1 /* SPDX-License-Identifier: GPL-2.0-or-later */
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 *
11 * Authors: Ross Biro
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>
15 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
16 * Bjorn Ekwall. <bj0rn@blox.se>
17 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 *
19 * Moved to /usr/include/linux for NET3
20 */
21 #ifndef _LINUX_NETDEVICE_H
22 #define _LINUX_NETDEVICE_H
23
24 #include <linux/timer.h>
25 #include <linux/bug.h>
26 #include <linux/delay.h>
27 #include <linux/atomic.h>
28 #include <linux/prefetch.h>
29 #include <asm/cache.h>
30 #include <asm/byteorder.h>
31 #include <asm/local.h>
32
33 #include <linux/percpu.h>
34 #include <linux/rculist.h>
35 #include <linux/workqueue.h>
36 #include <linux/dynamic_queue_limits.h>
37
38 #include <net/net_namespace.h>
39 #ifdef CONFIG_DCB
40 #include <net/dcbnl.h>
41 #endif
42 #include <net/netprio_cgroup.h>
43
44 #include <linux/netdev_features.h>
45 #include <linux/neighbour.h>
46 #include <uapi/linux/netdevice.h>
47 #include <uapi/linux/if_bonding.h>
48 #include <uapi/linux/pkt_cls.h>
49 #include <uapi/linux/netdev.h>
50 #include <linux/hashtable.h>
51 #include <linux/rbtree.h>
52 #include <net/net_trackers.h>
53 #include <net/net_debug.h>
54 #include <net/dropreason-core.h>
55
56 struct netpoll_info;
57 struct device;
58 struct ethtool_ops;
59 struct kernel_hwtstamp_config;
60 struct phy_device;
61 struct dsa_port;
62 struct ip_tunnel_parm;
63 struct macsec_context;
64 struct macsec_ops;
65 struct netdev_name_node;
66 struct sd_flow_limit;
67 struct sfp_bus;
68 /* 802.11 specific */
69 struct wireless_dev;
70 /* 802.15.4 specific */
71 struct wpan_dev;
72 struct mpls_dev;
73 /* UDP Tunnel offloads */
74 struct udp_tunnel_info;
75 struct udp_tunnel_nic_info;
76 struct udp_tunnel_nic;
77 struct bpf_prog;
78 struct xdp_buff;
79 struct xdp_frame;
80 struct xdp_metadata_ops;
81 struct xdp_md;
82
83 typedef u32 xdp_features_t;
84
85 void synchronize_net(void);
86 void netdev_set_default_ethtool_ops(struct net_device *dev,
87 const struct ethtool_ops *ops);
88 void netdev_sw_irq_coalesce_default_on(struct net_device *dev);
89
90 /* Backlog congestion levels */
91 #define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
92 #define NET_RX_DROP 1 /* packet dropped */
93
94 #define MAX_NEST_DEV 8
95
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
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
110 * others are propagated to higher layers.
111 */
112
113 /* qdisc ->enqueue() return codes. */
114 #define NET_XMIT_SUCCESS 0x00
115 #define NET_XMIT_DROP 0x01 /* skb dropped */
116 #define NET_XMIT_CN 0x02 /* congestion notification */
117 #define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
118
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. */
122 #define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
123 #define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
124
125 /* Driver transmit return codes */
126 #define NETDEV_TX_MASK 0xf0
127
128 enum netdev_tx {
129 __NETDEV_TX_MIN = INT_MIN, /* make sure enum is signed */
130 NETDEV_TX_OK = 0x00, /* driver took care of packet */
131 NETDEV_TX_BUSY = 0x10, /* driver tx path was busy*/
132 };
133 typedef enum netdev_tx netdev_tx_t;
134
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 */
dev_xmit_complete(int rc)139 static 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
153 /*
154 * Compute the worst-case header length according to the protocols
155 * used.
156 */
157
158 #if defined(CONFIG_HYPERV_NET)
159 # define LL_MAX_HEADER 128
160 #elif defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
161 # if defined(CONFIG_MAC80211_MESH)
162 # define LL_MAX_HEADER 128
163 # else
164 # define LL_MAX_HEADER 96
165 # endif
166 #else
167 # define LL_MAX_HEADER 32
168 #endif
169
170 #if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
171 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
172 #define MAX_HEADER LL_MAX_HEADER
173 #else
174 #define MAX_HEADER (LL_MAX_HEADER + 48)
175 #endif
176
177 /*
178 * Old network device statistics. Fields are native words
179 * (unsigned long) so they can be read and written atomically.
180 */
181
182 #define NET_DEV_STAT(FIELD) \
183 union { \
184 unsigned long FIELD; \
185 atomic_long_t __##FIELD; \
186 }
187
188 struct net_device_stats {
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);
212 };
213 #undef NET_DEV_STAT
214
215 /* per-cpu stats, allocated on demand.
216 * Try to fit them in a single cache line, for dev_get_stats() sake.
217 */
218 struct net_device_core_stats {
219 unsigned long rx_dropped;
220 unsigned long tx_dropped;
221 unsigned long rx_nohandler;
222 unsigned long rx_otherhost_dropped;
223 } __aligned(4 * sizeof(unsigned long));
224
225 #include <linux/cache.h>
226 #include <linux/skbuff.h>
227
228 #ifdef CONFIG_RPS
229 #include <linux/static_key.h>
230 extern struct static_key_false rps_needed;
231 extern struct static_key_false rfs_needed;
232 #endif
233
234 struct neighbour;
235 struct neigh_parms;
236 struct sk_buff;
237
238 struct netdev_hw_addr {
239 struct list_head list;
240 struct rb_node node;
241 unsigned char addr[MAX_ADDR_LEN];
242 unsigned char type;
243 #define NETDEV_HW_ADDR_T_LAN 1
244 #define NETDEV_HW_ADDR_T_SAN 2
245 #define NETDEV_HW_ADDR_T_UNICAST 3
246 #define NETDEV_HW_ADDR_T_MULTICAST 4
247 bool global_use;
248 int sync_cnt;
249 int refcount;
250 int synced;
251 struct rcu_head rcu_head;
252 };
253
254 struct netdev_hw_addr_list {
255 struct list_head list;
256 int count;
257
258 /* Auxiliary tree for faster lookup on addition and deletion */
259 struct rb_root tree;
260 };
261
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)
266
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)
271 #define netdev_for_each_synced_uc_addr(_ha, _dev) \
272 netdev_for_each_uc_addr((_ha), (_dev)) \
273 if ((_ha)->sync_cnt)
274
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)
277 #define netdev_for_each_mc_addr(ha, dev) \
278 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
279 #define netdev_for_each_synced_mc_addr(_ha, _dev) \
280 netdev_for_each_mc_addr((_ha), (_dev)) \
281 if ((_ha)->sync_cnt)
282
283 struct hh_cache {
284 unsigned int hh_len;
285 seqlock_t hh_lock;
286
287 /* cached hardware header; allow for machine alignment needs. */
288 #define HH_DATA_MOD 16
289 #define HH_DATA_OFF(__len) \
290 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
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
296 /* Reserve HH_DATA_MOD byte-aligned hard_header_len, but at least that much.
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) \
305 ((((dev)->hard_header_len + READ_ONCE((dev)->needed_headroom)) \
306 & ~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
307 #define LL_RESERVED_SPACE_EXTRA(dev,extra) \
308 ((((dev)->hard_header_len + READ_ONCE((dev)->needed_headroom) + (extra)) \
309 & ~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
310
311 struct header_ops {
312 int (*create) (struct sk_buff *skb, struct net_device *dev,
313 unsigned short type, const void *daddr,
314 const void *saddr, unsigned int len);
315 int (*parse)(const struct sk_buff *skb, unsigned char *haddr);
316 int (*cache)(const struct neighbour *neigh, struct hh_cache *hh, __be16 type);
317 void (*cache_update)(struct hh_cache *hh,
318 const struct net_device *dev,
319 const unsigned char *haddr);
320 bool (*validate)(const char *ll_header, unsigned int len);
321 __be16 (*parse_protocol)(const struct sk_buff *skb);
322 };
323
324 /* These flag bits are private to the generic network queueing
325 * layer; they may not be explicitly referenced by any other
326 * code.
