_add)(struct net_device *dev, struct nlattr *tb[],
1301 * u16 nlmsg_flags, struct netlink_ext_ack *extack);
1302 * Adds an MDB entry to dev.
1303 * int (*ndo_mdb_del)(struct net_device *dev, struct nlattr *tb[],
1304 * struct netlink_ext_ack *extack);
1305 * Deletes the MDB entry from dev.
1306 * int (*ndo_mdb_dump)(struct net_device *dev, struct sk_buff *skb,
1307 * struct netlink_callback *cb);
1308 * Dumps MDB entries from dev. The first argument (marker) in the netlink
1309 * callback is used by core rtnetlink code.
1310 *
ad41faa8 1311 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
2fd527b7 1312 * u16 flags, struct netlink_ext_ack *extack)
e5a55a89 1313 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
46c264da
ND
1314 * struct net_device *dev, u32 filter_mask,
1315 * int nlflags)
ad41faa8
ND
1316 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1317 * u16 flags);
4bf84c35
JP
1318 *
1319 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
1320 * Called to change device carrier. Soft-devices (like dummy, team, etc)
1321 * which do not represent real hardware may define this to allow their
1322 * userspace components to manage their virtual carrier state. Devices
1323 * that determine carrier state from physical hardware properties (eg
1324 * network cables) or protocol-dependent mechanisms (eg
1325 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
66b52b0d
JP
1326 *
1327 * int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1328 * struct netdev_phys_item_id *ppid);
66b52b0d
JP
1329 * Called to get ID of physical port of this device. If driver does
1330 * not implement this, it is assumed that the hw is not able to have
1331 * multiple net devices on single physical port.
53cf5275 1332 *
d6abc596
FF
1333 * int (*ndo_get_port_parent_id)(struct net_device *dev,
1334 * struct netdev_phys_item_id *ppid)
1335 * Called to get the parent ID of the physical port of this device.
1336 *
a6cc0cfa
JF
1337 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1338 * struct net_device *dev)
1339 * Called by upper layer devices to accelerate switching or other
1340 * station functionality into hardware. 'pdev is the lowerdev
1341 * to use for the offload and 'dev' is the net device that will
1342 * back the offload. Returns a pointer to the private structure
1343 * the upper layer will maintain.
1344 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1345 * Called by upper layer device to delete the station created
1346 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1347 * the station and priv is the structure returned by the add
1348 * operation.
822b3b2e
JF
1349 * int (*ndo_set_tx_maxrate)(struct net_device *dev,
1350 * int queue_index, u32 maxrate);
1351 * Called when a user wants to set a max-rate limitation of specific
1352 * TX queue.
a54acb3a
ND
1353 * int (*ndo_get_iflink)(const struct net_device *dev);
1354 * Called to get the iflink value of this device.
fc4099f1
PS
1355 * int (*ndo_fill_metadata_dst)(struct net_device *dev, struct sk_buff *skb);
1356 * This function is used to get egress tunnel information for given skb.
1357 * This is useful for retrieving outer tunnel header parameters while
1358 * sampling packet.
871b642a
PA
1359 * void (*ndo_set_rx_headroom)(struct net_device *dev, int needed_headroom);
1360 * This function is used to specify the headroom that the skb must
1361 * consider when allocation skb during packet reception. Setting
1362 * appropriate rx headroom value allows avoiding skb head copy on
5e82b4b2 1363 * forward. Setting a negative value resets the rx headroom to the
871b642a 1364 * default value.
f4e63525 1365 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
a7862b45 1366 * This function is used to set or query state related to XDP on the
f4e63525
JK
1367 * netdevice and manage BPF offload. See definition of
1368 * enum bpf_netdev_command for details.
42b33468
JDB
1369 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1370 * u32 flags);
735fc405
JDB
1371 * This function is used to submit @n XDP packets for transmit on a
1372 * netdevice. Returns number of frames successfully transmitted, frames
1373 * that got dropped are freed/returned via xdp_return_frame().
1374 * Returns negative number, means general error invoking ndo, meaning
1375 * no frames were xmit'ed and core-caller will free all frames.
879af96f
JM
1376 * struct net_device *(*ndo_xdp_get_xmit_slave)(struct net_device *dev,
1377 * struct xdp_buff *xdp);
1378 * Get the xmit slave of master device based on the xdp_buff.
9116e5e2
MK
1379 * int (*ndo_xsk_wakeup)(struct net_device *dev, u32 queue_id, u32 flags);
1380 * This function is used to wake up the softirq, ksoftirqd or kthread
1381 * responsible for sending and/or receiving packets on a specific
1382 * queue id bound to an AF_XDP socket. The flags field specifies if
1383 * only RX, only Tx, or both should be woken up using the flags
1384 * XDP_WAKEUP_RX and XDP_WAKEUP_TX.
607259a6
CH
1385 * int (*ndo_tunnel_ctl)(struct net_device *dev, struct ip_tunnel_parm *p,
1386 * int cmd);
1387 * Add, change, delete or get information on an IPv4 tunnel.
9aa1206e
DB
1388 * struct net_device *(*ndo_get_peer_dev)(struct net_device *dev);
1389 * If a device is paired with a peer device, return the peer instance.
1390 * The caller must be under RCU read context.
ddb94eaf
PNA
1391 * int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx, struct net_device_path *path);
1392 * Get the forwarding path to reach the real device from the HW destination address
97dc7cd9
GE
1393 * ktime_t (*ndo_get_tstamp)(struct net_device *dev,
1394 * const struct skb_shared_hwtstamps *hwtstamps,
1395 * bool cycles);
1396 * Get hardware timestamp based on normal/adjustable time or free running
1397 * cycle counter. This function is required if physical clock supports a
1398 * free running cycle counter.
66f72230
MG
1399 *
1400 * int (*ndo_hwtstamp_get)(struct net_device *dev,
1401 * struct kernel_hwtstamp_config *kernel_config);
1402 * Get the currently configured hardware timestamping parameters for the
1403 * NIC device.
1404 *
1405 * int (*ndo_hwtstamp_set)(struct net_device *dev,
1406 * struct kernel_hwtstamp_config *kernel_config,
1407 * struct netlink_ext_ack *extack);
1408 * Change the hardware timestamping parameters for NIC device.
d314774c
SH
1409 */
1410struct net_device_ops {
1411 int (*ndo_init)(struct net_device *dev);
1412 void (*ndo_uninit)(struct net_device *dev);
1413 int (*ndo_open)(struct net_device *dev);
1414 int (*ndo_stop)(struct net_device *dev);
cdba756f
ED
1415 netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1416 struct net_device *dev);
1417 netdev_features_t (*ndo_features_check)(struct sk_buff *skb,
1418 struct net_device *dev,
1419 netdev_features_t features);
00829823 1420 u16 (*ndo_select_queue)(struct net_device *dev,
f663dd9a 1421 struct sk_buff *skb,
a350ecce 1422 struct net_device *sb_dev);
d314774c
SH
1423 void (*ndo_change_rx_flags)(struct net_device *dev,
1424 int flags);
d314774c 1425 void (*ndo_set_rx_mode)(struct net_device *dev);
d314774c
SH
1426 int (*ndo_set_mac_address)(struct net_device *dev,
1427 void *addr);
d314774c 1428 int (*ndo_validate_addr)(struct net_device *dev);
d314774c
SH
1429 int (*ndo_do_ioctl)(struct net_device *dev,
1430 struct ifreq *ifr, int cmd);
a7605370
AB
1431 int (*ndo_eth_ioctl)(struct net_device *dev,
1432 struct ifreq *ifr, int cmd);
3d9d00bd
AB
1433 int (*ndo_siocbond)(struct net_device *dev,
1434 struct ifreq *ifr, int cmd);
ad7eab2a
AB
1435 int (*ndo_siocwandev)(struct net_device *dev,
1436 struct if_settings *ifs);
b9067f5d
AB
1437 int (*ndo_siocdevprivate)(struct net_device *dev,
1438 struct ifreq *ifr,
1439 void __user *data, int cmd);
d314774c
SH
1440 int (*ndo_set_config)(struct net_device *dev,
1441 struct ifmap *map);
00829823
SH
1442 int (*ndo_change_mtu)(struct net_device *dev,
1443 int new_mtu);
1444 int (*ndo_neigh_setup)(struct net_device *dev,
1445 struct neigh_parms *);
0290bd29
MT
1446 void (*ndo_tx_timeout) (struct net_device *dev,
1447 unsigned int txqueue);
d314774c 1448
bc1f4470 1449 void (*ndo_get_stats64)(struct net_device *dev,
1450 struct rtnl_link_stats64 *storage);
3df5b3c6 1451 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
2c9d85d4
NF
1452 int (*ndo_get_offload_stats)(int attr_id,
1453 const struct net_device *dev,
1454 void *attr_data);
d314774c
SH
1455 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1456
8e586137 1457 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
80d5c368 1458 __be16 proto, u16 vid);
8e586137 1459 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
80d5c368 1460 __be16 proto, u16 vid);
d314774c 1461#ifdef CONFIG_NET_POLL_CONTROLLER
d314774c 1462 void (*ndo_poll_controller)(struct net_device *dev);
4247e161 1463 int (*ndo_netpoll_setup)(struct net_device *dev,
a8779ec1 1464 struct netpoll_info *info);
0e34e931 1465 void (*ndo_netpoll_cleanup)(struct net_device *dev);
d314774c 1466#endif
95c26df8
WM
1467 int (*ndo_set_vf_mac)(struct net_device *dev,
1468 int queue, u8 *mac);
1469 int (*ndo_set_vf_vlan)(struct net_device *dev,
79aab093
MS
1470 int queue, u16 vlan,
1471 u8 qos, __be16 proto);
ed616689
SC
1472 int (*ndo_set_vf_rate)(struct net_device *dev,
1473 int vf, int min_tx_rate,
1474 int max_tx_rate);
5f8444a3
GR
1475 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1476 int vf, bool setting);
dd461d6a
HS
1477 int (*ndo_set_vf_trust)(struct net_device *dev,
1478 int vf, bool setting);
95c26df8
WM
1479 int (*ndo_get_vf_config)(struct net_device *dev,
1480 int vf,
1481 struct ifla_vf_info *ivf);
1d8faf48
RE
1482 int (*ndo_set_vf_link_state)(struct net_device *dev,
1483 int vf, int link_state);
3b766cd8
EBE
1484 int (*ndo_get_vf_stats)(struct net_device *dev,
1485 int vf,
1486 struct ifla_vf_stats
1487 *vf_stats);
57b61080
SF
1488 int (*ndo_set_vf_port)(struct net_device *dev,
1489 int vf,
1490 struct nlattr *port[]);
1491 int (*ndo_get_vf_port)(struct net_device *dev,
1492 int vf, struct sk_buff *skb);
30aad417
DG
1493 int (*ndo_get_vf_guid)(struct net_device *dev,
1494 int vf,
1495 struct ifla_vf_guid *node_guid,
1496 struct ifla_vf_guid *port_guid);
cc8e27cc
EC
1497 int (*ndo_set_vf_guid)(struct net_device *dev,
1498 int vf, u64 guid,
1499 int guid_type);
01a3d796
VZ
1500 int (*ndo_set_vf_rss_query_en)(
1501 struct net_device *dev,
1502 int vf, bool setting);
16e5cc64 1503 int (*ndo_setup_tc)(struct net_device *dev,
2572ac53 1504 enum tc_setup_type type,
de4784ca 1505 void *type_data);
d11ead75 1506#if IS_ENABLED(CONFIG_FCOE)
cb454399
YZ
1507 int (*ndo_fcoe_enable)(struct net_device *dev);
1508 int (*ndo_fcoe_disable)(struct net_device *dev);
4d288d57
YZ
1509 int (*ndo_fcoe_ddp_setup)(struct net_device *dev,
1510 u16 xid,
1511 struct scatterlist *sgl,
1512 unsigned int sgc);
1513 int (*ndo_fcoe_ddp_done)(struct net_device *dev,
1514 u16 xid);
6247e086
YZ
1515 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1516 u16 xid,
1517 struct scatterlist *sgl,
1518 unsigned int sgc);
68bad94e
NP
1519 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1520 struct netdev_fcoe_hbainfo *hbainfo);
3c9c36bc
BPG
1521#endif
1522
d11ead75 1523#if IS_ENABLED(CONFIG_LIBFCOE)
df5c7945
YZ
1524#define NETDEV_FCOE_WWNN 0
1525#define NETDEV_FCOE_WWPN 1
1526 int (*ndo_fcoe_get_wwn)(struct net_device *dev,
1527 u64 *wwn, int type);
4d288d57 1528#endif
3c9c36bc 1529
c445477d
BH
1530#ifdef CONFIG_RFS_ACCEL
1531 int (*ndo_rx_flow_steer)(struct net_device *dev,
1532 const struct sk_buff *skb,
1533 u16 rxq_index,
1534 u32 flow_id);
1535#endif
fbaec0ea 1536 int (*ndo_add_slave)(struct net_device *dev,
33eaf2a6
DA
1537 struct net_device *slave_dev,
1538 struct netlink_ext_ack *extack);
fbaec0ea
JP
1539 int (*ndo_del_slave)(struct net_device *dev,
1540 struct net_device *slave_dev);
cff9f12b
MG
1541 struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
1542 struct sk_buff *skb,
1543 bool all_slaves);
719a402c
TT
1544 struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
1545 struct sock *sk);
c8f44aff
MM
1546 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1547 netdev_features_t features);
5455c699 1548 int (*ndo_set_features)(struct net_device *dev,
c8f44aff 1549 netdev_features_t features);
503eebc2
JP
1550 int (*ndo_neigh_construct)(struct net_device *dev,
1551 struct neighbour *n);
1552 void (*ndo_neigh_destroy)(struct net_device *dev,
1553 struct neighbour *n);
77162022
JF
1554
1555 int (*ndo_fdb_add)(struct ndmsg *ndm,
edc7d573 1556 struct nlattr *tb[],
77162022 1557 struct net_device *dev,
6b6e2725 1558 const unsigned char *addr,
f6f6424b 1559 u16 vid,
87b0984e
PM
1560 u16 flags,
1561 struct netlink_ext_ack *extack);
77162022 1562 int (*ndo_fdb_del)(struct ndmsg *ndm,
1690be63 1563 struct nlattr *tb[],
77162022 1564 struct net_device *dev,
f6f6424b 1565 const unsigned char *addr,
ca4567f1 1566 u16 vid, struct netlink_ext_ack *extack);
1306d536
NA
1567 int (*ndo_fdb_del_bulk)(struct ndmsg *ndm,
1568 struct nlattr *tb[],
1569 struct net_device *dev,
1570 u16 vid,
1571 struct netlink_ext_ack *extack);
77162022
JF
1572 int (*ndo_fdb_dump)(struct sk_buff *skb,
1573 struct netlink_callback *cb,
1574 struct net_device *dev,
5d5eacb3 1575 struct net_device *filter_dev,
d297653d 1576 int *idx);
5b2f94b2
RP
1577 int (*ndo_fdb_get)(struct sk_buff *skb,
1578 struct nlattr *tb[],
1579 struct net_device *dev,
1580 const unsigned char *addr,
1581 u16 vid, u32 portid, u32 seq,
1582 struct netlink_ext_ack *extack);
8c44fa12
IS
1583 int (*ndo_mdb_add)(struct net_device *dev,
1584 struct nlattr *tb[],
1585 u16 nlmsg_flags,
1586 struct netlink_ext_ack *extack);
1587 int (*ndo_mdb_del)(struct net_device *dev,
1588 struct nlattr *tb[],
1589 struct netlink_ext_ack *extack);
1590 int (*ndo_mdb_dump)(struct net_device *dev,
1591 struct sk_buff *skb,
1592 struct netlink_callback *cb);
e5a55a89 1593 int (*ndo_bridge_setlink)(struct net_device *dev,
add511b3 1594 struct nlmsghdr *nlh,
2fd527b7
PM
1595 u16 flags,
1596 struct netlink_ext_ack *extack);
e5a55a89
JF
1597 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1598 u32 pid, u32 seq,
6cbdceeb 1599 struct net_device *dev,
46c264da
ND
1600 u32 filter_mask,
1601 int nlflags);
407af329 1602 int (*ndo_bridge_dellink)(struct net_device *dev,
add511b3
RP
1603 struct nlmsghdr *nlh,
1604 u16 flags);
4bf84c35
JP
1605 int (*ndo_change_carrier)(struct net_device *dev,
1606 bool new_carrier);
66b52b0d 1607 int (*ndo_get_phys_port_id)(struct net_device *dev,
02637fce 1608 struct netdev_phys_item_id *ppid);
d6abc596
FF
1609 int (*ndo_get_port_parent_id)(struct net_device *dev,
1610 struct netdev_phys_item_id *ppid);
db24a904
DA
1611 int (*ndo_get_phys_port_name)(struct net_device *dev,
1612 char *name, size_t len);
a6cc0cfa
JF
1613 void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1614 struct net_device *dev);
1615 void (*ndo_dfwd_del_station)(struct net_device *pdev,
1616 void *priv);
1617
822b3b2e
JF
1618 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1619 int queue_index,
1620 u32 maxrate);
a54acb3a 1621 int (*ndo_get_iflink)(const struct net_device *dev);
fc4099f1
PS
1622 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1623 struct sk_buff *skb);
871b642a
PA
1624 void (*ndo_set_rx_headroom)(struct net_device *dev,
1625 int needed_headroom);
f4e63525
JK
1626 int (*ndo_bpf)(struct net_device *dev,
1627 struct netdev_bpf *bpf);
735fc405 1628 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
42b33468
JDB
1629 struct xdp_frame **xdp,
1630 u32 flags);
879af96f
JM
1631 struct net_device * (*ndo_xdp_get_xmit_slave)(struct net_device *dev,
1632 struct xdp_buff *xdp);
9116e5e2
MK
1633 int (*ndo_xsk_wakeup)(struct net_device *dev,
1634 u32 queue_id, u32 flags);
607259a6
CH
1635 int (*ndo_tunnel_ctl)(struct net_device *dev,
1636 struct ip_tunnel_parm *p, int cmd);
9aa1206e 1637 struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
ddb94eaf
PNA
1638 int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
1639 struct net_device_path *path);
97dc7cd9
GE
1640 ktime_t (*ndo_get_tstamp)(struct net_device *dev,
1641 const struct skb_shared_hwtstamps *hwtstamps,
1642 bool cycles);
66f72230
MG
1643 int (*ndo_hwtstamp_get)(struct net_device *dev,
1644 struct kernel_hwtstamp_config *kernel_config);
1645 int (*ndo_hwtstamp_set)(struct net_device *dev,
1646 struct kernel_hwtstamp_config *kernel_config,
1647 struct netlink_ext_ack *extack);
d314774c
SH
1648};
1649
7aa98047 1650/**
270f3385 1651 * enum netdev_priv_flags - &struct net_device priv_flags
7aa98047
LR
1652 *
1653 * These are the &struct net_device, they are only set internally
1654 * by drivers and used in the kernel. These flags are invisible to
5e82b4b2 1655 * userspace; this means that the order of these flags can change
7aa98047
LR
1656 * during any kernel release.
1657 *
1658 * You should have a pretty good reason to be extending these flags.
1659 *
1660 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1661 * @IFF_EBRIDGE: Ethernet bridging device
7aa98047 1662 * @IFF_BONDING: bonding master or slave
7aa98047 1663 * @IFF_ISATAP: ISATAP interface (RFC4214)
7aa98047
LR
1664 * @IFF_WAN_HDLC: WAN HDLC device
1665 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1666 * release skb->dst
1667 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1668 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1669 * @IFF_MACVLAN_PORT: device used as macvlan port
1670 * @IFF_BRIDGE_PORT: device used as bridge port
1671 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1672 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1673 * @IFF_UNICAST_FLT: Supports unicast filtering
1674 * @IFF_TEAM_PORT: device used as team port
1675 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1676 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1677 * change when it's running
1678 * @IFF_MACVLAN: Macvlan device
6d0e24cd
LB
1679 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1680 * underlying stacked devices
007979ea 1681 * @IFF_L3MDEV_MASTER: device is an L3 master device
fa8187c9 1682 * @IFF_NO_QUEUE: device can run without qdisc attached
35d4e172 1683 * @IFF_OPENVSWITCH: device is a Open vSwitch master
fee6d4c7 1684 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
c981e421 1685 * @IFF_TEAM: device is a team device
d4ab4286 1686 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
871b642a
PA
1687 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1688 * entity (i.e. the master device for bridged veth)
3c175784 1689 * @IFF_MACSEC: device is a MACsec device
f5426250 1690 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
30c8bd5a
SS
1691 * @IFF_FAILOVER: device is a failover master device
1692 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
d5256083 1693 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
8a321cf7 1694 * @IFF_NO_ADDRCONF: prevent ipv6 addrconf
c2ff53d8
XZ
1695 * @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1696 * skb_headlen(skb) == 0 (data starts from frag0)
2106efda 1697 * @IFF_CHANGE_PROTO_DOWN: device supports setting carrier via IFLA_PROTO_DOWN
fd770e85
VO
1698 * @IFF_SEE_ALL_HWTSTAMP_REQUESTS: device wants to see calls to
1699 * ndo_hwtstamp_set() for all timestamp requests regardless of source,
1700 * even if those aren't HWTSTAMP_SOURCE_NETDEV.