327 */
328
329 enum netdev_state_t {
330 __LINK_STATE_START,
331 __LINK_STATE_PRESENT,
332 __LINK_STATE_NOCARRIER,
333 __LINK_STATE_LINKWATCH_PENDING,
334 __LINK_STATE_DORMANT,
335 __LINK_STATE_TESTING,
336 };
337
338 struct gro_list {
339 struct list_head list;
340 int count;
341 };
342
343 /*
344 * size of gro hash buckets, must less than bit number of
345 * napi_struct::gro_bitmask
346 */
347 #define GRO_HASH_BUCKETS 8
348
349 /*
350 * Structure for NAPI scheduling similar to tasklet but with weighting
351 */
352 struct 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
356 * to the per-CPU poll_list, and whoever clears that bit
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;
363 int defer_hard_irqs_count;
364 unsigned long gro_bitmask;
365 int (*poll)(struct napi_struct *, int);
366 #ifdef CONFIG_NETPOLL
367 /* CPU actively polling if netpoll is configured */
368 int poll_owner;
369 #endif
370 /* CPU on which NAPI has been scheduled for processing */
371 int list_owner;
372 struct net_device *dev;
373 struct gro_list gro_hash[GRO_HASH_BUCKETS];
374 struct sk_buff *skb;
375 struct list_head rx_list; /* Pending GRO_NORMAL skbs */
376 int rx_count; /* length of rx_list */
377 unsigned int napi_id;
378 struct hrtimer timer;
379 struct task_struct *thread;
380 /* control-path-only fields follow */
381 struct list_head dev_list;
382 struct hlist_node napi_hash_node;
383 };
384
385 enum {
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*/
394 NAPI_STATE_THREADED, /* The poll is performed inside its own thread*/
395 NAPI_STATE_SCHED_THREADED, /* Napi is currently scheduled in threaded mode */
396 };
397
398 enum {
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),
407 NAPIF_STATE_THREADED = BIT(NAPI_STATE_THREADED),
408 NAPIF_STATE_SCHED_THREADED = BIT(NAPI_STATE_SCHED_THREADED),
409 };
410
411 enum gro_result {
412 GRO_MERGED,
413 GRO_MERGED_FREE,
414 GRO_HELD,
415 GRO_NORMAL,
416 GRO_CONSUMED,
417 };
418 typedef enum gro_result gro_result_t;
419
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.
427 * @RX_HANDLER_PASS: Do nothing, pass the skb as if no rx_handler was called.
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 *
442 * If the rx_handler consumed the skb in some way, it should return
443 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
444 * the skb to be delivered in some other way.
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 *
450 * If the rx_handler decides the skb should be ignored, it should return
451 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
452 * are registered on exact device (ptype->dev == skb->dev).
453 *
454 * If the rx_handler didn't change skb->dev, but wants the skb to be normally
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
461 enum rx_handler_result {
462 RX_HANDLER_CONSUMED,
463 RX_HANDLER_ANOTHER,
464 RX_HANDLER_EXACT,
465 RX_HANDLER_PASS,
466 };
467 typedef enum rx_handler_result rx_handler_result_t;
468 typedef rx_handler_result_t rx_handler_func_t(struct sk_buff **pskb);
469
470 void __napi_schedule(struct napi_struct *n);
471 void __napi_schedule_irqoff(struct napi_struct *n);
472
napi_disable_pending(struct napi_struct * n)473 static inline bool napi_disable_pending(struct napi_struct *n)
474 {
475 return test_bit(NAPI_STATE_DISABLE, &n->state);
476 }
477
napi_prefer_busy_poll(struct napi_struct * n)478 static inline bool napi_prefer_busy_poll(struct napi_struct *n)
479 {
480 return test_bit(NAPI_STATE_PREFER_BUSY_POLL, &n->state);
481 }
482
483 bool napi_schedule_prep(struct napi_struct *n);
484
485 /**
486 * napi_schedule - schedule NAPI poll
487 * @n: NAPI context
488 *
489 * Schedule NAPI poll routine to be called if it is not already
490 * running.
491 */
napi_schedule(struct napi_struct * n)492 static inline void napi_schedule(struct napi_struct *n)
493 {
494 if (napi_schedule_prep(n))
495 __napi_schedule(n);
496 }
497
498 /**
499 * napi_schedule_irqoff - schedule NAPI poll
500 * @n: NAPI context
501 *
502 * Variant of napi_schedule(), assuming hard irqs are masked.
503 */
napi_schedule_irqoff(struct napi_struct * n)504 static inline void napi_schedule_irqoff(struct napi_struct *n)
505 {
506 if (napi_schedule_prep(n))
507 __napi_schedule_irqoff(n);
508 }
509
510 /* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
napi_reschedule(struct napi_struct * napi)511 static inline bool napi_reschedule(struct napi_struct *napi)
512 {
513 if (napi_schedule_prep(napi)) {
514 __napi_schedule(napi);
515 return true;
516 }
517 return false;
518 }
519
520 /**
521 * napi_complete_done - NAPI processing complete
522 * @n: NAPI context
523 * @work_done: number of packets processed
524 *
525 * Mark NAPI processing as complete. Should only be called if poll budget
526 * has not been completely consumed.
527 * Prefer over napi_complete().
528 * Return false if device should avoid rearming interrupts.
529 */
530 bool napi_complete_done(struct napi_struct *n, int work_done);
531
napi_complete(struct napi_struct * n)532 static inline bool napi_complete(struct napi_struct *n)
533 {
534 return napi_complete_done(n, 0);
535 }
536
537 int dev_set_threaded(struct net_device *dev, bool threaded);
538
539 /**
540 * napi_disable - prevent NAPI from scheduling
541 * @n: NAPI context
542 *
543 * Stop NAPI from being scheduled on this context.
544 * Waits till any outstanding processing completes.
545 */
546 void napi_disable(struct napi_struct *n);
547
548 void napi_enable(struct napi_struct *n);
549
550 /**
551 * napi_synchronize - wait until NAPI is not running
552 * @n: NAPI context
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 */
napi_synchronize(const struct napi_struct * n)558 static inline void napi_synchronize(const struct napi_struct *n)
559 {
560 if (IS_ENABLED(CONFIG_SMP))
561 while (test_bit(NAPI_STATE_SCHED, &n->state))
562 msleep(1);
563 else
564 barrier();
565 }
566
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 **/
napi_if_scheduled_mark_missed(struct napi_struct * n)575 static inline bool napi_if_scheduled_mark_missed(struct napi_struct *n)
576 {
577 unsigned long val, new;
578
579 val = READ_ONCE(n->state);
580 do {
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;
588 } while (!try_cmpxchg(&n->state, &val, new));
589
590 return true;
591 }
592
593 enum netdev_queue_state_t {
594 __QUEUE_STATE_DRV_XOFF,
595 __QUEUE_STATE_STACK_XOFF,
596 __QUEUE_STATE_FROZEN,
597 };
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
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 */
618
619 struct netdev_queue {
620 /*
621 * read-mostly part
622 */
623 struct net_device *dev;
624 netdevice_tracker dev_tracker;
625
626 struct Qdisc __rcu *qdisc;
627 struct Qdisc __rcu *qdisc_sleeping;
628 #ifdef CONFIG_SYSFS
629 struct kobject kobj;
630 #endif
631 #if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
632 int numa_node;
633 #endif
634 unsigned long tx_maxrate;
635 /*
636 * Number of TX timeouts for this queue
637 * (/sys/class/net/DEV/Q/trans_timeout)
638 */
639 atomic_long_t trans_timeout;
640
641 /* Subordinate device that the queue has been assigned to */
642 struct net_device *sb_dev;
643 #ifdef CONFIG_XDP_SOCKETS
644 struct xsk_buff_pool *pool;
645 #endif
646 /*
647 * write-mostly part
648 */
649 spinlock_t _xmit_lock ____cacheline_aligned_in_smp;
650 int xmit_lock_owner;
651 /*
652 * Time (in jiffies) of last Tx
653 */
654 unsigned long trans_start;
655
656 unsigned long state;
657
658 #ifdef CONFIG_BQL
659 struct dql dql;
660 #endif
661 } ____cacheline_aligned_in_smp;
662
663 extern int sysctl_fb_tunnels_only_for_init_net;
664 extern int sysctl_devconf_inherit_init_net;
665
666 /*
667 * sysctl_fb_tunnels_only_for_init_net == 0 : For all netns
668 * == 1 : For initns only
669 * == 2 : For none.
670 */
net_has_fallback_tunnels(const struct net * net)671 static inline bool net_has_fallback_tunnels(const struct net *net)
672 {
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
681 }
682
net_inherit_devconf(void)683 static 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
690 }
691
netdev_queue_numa_node_read(const struct netdev_queue * q)692 static 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
697 return NUMA_NO_NODE;
698 #endif
699 }
700
netdev_queue_numa_node_write(struct netdev_queue * q,int node)701 static 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
708 #ifdef CONFIG_RPS
709 /*
710 * This structure holds an RPS map which can be of variable length. The
711 * map is an array of CPUs.
712 */
713 struct rps_map {
714 unsigned int len;
715 struct rcu_head rcu;
716 u16 cpus[];
717 };
718 #define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
719
720 /*
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.
724 */
725 struct rps_dev_flow {
726 u16 cpu;
727 u16 filter;
728 unsigned int last_qtail;
729 };
730 #define RPS_NO_FILTER 0xffff
731
732 /*
733 * The rps_dev_flow_table structure contains a table of flow mappings.
734 */
735 struct rps_dev_flow_table {
736 unsigned int mask;
737 struct rcu_head rcu;
738 struct rps_dev_flow flows[];
739 };
740 #define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
741 ((_num) * sizeof(struct rps_dev_flow)))
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).
746 * Each entry is a 32bit value. Upper part is the high-order bits
747 * of flow hash, lower part is CPU number.
748 * rps_cpu_mask is used to partition the space, depending on number of
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,
751 * meaning we use 32-6=26 bits for the hash.
752 */
753 struct rps_sock_flow_table {
754 u32 mask;
755
756 u32 ents[] ____cacheline_aligned_in_smp;
757 };
758 #define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
759
760 #define RPS_NO_CPU 0xffff
761
762 extern u32 rps_cpu_mask;
763 extern struct rps_sock_flow_table __rcu *rps_sock_flow_table;
764
rps_record_sock_flow(struct rps_sock_flow_table * table,u32 hash)765 static inline void rps_record_sock_flow(struct rps_sock_flow_table *table,
766 u32 hash)
767 {
768 if (table && hash) {
769 unsigned int index = hash & table->mask;
770 u32 val = hash & ~rps_cpu_mask;
771
772 /* We only give a hint, preemption can change CPU under us */
773 val |= raw_smp_processor_id();
774
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);
780 }
781 }
782
783 #ifdef CONFIG_RFS_ACCEL
784 bool rps_may_expire_flow(struct net_device *dev, u16 rxq_index, u32 flow_id,
785 u16 filter_id);
786 #endif
787 #endif /* CONFIG_RPS */
788
789 /* XPS map type and offset of the xps map within net_device->xps_maps[]. */
790 enum xps_map_type {
791 XPS_CPUS = 0,
792 XPS_RXQS,
793 XPS_MAPS_MAX,
794 };
795
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 */
801 struct xps_map {
802 unsigned int len;
803 unsigned int alloc_len;
804 struct rcu_head rcu;
805 u16 queues[];
806 };
807 #define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
808 #define XPS_MIN_MAP_ALLOC ((L1_CACHE_ALIGN(offsetof(struct xps_map, queues[1])) \
809 - sizeof(struct xps_map)) / sizeof(u16))
810
811 /*
812 * This structure holds all XPS maps for device. Maps are indexed by CPU.
813 *
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 *
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.
821 */
822 struct xps_dev_maps {
823 struct rcu_head rcu;
824 unsigned int nr_ids;
825 s16 num_tc;
826 struct xps_map __rcu *attr_map[]; /* Either CPUs map or RXQs map */
827 };
828
829 #define XPS_CPU_DEV_MAPS_SIZE(_tcs) (sizeof(struct xps_dev_maps) + \
830 (nr_cpu_ids * (_tcs) * sizeof(struct xps_map *)))
831
832 #define XPS_RXQ_DEV_MAPS_SIZE(_tcs, _rxqs) (sizeof(struct xps_dev_maps) +\
833 (_rxqs * (_tcs) * sizeof(struct xps_map *)))
834
835 #endif /* CONFIG_XPS */
836
837 #define TC_MAX_QUEUE 16
838 #define TC_BITMASK 15
839 /* HW offloaded queuing disciplines txq count and offset maps */
840 struct netdev_tc_txq {
841 u16 count;
842 u16 offset;
843 };
844
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 */
850 struct 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
862 #define MAX_PHYS_ITEM_ID_LEN 32
863
864 /* This structure holds a unique identifier to identify some
865 * physical item (port for example) used by a netdevice.
866 */
867 struct netdev_phys_item_id {
868 unsigned char id[MAX_PHYS_ITEM_ID_LEN];
869 unsigned char id_len;
870 };
871
netdev_phys_item_id_same(struct netdev_phys_item_id * a,struct netdev_phys_item_id * b)872 static 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
879 typedef u16 (*select_queue_fallback_t)(struct net_device *dev,
880 struct sk_buff *skb,
881 struct net_device *sb_dev);
882
883 enum net_device_path_type {
884 DEV_PATH_ETHERNET = 0,
885 DEV_PATH_VLAN,
886 DEV_PATH_BRIDGE,
887 DEV_PATH_PPPOE,
888 DEV_PATH_DSA,
889 DEV_PATH_MTK_WDMA,
890 };
891
892 struct net_device_path {
893 enum net_device_path_type type;
894 const struct net_device *dev;
895 union {
896 struct {
897 u16 id;
898 __be16 proto;
899 u8 h_dest[ETH_ALEN];
900 } encap;
901 struct {
902 enum {
903 DEV_PATH_BR_VLAN_KEEP,
904 DEV_PATH_BR_VLAN_TAG,
905 DEV_PATH_BR_VLAN_UNTAG,
906 DEV_PATH_BR_VLAN_UNTAG_HW,
907 } vlan_mode;
908 u16 vlan_id;
909 __be16 vlan_proto;
910 } bridge;
911 struct {
912 int port;
913 u16 proto;
914 } dsa;
915 struct {
916 u8 wdma_idx;
917 u8 queue;
918 u16 wcid;
919 u8 bss;
920 } mtk_wdma;
921 };
922 };
923
924 #define NET_DEVICE_PATH_STACK_MAX 5
925 #define NET_DEVICE_PATH_VLAN_MAX 2
926
927 struct net_device_path_stack {
928 int num_paths;
929 struct net_device_path path[NET_DEVICE_PATH_STACK_MAX];
930 };
931
932 struct net_device_path_ctx {
933 const struct net_device *dev;
934 u8 daddr[ETH_ALEN];
935
936 int num_vlans;
937 struct {
938 u16 id;
939 __be16 proto;
940 } vlan[NET_DEVICE_PATH_VLAN_MAX];
941 };
942
943 enum tc_setup_type {
944 TC_QUERY_CAPS,
945 TC_SETUP_QDISC_MQPRIO,
946 TC_SETUP_CLSU32,
947 TC_SETUP_CLSFLOWER,
948 TC_SETUP_CLSMATCHALL,
949 TC_SETUP_CLSBPF,
950 TC_SETUP_BLOCK,
951 TC_SETUP_QDISC_CBS,
952 TC_SETUP_QDISC_RED,
953 TC_SETUP_QDISC_PRIO,
954 TC_SETUP_QDISC_MQ,
955 TC_SETUP_QDISC_ETF,
956 TC_SETUP_ROOT_QDISC,
957 TC_SETUP_QDISC_GRED,
958 TC_SETUP_QDISC_TAPRIO,
959 TC_SETUP_FT,
960 TC_SETUP_QDISC_ETS,
961 TC_SETUP_QDISC_TBF,
962 TC_SETUP_QDISC_FIFO,
963 TC_SETUP_QDISC_HTB,
964 TC_SETUP_ACT,
965 };
966
967 /* These structures hold the attributes of bpf state that are being passed
968 * to the netdevice through the bpf op.
969 */
970 enum bpf_netdev_command {
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,
979 XDP_SETUP_PROG_HW,
980 /* BPF program for offload callbacks, invoked at program load time. */
981 BPF_OFFLOAD_MAP_ALLOC,
982 BPF_OFFLOAD_MAP_FREE,
983 XDP_SETUP_XSK_POOL,
984 };
985
986 struct bpf_prog_offload_ops;
987 struct netlink_ext_ack;
988 struct xdp_umem;
989 struct xdp_dev_bulk_queue;
990 struct bpf_xdp_link;
991
992 enum bpf_xdp_mode {
993 XDP_MODE_SKB = 0,
994 XDP_MODE_DRV = 1,
995 XDP_MODE_HW = 2,
996 __MAX_XDP_MODE
997 };
998
999 struct bpf_xdp_entity {
1000 struct bpf_prog *prog;
1001 struct bpf_xdp_link *link;
1002 };
1003
1004 struct netdev_bpf {
1005 enum bpf_netdev_command command;
1006 union {
1007 /* XDP_SETUP_PROG */
1008 struct {
1009 u32 flags;
1010 struct bpf_prog *prog;
1011 struct netlink_ext_ack *extack;
1012 };
1013 /* BPF_OFFLOAD_MAP_ALLOC, BPF_OFFLOAD_MAP_FREE */
1014 struct {
1015 struct bpf_offloaded_map *offmap;
1016 };
1017 /* XDP_SETUP_XSK_POOL */
1018 struct {
1019 struct xsk_buff_pool *pool;
1020 u16 queue_id;
1021 } xsk;
1022 };
1023 };
1024
1025 /* Flags for ndo_xsk_wakeup. */
1026 #define XDP_WAKEUP_RX (1 << 0)
1027 #define XDP_WAKEUP_TX (1 << 1)
1028
1029 #ifdef CONFIG_XFRM_OFFLOAD
1030 struct xfrmdev_ops {
1031 int (*xdo_dev_state_add) (struct xfrm_state *x, struct netlink_ext_ack *extack);
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);
1036 void (*xdo_dev_state_advance_esn) (struct xfrm_state *x);
1037 void (*xdo_dev_state_update_curlft) (struct xfrm_state *x);
1038 int (*xdo_dev_policy_add) (struct xfrm_policy *x, struct netlink_ext_ack *extack);
1039 void (*xdo_dev_policy_delete) (struct xfrm_policy *x);
1040 void (*xdo_dev_policy_free) (struct xfrm_policy *x);
1041 };
1042 #endif
1043
1044 struct dev_ifalias {
1045 struct rcu_head rcuhead;
1046 char ifalias[];
1047 };
1048
1049 struct devlink;
1050 struct tlsdev_ops;
1051
1052 struct netdev_net_notifier {
1053 struct list_head list;
1054 struct notifier_block *nb;
1055 };
1056
1057 /*
1058 * This structure defines the management hooks for network devices.
1059 * The following hooks can be defined; unless noted otherwise, they are
1060 * optional and can be filled with a null pointer.
1061 *
1062 * int (*ndo_init)(struct net_device *dev);
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.
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);
1073 * This function is called when a network device transitions to the up
1074 * state.
1075 *
1076 * int (*ndo_stop)(struct net_device *dev);
1077 * This function is called when a network device transitions to the down
1078 * state.