7aa98047
LR
1701 */
1702enum netdev_priv_flags {
1703 IFF_802_1Q_VLAN = 1<<0,
1704 IFF_EBRIDGE = 1<<1,
0dc1549b
JP
1705 IFF_BONDING = 1<<2,
1706 IFF_ISATAP = 1<<3,
1707 IFF_WAN_HDLC = 1<<4,
1708 IFF_XMIT_DST_RELEASE = 1<<5,
1709 IFF_DONT_BRIDGE = 1<<6,
1710 IFF_DISABLE_NETPOLL = 1<<7,
1711 IFF_MACVLAN_PORT = 1<<8,
1712 IFF_BRIDGE_PORT = 1<<9,
1713 IFF_OVS_DATAPATH = 1<<10,
1714 IFF_TX_SKB_SHARING = 1<<11,
1715 IFF_UNICAST_FLT = 1<<12,
1716 IFF_TEAM_PORT = 1<<13,
1717 IFF_SUPP_NOFCS = 1<<14,
1718 IFF_LIVE_ADDR_CHANGE = 1<<15,
1719 IFF_MACVLAN = 1<<16,
1720 IFF_XMIT_DST_RELEASE_PERM = 1<<17,
1ec54cb4
PA
1721 IFF_L3MDEV_MASTER = 1<<18,
1722 IFF_NO_QUEUE = 1<<19,
1723 IFF_OPENVSWITCH = 1<<20,
1724 IFF_L3MDEV_SLAVE = 1<<21,
1725 IFF_TEAM = 1<<22,
1726 IFF_RXFH_CONFIGURED = 1<<23,
1727 IFF_PHONY_HEADROOM = 1<<24,
1728 IFF_MACSEC = 1<<25,
f5426250 1729 IFF_NO_RX_HANDLER = 1<<26,
30c8bd5a
SS
1730 IFF_FAILOVER = 1<<27,
1731 IFF_FAILOVER_SLAVE = 1<<28,
d5256083 1732 IFF_L3MDEV_RX_HANDLER = 1<<29,
8a321cf7 1733 IFF_NO_ADDRCONF = BIT_ULL(30),
3b89b511 1734 IFF_TX_SKB_NO_LINEAR = BIT_ULL(31),
2106efda 1735 IFF_CHANGE_PROTO_DOWN = BIT_ULL(32),
fd770e85 1736 IFF_SEE_ALL_HWTSTAMP_REQUESTS = BIT_ULL(33),
7aa98047
LR
1737};
1738
1739#define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1740#define IFF_EBRIDGE IFF_EBRIDGE
7aa98047 1741#define IFF_BONDING IFF_BONDING
7aa98047 1742#define IFF_ISATAP IFF_ISATAP
7aa98047
LR
1743#define IFF_WAN_HDLC IFF_WAN_HDLC
1744#define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1745#define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1746#define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1747#define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1748#define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1749#define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1750#define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1751#define IFF_UNICAST_FLT IFF_UNICAST_FLT
1752#define IFF_TEAM_PORT IFF_TEAM_PORT
1753#define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1754#define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1755#define IFF_MACVLAN IFF_MACVLAN
02875878 1756#define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
007979ea 1757#define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
fa8187c9 1758#define IFF_NO_QUEUE IFF_NO_QUEUE
35d4e172 1759#define IFF_OPENVSWITCH IFF_OPENVSWITCH
8f25348b 1760#define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
c981e421 1761#define IFF_TEAM IFF_TEAM
d4ab4286 1762#define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
2463e073 1763#define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
3c175784 1764#define IFF_MACSEC IFF_MACSEC
f5426250 1765#define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
30c8bd5a
SS
1766#define IFF_FAILOVER IFF_FAILOVER
1767#define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
d5256083 1768#define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
c2ff53d8 1769#define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
7aa98047 1770
4e096a18
OR
1771/* Specifies the type of the struct net_device::ml_priv pointer */
1772enum netdev_ml_priv_type {
1773 ML_PRIV_NONE,
1774 ML_PRIV_CAN,
1775};
1776
536721b1
KK
1777/**
1778 * struct net_device - The DEVICE structure.
d651983d
MCC
1779 *
1780 * Actually, this whole structure is a big mistake. It mixes I/O
1781 * data with strictly "high-level" data, and it has to know about
1782 * almost every data structure used in the INET module.
536721b1
KK
1783 *
1784 * @name: This is the first field of the "visible" part of this structure
1785 * (i.e. as seen by users in the "Space.c" file). It is the name
d651983d 1786 * of the interface.
536721b1 1787 *
ff927412 1788 * @name_node: Name hashlist node
536721b1
KK
1789 * @ifalias: SNMP alias
1790 * @mem_end: Shared memory end
1791 * @mem_start: Shared memory start
1792 * @base_addr: Device I/O address
1793 * @irq: Device IRQ number
1794 *
1795 * @state: Generic network queuing layer state, see netdev_state_t
1796 * @dev_list: The global list of network devices
5e82b4b2
BH
1797 * @napi_list: List entry used for polling NAPI devices
1798 * @unreg_list: List entry when we are unregistering the
1799 * device; see the function unregister_netdev
1800 * @close_list: List entry used when we are closing the device
62d885fe
BP
1801 * @ptype_all: Device-specific packet handlers for all protocols
1802 * @ptype_specific: Device-specific, protocol-specific packet handlers
536721b1
KK
1803 *
1804 * @adj_list: Directly linked devices, like slaves for bonding
536721b1
KK
1805 * @features: Currently active device features
1806 * @hw_features: User-changeable features
1807 *
1808 * @wanted_features: User-requested features
1809 * @vlan_features: Mask of features inheritable by VLAN devices
1810 *
1811 * @hw_enc_features: Mask of features inherited by encapsulating devices
1812 * This field indicates what encapsulation
1813 * offloads the hardware is capable of doing,
1814 * and drivers will need to set them appropriately.
1815 *
1816 * @mpls_features: Mask of features inheritable by MPLS
a1fa83bd 1817 * @gso_partial_features: value(s) from NETIF_F_GSO\*
536721b1
KK
1818 *
1819 * @ifindex: interface index
5e82b4b2 1820 * @group: The group the device belongs to
536721b1
KK
1821 *
1822 * @stats: Statistics struct, which was left as a legacy, use
1823 * rtnl_link_stats64 instead
1824 *
625788b5 1825 * @core_stats: core networking counters,
536721b1 1826 * do not use this in drivers
9e55e5d3
FF
1827 * @carrier_up_count: Number of times the carrier has been up
1828 * @carrier_down_count: Number of times the carrier has been down
536721b1 1829 *
536721b1
KK
1830 * @wireless_handlers: List of functions to handle Wireless Extensions,
1831 * instead of ioctl,
1832 * see <net/iw_handler.h> for details.
1833 * @wireless_data: Instance data managed by the core of wireless extensions
1834 *
1835 * @netdev_ops: Includes several pointers to callbacks,
1836 * if one wants to override the ndo_*() functions
3d76a4d3 1837 * @xdp_metadata_ops: Includes pointers to XDP metadata callbacks.
536721b1 1838 * @ethtool_ops: Management operations
a1fa83bd 1839 * @l3mdev_ops: Layer 3 master device operations
f997c55c
AA
1840 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1841 * discovery handling. Necessary for e.g. 6LoWPAN.
a1fa83bd
RD
1842 * @xfrmdev_ops: Transformation offload operations
1843 * @tlsdev_ops: Transport Layer Security offload operations
d476059e 1844 * @header_ops: Includes callbacks for creating,parsing,caching,etc
536721b1
KK
1845 * of Layer 2 headers.
1846 *
1847 * @flags: Interface flags (a la BSD)
26759bee 1848 * @xdp_features: XDP capability supported by the device
536721b1
KK
1849 * @priv_flags: Like 'flags' but invisible to userspace,
1850 * see if.h for the definitions
1851 * @gflags: Global flags ( kept as legacy )
1852 * @padded: How much padding added by alloc_netdev()
1853 * @operstate: RFC2863 operstate
1854 * @link_mode: Mapping policy to operstate
1855 * @if_port: Selectable AUI, TP, ...
1856 * @dma: DMA channel
1857 * @mtu: Interface MTU value
61e84623
JW
1858 * @min_mtu: Interface Minimum MTU value
1859 * @max_mtu: Interface Maximum MTU value
536721b1 1860 * @type: Interface hardware type
2793a23a 1861 * @hard_header_len: Maximum hardware header length.
217e6fa2 1862 * @min_header_len: Minimum hardware header length
536721b1
KK
1863 *
1864 * @needed_headroom: Extra headroom the hardware may need, but not in all
1865 * cases can this be guaranteed
1866 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1867 * cases can this be guaranteed. Some cases also use
1868 * LL_MAX_HEADER instead to allocate the skb
1869 *
1870 * interface address info:
1871 *
1872 * @perm_addr: Permanent hw address
1873 * @addr_assign_type: Hw address assignment type
1874 * @addr_len: Hardware address length
5343da4c
TY
1875 * @upper_level: Maximum depth level of upper devices.
1876 * @lower_level: Maximum depth level of lower devices.
8626a0c8 1877 * @neigh_priv_len: Used in neigh_alloc()
536721b1
KK
1878 * @dev_id: Used to differentiate devices that share
1879 * the same link layer address
1880 * @dev_port: Used to differentiate devices that share
1881 * the same function
1882 * @addr_list_lock: XXX: need comments on this one
a1fa83bd 1883 * @name_assign_type: network interface name assignment type
5e82b4b2 1884 * @uc_promisc: Counter that indicates promiscuous mode
536721b1
KK
1885 * has been enabled due to the need to listen to
1886 * additional unicast addresses in a device that
1887 * does not implement ndo_set_rx_mode()
14ffbbb8
TG
1888 * @uc: unicast mac addresses
1889 * @mc: multicast mac addresses
1890 * @dev_addrs: list of device hw addresses
1891 * @queues_kset: Group of all Kobjects in the Tx and RX queues
5e82b4b2
BH
1892 * @promiscuity: Number of times the NIC is told to work in
1893 * promiscuous mode; if it becomes 0 the NIC will
1894 * exit promiscuous mode
536721b1
KK
1895 * @allmulti: Counter, enables or disables allmulticast mode
1896 *
1897 * @vlan_info: VLAN info
1898 * @dsa_ptr: dsa specific data
1899 * @tipc_ptr: TIPC specific data
1900 * @atalk_ptr: AppleTalk link
1901 * @ip_ptr: IPv4 specific data
536721b1
KK
1902 * @ip6_ptr: IPv6 specific data
1903 * @ax25_ptr: AX.25 specific data
1904 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
a1fa83bd
RD
1905 * @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
1906 * device struct
1907 * @mpls_ptr: mpls_dev struct pointer
583be982 1908 * @mctp_ptr: MCTP specific data
536721b1 1909 *
536721b1
KK
1910 * @dev_addr: Hw address (before bcast,
1911 * because most packets are unicast)
1912 *
1913 * @_rx: Array of RX queues
1914 * @num_rx_queues: Number of RX queues
1915 * allocated at register_netdev() time
1916 * @real_num_rx_queues: Number of RX queues currently active in device
a1fa83bd
RD
1917 * @xdp_prog: XDP sockets filter program pointer
1918 * @gro_flush_timeout: timeout for GRO layer in NAPI
5c45a918
MCC
1919 * @napi_defer_hard_irqs: If not zero, provides a counter that would
1920 * allow to avoid NIC hard IRQ, on busy queues.
536721b1
KK
1921 *
1922 * @rx_handler: handler for received packets
1923 * @rx_handler_data: XXX: need comments on this one
e420bed0 1924 * @tcx_ingress: BPF & clsact qdisc specific data for ingress processing
536721b1 1925 * @ingress_queue: XXX: need comments on this one
2f5e70c8 1926 * @nf_hooks_ingress: netfilter hooks executed for ingress packets
536721b1
KK
1927 * @broadcast: hw bcast address
1928 *
14ffbbb8
TG
1929 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1930 * indexed by RX queue number. Assigned by driver.
1931 * This must only be set if the ndo_rx_flow_steer
1932 * operation is defined
1933 * @index_hlist: Device index hash chain
1934 *
536721b1
KK
1935 * @_tx: Array of TX queues
1936 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1937 * @real_num_tx_queues: Number of TX queues currently active in device
1938 * @qdisc: Root qdisc from userspace point of view
1939 * @tx_queue_len: Max frames per queue allowed
1940 * @tx_global_lock: XXX: need comments on this one
a1fa83bd 1941 * @xdp_bulkq: XDP device bulk queue
044ab86d 1942 * @xps_maps: all CPUs/RXQs maps for XPS device
536721b1
KK
1943 *
1944 * @xps_maps: XXX: need comments on this one
e420bed0 1945 * @tcx_egress: BPF & clsact qdisc specific data for egress processing
42df6e1d 1946 * @nf_hooks_egress: netfilter hooks executed for egress packets
a1fa83bd 1947 * @qdisc_hash: qdisc hash table
536721b1 1948 * @watchdog_timeo: Represents the timeout that is used by
5e82b4b2 1949 * the watchdog (see dev_watchdog())
536721b1
KK
1950 * @watchdog_timer: List of timers
1951 *
eb02d39a 1952 * @proto_down_reason: reason a netdev interface is held down
536721b1 1953 * @pcpu_refcnt: Number of references to this device
add2d736 1954 * @dev_refcnt: Number of references to this device
4d92b95f 1955 * @refcnt_tracker: Tracker directory for tracked references to this device
536721b1 1956 * @todo_list: Delayed register/unregister
536721b1
KK
1957 * @link_watch_list: XXX: need comments on this one
1958 *
1959 * @reg_state: Register/unregister state machine
1960 * @dismantle: Device is going to be freed
1961 * @rtnl_link_state: This enum represents the phases of creating
1962 * a new link
1963 *
cf124db5
DM
1964 * @needs_free_netdev: Should unregister perform free_netdev?
1965 * @priv_destructor: Called from unregister
536721b1
KK
1966 * @npinfo: XXX: need comments on this one
1967 * @nd_net: Network namespace this network device is inside
1968 *
1969 * @ml_priv: Mid-layer private
4e096a18 1970 * @ml_priv_type: Mid-layer private type
536721b1
KK
1971 * @lstats: Loopback statistics
1972 * @tstats: Tunnel statistics
1973 * @dstats: Dummy statistics
1974 * @vstats: Virtual ethernet statistics
1975 *
1976 * @garp_port: GARP
1977 * @mrp_port: MRP
1978 *
4a0cb83b
JK
1979 * @dm_private: Drop monitor private
1980 *
536721b1
KK
1981 * @dev: Class/net/name entry
1982 * @sysfs_groups: Space for optional device, statistics and wireless
1983 * sysfs groups
1984 *
1985 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1986 * @rtnl_link_ops: Rtnl_link_ops
1987 *
1988 * @gso_max_size: Maximum size of generic segmentation offload
14d7b812 1989 * @tso_max_size: Device (as in HW) limit on the max TSO request size
536721b1
KK
1990 * @gso_max_segs: Maximum number of segments that can be passed to the
1991 * NIC for GSO
14d7b812 1992 * @tso_max_segs: Device (as in HW) limit on the max TSO segment count
9eefedd5
XL
1993 * @gso_ipv4_max_size: Maximum size of generic segmentation offload,
1994 * for IPv4.
536721b1
KK
1995 *
1996 * @dcbnl_ops: Data Center Bridging netlink ops
1997 * @num_tc: Number of traffic classes in the net device
1998 * @tc_to_txq: XXX: need comments on this one
920c1cd3 1999 * @prio_tc_map: XXX: need comments on this one
536721b1
KK
2000 *
2001 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
2002 *
2003 * @priomap: XXX: need comments on this one
2004 * @phydev: Physical device may attach itself
2005 * for hardware timestamping
e679c9c1 2006 * @sfp_bus: attached &struct sfp_bus structure.
1a33e10e 2007 *
1a33e10e 2008 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
536721b1 2009 *
d746d707
AK
2010 * @proto_down: protocol port state information can be sent to the
2011 * switch driver and used to set the phys state of the
2012 * switch port.
2013 *
61941143
HK
2014 * @wol_enabled: Wake-on-LAN is enabled
2015 *
29863d41
WW
2016 * @threaded: napi threaded mode is enabled
2017 *
93642e14
JP
2018 * @net_notifier_list: List of per-net netdev notifier block
2019 * that follow this device when it is moved
2020 * to another network namespace.
2021 *
30e9bb84
AT
2022 * @macsec_ops: MACsec offloading ops
2023 *
cc4e3835
JK
2024 * @udp_tunnel_nic_info: static structure describing the UDP tunnel
2025 * offload capabilities of the device
2026 * @udp_tunnel_nic: UDP tunnel offload state
ffa59b0b 2027 * @xdp_state: stores info on attached XDP BPF programs
cc4e3835 2028 *
b62e3317 2029 * @nested_level: Used as a parameter of spin_lock_nested() of
a93bdcb9
MCC
2030 * dev->addr_list_lock.
2031 * @unlink_list: As netif_addr_lock() can be called recursively,
2032 * keep a list of interfaces to be deleted.
eac1b93c
CL
2033 * @gro_max_size: Maximum size of aggregated packet in generic
2034 * receive offload (GRO)
9eefedd5
XL
2035 * @gro_ipv4_max_size: Maximum size of aggregated packet in generic
2036 * receive offload (GRO), for IPv4.
a097627d
MF
2037 * @xdp_zc_max_segs: Maximum number of segments supported by AF_XDP
2038 * zero copy driver
a93bdcb9 2039 *
d07b26f5 2040 * @dev_addr_shadow: Copy of @dev_addr to catch direct writes.
63f13937 2041 * @linkwatch_dev_tracker: refcount tracker used by linkwatch.
f12bf6f3 2042 * @watchdog_dev_tracker: refcount tracker used by watchdog.
b2309a71
ED
2043 * @dev_registered_tracker: tracker for reference held while
2044 * registered
9309f97a 2045 * @offload_xstats_l3: L3 HW stats for this netdevice.
d07b26f5 2046 *
02a68a47
JP
2047 * @devlink_port: Pointer to related devlink port structure.
2048 * Assigned by a driver before netdev registration using
2049 * SET_NETDEV_DEVLINK_PORT macro. This pointer is static
2050 * during the time netdevice is registered.
2051 *
1da177e4
LT
2052 * FIXME: cleanup struct net_device such that network protocol info
2053 * moves out.
2054 */
2055
d94d9fee 2056struct net_device {
1da177e4 2057 char name[IFNAMSIZ];
ff927412 2058 struct netdev_name_node *name_node;
6c557001 2059 struct dev_ifalias __rcu *ifalias;
1da177e4
LT
2060 /*
2061 * I/O specific fields
2062 * FIXME: Merge these and struct ifmap into one
2063 */
536721b1
KK
2064 unsigned long mem_end;
2065 unsigned long mem_start;
2066 unsigned long base_addr;
1da177e4
LT
2067
2068 /*
536721b1
KK
2069 * Some hardware also needs these fields (state,dev_list,
2070 * napi_list,unreg_list,close_list) but they are not
1da177e4
LT
2071 * part of the usual set specified in Space.c.
2072 */
2073
1da177e4
LT
2074 unsigned long state;
2075
7562f876 2076 struct list_head dev_list;
bea3348e 2077 struct list_head napi_list;
44a0873d 2078 struct list_head unreg_list;
5cde2829 2079 struct list_head close_list;
7866a621
SN
2080 struct list_head ptype_all;
2081 struct list_head ptype_specific;
2f268f12 2082
2f268f12
VF
2083 struct {
2084 struct list_head upper;
2085 struct list_head lower;
2086 } adj_list;
2087
28af22c6
JDB
2088 /* Read-mostly cache-line for fast-path access */
2089 unsigned int flags;
d3d854fd 2090 xdp_features_t xdp_features;
2106efda 2091 unsigned long long priv_flags;
28af22c6 2092 const struct net_device_ops *netdev_ops;
3d76a4d3 2093 const struct xdp_metadata_ops *xdp_metadata_ops;
28af22c6
JDB
2094 int ifindex;
2095 unsigned short gflags;
2096 unsigned short hard_header_len;
2097
2098 /* Note : dev->mtu is often read without holding a lock.
2099 * Writers usually hold RTNL.
2100 * It is recommended to use READ_ONCE() to annotate the reads,
2101 * and to use WRITE_ONCE() to annotate the writes.
2102 */
2103 unsigned int mtu;
2104 unsigned short needed_headroom;
2105 unsigned short needed_tailroom;
2106
c8f44aff 2107 netdev_features_t features;
c8f44aff 2108 netdev_features_t hw_features;
c8f44aff 2109 netdev_features_t wanted_features;
c8f44aff 2110 netdev_features_t vlan_features;
6a674e9c 2111 netdev_features_t hw_enc_features;
0d89d203 2112 netdev_features_t mpls_features;
802ab55a 2113 netdev_features_t gso_partial_features;
04ed3e74 2114
28af22c6
JDB
2115 unsigned int min_mtu;
2116 unsigned int max_mtu;
2117 unsigned short type;
2118 unsigned char min_header_len;
2119 unsigned char name_assign_type;
2120
7a66bbc9 2121 int group;
1da177e4 2122
28af22c6 2123 struct net_device_stats stats; /* not used by modern drivers */
015f0688 2124
625788b5 2125 struct net_device_core_stats __percpu *core_stats;
1da177e4 2126
b2d3bcfa
DD
2127 /* Stats to monitor link on/off, flapping */
2128 atomic_t carrier_up_count;
2129 atomic_t carrier_down_count;
2130
b86e0280 2131#ifdef CONFIG_WIRELESS_EXT
5e82b4b2
BH
2132 const struct iw_handler_def *wireless_handlers;
2133 struct iw_public_data *wireless_data;
b86e0280 2134#endif
76fd8593 2135 const struct ethtool_ops *ethtool_ops;
1b69c6d0
DA
2136#ifdef CONFIG_NET_L3_MASTER_DEV
2137 const struct l3mdev_ops *l3mdev_ops;
2138#endif
f997c55c
AA
2139#if IS_ENABLED(CONFIG_IPV6)
2140 const struct ndisc_ops *ndisc_ops;
2141#endif
1da177e4 2142
9cb0d21d 2143#ifdef CONFIG_XFRM_OFFLOAD
d77e38e6
SK
2144 const struct xfrmdev_ops *xfrmdev_ops;
2145#endif
2146
a5c37c63
IL
2147#if IS_ENABLED(CONFIG_TLS_DEVICE)
2148 const struct tlsdev_ops *tlsdev_ops;
2149#endif
2150
3b04ddde
SH
2151 const struct header_ops *header_ops;
2152
536721b1
KK
2153 unsigned char operstate;
2154 unsigned char link_mode;
b00055aa 2155
536721b1
KK
2156 unsigned char if_port;
2157 unsigned char dma;
bdc220da 2158
1da177e4 2159 /* Interface address info. */
536721b1
KK
2160 unsigned char perm_addr[MAX_ADDR_LEN];
2161 unsigned char addr_assign_type;
2162 unsigned char addr_len;
5343da4c
TY
2163 unsigned char upper_level;
2164 unsigned char lower_level;
1fc70edb 2165
a0a9663d 2166 unsigned short neigh_priv_len;
536721b1
KK
2167 unsigned short dev_id;
2168 unsigned short dev_port;
28af22c6
JDB
2169 unsigned short padded;
2170
ccffad25 2171 spinlock_t addr_list_lock;
28af22c6 2172 int irq;
1fc70edb 2173
536721b1
KK
2174 struct netdev_hw_addr_list uc;
2175 struct netdev_hw_addr_list mc;
2176 struct netdev_hw_addr_list dev_addrs;
2177
4c3d5e7b
ED
2178#ifdef CONFIG_SYSFS
2179 struct kset *queues_kset;
1fc70edb
TY
2180#endif
2181#ifdef CONFIG_LOCKDEP
2182 struct list_head unlink_list;
4c3d5e7b 2183#endif
9d45abe1
WC
2184 unsigned int promiscuity;
2185 unsigned int allmulti;
1fc70edb
TY
2186 bool uc_promisc;
2187#ifdef CONFIG_LOCKDEP
2188 unsigned char nested_level;
2189#endif
1da177e4 2190
1da177e4 2191
5e82b4b2 2192 /* Protocol-specific pointers */
65ac6a5f 2193
c304eddc
JK
2194 struct in_device __rcu *ip_ptr;
2195 struct inet6_dev __rcu *ip6_ptr;
d11ead75 2196#if IS_ENABLED(CONFIG_VLAN_8021Q)
536721b1 2197 struct vlan_info __rcu *vlan_info;
65ac6a5f 2198#endif
34a430d7 2199#if IS_ENABLED(CONFIG_NET_DSA)
2f657a60 2200 struct dsa_port *dsa_ptr;
37cb0620
YX
2201#endif
2202#if IS_ENABLED(CONFIG_TIPC)
536721b1 2203 struct tipc_bearer __rcu *tipc_ptr;
91da11f8 2204#endif
d6c6d0bb 2205#if IS_ENABLED(CONFIG_ATALK)
536721b1 2206 void *atalk_ptr;
89e58148 2207#endif
19ff13f2 2208#if IS_ENABLED(CONFIG_AX25)
536721b1 2209 void *ax25_ptr;
19ff13f2 2210#endif
c304eddc 2211#if IS_ENABLED(CONFIG_CFG80211)
536721b1 2212 struct wireless_dev *ieee80211_ptr;
c304eddc
JK
2213#endif
2214#if IS_ENABLED(CONFIG_IEEE802154) || IS_ENABLED(CONFIG_6LOWPAN)
98a18b6f 2215 struct wpan_dev *ieee802154_ptr;
c304eddc 2216#endif
03c57747
RS
2217#if IS_ENABLED(CONFIG_MPLS_ROUTING)
2218 struct mpls_dev __rcu *mpls_ptr;
2219#endif
583be982
JK
2220#if IS_ENABLED(CONFIG_MCTP)
2221 struct mctp_dev __rcu *mctp_ptr;
2222#endif
1da177e4 2223
9356b8fc 2224/*
cd13539b 2225 * Cache lines mostly used on receive path (including eth_type_trans())
9356b8fc 2226 */
9356b8fc 2227 /* Interface address info used in eth_type_trans() */
adeef3e3 2228 const unsigned char *dev_addr;
f001fde5 2229
0a9627f2 2230 struct netdev_rx_queue *_rx;
0a9627f2 2231 unsigned int num_rx_queues;
62fe0b40 2232 unsigned int real_num_rx_queues;
0a9627f2 2233
7acedaf5 2234 struct bpf_prog __rcu *xdp_prog;
3b47d303 2235 unsigned long gro_flush_timeout;
6f8b12d6 2236 int napi_defer_hard_irqs;
0fe79f28
AD
2237#define GRO_LEGACY_MAX_SIZE 65536u
2238/* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
2239 * and shinfo->gso_segs is a 16bit field.
2240 */
2241#define GRO_MAX_SIZE (8 * 65535u)
eac1b93c 2242 unsigned int gro_max_size;
9eefedd5 2243 unsigned int gro_ipv4_max_size;
13ce2daa 2244 unsigned int xdp_zc_max_segs;
61391cde 2245 rx_handler_func_t __rcu *rx_handler;
2246 void __rcu *rx_handler_data;
e420bed0
DB
2247#ifdef CONFIG_NET_XGRESS
2248 struct bpf_mprog_entry __rcu *tcx_ingress;
d2788d34 2249#endif
24824a09 2250 struct netdev_queue __rcu *ingress_queue;
e687ad60 2251#ifdef CONFIG_NETFILTER_INGRESS
960632ec 2252 struct nf_hook_entries __rcu *nf_hooks_ingress;
e687ad60 2253#endif
d2788d34 2254
536721b1 2255 unsigned char broadcast[MAX_ADDR_LEN];
14ffbbb8
TG
2256#ifdef CONFIG_RFS_ACCEL
2257 struct cpu_rmap *rx_cpu_rmap;
2258#endif
2259 struct hlist_node index_hlist;
cd13539b
ED
2260
2261/*
2262 * Cache lines mostly used on transmit path
2263 */
e8a0464c
DM
2264 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
2265 unsigned int num_tx_queues;
fd2ea0a7 2266 unsigned int real_num_tx_queues;
5891cd5e 2267 struct Qdisc __rcu *qdisc;
0cd29503 2268 unsigned int tx_queue_len;
c3f26a26 2269 spinlock_t tx_global_lock;
75ccae62
THJ
2270
2271 struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
cd13539b 2272
bf264145 2273#ifdef CONFIG_XPS
044ab86d 2274 struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
bf264145 2275#endif
e420bed0
DB
2276#ifdef CONFIG_NET_XGRESS
2277 struct bpf_mprog_entry __rcu *tcx_egress;
1f211a1b 2278#endif
42df6e1d
LW
2279#ifdef CONFIG_NETFILTER_EGRESS
2280 struct nf_hook_entries __rcu *nf_hooks_egress;
2281#endif
0c4f691f 2282
75ccae62
THJ
2283#ifdef CONFIG_NET_SCHED
2284 DECLARE_HASHTABLE (qdisc_hash, 4);
2285#endif
9356b8fc 2286 /* These may be needed for future network-power-down code. */
9356b8fc 2287 struct timer_list watchdog_timer;
75ccae62 2288 int watchdog_timeo;
9356b8fc 2289
829eb208
RP
2290 u32 proto_down_reason;
2291
1da177e4 2292 struct list_head todo_list;
919067cc
ED
2293
2294#ifdef CONFIG_PCPU_DEV_REFCNT
75ccae62 2295 int __percpu *pcpu_refcnt;
919067cc
ED
2296#else
2297 refcount_t dev_refcnt;
2298#endif
4d92b95f 2299 struct ref_tracker_dir refcnt_tracker;
1da177e4 2300
e014debe 2301 struct list_head link_watch_list;
572a103d 2302
1da177e4 2303 enum { NETREG_UNINITIALIZED=0,
b17a7c17 2304 NETREG_REGISTERED, /* completed register_netdevice */
1da177e4
LT
2305 NETREG_UNREGISTERING, /* called unregister_netdevice */
2306 NETREG_UNREGISTERED, /* completed unregister todo */
2307 NETREG_RELEASED, /* called free_netdev */
937f1ba5 2308 NETREG_DUMMY, /* dummy device for NAPI poll */
449f4544
ED
2309 } reg_state:8;
2310
536721b1 2311 bool dismantle;
a2835763
PM
2312
2313 enum {
2314 RTNL_LINK_INITIALIZED,
2315 RTNL_LINK_INITIALIZING,
2316 } rtnl_link_state:16;
1da177e4 2317
cf124db5
DM
2318 bool needs_free_netdev;
2319 void (*priv_destructor)(struct net_device *dev);
1da177e4 2320
1da177e4 2321#ifdef CONFIG_NETPOLL
5fbee843 2322 struct netpoll_info __rcu *npinfo;
1da177e4 2323#endif
eae792b7 2324
0c5c9fb5 2325 possible_net_t nd_net;
4a1c5371 2326
4951704b 2327 /* mid-layer private */
4e096a18
OR
2328 void *ml_priv;
2329 enum netdev_ml_priv_type ml_priv_type;
2330
a7855c78 2331 union {
536721b1 2332 struct pcpu_lstats __percpu *lstats;
8f84985f 2333 struct pcpu_sw_netstats __percpu *tstats;
536721b1 2334 struct pcpu_dstats __percpu *dstats;
a7855c78 2335 };
536721b1 2336
fb585b44 2337#if IS_ENABLED(CONFIG_GARP)
3cc77ec7 2338 struct garp_port __rcu *garp_port;
fb585b44
TK
2339#endif
2340#if IS_ENABLED(CONFIG_MRP)
febf018d 2341 struct mrp_port __rcu *mrp_port;
fb585b44 2342#endif
b26ef81c
ED
2343#if IS_ENABLED(CONFIG_NET_DROP_MONITOR)
2344 struct dm_hw_stat_delta __rcu *dm_private;
2345#endif
5e82b4b2 2346 struct device dev;
0c509a6c 2347 const struct attribute_group *sysfs_groups[4];
a953be53 2348 const struct attribute_group *sysfs_rx_queue_group;
38f7b870 2349
38f7b870 2350 const struct rtnl_link_ops *rtnl_link_ops;
f25f4e44 2351
82cc1a7a 2352 /* for setting kernel sock attribute on TCP connection setup */
34b92e8d 2353#define GSO_MAX_SEGS 65535u
7c4e983c 2354#define GSO_LEGACY_MAX_SIZE 65536u
34b92e8d
ED
2355/* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
2356 * and shinfo->gso_segs is a 16bit field.
2357 */
2358#define GSO_MAX_SIZE (8 * GSO_MAX_SEGS)
7c4e983c 2359
82cc1a7a 2360 unsigned int gso_max_size;
14d7b812
JK
2361#define TSO_LEGACY_MAX_SIZE 65536
2362#define TSO_MAX_SIZE UINT_MAX
2363 unsigned int tso_max_size;
30b678d8 2364 u16 gso_max_segs;
14d7b812
JK
2365#define TSO_MAX_SEGS U16_MAX
2366 u16 tso_max_segs;
9eefedd5 2367 unsigned int gso_ipv4_max_size;
743b03a8 2368
7a6b6f51 2369#ifdef CONFIG_DCB
32953543 2370 const struct dcbnl_rtnl_ops *dcbnl_ops;
2f90b865 2371#endif
ffcfe25b 2372 s16 num_tc;
5e82b4b2
BH
2373 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2374 u8 prio_tc_map[TC_BITMASK + 1];
2f90b865 2375
d11ead75 2376#if IS_ENABLED(CONFIG_FCOE)
4d288d57 2377 unsigned int fcoe_ddp_xid;
5bc1421e 2378#endif
86f8515f 2379#if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
5bc1421e 2380 struct netprio_map __rcu *priomap;
4d288d57 2381#endif
5e82b4b2 2382 struct phy_device *phydev;
e679c9c1 2383 struct sfp_bus *sfp_bus;
1a33e10e 2384 struct lock_class_key *qdisc_tx_busylock;
5e82b4b2 2385 bool proto_down;
61941143 2386 unsigned wol_enabled:1;
29863d41 2387 unsigned threaded:1;
93642e14
JP
2388
2389 struct list_head net_notifier_list;
30e9bb84
AT
2390
2391#if IS_ENABLED(CONFIG_MACSEC)
2392 /* MACsec management functions */
2393 const struct macsec_ops *macsec_ops;
2394#endif
cc4e3835
JK
2395 const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
2396 struct udp_tunnel_nic *udp_tunnel_nic;
7f0a8382
AN
2397
2398 /* protected by rtnl_lock */
2399 struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
d07b26f5
JK
2400
2401 u8 dev_addr_shadow[MAX_ADDR_LEN];
63f13937 2402 netdevice_tracker linkwatch_dev_tracker;
f12bf6f3 2403 netdevice_tracker watchdog_dev_tracker;
b2309a71 2404 netdevice_tracker dev_registered_tracker;
9309f97a 2405 struct rtnl_hw_stats64 *offload_xstats_l3;
02a68a47
JP
2406
2407 struct devlink_port *devlink_port;
1da177e4 2408};
43cb76d9 2409#define to_net_dev(d) container_of(d, struct net_device, dev)
1da177e4 2410
02a68a47
JP
2411/*
2412 * Driver should use this to assign devlink port instance to a netdevice
2413 * before it registers the netdevice. Therefore devlink_port is static
2414 * during the netdev lifetime after it is registered.
2415 */
2416#define SET_NETDEV_DEVLINK_PORT(dev, port) \
2417({ \
2418 WARN_ON((dev)->reg_state != NETREG_UNINITIALIZED); \
2419 ((dev)->devlink_port = (port)); \
2420})
2421
b5cdae32
DM
2422static inline bool netif_elide_gro(const struct net_device *dev)
2423{
2424 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2425 return true;
2426 return false;
2427}
2428
1da177e4 2429#define NETDEV_ALIGN 32
1da177e4 2430
4f57c087
JF
2431static inline
2432int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2433{
2434 return dev->prio_tc_map[prio & TC_BITMASK];
2435}
2436
2437static inline
2438int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2439{
2440 if (tc >= dev->num_tc)
2441 return -EINVAL;
2442
2443 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2444 return 0;
2445}
2446
8d059b0f 2447int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
9cf1f6a8
AD
2448void netdev_reset_tc(struct net_device *dev);
2449int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2450int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
4f57c087
JF
2451
2452static inline
2453int netdev_get_num_tc(struct net_device *dev)
2454{
2455 return dev->num_tc;
2456}
2457
f468f21b
TT
2458static inline void net_prefetch(void *p)
2459{
2460 prefetch(p);
2461#if L1_CACHE_BYTES < 128
2462 prefetch((u8 *)p + L1_CACHE_BYTES);
2463#endif
2464}
2465
2466static inline void net_prefetchw(void *p)
2467{
2468 prefetchw(p);
2469#if L1_CACHE_BYTES < 128
2470 prefetchw((u8 *)p + L1_CACHE_BYTES);
2471#endif
2472}
2473
ffcfe25b
AD
2474void netdev_unbind_sb_channel(struct net_device *dev,
2475 struct net_device *sb_dev);
2476int netdev_bind_sb_channel_queue(struct net_device *dev,
2477 struct net_device *sb_dev,
2478 u8 tc, u16 count, u16 offset);
2479int netdev_set_sb_channel(struct net_device *dev, u16 channel);
2480static inline int netdev_get_sb_channel(struct net_device *dev)
2481{
2482 return max_t(int, -dev->num_tc, 0);
2483}
2484
e8a0464c
DM
2485static inline
2486struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2487 unsigned int index)
2488{
1cc6571f 2489 DEBUG_NET_WARN_ON_ONCE(index >= dev->num_tx_queues);
e8a0464c
DM
2490 return &dev->_tx[index];
2491}
2492
10c51b56
DB
2493static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2494 const struct sk_buff *skb)
2495{
2496 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2497}
2498
e8a0464c
DM
2499static inline void netdev_for_each_tx_queue(struct net_device *dev,
2500 void (*f)(struct net_device *,
2501 struct netdev_queue *,
2502 void *),
2503 void *arg)
2504{
2505 unsigned int i;
2506
2507 for (i = 0; i < dev->num_tx_queues; i++)
2508 f(dev, &dev->_tx[i], arg);
2509}
2510
1a33e10e
CW
2511#define netdev_lockdep_set_classes(dev) \
2512{ \
2513 static struct lock_class_key qdisc_tx_busylock_key; \
1a33e10e 2514 static struct lock_class_key qdisc_xmit_lock_key; \
845e0ebb 2515 static struct lock_class_key dev_addr_list_lock_key; \
1a33e10e
CW
2516 unsigned int i; \
2517 \
2518 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
845e0ebb
CW
2519 lockdep_set_class(&(dev)->addr_list_lock, \
2520 &dev_addr_list_lock_key); \
1a33e10e
CW
2521 for (i = 0; i < (dev)->num_tx_queues; i++) \
2522 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2523 &qdisc_xmit_lock_key); \
2524}
2525
b71b5837
PA
2526u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2527 struct net_device *sb_dev);
4bd97d51
PA
2528struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2529 struct sk_buff *skb,
2530 struct net_device *sb_dev);
8c4c49df 2531
871b642a
PA
2532/* returns the headroom that the master device needs to take in account
2533 * when forwarding to this dev
2534 */
2535static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2536{
2537 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2538}
2539
2540static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2541{
2542 if (dev->netdev_ops->ndo_set_rx_headroom)
2543 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2544}
2545
2546/* set the device rx headroom to the dev's default */
2547static inline void netdev_reset_rx_headroom(struct net_device *dev)
2548{
2549 netdev_set_rx_headroom(dev, -1);
2550}
2551
4e096a18
OR
2552static inline void *netdev_get_ml_priv(struct net_device *dev,
2553 enum netdev_ml_priv_type type)
2554{
2555 if (dev->ml_priv_type != type)
2556 return NULL;
2557
2558 return dev->ml_priv;
2559}
2560
2561static inline void netdev_set_ml_priv(struct net_device *dev,
2562 void *ml_priv,
2563 enum netdev_ml_priv_type type)
2564{
2565 WARN(dev->ml_priv_type && dev->ml_priv_type != type,
2566 "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
2567 dev->ml_priv_type, type);
2568 WARN(!dev->ml_priv_type && dev->ml_priv,
2569 "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
2570
2571 dev->ml_priv = ml_priv;
2572 dev->ml_priv_type = type;
2573}
2574
c346dca1
YH
2575/*
2576 * Net namespace inlines
2577 */
2578static inline
2579struct net *dev_net(const struct net_device *dev)
2580{
c2d9ba9b 2581 return read_pnet(&dev->nd_net);
c346dca1
YH
2582}
2583
2584static inline
f5aa23fd 2585void dev_net_set(struct net_device *dev, struct net *net)
c346dca1 2586{
0c5c9fb5 2587 write_pnet(&dev->nd_net, net);
c346dca1
YH
2588}
2589
bea3348e
SH
2590/**
2591 * netdev_priv - access network device private data
2592 * @dev: network device
2593 *
2594 * Get network device private data
2595 */
6472ce60 2596static inline void *netdev_priv(const struct net_device *dev)
1da177e4 2597{
1ce8e7b5 2598 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
1da177e4
LT
2599}
2600
1da177e4
LT
2601/* Set the sysfs physical device reference for the network logical device
2602 * if set prior to registration will cause a symlink during initialization.