1079 *
1080 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
1081 * struct net_device *dev);
1082 * Called when a packet needs to be transmitted.
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.
1087 * Required; cannot be NULL.
1088 *
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.
1098 *
1099 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
1100 * struct net_device *sb_dev);
1101 * Called to decide which queue to use when device supports multiple
1102 * transmit queues.
1103 *
1104 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
1105 * This function is called to allow device receiver to make
1106 * changes to configuration when multicast or promiscuous is enabled.
1107 *
1108 * void (*ndo_set_rx_mode)(struct net_device *dev);
1109 * This function is called device changes address list filtering.
1110 * If driver handles unicast address filtering, it should set
1111 * IFF_UNICAST_FLT in its priv_flags.
1112 *
1113 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
1114 * This function is called when the Media Access Control address
1115 * needs to be changed. If this interface is not defined, the
1116 * MAC address can not be changed.
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);
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
1130 *
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 *
1135 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
1136 * Used to set network devices bus interface parameters. This interface
1137 * is retained for legacy reasons; new devices should use the bus
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
1142 * of a device.
1143 *
1144 * void (*ndo_tx_timeout)(struct net_device *dev, unsigned int txqueue);
1145 * Callback used when the transmitter has not made any progress
1146 * for dev->watchdog ticks.
1147 *
1148 * void (*ndo_get_stats64)(struct net_device *dev,
1149 * struct rtnl_link_stats64 *storage);
1150 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1151 * Called when a user wants to get the network device usage
1152 * statistics. Drivers must do one of the following:
1153 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
1154 * rtnl_link_stats64 structure passed by the caller.
1155 * 2. Define @ndo_get_stats to update a net_device_stats structure
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.
1161 *
1162 * bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id)
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 *
1170 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
1171 * If device supports VLAN filtering this function is called when a
1172 * VLAN id is registered.
1173 *
1174 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
1175 * If device supports VLAN filtering this function is called when a
1176 * VLAN id is unregistered.
1177 *
1178 * void (*ndo_poll_controller)(struct net_device *dev);
1179 *
1180 * SR-IOV management functions.
1181 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
1182 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan,
1183 * u8 qos, __be16 proto);
1184 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
1185 * int max_tx_rate);
1186 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
1187 * int (*ndo_set_vf_trust)(struct net_device *dev, int vf, bool setting);
1188 * int (*ndo_get_vf_config)(struct net_device *dev,
1189 * int vf, struct ifla_vf_info *ivf);
1190 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
1191 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
1192 * struct nlattr *port[]);
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
1196 * with PF and querying it may introduce a theoretical security risk.
1197 * int (*ndo_set_vf_rss_query_en)(struct net_device *dev, int vf, bool setting);
1198 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
1199 * int (*ndo_setup_tc)(struct net_device *dev, enum tc_setup_type type,
1200 * void *type_data);
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.
1205 *
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 *
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 *
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 *
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.
1255 *
1256 * Slave management functions (for bridge, bonding, etc).
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.
1262 *
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 *
1269 * Feature/offload setting functions.
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 *
1276 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
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 *
1281 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
1282 * struct net_device *dev,
1283 * const unsigned char *addr, u16 vid, u16 flags,
1284 * struct netlink_ext_ack *extack);
1285 * Adds an FDB entry to dev for addr.
1286 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
1287 * struct net_device *dev,
1288 * const unsigned char *addr, u16 vid)
1289 * Deletes the FDB entry from dev coresponding to addr.
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);
1294 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
1295 * struct net_device *dev, struct net_device *filter_dev,
1296 * int *idx)
1297 * Used to add FDB entries to dump requests. Implementers should add
1298 * entries to skb and update idx with the number of entries.
1299 *
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 *
1311 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh,
1312 * u16 flags, struct netlink_ext_ack *extack)
1313 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
1314 * struct net_device *dev, u32 filter_mask,
1315 * int nlflags)
1316 * int (*ndo_bridge_dellink)(struct net_device *dev, struct nlmsghdr *nlh,
1317 * u16 flags);
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.
1326 *
1327 * int (*ndo_get_phys_port_id)(struct net_device *dev,
1328 * struct netdev_phys_item_id *ppid);
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.
1332 *
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 *
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.
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.
1353 * int (*ndo_get_iflink)(const struct net_device *dev);
1354 * Called to get the iflink value of this device.
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.
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
1363 * forward. Setting a negative value resets the rx headroom to the
1364 * default value.
1365 * int (*ndo_bpf)(struct net_device *dev, struct netdev_bpf *bpf);
1366 * This function is used to set or query state related to XDP on the
1367 * netdevice and manage BPF offload. See definition of
1368 * enum bpf_netdev_command for details.
1369 * int (*ndo_xdp_xmit)(struct net_device *dev, int n, struct xdp_frame **xdp,
1370 * u32 flags);
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.
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.
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.
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.
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.
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
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.
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.
1409 */
1410 struct 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);
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);
1420 u16 (*ndo_select_queue)(struct net_device *dev,
1421 struct sk_buff *skb,
1422 struct net_device *sb_dev);
1423 void (*ndo_change_rx_flags)(struct net_device *dev,
1424 int flags);
1425 void (*ndo_set_rx_mode)(struct net_device *dev);
1426 int (*ndo_set_mac_address)(struct net_device *dev,
1427 void *addr);
1428 int (*ndo_validate_addr)(struct net_device *dev);
1429 int (*ndo_do_ioctl)(struct net_device *dev,
1430 struct ifreq *ifr, int cmd);
1431 int (*ndo_eth_ioctl)(struct net_device *dev,
1432 struct ifreq *ifr, int cmd);
1433 int (*ndo_siocbond)(struct net_device *dev,
1434 struct ifreq *ifr, int cmd);
1435 int (*ndo_siocwandev)(struct net_device *dev,
1436 struct if_settings *ifs);
1437 int (*ndo_siocdevprivate)(struct net_device *dev,
1438 struct ifreq *ifr,
1439 void __user *data, int cmd);
1440 int (*ndo_set_config)(struct net_device *dev,
1441 struct ifmap *map);
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 *);
1446 void (*ndo_tx_timeout) (struct net_device *dev,
1447 unsigned int txqueue);
1448
1449 void (*ndo_get_stats64)(struct net_device *dev,
1450 struct rtnl_link_stats64 *storage);
1451 bool (*ndo_has_offload_stats)(const struct net_device *dev, int attr_id);
1452 int (*ndo_get_offload_stats)(int attr_id,
1453 const struct net_device *dev,
1454 void *attr_data);
1455 struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
1456
1457 int (*ndo_vlan_rx_add_vid)(struct net_device *dev,
1458 __be16 proto, u16 vid);
1459 int (*ndo_vlan_rx_kill_vid)(struct net_device *dev,
1460 __be16 proto, u16 vid);
1461 #ifdef CONFIG_NET_POLL_CONTROLLER
1462 void (*ndo_poll_controller)(struct net_device *dev);
1463 int (*ndo_netpoll_setup)(struct net_device *dev,
1464 struct netpoll_info *info);
1465 void (*ndo_netpoll_cleanup)(struct net_device *dev);
1466 #endif
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,
1470 int queue, u16 vlan,
1471 u8 qos, __be16 proto);
1472 int (*ndo_set_vf_rate)(struct net_device *dev,
1473 int vf, int min_tx_rate,
1474 int max_tx_rate);
1475 int (*ndo_set_vf_spoofchk)(struct net_device *dev,
1476 int vf, bool setting);
1477 int (*ndo_set_vf_trust)(struct net_device *dev,
1478 int vf, bool setting);
1479 int (*ndo_get_vf_config)(struct net_device *dev,
1480 int vf,
1481 struct ifla_vf_info *ivf);
1482 int (*ndo_set_vf_link_state)(struct net_device *dev,
1483 int vf, int link_state);
1484 int (*ndo_get_vf_stats)(struct net_device *dev,
1485 int vf,
1486 struct ifla_vf_stats
1487 *vf_stats);
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);
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);
1497 int (*ndo_set_vf_guid)(struct net_device *dev,
1498 int vf, u64 guid,
1499 int guid_type);
1500 int (*ndo_set_vf_rss_query_en)(
1501 struct net_device *dev,
1502 int vf, bool setting);
1503 int (*ndo_setup_tc)(struct net_device *dev,
1504 enum tc_setup_type type,
1505 void *type_data);
1506 #if IS_ENABLED(CONFIG_FCOE)
1507 int (*ndo_fcoe_enable)(struct net_device *dev);
1508 int (*ndo_fcoe_disable)(struct net_device *dev);
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);
1515 int (*ndo_fcoe_ddp_target)(struct net_device *dev,
1516 u16 xid,
1517 struct scatterlist *sgl,
1518 unsigned int sgc);
1519 int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
1520 struct netdev_fcoe_hbainfo *hbainfo);
1521 #endif
1522
1523 #if IS_ENABLED(CONFIG_LIBFCOE)
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);
1528 #endif
1529
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
1536 int (*ndo_add_slave)(struct net_device *dev,
1537 struct net_device *slave_dev,
1538 struct netlink_ext_ack *extack);
1539 int (*ndo_del_slave)(struct net_device *dev,
1540 struct net_device *slave_dev);
1541 struct net_device* (*ndo_get_xmit_slave)(struct net_device *dev,
1542 struct sk_buff *skb,
1543 bool all_slaves);
1544 struct net_device* (*ndo_sk_get_lower_dev)(struct net_device *dev,
1545 struct sock *sk);
1546 netdev_features_t (*ndo_fix_features)(struct net_device *dev,
1547 netdev_features_t features);
1548 int (*ndo_set_features)(struct net_device *dev,
1549 netdev_features_t features);
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);
1554
1555 int (*ndo_fdb_add)(struct ndmsg *ndm,
1556 struct nlattr *tb[],
1557 struct net_device *dev,
1558 const unsigned char *addr,
1559 u16 vid,
1560 u16 flags,
1561 struct netlink_ext_ack *extack);
1562 int (*ndo_fdb_del)(struct ndmsg *ndm,
1563 struct nlattr *tb[],
1564 struct net_device *dev,
1565 const unsigned char *addr,
1566 u16 vid, struct netlink_ext_ack *extack);
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);
1572 int (*ndo_fdb_dump)(struct sk_buff *skb,
1573 struct netlink_callback *cb,
1574 struct net_device *dev,
1575 struct net_device *filter_dev,
1576 int *idx);
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);
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);
1593 int (*ndo_bridge_setlink)(struct net_device *dev,
1594 struct nlmsghdr *nlh,
1595 u16 flags,
1596 struct netlink_ext_ack *extack);
1597 int (*ndo_bridge_getlink)(struct sk_buff *skb,
1598 u32 pid, u32 seq,
1599 struct net_device *dev,
1600 u32 filter_mask,
1601 int nlflags);
1602 int (*ndo_bridge_dellink)(struct net_device *dev,
1603 struct nlmsghdr *nlh,
1604 u16 flags);
1605 int (*ndo_change_carrier)(struct net_device *dev,
1606 bool new_carrier);
1607 int (*ndo_get_phys_port_id)(struct net_device *dev,
1608 struct netdev_phys_item_id *ppid);
1609 int (*ndo_get_port_parent_id)(struct net_device *dev,
1610 struct netdev_phys_item_id *ppid);
1611 int (*ndo_get_phys_port_name)(struct net_device *dev,
1612 char *name, size_t len);
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
1618 int (*ndo_set_tx_maxrate)(struct net_device *dev,
1619 int queue_index,
1620 u32 maxrate);
1621 int (*ndo_get_iflink)(const struct net_device *dev);
1622 int (*ndo_fill_metadata_dst)(struct net_device *dev,
1623 struct sk_buff *skb);
1624 void (*ndo_set_rx_headroom)(struct net_device *dev,
1625 int needed_headroom);
1626 int (*ndo_bpf)(struct net_device *dev,
1627 struct netdev_bpf *bpf);
1628 int (*ndo_xdp_xmit)(struct net_device *dev, int n,
1629 struct xdp_frame **xdp,
1630 u32 flags);
1631 struct net_device * (*ndo_xdp_get_xmit_slave)(struct net_device *dev,
1632 struct xdp_buff *xdp);
1633 int (*ndo_xsk_wakeup)(struct net_device *dev,
1634 u32 queue_id, u32 flags);
1635 int (*ndo_tunnel_ctl)(struct net_device *dev,
1636 struct ip_tunnel_parm *p, int cmd);
1637 struct net_device * (*ndo_get_peer_dev)(struct net_device *dev);
1638 int (*ndo_fill_forward_path)(struct net_device_path_ctx *ctx,
1639 struct net_device_path *path);
1640 ktime_t (*ndo_get_tstamp)(struct net_device *dev,
1641 const struct skb_shared_hwtstamps *hwtstamps,
1642 bool cycles);
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);
1648 };
1649
1650 /**
1651 * enum netdev_priv_flags - &struct net_device priv_flags
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
1655 * userspace; this means that the order of these flags can change
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
1662 * @IFF_BONDING: bonding master or slave
1663 * @IFF_ISATAP: ISATAP interface (RFC4214)
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
1679 * @IFF_XMIT_DST_RELEASE_PERM: IFF_XMIT_DST_RELEASE not taking into account
1680 * underlying stacked devices
1681 * @IFF_L3MDEV_MASTER: device is an L3 master device
1682 * @IFF_NO_QUEUE: device can run without qdisc attached
1683 * @IFF_OPENVSWITCH: device is a Open vSwitch master
1684 * @IFF_L3MDEV_SLAVE: device is enslaved to an L3 master device
1685 * @IFF_TEAM: device is a team device
1686 * @IFF_RXFH_CONFIGURED: device has had Rx Flow indirection table configured
1687 * @IFF_PHONY_HEADROOM: the headroom value is controlled by an external
1688 * entity (i.e. the master device for bridged veth)
1689 * @IFF_MACSEC: device is a MACsec device
1690 * @IFF_NO_RX_HANDLER: device doesn't support the rx_handler hook
1691 * @IFF_FAILOVER: device is a failover master device
1692 * @IFF_FAILOVER_SLAVE: device is lower dev of a failover master device
1693 * @IFF_L3MDEV_RX_HANDLER: only invoke the rx handler of L3 master device
1694 * @IFF_NO_ADDRCONF: prevent ipv6 addrconf
1695 * @IFF_TX_SKB_NO_LINEAR: device/driver is capable of xmitting frames with
1696 * skb_headlen(skb) == 0 (data starts from frag0)
1697 * @IFF_CHANGE_PROTO_DOWN: device supports setting carrier via IFLA_PROTO_DOWN
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.
1701 */
1702 enum netdev_priv_flags {
1703 IFF_802_1Q_VLAN = 1<<0,
1704 IFF_EBRIDGE = 1<<1,
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,
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,
1729 IFF_NO_RX_HANDLER = 1<<26,
1730 IFF_FAILOVER = 1<<27,
1731 IFF_FAILOVER_SLAVE = 1<<28,
1732 IFF_L3MDEV_RX_HANDLER = 1<<29,
1733 IFF_NO_ADDRCONF = BIT_ULL(30),
1734 IFF_TX_SKB_NO_LINEAR = BIT_ULL(31),
1735 IFF_CHANGE_PROTO_DOWN = BIT_ULL(32),
1736 IFF_SEE_ALL_HWTSTAMP_REQUESTS = BIT_ULL(33),
1737 };
1738
1739 #define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1740 #define IFF_EBRIDGE IFF_EBRIDGE
1741 #define IFF_BONDING IFF_BONDING
1742 #define IFF_ISATAP IFF_ISATAP
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
1756 #define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
1757 #define IFF_L3MDEV_MASTER IFF_L3MDEV_MASTER
1758 #define IFF_NO_QUEUE IFF_NO_QUEUE
1759 #define IFF_OPENVSWITCH IFF_OPENVSWITCH
1760 #define IFF_L3MDEV_SLAVE IFF_L3MDEV_SLAVE
1761 #define IFF_TEAM IFF_TEAM
1762 #define IFF_RXFH_CONFIGURED IFF_RXFH_CONFIGURED
1763 #define IFF_PHONY_HEADROOM IFF_PHONY_HEADROOM
1764 #define IFF_MACSEC IFF_MACSEC
1765 #define IFF_NO_RX_HANDLER IFF_NO_RX_HANDLER
1766 #define IFF_FAILOVER IFF_FAILOVER
1767 #define IFF_FAILOVER_SLAVE IFF_FAILOVER_SLAVE
1768 #define IFF_L3MDEV_RX_HANDLER IFF_L3MDEV_RX_HANDLER
1769 #define IFF_TX_SKB_NO_LINEAR IFF_TX_SKB_NO_LINEAR
1770
1771 /* Specifies the type of the struct net_device::ml_priv pointer */
1772 enum netdev_ml_priv_type {
1773 ML_PRIV_NONE,
1774 ML_PRIV_CAN,
1775 };
1776
1777 enum netdev_stat_type {
1778 NETDEV_PCPU_STAT_NONE,
1779 NETDEV_PCPU_STAT_LSTATS, /* struct pcpu_lstats */
1780 NETDEV_PCPU_STAT_TSTATS, /* struct pcpu_sw_netstats */
1781 NETDEV_PCPU_STAT_DSTATS, /* struct pcpu_dstats */
1782 };
1783
1784 /**
1785 * struct net_device - The DEVICE structure.
1786 *
1787 * Actually, this whole structure is a big mistake. It mixes I/O
1788 * data with strictly "high-level" data, and it has to know about
1789 * almost every data structure used in the INET module.
1790 *
1791 * @name: This is the first field of the "visible" part of this structure
1792 * (i.e. as seen by users in the "Space.c" file). It is the name
1793 * of the interface.
1794 *
1795 * @name_node: Name hashlist node
1796 * @ifalias: SNMP alias
1797 * @mem_end: Shared memory end
1798 * @mem_start: Shared memory start
1799 * @base_addr: Device I/O address
1800 * @irq: Device IRQ number
1801 *
1802 * @state: Generic network queuing layer state, see netdev_state_t
1803 * @dev_list: The global list of network devices
1804 * @napi_list: List entry used for polling NAPI devices
1805 * @unreg_list: List entry when we are unregistering the
1806 * device; see the function unregister_netdev
1807 * @close_list: List entry used when we are closing the device
1808 * @ptype_all: Device-specific packet handlers for all protocols
1809 * @ptype_specific: Device-specific, protocol-specific packet handlers
1810 *
1811 * @adj_list: Directly linked devices, like slaves for bonding
1812 * @features: Currently active device features
1813 * @hw_features: User-changeable features
1814 *
1815 * @wanted_features: User-requested features
1816 * @vlan_features: Mask of features inheritable by VLAN devices
1817 *
1818 * @hw_enc_features: Mask of features inherited by encapsulating devices
1819 * This field indicates what encapsulation
1820 * offloads the hardware is capable of doing,
1821 * and drivers will need to set them appropriately.