2603 */
43cb76d9 2604#define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1da177e4 2605
384912ed 2606/* Set the sysfs device type for the network logical device to allow
3f79410c 2607 * fine-grained identification of different network device types. For
5e82b4b2 2608 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
384912ed
MH
2609 */
2610#define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2611
82dc3c63
ED
2612/* Default NAPI poll() weight
2613 * Device drivers are strongly advised to not use bigger value
2614 */
2615#define NAPI_POLL_WEIGHT 64
2616
58caed3d
JK
2617void netif_napi_add_weight(struct net_device *dev, struct napi_struct *napi,
2618 int (*poll)(struct napi_struct *, int), int weight);
2619
3b582cc1 2620/**
58caed3d
JK
2621 * netif_napi_add() - initialize a NAPI context
2622 * @dev: network device
2623 * @napi: NAPI context
2624 * @poll: polling function
3b582cc1 2625 *
5e82b4b2
BH
2626 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2627 * *any* of the other NAPI-related functions.
3b582cc1 2628 */
58caed3d
JK
2629static inline void
2630netif_napi_add(struct net_device *dev, struct napi_struct *napi,
b48b89f9 2631 int (*poll)(struct napi_struct *, int))
58caed3d 2632{
b48b89f9 2633 netif_napi_add_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
58caed3d
JK
2634}
2635
2636static inline void
2637netif_napi_add_tx_weight(struct net_device *dev,
2638 struct napi_struct *napi,
2639 int (*poll)(struct napi_struct *, int),
2640 int weight)
2641{
2642 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2643 netif_napi_add_weight(dev, napi, poll, weight);
2644}
2645
d64b5e85 2646/**
58caed3d
JK
2647 * netif_napi_add_tx() - initialize a NAPI context to be used for Tx only
2648 * @dev: network device
2649 * @napi: NAPI context
2650 * @poll: polling function
d64b5e85
ED
2651 *
2652 * This variant of netif_napi_add() should be used from drivers using NAPI
2653 * to exclusively poll a TX queue.
2654 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2655 */
58caed3d 2656static inline void netif_napi_add_tx(struct net_device *dev,
d64b5e85 2657 struct napi_struct *napi,
58caed3d 2658 int (*poll)(struct napi_struct *, int))
d64b5e85 2659{
58caed3d 2660 netif_napi_add_tx_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
d64b5e85
ED
2661}
2662
5198d545
JK
2663/**
2664 * __netif_napi_del - remove a NAPI context
2665 * @napi: NAPI context
2666 *
2667 * Warning: caller must observe RCU grace period before freeing memory
2668 * containing @napi. Drivers might want to call this helper to combine
2669 * all the needed RCU grace periods into a single one.
2670 */
2671void __netif_napi_del(struct napi_struct *napi);
2672
d8156534 2673/**
5e82b4b2
BH
2674 * netif_napi_del - remove a NAPI context
2675 * @napi: NAPI context
d8156534 2676 *
5e82b4b2 2677 * netif_napi_del() removes a NAPI context from the network device NAPI list
d8156534 2678 */
5198d545
JK
2679static inline void netif_napi_del(struct napi_struct *napi)
2680{
2681 __netif_napi_del(napi);
2682 synchronize_net();
2683}
d565b0a1 2684
1da177e4 2685struct packet_type {
f2ccd8fa 2686 __be16 type; /* This is really htons(ether_type). */
fa788d98 2687 bool ignore_outgoing;
f2ccd8fa 2688 struct net_device *dev; /* NULL is wildcarded here */
f1d9268e 2689 netdevice_tracker dev_tracker;
f2ccd8fa
DM
2690 int (*func) (struct sk_buff *,
2691 struct net_device *,
2692 struct packet_type *,
2693 struct net_device *);
17266ee9
EC
2694 void (*list_func) (struct list_head *,
2695 struct packet_type *,
2696 struct net_device *);
c0de08d0
EL
2697 bool (*id_match)(struct packet_type *ptype,
2698 struct sock *sk);
47934e06 2699 struct net *af_packet_net;
1da177e4
LT
2700 void *af_packet_priv;
2701 struct list_head list;
2702};
2703
f191a1d1 2704struct offload_callbacks {
576a30eb 2705 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
c8f44aff 2706 netdev_features_t features);
d4546c25
DM
2707 struct sk_buff *(*gro_receive)(struct list_head *head,
2708 struct sk_buff *skb);
299603e8 2709 int (*gro_complete)(struct sk_buff *skb, int nhoff);
f191a1d1
VY
2710};
2711
2712struct packet_offload {
2713 __be16 type; /* This is really htons(ether_type). */
bdef7de4 2714 u16 priority;
f191a1d1
VY
2715 struct offload_callbacks callbacks;
2716 struct list_head list;
1da177e4
LT
2717};
2718
5e82b4b2 2719/* often modified stats are per-CPU, other are shared (netdev->stats) */
8f84985f 2720struct pcpu_sw_netstats {
9962acef
ED
2721 u64_stats_t rx_packets;
2722 u64_stats_t rx_bytes;
2723 u64_stats_t tx_packets;
2724 u64_stats_t tx_bytes;
8f84985f 2725 struct u64_stats_sync syncp;
9a5ee462 2726} __aligned(4 * sizeof(u64));
52bb6677
LR
2727
2728struct pcpu_lstats {
fd2f4737
ED
2729 u64_stats_t packets;
2730 u64_stats_t bytes;
52bb6677 2731 struct u64_stats_sync syncp;
9a5ee462 2732} __aligned(2 * sizeof(u64));
8f84985f 2733
de7d5084
ED
2734void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2735
451b05f4
FF
2736static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
2737{
2738 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2739
2740 u64_stats_update_begin(&tstats->syncp);
9962acef
ED
2741 u64_stats_add(&tstats->rx_bytes, len);
2742 u64_stats_inc(&tstats->rx_packets);
451b05f4
FF
2743 u64_stats_update_end(&tstats->syncp);
2744}
2745
d3fd6548
HK
2746static inline void dev_sw_netstats_tx_add(struct net_device *dev,
2747 unsigned int packets,
2748 unsigned int len)
2749{
2750 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2751
2752 u64_stats_update_begin(&tstats->syncp);
9962acef
ED
2753 u64_stats_add(&tstats->tx_bytes, len);
2754 u64_stats_add(&tstats->tx_packets, packets);
d3fd6548
HK
2755 u64_stats_update_end(&tstats->syncp);
2756}
2757
dd5382a0
ED
2758static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
2759{
2760 struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
2761
2762 u64_stats_update_begin(&lstats->syncp);
fd2f4737
ED
2763 u64_stats_add(&lstats->bytes, len);
2764 u64_stats_inc(&lstats->packets);
dd5382a0
ED
2765 u64_stats_update_end(&lstats->syncp);
2766}
2767
aabc92bb
PNA
2768#define __netdev_alloc_pcpu_stats(type, gfp) \
2769({ \
2770 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2771 if (pcpu_stats) { \
2772 int __cpu; \
2773 for_each_possible_cpu(__cpu) { \
2774 typeof(type) *stat; \
2775 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2776 u64_stats_init(&stat->syncp); \
2777 } \
2778 } \
2779 pcpu_stats; \
1c213bd2
WC
2780})
2781
aabc92bb 2782#define netdev_alloc_pcpu_stats(type) \
326fcfa5 2783 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
aabc92bb 2784
81b01894
HK
2785#define devm_netdev_alloc_pcpu_stats(dev, type) \
2786({ \
2787 typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
2788 if (pcpu_stats) { \
2789 int __cpu; \
2790 for_each_possible_cpu(__cpu) { \
2791 typeof(type) *stat; \
2792 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2793 u64_stats_init(&stat->syncp); \
2794 } \
2795 } \
2796 pcpu_stats; \
2797})
2798
764f5e54
JP
2799enum netdev_lag_tx_type {
2800 NETDEV_LAG_TX_TYPE_UNKNOWN,
2801 NETDEV_LAG_TX_TYPE_RANDOM,
2802 NETDEV_LAG_TX_TYPE_BROADCAST,
2803 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2804 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2805 NETDEV_LAG_TX_TYPE_HASH,
2806};
2807
f44aa9ef
JH
2808enum netdev_lag_hash {
2809 NETDEV_LAG_HASH_NONE,
2810 NETDEV_LAG_HASH_L2,
2811 NETDEV_LAG_HASH_L34,
2812 NETDEV_LAG_HASH_L23,
2813 NETDEV_LAG_HASH_E23,
2814 NETDEV_LAG_HASH_E34,
7b8fc010 2815 NETDEV_LAG_HASH_VLAN_SRCMAC,
f44aa9ef
JH
2816 NETDEV_LAG_HASH_UNKNOWN,
2817};
2818
764f5e54
JP
2819struct netdev_lag_upper_info {
2820 enum netdev_lag_tx_type tx_type;
f44aa9ef 2821 enum netdev_lag_hash hash_type;
764f5e54
JP
2822};
2823
fb1b2e3c
JP
2824struct netdev_lag_lower_state_info {
2825 u8 link_up : 1,
2826 tx_enabled : 1;
2827};
2828
1da177e4
LT
2829#include <linux/notifier.h>
2830
ede2762d
KT
2831/* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2832 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2833 * adding new types.
dcfe1421 2834 */
ede2762d
KT
2835enum netdev_cmd {
2836 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2837 NETDEV_DOWN,
2838 NETDEV_REBOOT, /* Tell a protocol stack a network interface
dcfe1421
AW
2839 detected a hardware crash and restarted
2840 - we can use this eg to kick tcp sessions
2841 once done */
ede2762d
KT
2842 NETDEV_CHANGE, /* Notify device state change */
2843 NETDEV_REGISTER,
2844 NETDEV_UNREGISTER,
2845 NETDEV_CHANGEMTU, /* notify after mtu change happened */
1570415f
PM
2846 NETDEV_CHANGEADDR, /* notify after the address change */
2847 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
ede2762d
KT
2848 NETDEV_GOING_DOWN,
2849 NETDEV_CHANGENAME,
2850 NETDEV_FEAT_CHANGE,
2851 NETDEV_BONDING_FAILOVER,
2852 NETDEV_PRE_UP,
2853 NETDEV_PRE_TYPE_CHANGE,
2854 NETDEV_POST_TYPE_CHANGE,
2855 NETDEV_POST_INIT,
02a68a47 2856 NETDEV_PRE_UNINIT,
ede2762d
KT
2857 NETDEV_RELEASE,
2858 NETDEV_NOTIFY_PEERS,
2859 NETDEV_JOIN,
2860 NETDEV_CHANGEUPPER,
2861 NETDEV_RESEND_IGMP,
2862 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2863 NETDEV_CHANGEINFODATA,
2864 NETDEV_BONDING_INFO,
2865 NETDEV_PRECHANGEUPPER,
2866 NETDEV_CHANGELOWERSTATE,
2867 NETDEV_UDP_TUNNEL_PUSH_INFO,
2868 NETDEV_UDP_TUNNEL_DROP_INFO,
2869 NETDEV_CHANGE_TX_QUEUE_LEN,
9daae9bd
GP
2870 NETDEV_CVLAN_FILTER_PUSH_INFO,
2871 NETDEV_CVLAN_FILTER_DROP_INFO,
2872 NETDEV_SVLAN_FILTER_PUSH_INFO,
2873 NETDEV_SVLAN_FILTER_DROP_INFO,
9309f97a
PM
2874 NETDEV_OFFLOAD_XSTATS_ENABLE,
2875 NETDEV_OFFLOAD_XSTATS_DISABLE,
2876 NETDEV_OFFLOAD_XSTATS_REPORT_USED,
2877 NETDEV_OFFLOAD_XSTATS_REPORT_DELTA,
d3d854fd 2878 NETDEV_XDP_FEAT_CHANGE,
ede2762d
KT
2879};
2880const char *netdev_cmd_to_name(enum netdev_cmd cmd);
dcfe1421 2881
f629d208
JP
2882int register_netdevice_notifier(struct notifier_block *nb);
2883int unregister_netdevice_notifier(struct notifier_block *nb);
a30c7b42
JP
2884int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2885int unregister_netdevice_notifier_net(struct net *net,
2886 struct notifier_block *nb);
93642e14
JP
2887int register_netdevice_notifier_dev_net(struct net_device *dev,
2888 struct notifier_block *nb,
2889 struct netdev_net_notifier *nn);
2890int unregister_netdevice_notifier_dev_net(struct net_device *dev,
2891 struct notifier_block *nb,
2892 struct netdev_net_notifier *nn);
351638e7
JP
2893
2894struct netdev_notifier_info {
51d0c047
DA
2895 struct net_device *dev;
2896 struct netlink_ext_ack *extack;
351638e7
JP
2897};
2898
af7d6cce
SD
2899struct netdev_notifier_info_ext {
2900 struct netdev_notifier_info info; /* must be first */
2901 union {
2902 u32 mtu;
2903 } ext;
2904};
2905
be9efd36
JP
2906struct netdev_notifier_change_info {
2907 struct netdev_notifier_info info; /* must be first */
2908 unsigned int flags_changed;
2909};
2910
0e4ead9d
JP
2911struct netdev_notifier_changeupper_info {
2912 struct netdev_notifier_info info; /* must be first */
2913 struct net_device *upper_dev; /* new upper dev */
2914 bool master; /* is upper dev master */
5e82b4b2 2915 bool linking; /* is the notification for link or unlink */
29bf24af 2916 void *upper_info; /* upper dev info */
0e4ead9d
JP
2917};
2918
04d48266
JP
2919struct netdev_notifier_changelowerstate_info {
2920 struct netdev_notifier_info info; /* must be first */
2921 void *lower_state_info; /* is lower dev state */
2922};
2923
1570415f
PM
2924struct netdev_notifier_pre_changeaddr_info {
2925 struct netdev_notifier_info info; /* must be first */
2926 const unsigned char *dev_addr;
2927};
2928
9309f97a
PM
2929enum netdev_offload_xstats_type {
2930 NETDEV_OFFLOAD_XSTATS_TYPE_L3 = 1,
2931};
2932
2933struct netdev_notifier_offload_xstats_info {
2934 struct netdev_notifier_info info; /* must be first */
2935 enum netdev_offload_xstats_type type;
2936
2937 union {
2938 /* NETDEV_OFFLOAD_XSTATS_REPORT_DELTA */
2939 struct netdev_notifier_offload_xstats_rd *report_delta;
2940 /* NETDEV_OFFLOAD_XSTATS_REPORT_USED */
2941 struct netdev_notifier_offload_xstats_ru *report_used;
2942 };
2943};
2944
2945int netdev_offload_xstats_enable(struct net_device *dev,
2946 enum netdev_offload_xstats_type type,
2947 struct netlink_ext_ack *extack);
2948int netdev_offload_xstats_disable(struct net_device *dev,
2949 enum netdev_offload_xstats_type type);
2950bool netdev_offload_xstats_enabled(const struct net_device *dev,
2951 enum netdev_offload_xstats_type type);
2952int netdev_offload_xstats_get(struct net_device *dev,
2953 enum netdev_offload_xstats_type type,
2954 struct rtnl_hw_stats64 *stats, bool *used,
2955 struct netlink_ext_ack *extack);
2956void
2957netdev_offload_xstats_report_delta(struct netdev_notifier_offload_xstats_rd *rd,
2958 const struct rtnl_hw_stats64 *stats);
2959void
2960netdev_offload_xstats_report_used(struct netdev_notifier_offload_xstats_ru *ru);
2961void netdev_offload_xstats_push_delta(struct net_device *dev,
2962 enum netdev_offload_xstats_type type,
2963 const struct rtnl_hw_stats64 *stats);
2964
75538c2b
CW
2965static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2966 struct net_device *dev)
2967{
2968 info->dev = dev;
51d0c047 2969 info->extack = NULL;
75538c2b
CW
2970}
2971
351638e7
JP
2972static inline struct net_device *
2973netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2974{
2975 return info->dev;
2976}
2977
51d0c047
DA
2978static inline struct netlink_ext_ack *
2979netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
2980{
2981 return info->extack;
2982}
2983
f629d208 2984int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
88c0a6b5
VO
2985int call_netdevice_notifiers_info(unsigned long val,
2986 struct netdev_notifier_info *info);
dcfe1421 2987
1da177e4
LT
2988extern rwlock_t dev_base_lock; /* Device list lock */
2989
881d966b
EB
2990#define for_each_netdev(net, d) \
2991 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
dcbccbd4
EB
2992#define for_each_netdev_reverse(net, d) \
2993 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
c6d14c84
ED
2994#define for_each_netdev_rcu(net, d) \
2995 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
881d966b
EB
2996#define for_each_netdev_safe(net, d, n) \
2997 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2998#define for_each_netdev_continue(net, d) \
2999 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
afa0df59
JP
3000#define for_each_netdev_continue_reverse(net, d) \
3001 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
3002 dev_list)
254245d2 3003#define for_each_netdev_continue_rcu(net, d) \
3004 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
8a7fbfab 3005#define for_each_netdev_in_bond_rcu(bond, slave) \
3006 for_each_netdev_rcu(&init_net, slave) \
4ccce02e 3007 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
881d966b 3008#define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
7562f876 3009
84e00d9b
JK
3010#define for_each_netdev_dump(net, d, ifindex) \
3011 xa_for_each_start(&(net)->dev_by_index, (ifindex), (d), (ifindex))
3012
a050c33f
DL
3013static inline struct net_device *next_net_device(struct net_device *dev)
3014{
3015 struct list_head *lh;
3016 struct net *net;
3017
c346dca1 3018 net = dev_net(dev);
a050c33f
DL
3019 lh = dev->dev_list.next;
3020 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3021}
3022
ce81b76a
ED
3023static inline struct net_device *next_net_device_rcu(struct net_device *dev)
3024{
3025 struct list_head *lh;
3026 struct net *net;
3027
3028 net = dev_net(dev);
ccf43438 3029 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
ce81b76a
ED
3030 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3031}
3032
a050c33f
DL
3033static inline struct net_device *first_net_device(struct net *net)
3034{
3035 return list_empty(&net->dev_base_head) ? NULL :
3036 net_device_entry(net->dev_base_head.next);
3037}
7562f876 3038
ccf43438
ED
3039static inline struct net_device *first_net_device_rcu(struct net *net)
3040{
3041 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
3042
3043 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3044}
3045
f629d208 3046int netdev_boot_setup_check(struct net_device *dev);
f629d208
JP
3047struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
3048 const char *hwaddr);
3049struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
f629d208
JP
3050void dev_add_pack(struct packet_type *pt);
3051void dev_remove_pack(struct packet_type *pt);
3052void __dev_remove_pack(struct packet_type *pt);
3053void dev_add_offload(struct packet_offload *po);
3054void dev_remove_offload(struct packet_offload *po);
f629d208 3055
a54acb3a 3056int dev_get_iflink(const struct net_device *dev);
fc4099f1 3057int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
ddb94eaf
PNA
3058int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
3059 struct net_device_path_stack *stack);
6c555490
WC
3060struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
3061 unsigned short mask);
f629d208
JP
3062struct net_device *dev_get_by_name(struct net *net, const char *name);
3063struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
3064struct net_device *__dev_get_by_name(struct net *net, const char *name);
75ea27d0 3065bool netdev_name_in_use(struct net *net, const char *name);
f629d208 3066int dev_alloc_name(struct net_device *dev, const char *name);
00f54e68 3067int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
7051b88a 3068void dev_close(struct net_device *dev);
3069void dev_close_many(struct list_head *head, bool unlink);
f629d208 3070void dev_disable_lro(struct net_device *dev);
0c4b51f0 3071int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
a4ea8a3d 3072u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
a350ecce 3073 struct net_device *sb_dev);
a4ea8a3d 3074u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
a350ecce 3075 struct net_device *sb_dev);
36ccdf85 3076
c526fd8f 3077int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev);
36ccdf85
BT
3078int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
3079
c526fd8f
PB
3080static inline int dev_queue_xmit(struct sk_buff *skb)
3081{
3082 return __dev_queue_xmit(skb, NULL);
3083}
3084
3085static inline int dev_queue_xmit_accel(struct sk_buff *skb,
3086 struct net_device *sb_dev)
3087{
3088 return __dev_queue_xmit(skb, sb_dev);
3089}
3090
36ccdf85
BT
3091static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
3092{
3093 int ret;
3094
3095 ret = __dev_direct_xmit(skb, queue_id);
3096 if (!dev_xmit_complete(ret))
3097 kfree_skb(skb);
3098 return ret;
3099}
3100
f629d208
JP
3101int register_netdevice(struct net_device *dev);
3102void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
3103void unregister_netdevice_many(struct list_head *head);
44a0873d
ED
3104static inline void unregister_netdevice(struct net_device *dev)
3105{
3106 unregister_netdevice_queue(dev, NULL);
3107}
3108
f629d208
JP
3109int netdev_refcnt_read(const struct net_device *dev);
3110void free_netdev(struct net_device *dev);
74d332c1 3111void netdev_freemem(struct net_device *dev);
f629d208 3112int init_dummy_netdev(struct net_device *dev);
937f1ba5 3113
cff9f12b
MG
3114struct net_device *netdev_get_xmit_slave(struct net_device *dev,
3115 struct sk_buff *skb,
3116 bool all_slaves);
719a402c
TT
3117struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
3118 struct sock *sk);
f629d208
JP
3119struct net_device *dev_get_by_index(struct net *net, int ifindex);
3120struct net_device *__dev_get_by_index(struct net *net, int ifindex);
70f7457a
JK
3121struct net_device *netdev_get_by_index(struct net *net, int ifindex,
3122 netdevice_tracker *tracker, gfp_t gfp);
3123struct net_device *netdev_get_by_name(struct net *net, const char *name,
3124 netdevice_tracker *tracker, gfp_t gfp);
f629d208 3125struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
90b602f8 3126struct net_device *dev_get_by_napi_id(unsigned int napi_id);
5f114163 3127
0c4e8581
SH
3128static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
3129 unsigned short type,
3b04ddde 3130 const void *daddr, const void *saddr,
95c96174 3131 unsigned int len)
0c4e8581 3132{
f1ecfd5d 3133 if (!dev->header_ops || !dev->header_ops->create)
0c4e8581 3134 return 0;
3b04ddde
SH
3135
3136 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
0c4e8581
SH
3137}
3138
b95cce35
SH
3139static inline int dev_parse_header(const struct sk_buff *skb,
3140 unsigned char *haddr)
3141{
3142 const struct net_device *dev = skb->dev;
3143
1b83336b 3144 if (!dev->header_ops || !dev->header_ops->parse)
b95cce35 3145 return 0;
3b04ddde 3146 return dev->header_ops->parse(skb, haddr);
b95cce35
SH
3147}
3148
e78b2915
MM
3149static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
3150{
3151 const struct net_device *dev = skb->dev;
3152
3153 if (!dev->header_ops || !dev->header_ops->parse_protocol)
3154 return 0;
3155 return dev->header_ops->parse_protocol(skb);
3156}
3157
2793a23a
WB
3158/* ll_header must have at least hard_header_len allocated */
3159static inline bool dev_validate_header(const struct net_device *dev,
3160 char *ll_header, int len)
3161{
3162 if (likely(len >= dev->hard_header_len))
3163 return true;
217e6fa2
WB
3164 if (len < dev->min_header_len)
3165 return false;
2793a23a
WB
3166
3167 if (capable(CAP_SYS_RAWIO)) {
3168 memset(ll_header + len, 0, dev->hard_header_len - len);
3169 return true;
3170 }
3171
3172 if (dev->header_ops && dev->header_ops->validate)
3173 return dev->header_ops->validate(ll_header, len);
3174
3175 return false;
3176}
3177
d5496990
EB
3178static inline bool dev_has_header(const struct net_device *dev)
3179{
3180 return dev->header_ops && dev->header_ops->create;
3181}
3182
1da177e4 3183/*
5e82b4b2 3184 * Incoming packets are placed on per-CPU queues
1da177e4 3185 */
d94d9fee 3186struct softnet_data {
1da177e4 3187 struct list_head poll_list;
6e7676c1 3188 struct sk_buff_head process_queue;
1da177e4 3189
dee42870 3190 /* stats */
cd7b5396
DM
3191 unsigned int processed;
3192 unsigned int time_squeeze;
fd793d89 3193#ifdef CONFIG_RPS
88751275 3194 struct softnet_data *rps_ipi_list;
4cdb1e2e 3195#endif
87eff2ec 3196
c59647c0 3197 bool in_net_rx_action;
87eff2ec
ED
3198 bool in_napi_threaded_poll;
3199
4cdb1e2e
ED
3200#ifdef CONFIG_NET_FLOW_LIMIT
3201 struct sd_flow_limit __rcu *flow_limit;
3202#endif
3203 struct Qdisc *output_queue;
3204 struct Qdisc **output_queue_tailp;
3205 struct sk_buff *completion_queue;
f53c7239
SK
3206#ifdef CONFIG_XFRM_OFFLOAD
3207 struct sk_buff_head xfrm_backlog;
3208#endif
97cdcf37
FW
3209 /* written and read only by owning cpu: */
3210 struct {
3211 u16 recursion;
3212 u8 more;
2f1e85b1
TZ
3213#ifdef CONFIG_NET_EGRESS
3214 u8 skip_txqueue;
3215#endif
97cdcf37 3216 } xmit;
4cdb1e2e 3217#ifdef CONFIG_RPS
501e7ef5
ED
3218 /* input_queue_head should be written by cpu owning this struct,
3219 * and only read by other cpus. Worth using a cache line.
3220 */
3221 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3222
3223 /* Elements below can be accessed between CPUs for RPS/RFS */
966a9671 3224 call_single_data_t csd ____cacheline_aligned_in_smp;
88751275
ED
3225 struct softnet_data *rps_ipi_next;
3226 unsigned int cpu;
76cc8b13 3227 unsigned int input_queue_tail;
1e94d72f 3228#endif
14d898f3 3229 unsigned int received_rps;
95c96174 3230 unsigned int dropped;
0a9627f2 3231 struct sk_buff_head input_pkt_queue;
bea3348e 3232 struct napi_struct backlog;
99bbc707 3233
68822bdf
ED
3234 /* Another possibly contended cache line */
3235 spinlock_t defer_lock ____cacheline_aligned_in_smp;
3236 int defer_count;
97e719a8 3237 int defer_ipi_scheduled;
68822bdf
ED
3238 struct sk_buff *defer_list;
3239 call_single_data_t defer_csd;
1da177e4
LT
3240};
3241
76cc8b13 3242static inline void input_queue_head_incr(struct softnet_data *sd)
fec5e652
TH
3243{
3244#ifdef CONFIG_RPS
76cc8b13
TH
3245 sd->input_queue_head++;
3246#endif
3247}
3248
3249static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3250 unsigned int *qtail)
3251{
3252#ifdef CONFIG_RPS
3253 *qtail = ++sd->input_queue_tail;
fec5e652
TH
3254#endif
3255}
3256
0a9627f2 3257DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
1da177e4 3258
97cdcf37
FW
3259static inline int dev_recursion_level(void)
3260{
28b05b92 3261 return this_cpu_read(softnet_data.xmit.recursion);
97cdcf37
FW
3262}
3263
fb7861d1 3264#define XMIT_RECURSION_LIMIT 8
97cdcf37
FW
3265static inline bool dev_xmit_recursion(void)
3266{
3267 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3268 XMIT_RECURSION_LIMIT);
3269}
3270
3271static inline void dev_xmit_recursion_inc(void)
3272{
3273 __this_cpu_inc(softnet_data.xmit.recursion);
3274}
3275
3276static inline void dev_xmit_recursion_dec(void)
3277{
3278 __this_cpu_dec(softnet_data.xmit.recursion);
3279}
3280
f629d208 3281void __netif_schedule(struct Qdisc *q);
46e5da40 3282void netif_schedule_queue(struct netdev_queue *txq);
86d804e1 3283
fd2ea0a7
DM
3284static inline void netif_tx_schedule_all(struct net_device *dev)
3285{
3286 unsigned int i;
3287
3288 for (i = 0; i < dev->num_tx_queues; i++)
3289 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3290}
3291
f9a7cbbf 3292static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
d29f749e 3293{
73466498 3294 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3295}
3296
bea3348e
SH
3297/**
3298 * netif_start_queue - allow transmit
3299 * @dev: network device
3300 *
3301 * Allow upper layers to call the device hard_start_xmit routine.
3302 */
1da177e4
LT
3303static inline void netif_start_queue(struct net_device *dev)
3304{
e8a0464c 3305 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3306}
3307
fd2ea0a7
DM
3308static inline void netif_tx_start_all_queues(struct net_device *dev)
3309{
3310 unsigned int i;
3311
3312 for (i = 0; i < dev->num_tx_queues; i++) {
3313 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3314 netif_tx_start_queue(txq);
3315 }
3316}
3317
46e5da40 3318void netif_tx_wake_queue(struct netdev_queue *dev_queue);
79d16385 3319
d29f749e
DJ
3320/**
3321 * netif_wake_queue - restart transmit
3322 * @dev: network device
3323 *
3324 * Allow upper layers to call the device hard_start_xmit routine.
3325 * Used for flow control when transmit resources are available.
3326 */
79d16385
DM
3327static inline void netif_wake_queue(struct net_device *dev)
3328{
e8a0464c 3329 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3330}
3331
fd2ea0a7
DM
3332static inline void netif_tx_wake_all_queues(struct net_device *dev)
3333{
3334 unsigned int i;
3335
3336 for (i = 0; i < dev->num_tx_queues; i++) {
3337 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3338 netif_tx_wake_queue(txq);
3339 }
3340}
3341
f9a7cbbf 3342static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
d29f749e 3343{
c91c46de 3344 /* Must be an atomic op see netif_txq_try_stop() */
73466498 3345 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3346}
3347
bea3348e
SH
3348/**
3349 * netif_stop_queue - stop transmitted packets
3350 * @dev: network device
3351 *
3352 * Stop upper layers calling the device hard_start_xmit routine.
3353 * Used for flow control when transmit resources are unavailable.
3354 */
1da177e4
LT
3355static inline void netif_stop_queue(struct net_device *dev)
3356{
e8a0464c 3357 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3358}
3359
a2029240 3360void netif_tx_stop_all_queues(struct net_device *dev);
fd2ea0a7 3361
4d29515f 3362static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
d29f749e 3363{
73466498 3364 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
d29f749e
DJ
3365}
3366
bea3348e
SH
3367/**
3368 * netif_queue_stopped - test if transmit queue is flowblocked
3369 * @dev: network device
3370 *
3371 * Test if transmit queue on device is currently unable to send.
3372 */
4d29515f 3373static inline bool netif_queue_stopped(const struct net_device *dev)
1da177e4 3374{
e8a0464c 3375 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
1da177e4
LT
3376}
3377
4d29515f 3378static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
c3f26a26 3379{
73466498
TH
3380 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3381}
3382
8e2f1a63
DB
3383static inline bool
3384netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
73466498
TH
3385{
3386 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3387}
3388
8e2f1a63
DB
3389static inline bool
3390netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3391{
3392 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3393}
3394
f57bac3c
VM
3395/**
3396 * netdev_queue_set_dql_min_limit - set dql minimum limit
3397 * @dev_queue: pointer to transmit queue
3398 * @min_limit: dql minimum limit
3399 *
3400 * Forces xmit_more() to return true until the minimum threshold
3401 * defined by @min_limit is reached (or until the tx queue is
3402 * empty). Warning: to be use with care, misuse will impact the
3403 * latency.
3404 */
3405static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
3406 unsigned int min_limit)
3407{
3408#ifdef CONFIG_BQL
3409 dev_queue->dql.min_limit = min_limit;
3410#endif
3411}
3412
53511453
ED
3413/**
3414 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3415 * @dev_queue: pointer to transmit queue
3416 *
3417 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
5e82b4b2 3418 * to give appropriate hint to the CPU.
53511453
ED
3419 */
3420static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3421{
3422#ifdef CONFIG_BQL
3423 prefetchw(&dev_queue->dql.num_queued);
3424#endif
3425}
3426
3427/**
3428 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3429 * @dev_queue: pointer to transmit queue
3430 *
3431 * BQL enabled drivers might use this helper in their TX completion path,
5e82b4b2 3432 * to give appropriate hint to the CPU.
53511453
ED
3433 */
3434static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3435{
3436#ifdef CONFIG_BQL
3437 prefetchw(&dev_queue->dql.limit);
3438#endif
3439}
3440
977f1aa5
ED
3441/**
3442 * netdev_tx_sent_queue - report the number of bytes queued to a given tx queue
3443 * @dev_queue: network device queue
3444 * @bytes: number of bytes queued to the device queue
3445 *
3446 * Report the number of bytes queued for sending/completion to the network
3447 * device hardware queue. @bytes should be a good approximation and should
3448 * exactly match netdev_completed_queue() @bytes.
3449 * This is typically called once per packet, from ndo_start_xmit().
3450 */
c5d67bd7
TH
3451static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3452 unsigned int bytes)
3453{
114cf580
TH
3454#ifdef CONFIG_BQL
3455 dql_queued(&dev_queue->dql, bytes);
b37c0fbe
AD
3456
3457 if (likely(dql_avail(&dev_queue->dql) >= 0))
3458 return;
3459
3460 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3461
3462 /*
3463 * The XOFF flag must be set before checking the dql_avail below,
3464 * because in netdev_tx_completed_queue we update the dql_completed
3465 * before checking the XOFF flag.
3466 */
3467 smp_mb();
3468
3469 /* check again in case another CPU has just made room avail */
3470 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3471 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
114cf580 3472#endif
c5d67bd7
TH
3473}
3474
3e59020a
ED
3475/* Variant of netdev_tx_sent_queue() for drivers that are aware
3476 * that they should not test BQL status themselves.
3477 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3478 * skb of a batch.
3479 * Returns true if the doorbell must be used to kick the NIC.
3480 */
3481static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3482 unsigned int bytes,
3483 bool xmit_more)
3484{
3485 if (xmit_more) {
3486#ifdef CONFIG_BQL
3487 dql_queued(&dev_queue->dql, bytes);
3488#endif
3489 return netif_tx_queue_stopped(dev_queue);
3490 }
3491 netdev_tx_sent_queue(dev_queue, bytes);
3492 return true;
3493}
3494
0042d0c8 3495/**
977f1aa5
ED
3496 * netdev_sent_queue - report the number of bytes queued to hardware
3497 * @dev: network device
3498 * @bytes: number of bytes queued to the hardware device queue
0042d0c8 3499 *
977f1aa5
ED
3500 * Report the number of bytes queued for sending/completion to the network
3501 * device hardware queue#0. @bytes should be a good approximation and should
3502 * exactly match netdev_completed_queue() @bytes.
3503 * This is typically called once per packet, from ndo_start_xmit().
0042d0c8 3504 */
c5d67bd7
TH
3505static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3506{
3507 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3508}
3509
620344c4
HK
3510static inline bool __netdev_sent_queue(struct net_device *dev,
3511 unsigned int bytes,
3512 bool xmit_more)
3513{
3514 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3515 xmit_more);
3516}
3517
977f1aa5
ED
3518/**
3519 * netdev_tx_completed_queue - report number of packets/bytes at TX completion.
3520 * @dev_queue: network device queue
3521 * @pkts: number of packets (currently ignored)
3522 * @bytes: number of bytes dequeued from the device queue
3523 *
3524 * Must be called at most once per TX completion round (and not per
3525 * individual packet), so that BQL can adjust its limits appropriately.
3526 */
c5d67bd7 3527static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
95c96174 3528 unsigned int pkts, unsigned int bytes)
c5d67bd7 3529{
114cf580 3530#ifdef CONFIG_BQL
b37c0fbe
AD
3531 if (unlikely(!bytes))
3532 return;
3533
3534 dql_completed(&dev_queue->dql, bytes);
3535
3536 /*
3537 * Without the memory barrier there is a small possiblity that
3538 * netdev_tx_sent_queue will miss the update and cause the queue to
3539 * be stopped forever
3540 */
301f227f 3541 smp_mb(); /* NOTE: netdev_txq_completed_mb() assumes this exists */
b37c0fbe 3542
f3acd33d 3543 if (unlikely(dql_avail(&dev_queue->dql) < 0))
b37c0fbe
AD
3544 return;
3545
3546 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3547 netif_schedule_queue(dev_queue);
114cf580 3548#endif
c5d67bd7
TH
3549}
3550
0042d0c8
FF
3551/**
3552 * netdev_completed_queue - report bytes and packets completed by device
3553 * @dev: network device
3554 * @pkts: actual number of packets sent over the medium
3555 * @bytes: actual number of bytes sent over the medium
3556 *
3557 * Report the number of bytes and packets transmitted by the network device
3558 * hardware queue over the physical medium, @bytes must exactly match the
3559 * @bytes amount passed to netdev_sent_queue()
3560 */
c5d67bd7 3561static inline void netdev_completed_queue(struct net_device *dev,
95c96174 3562 unsigned int pkts, unsigned int bytes)
c5d67bd7
TH
3563{
3564 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3565}
3566
3567static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3568{
114cf580 3569#ifdef CONFIG_BQL
5c490354 3570 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
114cf580
TH
3571 dql_reset(&q->dql);
3572#endif
c5d67bd7
TH
3573}
3574
0042d0c8
FF
3575/**
3576 * netdev_reset_queue - reset the packets and bytes count of a network device
3577 * @dev_queue: network device
3578 *
3579 * Reset the bytes and packet count of a network device and clear the
3580 * software flow control OFF bit for this network device
3581 */
c5d67bd7
TH
3582static inline void netdev_reset_queue(struct net_device *dev_queue)
3583{
3584 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
c3f26a26
DM
3585}
3586
b9507bda
DB
3587/**
3588 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3589 * @dev: network device
3590 * @queue_index: given tx queue index
3591 *
3592 * Returns 0 if given tx queue index >= number of device tx queues,
3593 * otherwise returns the originally passed tx queue index.
3594 */
3595static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3596{
3597 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3598 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3599 dev->name, queue_index,
3600 dev->real_num_tx_queues);
3601 return 0;
3602 }
3603
3604 return queue_index;
3605}
3606
bea3348e
SH
3607/**
3608 * netif_running - test if up
3609 * @dev: network device
3610 *
3611 * Test if the device has been brought up.