1822 *
1823 * @mpls_features: Mask of features inheritable by MPLS
1824 * @gso_partial_features: value(s) from NETIF_F_GSO\*
1825 *
1826 * @ifindex: interface index
1827 * @group: The group the device belongs to
1828 *
1829 * @stats: Statistics struct, which was left as a legacy, use
1830 * rtnl_link_stats64 instead
1831 *
1832 * @core_stats: core networking counters,
1833 * do not use this in drivers
1834 * @carrier_up_count: Number of times the carrier has been up
1835 * @carrier_down_count: Number of times the carrier has been down
1836 *
1837 * @wireless_handlers: List of functions to handle Wireless Extensions,
1838 * instead of ioctl,
1839 * see <net/iw_handler.h> for details.
1840 * @wireless_data: Instance data managed by the core of wireless extensions
1841 *
1842 * @netdev_ops: Includes several pointers to callbacks,
1843 * if one wants to override the ndo_*() functions
1844 * @xdp_metadata_ops: Includes pointers to XDP metadata callbacks.
1845 * @ethtool_ops: Management operations
1846 * @l3mdev_ops: Layer 3 master device operations
1847 * @ndisc_ops: Includes callbacks for different IPv6 neighbour
1848 * discovery handling. Necessary for e.g. 6LoWPAN.
1849 * @xfrmdev_ops: Transformation offload operations
1850 * @tlsdev_ops: Transport Layer Security offload operations
1851 * @header_ops: Includes callbacks for creating,parsing,caching,etc
1852 * of Layer 2 headers.
1853 *
1854 * @flags: Interface flags (a la BSD)
1855 * @xdp_features: XDP capability supported by the device
1856 * @priv_flags: Like 'flags' but invisible to userspace,
1857 * see if.h for the definitions
1858 * @gflags: Global flags ( kept as legacy )
1859 * @padded: How much padding added by alloc_netdev()
1860 * @operstate: RFC2863 operstate
1861 * @link_mode: Mapping policy to operstate
1862 * @if_port: Selectable AUI, TP, ...
1863 * @dma: DMA channel
1864 * @mtu: Interface MTU value
1865 * @min_mtu: Interface Minimum MTU value
1866 * @max_mtu: Interface Maximum MTU value
1867 * @type: Interface hardware type
1868 * @hard_header_len: Maximum hardware header length.
1869 * @min_header_len: Minimum hardware header length
1870 *
1871 * @needed_headroom: Extra headroom the hardware may need, but not in all
1872 * cases can this be guaranteed
1873 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1874 * cases can this be guaranteed. Some cases also use
1875 * LL_MAX_HEADER instead to allocate the skb
1876 *
1877 * interface address info:
1878 *
1879 * @perm_addr: Permanent hw address
1880 * @addr_assign_type: Hw address assignment type
1881 * @addr_len: Hardware address length
1882 * @upper_level: Maximum depth level of upper devices.
1883 * @lower_level: Maximum depth level of lower devices.
1884 * @neigh_priv_len: Used in neigh_alloc()
1885 * @dev_id: Used to differentiate devices that share
1886 * the same link layer address
1887 * @dev_port: Used to differentiate devices that share
1888 * the same function
1889 * @addr_list_lock: XXX: need comments on this one
1890 * @name_assign_type: network interface name assignment type
1891 * @uc_promisc: Counter that indicates promiscuous mode
1892 * has been enabled due to the need to listen to
1893 * additional unicast addresses in a device that
1894 * does not implement ndo_set_rx_mode()
1895 * @uc: unicast mac addresses
1896 * @mc: multicast mac addresses
1897 * @dev_addrs: list of device hw addresses
1898 * @queues_kset: Group of all Kobjects in the Tx and RX queues
1899 * @promiscuity: Number of times the NIC is told to work in
1900 * promiscuous mode; if it becomes 0 the NIC will
1901 * exit promiscuous mode
1902 * @allmulti: Counter, enables or disables allmulticast mode
1903 *
1904 * @vlan_info: VLAN info
1905 * @dsa_ptr: dsa specific data
1906 * @tipc_ptr: TIPC specific data
1907 * @atalk_ptr: AppleTalk link
1908 * @ip_ptr: IPv4 specific data
1909 * @ip6_ptr: IPv6 specific data
1910 * @ax25_ptr: AX.25 specific data
1911 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1912 * @ieee802154_ptr: IEEE 802.15.4 low-rate Wireless Personal Area Network
1913 * device struct
1914 * @mpls_ptr: mpls_dev struct pointer
1915 * @mctp_ptr: MCTP specific data
1916 *
1917 * @dev_addr: Hw address (before bcast,
1918 * because most packets are unicast)
1919 *
1920 * @_rx: Array of RX queues
1921 * @num_rx_queues: Number of RX queues
1922 * allocated at register_netdev() time
1923 * @real_num_rx_queues: Number of RX queues currently active in device
1924 * @xdp_prog: XDP sockets filter program pointer
1925 * @gro_flush_timeout: timeout for GRO layer in NAPI
1926 * @napi_defer_hard_irqs: If not zero, provides a counter that would
1927 * allow to avoid NIC hard IRQ, on busy queues.
1928 *
1929 * @rx_handler: handler for received packets
1930 * @rx_handler_data: XXX: need comments on this one
1931 * @tcx_ingress: BPF & clsact qdisc specific data for ingress processing
1932 * @ingress_queue: XXX: need comments on this one
1933 * @nf_hooks_ingress: netfilter hooks executed for ingress packets
1934 * @broadcast: hw bcast address
1935 *
1936 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1937 * indexed by RX queue number. Assigned by driver.
1938 * This must only be set if the ndo_rx_flow_steer
1939 * operation is defined
1940 * @index_hlist: Device index hash chain
1941 *
1942 * @_tx: Array of TX queues
1943 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1944 * @real_num_tx_queues: Number of TX queues currently active in device
1945 * @qdisc: Root qdisc from userspace point of view
1946 * @tx_queue_len: Max frames per queue allowed
1947 * @tx_global_lock: XXX: need comments on this one
1948 * @xdp_bulkq: XDP device bulk queue
1949 * @xps_maps: all CPUs/RXQs maps for XPS device
1950 *
1951 * @xps_maps: XXX: need comments on this one
1952 * @tcx_egress: BPF & clsact qdisc specific data for egress processing
1953 * @nf_hooks_egress: netfilter hooks executed for egress packets
1954 * @qdisc_hash: qdisc hash table
1955 * @watchdog_timeo: Represents the timeout that is used by
1956 * the watchdog (see dev_watchdog())
1957 * @watchdog_timer: List of timers
1958 *
1959 * @proto_down_reason: reason a netdev interface is held down
1960 * @pcpu_refcnt: Number of references to this device
1961 * @dev_refcnt: Number of references to this device
1962 * @refcnt_tracker: Tracker directory for tracked references to this device
1963 * @todo_list: Delayed register/unregister
1964 * @link_watch_list: XXX: need comments on this one
1965 *
1966 * @reg_state: Register/unregister state machine
1967 * @dismantle: Device is going to be freed
1968 * @rtnl_link_state: This enum represents the phases of creating
1969 * a new link
1970 *
1971 * @needs_free_netdev: Should unregister perform free_netdev?
1972 * @priv_destructor: Called from unregister
1973 * @npinfo: XXX: need comments on this one
1974 * @nd_net: Network namespace this network device is inside
1975 *
1976 * @ml_priv: Mid-layer private
1977 * @ml_priv_type: Mid-layer private type
1978 *
1979 * @pcpu_stat_type: Type of device statistics which the core should
1980 * allocate/free: none, lstats, tstats, dstats. none
1981 * means the driver is handling statistics allocation/
1982 * freeing internally.
1983 * @lstats: Loopback statistics: packets, bytes
1984 * @tstats: Tunnel statistics: RX/TX packets, RX/TX bytes
1985 * @dstats: Dummy statistics: RX/TX/drop packets, RX/TX bytes
1986 *
1987 * @garp_port: GARP
1988 * @mrp_port: MRP
1989 *
1990 * @dm_private: Drop monitor private
1991 *
1992 * @dev: Class/net/name entry
1993 * @sysfs_groups: Space for optional device, statistics and wireless
1994 * sysfs groups
1995 *
1996 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1997 * @rtnl_link_ops: Rtnl_link_ops
1998 *
1999 * @gso_max_size: Maximum size of generic segmentation offload
2000 * @tso_max_size: Device (as in HW) limit on the max TSO request size
2001 * @gso_max_segs: Maximum number of segments that can be passed to the
2002 * NIC for GSO
2003 * @tso_max_segs: Device (as in HW) limit on the max TSO segment count
2004 * @gso_ipv4_max_size: Maximum size of generic segmentation offload,
2005 * for IPv4.