3612 */
4d29515f 3613static inline bool netif_running(const struct net_device *dev)
1da177e4
LT
3614{
3615 return test_bit(__LINK_STATE_START, &dev->state);
3616}
3617
f25f4e44 3618/*
5e82b4b2 3619 * Routines to manage the subqueues on a device. We only need start,
f25f4e44
PWJ
3620 * stop, and a check if it's stopped. All other device management is
3621 * done at the overall netdevice level.
3622 * Also test the device if we're multiqueue.
3623 */
bea3348e
SH
3624
3625/**
3626 * netif_start_subqueue - allow sending packets on subqueue
3627 * @dev: network device
3628 * @queue_index: sub queue index
3629 *
3630 * Start individual transmit queue of a device with multiple transmit queues.
3631 */
f25f4e44
PWJ
3632static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3633{
fd2ea0a7 3634 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3635
3636 netif_tx_start_queue(txq);
f25f4e44
PWJ
3637}
3638
bea3348e
SH
3639/**
3640 * netif_stop_subqueue - stop sending packets on subqueue
3641 * @dev: network device
3642 * @queue_index: sub queue index
3643 *
3644 * Stop individual transmit queue of a device with multiple transmit queues.
3645 */
f25f4e44
PWJ
3646static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3647{
fd2ea0a7 3648 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f 3649 netif_tx_stop_queue(txq);
f25f4e44
PWJ
3650}
3651
bea3348e 3652/**
270f3385 3653 * __netif_subqueue_stopped - test status of subqueue
bea3348e
SH
3654 * @dev: network device
3655 * @queue_index: sub queue index
3656 *
3657 * Check individual transmit queue of a device with multiple transmit queues.
3658 */
4d29515f
DM
3659static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3660 u16 queue_index)
f25f4e44 3661{
fd2ea0a7 3662 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
7b3d3e4f
KK
3663
3664 return netif_tx_queue_stopped(txq);
f25f4e44
PWJ
3665}
3666
270f3385
MCC
3667/**
3668 * netif_subqueue_stopped - test status of subqueue
3669 * @dev: network device
3670 * @skb: sub queue buffer pointer
3671 *
3672 * Check individual transmit queue of a device with multiple transmit queues.
3673 */
4d29515f
DM
3674static inline bool netif_subqueue_stopped(const struct net_device *dev,
3675 struct sk_buff *skb)
668f895a
PE
3676{
3677 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3678}
bea3348e 3679
738b35cc
FF
3680/**
3681 * netif_wake_subqueue - allow sending packets on subqueue
3682 * @dev: network device
3683 * @queue_index: sub queue index
3684 *
3685 * Resume individual transmit queue of a device with multiple transmit queues.
3686 */
3687static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3688{
3689 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3690
3691 netif_tx_wake_queue(txq);
3692}
f25f4e44 3693
537c00de 3694#ifdef CONFIG_XPS
53af53ae 3695int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
f629d208 3696 u16 index);
80d19669 3697int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
044ab86d 3698 u16 index, enum xps_map_type type);
80d19669
AN
3699
3700/**
3701 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3702 * @j: CPU/Rx queue index
3703 * @mask: bitmask of all cpus/rx queues
3704 * @nr_bits: number of bits in the bitmask
3705 *
3706 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3707 */
3708static inline bool netif_attr_test_mask(unsigned long j,
3709 const unsigned long *mask,
3710 unsigned int nr_bits)
3711{
3712 cpu_max_bits_warn(j, nr_bits);
3713 return test_bit(j, mask);
3714}
3715
3716/**
3717 * netif_attr_test_online - Test for online CPU/Rx queue
3718 * @j: CPU/Rx queue index
3719 * @online_mask: bitmask for CPUs/Rx queues that are online
3720 * @nr_bits: number of bits in the bitmask
3721 *
3722 * Returns true if a CPU/Rx queue is online.
3723 */
3724static inline bool netif_attr_test_online(unsigned long j,
3725 const unsigned long *online_mask,
3726 unsigned int nr_bits)
3727{
3728 cpu_max_bits_warn(j, nr_bits);
3729
3730 if (online_mask)
3731 return test_bit(j, online_mask);
3732
3733 return (j < nr_bits);
3734}
3735
3736/**
3737 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3738 * @n: CPU/Rx queue index
3739 * @srcp: the cpumask/Rx queue mask pointer
3740 * @nr_bits: number of bits in the bitmask
3741 *
3742 * Returns >= nr_bits if no further CPUs/Rx queues set.
3743 */
3744static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3745 unsigned int nr_bits)
3746{
fc8695eb
JK
3747 /* -1 is a legal arg here. */
3748 if (n != -1)
3749 cpu_max_bits_warn(n, nr_bits);
80d19669
AN
3750
3751 if (srcp)
3752 return find_next_bit(srcp, nr_bits, n + 1);
3753
3754 return n + 1;
3755}
3756
3757/**
a1fa83bd 3758 * netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
80d19669
AN
3759 * @n: CPU/Rx queue index
3760 * @src1p: the first CPUs/Rx queues mask pointer
3761 * @src2p: the second CPUs/Rx queues mask pointer
3762 * @nr_bits: number of bits in the bitmask
3763 *
3764 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3765 */
3766static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3767 const unsigned long *src2p,
3768 unsigned int nr_bits)
3769{
fc8695eb
JK
3770 /* -1 is a legal arg here. */
3771 if (n != -1)
3772 cpu_max_bits_warn(n, nr_bits);
80d19669
AN
3773
3774 if (src1p && src2p)
3775 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3776 else if (src1p)
3777 return find_next_bit(src1p, nr_bits, n + 1);
3778 else if (src2p)
3779 return find_next_bit(src2p, nr_bits, n + 1);
3780
3781 return n + 1;
3782}
537c00de
AD
3783#else
3784static inline int netif_set_xps_queue(struct net_device *dev,
3573540c 3785 const struct cpumask *mask,
537c00de
AD
3786 u16 index)
3787{
3788 return 0;
3789}
c9fbb2d2
KK
3790
3791static inline int __netif_set_xps_queue(struct net_device *dev,
3792 const unsigned long *mask,
044ab86d 3793 u16 index, enum xps_map_type type)
c9fbb2d2
KK
3794{
3795 return 0;
3796}
537c00de
AD
3797#endif
3798
bea3348e
SH
3799/**
3800 * netif_is_multiqueue - test if device has multiple transmit queues
3801 * @dev: network device
3802 *
3803 * Check if device has multiple transmit queues
bea3348e 3804 */
4d29515f 3805static inline bool netif_is_multiqueue(const struct net_device *dev)
f25f4e44 3806{
a02cec21 3807 return dev->num_tx_queues > 1;
f25f4e44 3808}
1da177e4 3809
f629d208 3810int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
f0796d5c 3811
a953be53 3812#ifdef CONFIG_SYSFS
f629d208 3813int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
62fe0b40
BH
3814#else
3815static inline int netif_set_real_num_rx_queues(struct net_device *dev,
c29c2ebd 3816 unsigned int rxqs)
62fe0b40 3817{
c29c2ebd 3818 dev->real_num_rx_queues = rxqs;
62fe0b40
BH
3819 return 0;
3820}
3821#endif
271e5b7d
JK
3822int netif_set_real_num_queues(struct net_device *dev,
3823 unsigned int txq, unsigned int rxq);
62fe0b40 3824
f629d208 3825int netif_get_num_default_rss_queues(void);
16917b87 3826
40bbae58
ED
3827void dev_kfree_skb_irq_reason(struct sk_buff *skb, enum skb_drop_reason reason);
3828void dev_kfree_skb_any_reason(struct sk_buff *skb, enum skb_drop_reason reason);
1da177e4 3829
e6247027
ED
3830/*
3831 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3832 * interrupt context or with hardware interrupts being disabled.
afa79d08 3833 * (in_hardirq() || irqs_disabled())
e6247027
ED
3834 *
3835 * We provide four helpers that can be used in following contexts :
3836 *
3837 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3838 * replacing kfree_skb(skb)
3839 *
3840 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3841 * Typically used in place of consume_skb(skb) in TX completion path
3842 *
3843 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3844 * replacing kfree_skb(skb)
3845 *
3846 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3847 * and consumed a packet. Used in place of consume_skb(skb)
1da177e4 3848 */
e6247027
ED
3849static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3850{
40bbae58 3851 dev_kfree_skb_irq_reason(skb, SKB_DROP_REASON_NOT_SPECIFIED);
e6247027
ED
3852}
3853
3854static inline void dev_consume_skb_irq(struct sk_buff *skb)
3855{
40bbae58 3856 dev_kfree_skb_irq_reason(skb, SKB_CONSUMED);
e6247027
ED
3857}
3858
3859static inline void dev_kfree_skb_any(struct sk_buff *skb)
3860{
40bbae58 3861 dev_kfree_skb_any_reason(skb, SKB_DROP_REASON_NOT_SPECIFIED);
e6247027
ED
3862}
3863
3864static inline void dev_consume_skb_any(struct sk_buff *skb)
3865{
40bbae58 3866 dev_kfree_skb_any_reason(skb, SKB_CONSUMED);
e6247027 3867}
1da177e4 3868
fe21cb91
KKD
3869u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp,
3870 struct bpf_prog *xdp_prog);
7c497478
JW
3871void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3872int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
f629d208 3873int netif_rx(struct sk_buff *skb);
baebdf48
SAS
3874int __netif_rx(struct sk_buff *skb);
3875
04eb4489 3876int netif_receive_skb(struct sk_buff *skb);
1c601d82 3877int netif_receive_skb_core(struct sk_buff *skb);
587652bb 3878void netif_receive_skb_list_internal(struct list_head *head);
f6ad8c1b 3879void netif_receive_skb_list(struct list_head *head);
f629d208
JP
3880gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3881void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3882struct sk_buff *napi_get_frags(struct napi_struct *napi);
dbae2b06 3883void napi_get_frags_check(struct napi_struct *napi);
f629d208 3884gro_result_t napi_gro_frags(struct napi_struct *napi);
bf5a755f
JC
3885struct packet_offload *gro_find_receive_by_type(__be16 type);
3886struct packet_offload *gro_find_complete_by_type(__be16 type);
76620aaf
HX
3887
3888static inline void napi_free_frags(struct napi_struct *napi)
3889{
3890 kfree_skb(napi->skb);
3891 napi->skb = NULL;
3892}
3893
24b27fc4 3894bool netdev_is_rx_handler_busy(struct net_device *dev);
f629d208
JP
3895int netdev_rx_handler_register(struct net_device *dev,
3896 rx_handler_func_t *rx_handler,
3897 void *rx_handler_data);
3898void netdev_rx_handler_unregister(struct net_device *dev);
3899
3900bool dev_valid_name(const char *name);
d0efb162
PC
3901static inline bool is_socket_ioctl_cmd(unsigned int cmd)
3902{
3903 return _IOC_TYPE(cmd) == SOCK_IOC_TYPE;
3904}
29c49648
AB
3905int get_user_ifreq(struct ifreq *ifr, void __user **ifrdata, void __user *arg);
3906int put_user_ifreq(struct ifreq *ifr, void __user *arg);
44c02a2c 3907int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
a554bf96 3908 void __user *data, bool *need_copyout);
876f0bf9 3909int dev_ifconf(struct net *net, struct ifconf __user *ifc);
e47d01fe
MG
3910int generic_hwtstamp_get_lower(struct net_device *dev,
3911 struct kernel_hwtstamp_config *kernel_cfg);
3912int generic_hwtstamp_set_lower(struct net_device *dev,
3913 struct kernel_hwtstamp_config *kernel_cfg,
3914 struct netlink_ext_ack *extack);
a554bf96 3915int dev_ethtool(struct net *net, struct ifreq *ifr, void __user *userdata);
f629d208 3916unsigned int dev_get_flags(const struct net_device *);
6d040321
PM
3917int __dev_change_flags(struct net_device *dev, unsigned int flags,
3918 struct netlink_ext_ack *extack);
567c5e13
PM
3919int dev_change_flags(struct net_device *dev, unsigned int flags,
3920 struct netlink_ext_ack *extack);
f629d208 3921int dev_set_alias(struct net_device *, const char *, size_t);
6c557001 3922int dev_get_alias(const struct net_device *, char *, size_t);
0854fa82
AV
3923int __dev_change_net_namespace(struct net_device *dev, struct net *net,
3924 const char *pat, int new_ifindex);
3925static inline
eeb85a14 3926int dev_change_net_namespace(struct net_device *dev, struct net *net,
0854fa82
AV
3927 const char *pat)
3928{
3929 return __dev_change_net_namespace(dev, net, pat, 0);
3930}
f51048c3 3931int __dev_set_mtu(struct net_device *, int);
f629d208 3932int dev_set_mtu(struct net_device *, int);
d59cdf94
PM
3933int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
3934 struct netlink_ext_ack *extack);
3a37a963
PM
3935int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
3936 struct netlink_ext_ack *extack);
3b23a32a
CW
3937int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
3938 struct netlink_ext_ack *extack);
3939int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
d6abc596
FF
3940int dev_get_port_parent_id(struct net_device *dev,
3941 struct netdev_phys_item_id *ppid, bool recurse);
3942bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
f53c7239 3943struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
ce93718f
DM
3944struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3945 struct netdev_queue *txq, int *ret);
d67b9cd2 3946
aa8d3a71 3947int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
879af96f 3948u8 dev_xdp_prog_count(struct net_device *dev);
7f0a8382 3949u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
aa8d3a71 3950
a0265d28 3951int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
f629d208 3952int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
5f7d5728 3953int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
f4b05d27
NA
3954bool is_skb_forwardable(const struct net_device *dev,
3955 const struct sk_buff *skb);
1da177e4 3956
5f7d5728
JDB
3957static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
3958 const struct sk_buff *skb,
3959 const bool check_mtu)
3960{
3961 const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
3962 unsigned int len;
3963
3964 if (!(dev->flags & IFF_UP))
3965 return false;
3966
3967 if (!check_mtu)
3968 return true;
3969
3970 len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
3971 if (skb->len <= len)
3972 return true;
3973
3974 /* if TSO is enabled, we don't care about the length as the packet
3975 * could be forwarded without being segmented before
3976 */
3977 if (skb_is_gso(skb))
3978 return true;
3979
3980 return false;
3981}
3982
6510ea97 3983struct net_device_core_stats __percpu *netdev_core_stats_alloc(struct net_device *dev);
625788b5 3984
6510ea97 3985static inline struct net_device_core_stats __percpu *dev_core_stats(struct net_device *dev)
625788b5
ED
3986{
3987 /* This READ_ONCE() pairs with the write in netdev_core_stats_alloc() */
3988 struct net_device_core_stats __percpu *p = READ_ONCE(dev->core_stats);
3989
3990 if (likely(p))
6510ea97 3991 return p;
625788b5
ED
3992
3993 return netdev_core_stats_alloc(dev);
3994}
3995
3996#define DEV_CORE_STATS_INC(FIELD) \
3997static inline void dev_core_stats_##FIELD##_inc(struct net_device *dev) \
3998{ \
6510ea97 3999 struct net_device_core_stats __percpu *p; \
fc93db15 4000 \
fc93db15 4001 p = dev_core_stats(dev); \
625788b5 4002 if (p) \
6510ea97 4003 this_cpu_inc(p->FIELD); \
625788b5
ED
4004}
4005DEV_CORE_STATS_INC(rx_dropped)
4006DEV_CORE_STATS_INC(tx_dropped)
4007DEV_CORE_STATS_INC(rx_nohandler)
794c24e9 4008DEV_CORE_STATS_INC(rx_otherhost_dropped)
625788b5 4009
4e3264d2 4010static __always_inline int ____dev_forward_skb(struct net_device *dev,
5f7d5728
JDB
4011 struct sk_buff *skb,
4012 const bool check_mtu)
4e3264d2
MKL
4013{
4014 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
5f7d5728 4015 unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
625788b5 4016 dev_core_stats_rx_dropped_inc(dev);
4e3264d2
MKL
4017 kfree_skb(skb);
4018 return NET_RX_DROP;
4019 }
4020
ff70202b 4021 skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
4e3264d2
MKL
4022 skb->priority = 0;
4023 return 0;
4024}
4025
9f9a742d 4026bool dev_nit_active(struct net_device *dev);
74b20582
DA
4027void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
4028
4c6c11ea 4029static inline void __dev_put(struct net_device *dev)
1da177e4 4030{
b37a4668 4031 if (dev) {
919067cc 4032#ifdef CONFIG_PCPU_DEV_REFCNT
b37a4668 4033 this_cpu_dec(*dev->pcpu_refcnt);
919067cc 4034#else
b37a4668 4035 refcount_dec(&dev->dev_refcnt);
919067cc 4036#endif
b37a4668 4037 }
1da177e4
LT
4038}
4039
4c6c11ea 4040static inline void __dev_hold(struct net_device *dev)
15333061 4041{
b37a4668 4042 if (dev) {
919067cc 4043#ifdef CONFIG_PCPU_DEV_REFCNT
b37a4668 4044 this_cpu_inc(*dev->pcpu_refcnt);
919067cc 4045#else
b37a4668 4046 refcount_inc(&dev->dev_refcnt);
919067cc 4047#endif
b37a4668 4048 }
15333061 4049}
1da177e4 4050
4c6c11ea
ED
4051static inline void __netdev_tracker_alloc(struct net_device *dev,
4052 netdevice_tracker *tracker,
4053 gfp_t gfp)
4054{
4055#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4056 ref_tracker_alloc(&dev->refcnt_tracker, tracker, gfp);
4057#endif
4058}
4059
4060/* netdev_tracker_alloc() can upgrade a prior untracked reference
4061 * taken by dev_get_by_name()/dev_get_by_index() to a tracked one.
4062 */
4d92b95f
ED
4063static inline void netdev_tracker_alloc(struct net_device *dev,
4064 netdevice_tracker *tracker, gfp_t gfp)
4065{
4066#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4c6c11ea
ED
4067 refcount_dec(&dev->refcnt_tracker.no_tracker);
4068 __netdev_tracker_alloc(dev, tracker, gfp);
4d92b95f
ED
4069#endif
4070}
4071
4072static inline void netdev_tracker_free(struct net_device *dev,
4073 netdevice_tracker *tracker)
4074{
4075#ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4076 ref_tracker_free(&dev->refcnt_tracker, tracker);
4077#endif
4078}
4079
d62607c3
JK
4080static inline void netdev_hold(struct net_device *dev,
4081 netdevice_tracker *tracker, gfp_t gfp)
4d92b95f
ED
4082{
4083 if (dev) {
4c6c11ea
ED
4084 __dev_hold(dev);
4085 __netdev_tracker_alloc(dev, tracker, gfp);
4d92b95f
ED
4086 }
4087}
4088
d62607c3
JK
4089static inline void netdev_put(struct net_device *dev,
4090 netdevice_tracker *tracker)
4d92b95f
ED
4091{
4092 if (dev) {
4093 netdev_tracker_free(dev, tracker);
4c6c11ea 4094 __dev_put(dev);
4d92b95f
ED
4095 }
4096}
4097
4c6c11ea
ED
4098/**
4099 * dev_hold - get reference to device
4100 * @dev: network device
4101 *
4102 * Hold reference to device to keep it from being freed.
d62607c3 4103 * Try using netdev_hold() instead.
4c6c11ea
ED
4104 */
4105static inline void dev_hold(struct net_device *dev)
4106{
d62607c3 4107 netdev_hold(dev, NULL, GFP_ATOMIC);
4c6c11ea
ED
4108}
4109
4110/**
4111 * dev_put - release reference to device
4112 * @dev: network device
4113 *
4114 * Release reference to device to allow it to be freed.
d62607c3 4115 * Try using netdev_put() instead.
4c6c11ea
ED
4116 */
4117static inline void dev_put(struct net_device *dev)
4118{
d62607c3 4119 netdev_put(dev, NULL);
4c6c11ea
ED
4120}
4121
d62607c3
JK
4122static inline void netdev_ref_replace(struct net_device *odev,
4123 struct net_device *ndev,
4124 netdevice_tracker *tracker,
4125 gfp_t gfp)
9038c320 4126{
9038c320 4127 if (odev)
9280ac2e
ED
4128 netdev_tracker_free(odev, tracker);
4129
4c6c11ea
ED
4130 __dev_hold(ndev);
4131 __dev_put(odev);
9280ac2e 4132
9038c320 4133 if (ndev)
4c6c11ea 4134 __netdev_tracker_alloc(ndev, tracker, gfp);
9038c320
ED
4135}
4136
1da177e4
LT
4137/* Carrier loss detection, dial on demand. The functions netif_carrier_on
4138 * and _off may be called from IRQ context, but it is caller
4139 * who is responsible for serialization of these calls.
b00055aa
SR
4140 *
4141 * The name carrier is inappropriate, these functions should really be
4142 * called netif_lowerlayer_*() because they represent the state of any
4143 * kind of lower layer not just hardware media.
1da177e4 4144 */
f629d208 4145void linkwatch_fire_event(struct net_device *dev);
1da177e4 4146
bea3348e
SH
4147/**
4148 * netif_carrier_ok - test if carrier present
4149 * @dev: network device
4150 *
4151 * Check if carrier is present on device
4152 */
4d29515f 4153static inline bool netif_carrier_ok(const struct net_device *dev)
1da177e4
LT
4154{
4155 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
4156}
4157
f629d208 4158unsigned long dev_trans_start(struct net_device *dev);
9d21493b 4159
f629d208 4160void __netdev_watchdog_up(struct net_device *dev);
1da177e4 4161
f629d208 4162void netif_carrier_on(struct net_device *dev);
f629d208 4163void netif_carrier_off(struct net_device *dev);
490dceca 4164void netif_carrier_event(struct net_device *dev);
1da177e4 4165
bea3348e
SH
4166/**
4167 * netif_dormant_on - mark device as dormant.
4168 * @dev: network device
4169 *
4170 * Mark device as dormant (as per RFC2863).
4171 *
4172 * The dormant state indicates that the relevant interface is not
4173 * actually in a condition to pass packets (i.e., it is not 'up') but is
4174 * in a "pending" state, waiting for some external event. For "on-
4175 * demand" interfaces, this new state identifies the situation where the
4176 * interface is waiting for events to place it in the up state.
bea3348e 4177 */
b00055aa
SR
4178static inline void netif_dormant_on(struct net_device *dev)
4179{
4180 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
4181 linkwatch_fire_event(dev);
4182}
4183
bea3348e
SH
4184/**
4185 * netif_dormant_off - set device as not dormant.
4186 * @dev: network device
4187 *
4188 * Device is not in dormant state.
4189 */
b00055aa
SR
4190static inline void netif_dormant_off(struct net_device *dev)
4191{
4192 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
4193 linkwatch_fire_event(dev);
4194}
4195
bea3348e 4196/**
8ecbc40a 4197 * netif_dormant - test if device is dormant
bea3348e
SH
4198 * @dev: network device
4199 *
8ecbc40a 4200 * Check if device is dormant.
bea3348e 4201 */
4d29515f 4202static inline bool netif_dormant(const struct net_device *dev)
b00055aa
SR
4203{
4204 return test_bit(__LINK_STATE_DORMANT, &dev->state);
4205}
4206
4207
eec517cd
AL
4208/**
4209 * netif_testing_on - mark device as under test.
4210 * @dev: network device
4211 *
4212 * Mark device as under test (as per RFC2863).
4213 *
4214 * The testing state indicates that some test(s) must be performed on
4215 * the interface. After completion, of the test, the interface state
4216 * will change to up, dormant, or down, as appropriate.
4217 */
4218static inline void netif_testing_on(struct net_device *dev)
4219{
4220 if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
4221 linkwatch_fire_event(dev);
4222}
4223
4224/**
4225 * netif_testing_off - set device as not under test.
4226 * @dev: network device
4227 *
4228 * Device is not in testing state.
4229 */
4230static inline void netif_testing_off(struct net_device *dev)
4231{
4232 if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
4233 linkwatch_fire_event(dev);
4234}
4235
4236/**
4237 * netif_testing - test if device is under test
4238 * @dev: network device
4239 *
4240 * Check if device is under test
4241 */
4242static inline bool netif_testing(const struct net_device *dev)
4243{
4244 return test_bit(__LINK_STATE_TESTING, &dev->state);
4245}
4246
4247
bea3348e
SH
4248/**
4249 * netif_oper_up - test if device is operational
4250 * @dev: network device
4251 *
4252 * Check if carrier is operational
4253 */
4d29515f 4254static inline bool netif_oper_up(const struct net_device *dev)
d94d9fee 4255{
b00055aa
SR
4256 return (dev->operstate == IF_OPER_UP ||
4257 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
4258}
4259
bea3348e
SH
4260/**
4261 * netif_device_present - is device available or removed
4262 * @dev: network device
4263 *
4264 * Check if device has not been removed from system.
4265 */
7a126a43 4266static inline bool netif_device_present(const struct net_device *dev)
1da177e4
LT
4267{
4268 return test_bit(__LINK_STATE_PRESENT, &dev->state);
4269}
4270
f629d208 4271void netif_device_detach(struct net_device *dev);
1da177e4 4272
f629d208 4273void netif_device_attach(struct net_device *dev);
1da177e4
LT
4274
4275/*
4276 * Network interface message level settings
4277 */
1da177e4
LT
4278
4279enum {
6a94b8cc
MK
4280 NETIF_MSG_DRV_BIT,
4281 NETIF_MSG_PROBE_BIT,
4282 NETIF_MSG_LINK_BIT,
4283 NETIF_MSG_TIMER_BIT,
4284 NETIF_MSG_IFDOWN_BIT,
4285 NETIF_MSG_IFUP_BIT,
4286 NETIF_MSG_RX_ERR_BIT,
4287 NETIF_MSG_TX_ERR_BIT,
4288 NETIF_MSG_TX_QUEUED_BIT,
4289 NETIF_MSG_INTR_BIT,
4290 NETIF_MSG_TX_DONE_BIT,
4291 NETIF_MSG_RX_STATUS_BIT,
4292 NETIF_MSG_PKTDATA_BIT,
4293 NETIF_MSG_HW_BIT,
4294 NETIF_MSG_WOL_BIT,
4295
4296 /* When you add a new bit above, update netif_msg_class_names array
4297 * in net/ethtool/common.c
4298 */
4299 NETIF_MSG_CLASS_COUNT,
1da177e4 4300};
6a94b8cc
MK
4301/* Both ethtool_ops interface and internal driver implementation use u32 */
4302static_assert(NETIF_MSG_CLASS_COUNT <= 32);
4303
4304#define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
4305#define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
4306
4307#define NETIF_MSG_DRV __NETIF_MSG(DRV)
4308#define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
4309#define NETIF_MSG_LINK __NETIF_MSG(LINK)
4310#define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
4311#define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
4312#define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
4313#define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
4314#define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
4315#define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
4316#define NETIF_MSG_INTR __NETIF_MSG(INTR)
4317#define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
4318#define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
4319#define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
4320#define NETIF_MSG_HW __NETIF_MSG(HW)
4321#define NETIF_MSG_WOL __NETIF_MSG(WOL)
1da177e4
LT
4322
4323#define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
4324#define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
4325#define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
4326#define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
4327#define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
4328#define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
4329#define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
4330#define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
4331#define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
4332#define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
4333#define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
4334#define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
4335#define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
4336#define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
4337#define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
4338
4339static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
4340{
4341 /* use default */
4342 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
4343 return default_msg_enable_bits;
4344 if (debug_value == 0) /* no output */
4345 return 0;
4346 /* set low N bits */
f4d7b3e2 4347 return (1U << debug_value) - 1;
1da177e4
LT
4348}
4349
c773e847 4350static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
932ff279 4351{
c773e847 4352 spin_lock(&txq->_xmit_lock);
7a10d8c8
ED
4353 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4354 WRITE_ONCE(txq->xmit_lock_owner, cpu);
22dd7495
JHS
4355}
4356
5a717f4f
MT
4357static inline bool __netif_tx_acquire(struct netdev_queue *txq)
4358{
4359 __acquire(&txq->_xmit_lock);
4360 return true;
4361}
4362
4363static inline void __netif_tx_release(struct netdev_queue *txq)
4364{
4365 __release(&txq->_xmit_lock);
4366}
4367
fd2ea0a7
DM
4368static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
4369{
4370 spin_lock_bh(&txq->_xmit_lock);
7a10d8c8
ED
4371 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4372 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
fd2ea0a7
DM
4373}
4374
4d29515f 4375static inline bool __netif_tx_trylock(struct netdev_queue *txq)
c3f26a26 4376{
4d29515f 4377 bool ok = spin_trylock(&txq->_xmit_lock);
7a10d8c8
ED
4378
4379 if (likely(ok)) {
4380 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4381 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
4382 }
c3f26a26
DM
4383 return ok;
4384}
4385
4386static inline void __netif_tx_unlock(struct netdev_queue *txq)
4387{
7a10d8c8
ED
4388 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4389 WRITE_ONCE(txq->xmit_lock_owner, -1);
c3f26a26
DM
4390 spin_unlock(&txq->_xmit_lock);
4391}
4392
4393static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
4394{
7a10d8c8
ED
4395 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4396 WRITE_ONCE(txq->xmit_lock_owner, -1);
c3f26a26
DM
4397 spin_unlock_bh(&txq->_xmit_lock);
4398}
4399
5337824f
ED
4400/*
4401 * txq->trans_start can be read locklessly from dev_watchdog()
4402 */
08baf561
ED
4403static inline void txq_trans_update(struct netdev_queue *txq)
4404{
4405 if (txq->xmit_lock_owner != -1)
5337824f
ED
4406 WRITE_ONCE(txq->trans_start, jiffies);
4407}
4408
4409static inline void txq_trans_cond_update(struct netdev_queue *txq)
4410{
4411 unsigned long now = jiffies;
4412
4413 if (READ_ONCE(txq->trans_start) != now)
4414 WRITE_ONCE(txq->trans_start, now);
08baf561
ED
4415}
4416
ba162f8e
FW
4417/* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
4418static inline void netif_trans_update(struct net_device *dev)
4419{
9b36627a
FW
4420 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
4421
5337824f 4422 txq_trans_cond_update(txq);
ba162f8e
FW
4423}
4424
d29f749e
DJ
4425/**
4426 * netif_tx_lock - grab network device transmit lock
4427 * @dev: network device
d29f749e
DJ
4428 *
4429 * Get network device transmit lock
4430 */
dab8fe32 4431void netif_tx_lock(struct net_device *dev);
932ff279
HX
4432
4433static inline void netif_tx_lock_bh(struct net_device *dev)
4434{
e8a0464c
DM
4435 local_bh_disable();
4436 netif_tx_lock(dev);
932ff279
HX
4437}
4438
dab8fe32 4439void netif_tx_unlock(struct net_device *dev);
932ff279
HX
4440
4441static inline void netif_tx_unlock_bh(struct net_device *dev)
4442{
e8a0464c
DM
4443 netif_tx_unlock(dev);
4444 local_bh_enable();
932ff279
HX
4445}
4446
c773e847 4447#define HARD_TX_LOCK(dev, txq, cpu) { \
22dd7495 4448 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4449 __netif_tx_lock(txq, cpu); \
5a717f4f
MT
4450 } else { \
4451 __netif_tx_acquire(txq); \
22dd7495
JHS
4452 } \
4453}
4454
5efeac44
EB
4455#define HARD_TX_TRYLOCK(dev, txq) \
4456 (((dev->features & NETIF_F_LLTX) == 0) ? \
4457 __netif_tx_trylock(txq) : \
5a717f4f 4458 __netif_tx_acquire(txq))
5efeac44 4459
c773e847 4460#define HARD_TX_UNLOCK(dev, txq) { \
22dd7495 4461 if ((dev->features & NETIF_F_LLTX) == 0) { \
c773e847 4462 __netif_tx_unlock(txq); \
5a717f4f
MT
4463 } else { \
4464 __netif_tx_release(txq); \
22dd7495
JHS
4465 } \
4466}
4467
1da177e4
LT
4468static inline void netif_tx_disable(struct net_device *dev)
4469{
fd2ea0a7 4470 unsigned int i;
c3f26a26 4471 int cpu;
fd2ea0a7 4472
c3f26a26
DM
4473 local_bh_disable();
4474 cpu = smp_processor_id();
3aa6bce9 4475 spin_lock(&dev->tx_global_lock);
fd2ea0a7
DM
4476 for (i = 0; i < dev->num_tx_queues; i++) {
4477 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
c3f26a26
DM
4478
4479 __netif_tx_lock(txq, cpu);
fd2ea0a7 4480 netif_tx_stop_queue(txq);
c3f26a26 4481 __netif_tx_unlock(txq);
fd2ea0a7 4482 }
3aa6bce9 4483 spin_unlock(&dev->tx_global_lock);
c3f26a26 4484 local_bh_enable();
1da177e4
LT
4485}
4486
e308a5d8
DM
4487static inline void netif_addr_lock(struct net_device *dev)
4488{
1fc70edb 4489 unsigned char nest_level = 0;
e308a5d8 4490
1fc70edb
TY
4491#ifdef CONFIG_LOCKDEP
4492 nest_level = dev->nested_level;
4493#endif
4494 spin_lock_nested(&dev->addr_list_lock, nest_level);
845e0ebb
CW
4495}
4496
e308a5d8
DM
4497static inline void netif_addr_lock_bh(struct net_device *dev)
4498{
1fc70edb
TY
4499 unsigned char nest_level = 0;
4500
4501#ifdef CONFIG_LOCKDEP
4502 nest_level = dev->nested_level;
4503#endif
4504 local_bh_disable();
4505 spin_lock_nested(&dev->addr_list_lock, nest_level);
e308a5d8
DM
4506}
4507
4508static inline void netif_addr_unlock(struct net_device *dev)
4509{
4510 spin_unlock(&dev->addr_list_lock);
4511}
4512
4513static inline void netif_addr_unlock_bh(struct net_device *dev)
4514{
4515 spin_unlock_bh(&dev->addr_list_lock);
4516}
4517
f001fde5 4518/*
31278e71 4519 * dev_addrs walker. Should be used only for read access. Call with
f001fde5
JP
4520 * rcu_read_lock held.
4521 */
4522#define for_each_dev_addr(dev, ha) \
31278e71 4523 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
f001fde5 4524
1da177e4
LT
4525/* These functions live elsewhere (drivers/net/net_init.c, but related) */
4526
f629d208 4527void ether_setup(struct net_device *dev);
1da177e4
LT
4528
4529/* Support for loadable net-drivers */
f629d208 4530struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
c835a677 4531 unsigned char name_assign_type,
f629d208
JP
4532 void (*setup)(struct net_device *),
4533 unsigned int txqs, unsigned int rxqs);
c835a677
TG
4534#define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4535 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
36909ea4 4536
c835a677
TG
4537#define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4538 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4539 count)
36909ea4 4540
f629d208
JP
4541int register_netdev(struct net_device *dev);
4542void unregister_netdev(struct net_device *dev);
f001fde5 4543
cd16627f
BG
4544int devm_register_netdev(struct device *dev, struct net_device *ndev);
4545
22bedad3 4546/* General hardware address lists handling functions */
f629d208
JP
4547int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4548 struct netdev_hw_addr_list *from_list, int addr_len);
4549void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4550 struct netdev_hw_addr_list *from_list, int addr_len);
670e5b8e
AD
4551int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4552 struct net_device *dev,
4553 int (*sync)(struct net_device *, const unsigned char *),
4554 int (*unsync)(struct net_device *,
4555 const unsigned char *));
e7946760
IK
4556int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4557 struct net_device *dev,
4558 int (*sync)(struct net_device *,
4559 const unsigned char *, int),
4560 int (*unsync)(struct net_device *,
4561 const unsigned char *, int));
4562void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4563 struct net_device *dev,
4564 int (*unsync)(struct net_device *,
4565 const unsigned char *, int));
670e5b8e
AD
4566void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4567 struct net_device *dev,
4568 int (*unsync)(struct net_device *,
4569 const unsigned char *));
f629d208 4570void __hw_addr_init(struct netdev_hw_addr_list *list);
22bedad3 4571
f001fde5 4572/* Functions used for device addresses handling */
adeef3e3
JK
4573void dev_addr_mod(struct net_device *dev, unsigned int offset,
4574 const void *addr, size_t len);
4575
48eab831 4576static inline void
40af35fd 4577__dev_addr_set(struct net_device *dev, const void *addr, size_t len)
48eab831 4578{
adeef3e3 4579 dev_addr_mod(dev, 0, addr, len);
48eab831
JK
4580}
4581
4582static inline void dev_addr_set(struct net_device *dev, const u8 *addr)
4583{
4584 __dev_addr_set(dev, addr, dev->addr_len);
4585}
4586
f629d208
JP
4587int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4588 unsigned char addr_type);
4589int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4590 unsigned char addr_type);
a748ee24
JP
4591
4592/* Functions used for unicast addresses handling */
f629d208
JP
4593int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4594int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4595int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4596int dev_uc_sync(struct net_device *to, struct net_device *from);
4597int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4598void dev_uc_unsync(struct net_device *to, struct net_device *from);
4599void dev_uc_flush(struct net_device *dev);
4600void dev_uc_init(struct net_device *dev);
f001fde5 4601
670e5b8e
AD
4602/**
4603 * __dev_uc_sync - Synchonize device's unicast list
4604 * @dev: device to sync
4605 * @sync: function to call if address should be added
4606 * @unsync: function to call if address should be removed
4607 *
4608 * Add newly added addresses to the interface, and release
4609 * addresses that have been deleted.
5e82b4b2 4610 */
670e5b8e
AD
4611static inline int __dev_uc_sync(struct net_device *dev,
4612 int (*sync)(struct net_device *,
4613 const unsigned char *),
4614 int (*unsync)(struct net_device *,
4615 const unsigned char *))
4616{
4617 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4618}
4619
4620/**
e793c0f7 4621 * __dev_uc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4622 * @dev: device to sync
4623 * @unsync: function to call if address should be removed
4624 *
4625 * Remove all addresses that were added to the device by dev_uc_sync().
5e82b4b2 4626 */
670e5b8e
AD
4627static inline void __dev_uc_unsync(struct net_device *dev,
4628 int (*unsync)(struct net_device *,
4629 const unsigned char *))
4630{
4631 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4632}
4633
22bedad3 4634/* Functions used for multicast addresses handling */
f629d208
JP
4635int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4636int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4637int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4638int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4639int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4640int dev_mc_sync(struct net_device *to, struct net_device *from);
4641int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4642void dev_mc_unsync(struct net_device *to, struct net_device *from);
4643void dev_mc_flush(struct net_device *dev);
4644void dev_mc_init(struct net_device *dev);
f001fde5 4645
670e5b8e
AD
4646/**
4647 * __dev_mc_sync - Synchonize device's multicast list
4648 * @dev: device to sync
4649 * @sync: function to call if address should be added
4650 * @unsync: function to call if address should be removed
4651 *
4652 * Add newly added addresses to the interface, and release
4653 * addresses that have been deleted.
5e82b4b2 4654 */
670e5b8e
AD
4655static inline int __dev_mc_sync(struct net_device *dev,
4656 int (*sync)(struct net_device *,
4657 const unsigned char *),
4658 int (*unsync)(struct net_device *,
4659 const unsigned char *))
4660{
4661 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4662}
4663
4664/**
e793c0f7 4665 * __dev_mc_unsync - Remove synchronized addresses from device
670e5b8e
AD
4666 * @dev: device to sync
4667 * @unsync: function to call if address should be removed
4668 *
4669 * Remove all addresses that were added to the device by dev_mc_sync().