2006 *
2007 * @dcbnl_ops: Data Center Bridging netlink ops
2008 * @num_tc: Number of traffic classes in the net device
2009 * @tc_to_txq: XXX: need comments on this one
2010 * @prio_tc_map: XXX: need comments on this one
2011 *
2012 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
2013 *
2014 * @priomap: XXX: need comments on this one
2015 * @phydev: Physical device may attach itself
2016 * for hardware timestamping
2017 * @sfp_bus: attached &struct sfp_bus structure.
2018 *
2019 * @qdisc_tx_busylock: lockdep class annotating Qdisc->busylock spinlock
2020 *
2021 * @proto_down: protocol port state information can be sent to the
2022 * switch driver and used to set the phys state of the
2023 * switch port.
2024 *
2025 * @wol_enabled: Wake-on-LAN is enabled
2026 *
2027 * @threaded: napi threaded mode is enabled
2028 *
2029 * @net_notifier_list: List of per-net netdev notifier block
2030 * that follow this device when it is moved
2031 * to another network namespace.
2032 *
2033 * @macsec_ops: MACsec offloading ops
2034 *
2035 * @udp_tunnel_nic_info: static structure describing the UDP tunnel
2036 * offload capabilities of the device
2037 * @udp_tunnel_nic: UDP tunnel offload state
2038 * @xdp_state: stores info on attached XDP BPF programs
2039 *
2040 * @nested_level: Used as a parameter of spin_lock_nested() of
2041 * dev->addr_list_lock.
2042 * @unlink_list: As netif_addr_lock() can be called recursively,
2043 * keep a list of interfaces to be deleted.
2044 * @gro_max_size: Maximum size of aggregated packet in generic
2045 * receive offload (GRO)
2046 * @gro_ipv4_max_size: Maximum size of aggregated packet in generic
2047 * receive offload (GRO), for IPv4.
2048 * @xdp_zc_max_segs: Maximum number of segments supported by AF_XDP
2049 * zero copy driver
2050 *
2051 * @dev_addr_shadow: Copy of @dev_addr to catch direct writes.
2052 * @linkwatch_dev_tracker: refcount tracker used by linkwatch.
2053 * @watchdog_dev_tracker: refcount tracker used by watchdog.
2054 * @dev_registered_tracker: tracker for reference held while
2055 * registered
2056 * @offload_xstats_l3: L3 HW stats for this netdevice.
2057 *
2058 * @devlink_port: Pointer to related devlink port structure.
2059 * Assigned by a driver before netdev registration using
2060 * SET_NETDEV_DEVLINK_PORT macro. This pointer is static
2061 * during the time netdevice is registered.
2062 *
2063 * FIXME: cleanup struct net_device such that network protocol info
2064 * moves out.
2065 */
2066
2067 struct net_device {
2068 char name[IFNAMSIZ];
2069 struct netdev_name_node *name_node;
2070 struct dev_ifalias __rcu *ifalias;
2071 /*
2072 * I/O specific fields
2073 * FIXME: Merge these and struct ifmap into one
2074 */
2075 unsigned long mem_end;
2076 unsigned long mem_start;
2077 unsigned long base_addr;
2078
2079 /*
2080 * Some hardware also needs these fields (state,dev_list,
2081 * napi_list,unreg_list,close_list) but they are not
2082 * part of the usual set specified in Space.c.
2083 */
2084
2085 unsigned long state;
2086
2087 struct list_head dev_list;
2088 struct list_head napi_list;
2089 struct list_head unreg_list;
2090 struct list_head close_list;
2091 struct list_head ptype_all;
2092 struct list_head ptype_specific;
2093
2094 struct {
2095 struct list_head upper;
2096 struct list_head lower;
2097 } adj_list;
2098
2099 /* Read-mostly cache-line for fast-path access */
2100 unsigned int flags;
2101 xdp_features_t xdp_features;
2102 unsigned long long priv_flags;
2103 const struct net_device_ops *netdev_ops;
2104 const struct xdp_metadata_ops *xdp_metadata_ops;
2105 int ifindex;
2106 unsigned short gflags;
2107 unsigned short hard_header_len;
2108
2109 /* Note : dev->mtu is often read without holding a lock.
2110 * Writers usually hold RTNL.
2111 * It is recommended to use READ_ONCE() to annotate the reads,
2112 * and to use WRITE_ONCE() to annotate the writes.
2113 */
2114 unsigned int mtu;
2115 unsigned short needed_headroom;
2116 unsigned short needed_tailroom;
2117
2118 netdev_features_t features;
2119 netdev_features_t hw_features;
2120 netdev_features_t wanted_features;
2121 netdev_features_t vlan_features;
2122 netdev_features_t hw_enc_features;
2123 netdev_features_t mpls_features;
2124 netdev_features_t gso_partial_features;
2125
2126 unsigned int min_mtu;
2127 unsigned int max_mtu;
2128 unsigned short type;
2129 unsigned char min_header_len;
2130 unsigned char name_assign_type;
2131
2132 int group;
2133
2134 struct net_device_stats stats; /* not used by modern drivers */
2135
2136 struct net_device_core_stats __percpu *core_stats;
2137
2138 /* Stats to monitor link on/off, flapping */
2139 atomic_t carrier_up_count;
2140 atomic_t carrier_down_count;
2141
2142 #ifdef CONFIG_WIRELESS_EXT
2143 const struct iw_handler_def *wireless_handlers;
2144 struct iw_public_data *wireless_data;
2145 #endif
2146 const struct ethtool_ops *ethtool_ops;
2147 #ifdef CONFIG_NET_L3_MASTER_DEV
2148 const struct l3mdev_ops *l3mdev_ops;
2149 #endif
2150 #if IS_ENABLED(CONFIG_IPV6)
2151 const struct ndisc_ops *ndisc_ops;
2152 #endif
2153
2154 #ifdef CONFIG_XFRM_OFFLOAD
2155 const struct xfrmdev_ops *xfrmdev_ops;
2156 #endif
2157
2158 #if IS_ENABLED(CONFIG_TLS_DEVICE)
2159 const struct tlsdev_ops *tlsdev_ops;
2160 #endif
2161
2162 const struct header_ops *header_ops;
2163
2164 unsigned char operstate;
2165 unsigned char link_mode;
2166
2167 unsigned char if_port;
2168 unsigned char dma;
2169
2170 /* Interface address info. */
2171 unsigned char perm_addr[MAX_ADDR_LEN];
2172 unsigned char addr_assign_type;
2173 unsigned char addr_len;
2174 unsigned char upper_level;
2175 unsigned char lower_level;
2176
2177 unsigned short neigh_priv_len;
2178 unsigned short dev_id;
2179 unsigned short dev_port;
2180 unsigned short padded;
2181
2182 spinlock_t addr_list_lock;
2183 int irq;
2184
2185 struct netdev_hw_addr_list uc;
2186 struct netdev_hw_addr_list mc;
2187 struct netdev_hw_addr_list dev_addrs;
2188
2189 #ifdef CONFIG_SYSFS
2190 struct kset *queues_kset;
2191 #endif
2192 #ifdef CONFIG_LOCKDEP
2193 struct list_head unlink_list;
2194 #endif
2195 unsigned int promiscuity;
2196 unsigned int allmulti;
2197 bool uc_promisc;
2198 #ifdef CONFIG_LOCKDEP
2199 unsigned char nested_level;
2200 #endif
2201
2202
2203 /* Protocol-specific pointers */
2204
2205 struct in_device __rcu *ip_ptr;
2206 struct inet6_dev __rcu *ip6_ptr;
2207 #if IS_ENABLED(CONFIG_VLAN_8021Q)
2208 struct vlan_info __rcu *vlan_info;
2209 #endif
2210 #if IS_ENABLED(CONFIG_NET_DSA)
2211 struct dsa_port *dsa_ptr;
2212 #endif
2213 #if IS_ENABLED(CONFIG_TIPC)
2214 struct tipc_bearer __rcu *tipc_ptr;
2215 #endif
2216 #if IS_ENABLED(CONFIG_ATALK)
2217 void *atalk_ptr;
2218 #endif
2219 #if IS_ENABLED(CONFIG_AX25)
2220 void *ax25_ptr;
2221 #endif
2222 #if IS_ENABLED(CONFIG_CFG80211)
2223 struct wireless_dev *ieee80211_ptr;
2224 #endif
2225 #if IS_ENABLED(CONFIG_IEEE802154) || IS_ENABLED(CONFIG_6LOWPAN)
2226 struct wpan_dev *ieee802154_ptr;
2227 #endif
2228 #if IS_ENABLED(CONFIG_MPLS_ROUTING)
2229 struct mpls_dev __rcu *mpls_ptr;
2230 #endif
2231 #if IS_ENABLED(CONFIG_MCTP)
2232 struct mctp_dev __rcu *mctp_ptr;
2233 #endif
2234
2235 /*
2236 * Cache lines mostly used on receive path (including eth_type_trans())
2237 */
2238 /* Interface address info used in eth_type_trans() */
2239 const unsigned char *dev_addr;
2240
2241 struct netdev_rx_queue *_rx;
2242 unsigned int num_rx_queues;
2243 unsigned int real_num_rx_queues;
2244
2245 struct bpf_prog __rcu *xdp_prog;
2246 unsigned long gro_flush_timeout;
2247 int napi_defer_hard_irqs;
2248 #define GRO_LEGACY_MAX_SIZE 65536u
2249 /* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
2250 * and shinfo->gso_segs is a 16bit field.