5e82b4b2 4670 */
670e5b8e
AD
4671static inline void __dev_mc_unsync(struct net_device *dev,
4672 int (*unsync)(struct net_device *,
4673 const unsigned char *))
4674{
4675 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4676}
4677
4417da66 4678/* Functions used for secondary unicast and multicast support */
f629d208 4679void dev_set_rx_mode(struct net_device *dev);
f629d208
JP
4680int dev_set_promiscuity(struct net_device *dev, int inc);
4681int dev_set_allmulti(struct net_device *dev, int inc);
4682void netdev_state_change(struct net_device *dev);
7061eb8c 4683void __netdev_notify_peers(struct net_device *dev);
f629d208
JP
4684void netdev_notify_peers(struct net_device *dev);
4685void netdev_features_change(struct net_device *dev);
1da177e4 4686/* Load a device via the kmod */
f629d208
JP
4687void dev_load(struct net *net, const char *name);
4688struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4689 struct rtnl_link_stats64 *storage);
4690void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4691 const struct net_device_stats *netdev_stats);
44fa32f0
HK
4692void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
4693 const struct pcpu_sw_netstats __percpu *netstats);
a1839426 4694void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
eeda3fd6 4695
1da177e4 4696extern int netdev_max_backlog;
3d48b53f
MT
4697extern int dev_rx_weight;
4698extern int dev_tx_weight;
323ebb61 4699extern int gro_normal_batch;
9ff162a8 4700
1fc70edb
TY
4701enum {
4702 NESTED_SYNC_IMM_BIT,
4703 NESTED_SYNC_TODO_BIT,
4704};
4705
4706#define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
4707#define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
4708
4709#define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
4710#define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
4711
eff74233 4712struct netdev_nested_priv {
1fc70edb 4713 unsigned char flags;
eff74233
TY
4714 void *data;
4715};
4716
f629d208 4717bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
44a40855
VY
4718struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4719 struct list_head **iter);
8b5be856 4720
44a40855
VY
4721/* iterate through upper list, must be called under RCU read lock */
4722#define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4723 for (iter = &(dev)->adj_list.upper, \
4724 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4725 updev; \
4726 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4727
1a3f060c
DA
4728int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4729 int (*fn)(struct net_device *upper_dev,
eff74233
TY
4730 struct netdev_nested_priv *priv),
4731 struct netdev_nested_priv *priv);
1a3f060c
DA
4732
4733bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4734 struct net_device *upper_dev);
4735
25cc72a3
IS
4736bool netdev_has_any_upper_dev(struct net_device *dev);
4737
f629d208
JP
4738void *netdev_lower_get_next_private(struct net_device *dev,
4739 struct list_head **iter);
4740void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4741 struct list_head **iter);
31088a11
VF
4742
4743#define netdev_for_each_lower_private(dev, priv, iter) \
4744 for (iter = (dev)->adj_list.lower.next, \
4745 priv = netdev_lower_get_next_private(dev, &(iter)); \
4746 priv; \
4747 priv = netdev_lower_get_next_private(dev, &(iter)))
4748
4749#define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4750 for (iter = &(dev)->adj_list.lower, \
4751 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4752 priv; \
4753 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4754
4085ebe8
VY
4755void *netdev_lower_get_next(struct net_device *dev,
4756 struct list_head **iter);
7ce856aa 4757
4085ebe8 4758#define netdev_for_each_lower_dev(dev, ldev, iter) \
cfdd28be 4759 for (iter = (dev)->adj_list.lower.next, \
4085ebe8
VY
4760 ldev = netdev_lower_get_next(dev, &(iter)); \
4761 ldev; \
4762 ldev = netdev_lower_get_next(dev, &(iter)))
4763
7151affe 4764struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
7ce856aa 4765 struct list_head **iter);
1a3f060c
DA
4766int netdev_walk_all_lower_dev(struct net_device *dev,
4767 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4768 struct netdev_nested_priv *priv),
4769 struct netdev_nested_priv *priv);
1a3f060c
DA
4770int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4771 int (*fn)(struct net_device *lower_dev,
eff74233
TY
4772 struct netdev_nested_priv *priv),
4773 struct netdev_nested_priv *priv);
1a3f060c 4774
f629d208 4775void *netdev_adjacent_get_private(struct list_head *adj_list);
e001bfad 4776void *netdev_lower_get_first_private_rcu(struct net_device *dev);
f629d208
JP
4777struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4778struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
42ab19ee
DA
4779int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4780 struct netlink_ext_ack *extack);
f629d208 4781int netdev_master_upper_dev_link(struct net_device *dev,
6dffb044 4782 struct net_device *upper_dev,
42ab19ee
DA
4783 void *upper_priv, void *upper_info,
4784 struct netlink_ext_ack *extack);
f629d208
JP
4785void netdev_upper_dev_unlink(struct net_device *dev,
4786 struct net_device *upper_dev);
32b6d34f
TY
4787int netdev_adjacent_change_prepare(struct net_device *old_dev,
4788 struct net_device *new_dev,
4789 struct net_device *dev,
4790 struct netlink_ext_ack *extack);
4791void netdev_adjacent_change_commit(struct net_device *old_dev,
4792 struct net_device *new_dev,
4793 struct net_device *dev);
4794void netdev_adjacent_change_abort(struct net_device *old_dev,
4795 struct net_device *new_dev,
4796 struct net_device *dev);
5bb025fa 4797void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
f629d208
JP
4798void *netdev_lower_dev_get_private(struct net_device *dev,
4799 struct net_device *lower_dev);
04d48266
JP
4800void netdev_lower_state_changed(struct net_device *lower_dev,
4801 void *lower_state_info);
960fb622
ED
4802
4803/* RSS keys are 40 or 52 bytes long */
4804#define NETDEV_RSS_KEY_LEN 52
ba905f5e 4805extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
960fb622
ED
4806void netdev_rss_key_fill(void *buffer, size_t len);
4807
f629d208 4808int skb_checksum_help(struct sk_buff *skb);
b72b5bf6 4809int skb_crc32c_csum_help(struct sk_buff *skb);
43c26a1a
DC
4810int skb_csum_hwoffload_help(struct sk_buff *skb,
4811 const netdev_features_t features);
4812
61bd3857
MS
4813struct netdev_bonding_info {
4814 ifslave slave;
4815 ifbond master;
4816};
4817
4818struct netdev_notifier_bonding_info {
4819 struct netdev_notifier_info info; /* must be first */
4820 struct netdev_bonding_info bonding_info;
4821};
4822
4823void netdev_bonding_info_change(struct net_device *dev,
4824 struct netdev_bonding_info *bonding_info);
4825
6b08d6c1
MK
4826#if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
4827void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
4828#else
4829static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
4830 const void *data)
4831{
4832}
4833#endif
4834
53d6471c 4835__be16 skb_network_protocol(struct sk_buff *skb, int *depth);
ec5f0615
PS
4836
4837static inline bool can_checksum_protocol(netdev_features_t features,
4838 __be16 protocol)
4839{
c8cd0989
TH
4840 if (protocol == htons(ETH_P_FCOE))
4841 return !!(features & NETIF_F_FCOE_CRC);
4842
4843 /* Assume this is an IP checksum (not SCTP CRC) */
4844
4845 if (features & NETIF_F_HW_CSUM) {
4846 /* Can checksum everything */
4847 return true;
4848 }
4849
4850 switch (protocol) {
4851 case htons(ETH_P_IP):
4852 return !!(features & NETIF_F_IP_CSUM);
4853 case htons(ETH_P_IPV6):
4854 return !!(features & NETIF_F_IPV6_CSUM);
4855 default:
4856 return false;
4857 }
ec5f0615 4858}
12b0004d 4859
fb286bb2 4860#ifdef CONFIG_BUG
7fe50ac8 4861void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
fb286bb2 4862#else
7fe50ac8
CW
4863static inline void netdev_rx_csum_fault(struct net_device *dev,
4864 struct sk_buff *skb)
fb286bb2
HX
4865{
4866}
4867#endif
1da177e4 4868/* rx skb timestamps */
f629d208
JP
4869void net_enable_timestamp(void);
4870void net_disable_timestamp(void);
1da177e4 4871
97dc7cd9
GE
4872static inline ktime_t netdev_get_tstamp(struct net_device *dev,
4873 const struct skb_shared_hwtstamps *hwtstamps,
4874 bool cycles)
4875{
4876 const struct net_device_ops *ops = dev->netdev_ops;
4877
4878 if (ops->ndo_get_tstamp)
4879 return ops->ndo_get_tstamp(dev, hwtstamps, cycles);
4880
4881 return hwtstamps->hwtstamp;
4882}
4883
4798248e 4884static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
fa2dbdc2
DM
4885 struct sk_buff *skb, struct net_device *dev,
4886 bool more)
4798248e 4887{
6b16f9ee 4888 __this_cpu_write(softnet_data.xmit.more, more);
0b725a2c 4889 return ops->ndo_start_xmit(skb, dev);
4798248e
DM
4890}
4891
97cdcf37
FW
4892static inline bool netdev_xmit_more(void)
4893{
4894 return __this_cpu_read(softnet_data.xmit.more);
4895}
4896
10b3ad8c 4897static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
fa2dbdc2 4898 struct netdev_queue *txq, bool more)
4798248e
DM
4899{
4900 const struct net_device_ops *ops = dev->netdev_ops;
2183435c 4901 netdev_tx_t rc;
4798248e 4902
fa2dbdc2 4903 rc = __netdev_start_xmit(ops, skb, dev, more);
10b3ad8c
DM
4904 if (rc == NETDEV_TX_OK)
4905 txq_trans_update(txq);
4906
4907 return rc;
4798248e
DM
4908}
4909
b793dc5c 4910int netdev_class_create_file_ns(const struct class_attribute *class_attr,
42a2d923 4911 const void *ns);
b793dc5c 4912void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
42a2d923 4913 const void *ns);
58292cbe 4914
737aec57 4915extern const struct kobj_ns_type_operations net_ns_type_operations;
04600794 4916
f629d208 4917const char *netdev_drivername(const struct net_device *dev);
6579e57b 4918
da08143b
MK
4919static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4920 netdev_features_t f2)
4921{
c8cd0989
TH
4922 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4923 if (f1 & NETIF_F_HW_CSUM)
b6a0e72a 4924 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4925 else
b6a0e72a 4926 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
c8cd0989 4927 }
da08143b 4928
c8cd0989 4929 return f1 & f2;
da08143b
MK
4930}
4931
c8f44aff
MM
4932static inline netdev_features_t netdev_get_wanted_features(
4933 struct net_device *dev)
5455c699
MM
4934{
4935 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4936}
c8f44aff
MM
4937netdev_features_t netdev_increment_features(netdev_features_t all,
4938 netdev_features_t one, netdev_features_t mask);
b0ce3508
ED
4939
4940/* Allow TSO being used on stacked device :
4941 * Performing the GSO segmentation before last device
4942 * is a performance improvement.
4943 */
4944static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4945 netdev_features_t mask)
4946{
4947 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4948}
4949
6cb6a27c 4950int __netdev_update_features(struct net_device *dev);
5455c699 4951void netdev_update_features(struct net_device *dev);
afe12cc8 4952void netdev_change_features(struct net_device *dev);
7f353bf2 4953
fc4a7489
PM
4954void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4955 struct net_device *dev);
4956
e38f3025
TM
4957netdev_features_t passthru_features_check(struct sk_buff *skb,
4958 struct net_device *dev,
4959 netdev_features_t features);
c1e756bf 4960netdev_features_t netif_skb_features(struct sk_buff *skb);
d457a0e3 4961void skb_warn_bad_offload(const struct sk_buff *skb);
58e998c6 4962
4d29515f 4963static inline bool net_gso_ok(netdev_features_t features, int gso_type)
576a30eb 4964{
7b748340 4965 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
0345e186
MM
4966
4967 /* check flags correspondence */
4968 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
0345e186
MM
4969 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
4970 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
cbc53e08 4971 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
0345e186
MM
4972 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
4973 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4974 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
4975 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
7e13318d
TH
4976 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
4977 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
4b28252c
TH
4978 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
4979 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
802ab55a 4980 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
e585f236 4981 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
90017acc 4982 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
c7ef8f0c 4983 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
0c19f846 4984 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
83aa025f 4985 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
3b335832 4986 BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
0345e186 4987
d6b4991a 4988 return (features & feature) == feature;
576a30eb
HX
4989}
4990
4d29515f 4991static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
bcd76111 4992{
278b2513 4993 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
21dc3301 4994 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
bcd76111
HX
4995}
4996
8b86a61d 4997static inline bool netif_needs_gso(struct sk_buff *skb,
4d29515f 4998 netdev_features_t features)
7967168c 4999{
fc741216 5000 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
cdbee74c
YZ
5001 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
5002 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
7967168c
HX
5003}
5004
14d7b812
JK
5005void netif_set_tso_max_size(struct net_device *dev, unsigned int size);
5006void netif_set_tso_max_segs(struct net_device *dev, unsigned int segs);
6df6398f
JK
5007void netif_inherit_tso_max(struct net_device *to,
5008 const struct net_device *from);
5009
3c175784
SD
5010static inline bool netif_is_macsec(const struct net_device *dev)
5011{
5012 return dev->priv_flags & IFF_MACSEC;
5013}
5014
b618aaa9 5015static inline bool netif_is_macvlan(const struct net_device *dev)
a6cc0cfa
JF
5016{
5017 return dev->priv_flags & IFF_MACVLAN;
5018}
5019
b618aaa9 5020static inline bool netif_is_macvlan_port(const struct net_device *dev)
2f33e7d5
MB
5021{
5022 return dev->priv_flags & IFF_MACVLAN_PORT;
5023}
5024
b618aaa9 5025static inline bool netif_is_bond_master(const struct net_device *dev)
8a7fbfab 5026{
5027 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
5028}
5029
b618aaa9 5030static inline bool netif_is_bond_slave(const struct net_device *dev)
1765a575
JP
5031{
5032 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
5033}
5034
3bdc0eba
BG
5035static inline bool netif_supports_nofcs(struct net_device *dev)
5036{
5037 return dev->priv_flags & IFF_SUPP_NOFCS;
5038}
5039
d5256083
DB
5040static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
5041{
5042 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
5043}
5044
007979ea 5045static inline bool netif_is_l3_master(const struct net_device *dev)
4e3c8992 5046{
007979ea 5047 return dev->priv_flags & IFF_L3MDEV_MASTER;
4e3c8992
DA
5048}
5049
fee6d4c7
DA
5050static inline bool netif_is_l3_slave(const struct net_device *dev)
5051{
5052 return dev->priv_flags & IFF_L3MDEV_SLAVE;
5053}
5054
9a5cb797
GS
5055static inline int dev_sdif(const struct net_device *dev)
5056{
5057#ifdef CONFIG_NET_L3_MASTER_DEV
5058 if (netif_is_l3_slave(dev))
5059 return dev->ifindex;
5060#endif
5061 return 0;
5062}
5063
0894ae3f
JP
5064static inline bool netif_is_bridge_master(const struct net_device *dev)
5065{
5066 return dev->priv_flags & IFF_EBRIDGE;
5067}
5068
28f9ee22
VY
5069static inline bool netif_is_bridge_port(const struct net_device *dev)
5070{
5071 return dev->priv_flags & IFF_BRIDGE_PORT;
5072}
5073
35d4e172
JP
5074static inline bool netif_is_ovs_master(const struct net_device *dev)
5075{
5076 return dev->priv_flags & IFF_OPENVSWITCH;
5077}
5078
5be66141
JP
5079static inline bool netif_is_ovs_port(const struct net_device *dev)
5080{
5081 return dev->priv_flags & IFF_OVS_DATAPATH;
5082}
5083
f11e5bd1
MK
5084static inline bool netif_is_any_bridge_master(const struct net_device *dev)
5085{
5086 return netif_is_bridge_master(dev) || netif_is_ovs_master(dev);
5087}
5088
df23bb18
SB
5089static inline bool netif_is_any_bridge_port(const struct net_device *dev)
5090{
5091 return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
5092}
5093
b618aaa9 5094static inline bool netif_is_team_master(const struct net_device *dev)
c981e421
JP
5095{
5096 return dev->priv_flags & IFF_TEAM;
5097}
5098
b618aaa9 5099static inline bool netif_is_team_port(const struct net_device *dev)
f7f019ee
JP
5100{
5101 return dev->priv_flags & IFF_TEAM_PORT;
5102}
5103
b618aaa9 5104static inline bool netif_is_lag_master(const struct net_device *dev)
7be61833
JP
5105{
5106 return netif_is_bond_master(dev) || netif_is_team_master(dev);
5107}
5108
b618aaa9 5109static inline bool netif_is_lag_port(const struct net_device *dev)
e0ba1414
JP
5110{
5111 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
5112}
5113
d4ab4286
KJ
5114static inline bool netif_is_rxfh_configured(const struct net_device *dev)
5115{
5116 return dev->priv_flags & IFF_RXFH_CONFIGURED;
5117}
5118
30c8bd5a
SS
5119static inline bool netif_is_failover(const struct net_device *dev)
5120{
5121 return dev->priv_flags & IFF_FAILOVER;
5122}
5123
5124static inline bool netif_is_failover_slave(const struct net_device *dev)
5125{
5126 return dev->priv_flags & IFF_FAILOVER_SLAVE;
5127}
5128
02875878
ED
5129/* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
5130static inline void netif_keep_dst(struct net_device *dev)
5131{
5132 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
5133}
5134
18d3df3e
PA
5135/* return true if dev can't cope with mtu frames that need vlan tag insertion */
5136static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
5137{
5138 /* TODO: reserve and use an additional IFF bit, if we get more users */
01992152 5139 return netif_is_macsec(dev);
18d3df3e
PA
5140}
5141
505d4f73 5142extern struct pernet_operations __net_initdata loopback_net_ops;
b1b67dd4 5143
571ba423
JP
5144/* Logging, debugging and troubleshooting/diagnostic helpers. */
5145
5146/* netdev_printk helpers, similar to dev_printk */
5147
5148static inline const char *netdev_name(const struct net_device *dev)
5149{
c6f854d5
VF
5150 if (!dev->name[0] || strchr(dev->name, '%'))
5151 return "(unnamed net_device)";
571ba423
JP
5152 return dev->name;
5153}
5154
ccc7f496
VF
5155static inline const char *netdev_reg_state(const struct net_device *dev)
5156{
5157 switch (dev->reg_state) {
5158 case NETREG_UNINITIALIZED: return " (uninitialized)";
5159 case NETREG_REGISTERED: return "";
5160 case NETREG_UNREGISTERING: return " (unregistering)";
5161 case NETREG_UNREGISTERED: return " (unregistered)";
5162 case NETREG_RELEASED: return " (released)";
5163 case NETREG_DUMMY: return " (dummy)";
5164 }
5165
5166 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
5167 return " (unknown)";
5168}
5169
8909c9ad
VK
5170#define MODULE_ALIAS_NETDEV(device) \
5171 MODULE_ALIAS("netdev-" device)
5172
571ba423
JP
5173/*
5174 * netdev_WARN() acts like dev_printk(), but with the key difference
5175 * of using a WARN/WARN_ON to get the message out, including the
5176 * file/line information and a backtrace.
5177 */
5178#define netdev_WARN(dev, format, args...) \
e1cfe3d0 5179 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
ccc7f496 5180 netdev_reg_state(dev), ##args)
571ba423 5181
72dd831e 5182#define netdev_WARN_ONCE(dev, format, args...) \
e1cfe3d0 5183 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
375ef2b1
GP
5184 netdev_reg_state(dev), ##args)
5185
900ff8c6
CW
5186/*
5187 * The list of packet types we will receive (as opposed to discard)
5188 * and the routines to invoke.
5189 *
5190 * Why 16. Because with 16 the only overlap we get on a hash of the
5191 * low nibble of the protocol value is RARP/SNAP/X.25.
5192 *
900ff8c6 5193 * 0800 IP
900ff8c6
CW
5194 * 0001 802.3
5195 * 0002 AX.25
5196 * 0004 802.2
5197 * 8035 RARP
5198 * 0005 SNAP
5199 * 0805 X.25
5200 * 0806 ARP
5201 * 8137 IPX
5202 * 0009 Localtalk
5203 * 86DD IPv6
5204 */
5205#define PTYPE_HASH_SIZE (16)
5206#define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
5207
744b8376
VO
5208extern struct list_head ptype_all __read_mostly;
5209extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
5210
4de83b88
MB
5211extern struct net_device *blackhole_netdev;
5212
6c1c5097
ED
5213/* Note: Avoid these macros in fast path, prefer per-cpu or per-queue counters. */
5214#define DEV_STATS_INC(DEV, FIELD) atomic_long_inc(&(DEV)->stats.__##FIELD)
5215#define DEV_STATS_ADD(DEV, FIELD, VAL) \
5216 atomic_long_add((VAL), &(DEV)->stats.__##FIELD)
5217
385a154c 5218#endif /* _LINUX_NETDEVICE_H */