2251 */
2252 #define GRO_MAX_SIZE (8 * 65535u)
2253 unsigned int gro_max_size;
2254 unsigned int gro_ipv4_max_size;
2255 unsigned int xdp_zc_max_segs;
2256 rx_handler_func_t __rcu *rx_handler;
2257 void __rcu *rx_handler_data;
2258 #ifdef CONFIG_NET_XGRESS
2259 struct bpf_mprog_entry __rcu *tcx_ingress;
2260 #endif
2261 struct netdev_queue __rcu *ingress_queue;
2262 #ifdef CONFIG_NETFILTER_INGRESS
2263 struct nf_hook_entries __rcu *nf_hooks_ingress;
2264 #endif
2265
2266 unsigned char broadcast[MAX_ADDR_LEN];
2267 #ifdef CONFIG_RFS_ACCEL
2268 struct cpu_rmap *rx_cpu_rmap;
2269 #endif
2270 struct hlist_node index_hlist;
2271
2272 /*
2273 * Cache lines mostly used on transmit path
2274 */
2275 struct netdev_queue *_tx ____cacheline_aligned_in_smp;
2276 unsigned int num_tx_queues;
2277 unsigned int real_num_tx_queues;
2278 struct Qdisc __rcu *qdisc;
2279 unsigned int tx_queue_len;
2280 spinlock_t tx_global_lock;
2281
2282 struct xdp_dev_bulk_queue __percpu *xdp_bulkq;
2283
2284 #ifdef CONFIG_XPS
2285 struct xps_dev_maps __rcu *xps_maps[XPS_MAPS_MAX];
2286 #endif
2287 #ifdef CONFIG_NET_XGRESS
2288 struct bpf_mprog_entry __rcu *tcx_egress;
2289 #endif
2290 #ifdef CONFIG_NETFILTER_EGRESS
2291 struct nf_hook_entries __rcu *nf_hooks_egress;
2292 #endif
2293
2294 #ifdef CONFIG_NET_SCHED
2295 DECLARE_HASHTABLE (qdisc_hash, 4);
2296 #endif
2297 /* These may be needed for future network-power-down code. */
2298 struct timer_list watchdog_timer;
2299 int watchdog_timeo;
2300
2301 u32 proto_down_reason;
2302
2303 struct list_head todo_list;
2304
2305 #ifdef CONFIG_PCPU_DEV_REFCNT
2306 int __percpu *pcpu_refcnt;
2307 #else
2308 refcount_t dev_refcnt;
2309 #endif
2310 struct ref_tracker_dir refcnt_tracker;
2311
2312 struct list_head link_watch_list;
2313
2314 enum { NETREG_UNINITIALIZED=0,
2315 NETREG_REGISTERED, /* completed register_netdevice */
2316 NETREG_UNREGISTERING, /* called unregister_netdevice */
2317 NETREG_UNREGISTERED, /* completed unregister todo */
2318 NETREG_RELEASED, /* called free_netdev */
2319 NETREG_DUMMY, /* dummy device for NAPI poll */
2320 } reg_state:8;
2321
2322 bool dismantle;
2323
2324 enum {
2325 RTNL_LINK_INITIALIZED,
2326 RTNL_LINK_INITIALIZING,
2327 } rtnl_link_state:16;
2328
2329 bool needs_free_netdev;
2330 void (*priv_destructor)(struct net_device *dev);
2331
2332 #ifdef CONFIG_NETPOLL
2333 struct netpoll_info __rcu *npinfo;
2334 #endif
2335
2336 possible_net_t nd_net;
2337
2338 /* mid-layer private */
2339 void *ml_priv;
2340 enum netdev_ml_priv_type ml_priv_type;
2341
2342 enum netdev_stat_type pcpu_stat_type:8;
2343 union {
2344 struct pcpu_lstats __percpu *lstats;
2345 struct pcpu_sw_netstats __percpu *tstats;
2346 struct pcpu_dstats __percpu *dstats;
2347 };
2348
2349 #if IS_ENABLED(CONFIG_GARP)
2350 struct garp_port __rcu *garp_port;
2351 #endif
2352 #if IS_ENABLED(CONFIG_MRP)
2353 struct mrp_port __rcu *mrp_port;
2354 #endif
2355 #if IS_ENABLED(CONFIG_NET_DROP_MONITOR)
2356 struct dm_hw_stat_delta __rcu *dm_private;
2357 #endif
2358 struct device dev;
2359 const struct attribute_group *sysfs_groups[4];
2360 const struct attribute_group *sysfs_rx_queue_group;
2361
2362 const struct rtnl_link_ops *rtnl_link_ops;
2363
2364 /* for setting kernel sock attribute on TCP connection setup */
2365 #define GSO_MAX_SEGS 65535u
2366 #define GSO_LEGACY_MAX_SIZE 65536u
2367 /* TCP minimal MSS is 8 (TCP_MIN_GSO_SIZE),
2368 * and shinfo->gso_segs is a 16bit field.
2369 */
2370 #define GSO_MAX_SIZE (8 * GSO_MAX_SEGS)
2371
2372 unsigned int gso_max_size;
2373 #define TSO_LEGACY_MAX_SIZE 65536
2374 #define TSO_MAX_SIZE UINT_MAX
2375 unsigned int tso_max_size;
2376 u16 gso_max_segs;
2377 #define TSO_MAX_SEGS U16_MAX
2378 u16 tso_max_segs;
2379 unsigned int gso_ipv4_max_size;
2380
2381 #ifdef CONFIG_DCB
2382 const struct dcbnl_rtnl_ops *dcbnl_ops;
2383 #endif
2384 s16 num_tc;
2385 struct netdev_tc_txq tc_to_txq[TC_MAX_QUEUE];
2386 u8 prio_tc_map[TC_BITMASK + 1];
2387
2388 #if IS_ENABLED(CONFIG_FCOE)
2389 unsigned int fcoe_ddp_xid;
2390 #endif
2391 #if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
2392 struct netprio_map __rcu *priomap;
2393 #endif
2394 struct phy_device *phydev;
2395 struct sfp_bus *sfp_bus;
2396 struct lock_class_key *qdisc_tx_busylock;
2397 bool proto_down;
2398 unsigned wol_enabled:1;
2399 unsigned threaded:1;
2400
2401 struct list_head net_notifier_list;
2402
2403 #if IS_ENABLED(CONFIG_MACSEC)
2404 /* MACsec management functions */
2405 const struct macsec_ops *macsec_ops;
2406 #endif
2407 const struct udp_tunnel_nic_info *udp_tunnel_nic_info;
2408 struct udp_tunnel_nic *udp_tunnel_nic;
2409
2410 /* protected by rtnl_lock */
2411 struct bpf_xdp_entity xdp_state[__MAX_XDP_MODE];
2412
2413 u8 dev_addr_shadow[MAX_ADDR_LEN];
2414 netdevice_tracker linkwatch_dev_tracker;
2415 netdevice_tracker watchdog_dev_tracker;
2416 netdevice_tracker dev_registered_tracker;
2417 struct rtnl_hw_stats64 *offload_xstats_l3;
2418
2419 struct devlink_port *devlink_port;
2420 };
2421 #define to_net_dev(d) container_of(d, struct net_device, dev)
2422
2423 /*
2424 * Driver should use this to assign devlink port instance to a netdevice
2425 * before it registers the netdevice. Therefore devlink_port is static
2426 * during the netdev lifetime after it is registered.
2427 */
2428 #define SET_NETDEV_DEVLINK_PORT(dev, port) \
2429 ({ \
2430 WARN_ON((dev)->reg_state != NETREG_UNINITIALIZED); \
2431 ((dev)->devlink_port = (port)); \
2432 })
2433
netif_elide_gro(const struct net_device * dev)2434 static inline bool netif_elide_gro(const struct net_device *dev)
2435 {
2436 if (!(dev->features & NETIF_F_GRO) || dev->xdp_prog)
2437 return true;
2438 return false;
2439 }
2440
2441 #define NETDEV_ALIGN 32
2442
2443 static inline
netdev_get_prio_tc_map(const struct net_device * dev,u32 prio)2444 int netdev_get_prio_tc_map(const struct net_device *dev, u32 prio)
2445 {
2446 return dev->prio_tc_map[prio & TC_BITMASK];
2447 }
2448
2449 static inline
netdev_set_prio_tc_map(struct net_device * dev,u8 prio,u8 tc)2450 int netdev_set_prio_tc_map(struct net_device *dev, u8 prio, u8 tc)
2451 {
2452 if (tc >= dev->num_tc)
2453 return -EINVAL;
2454
2455 dev->prio_tc_map[prio & TC_BITMASK] = tc & TC_BITMASK;
2456 return 0;
2457 }
2458
2459 int netdev_txq_to_tc(struct net_device *dev, unsigned int txq);
2460 void netdev_reset_tc(struct net_device *dev);
2461 int netdev_set_tc_queue(struct net_device *dev, u8 tc, u16 count, u16 offset);
2462 int netdev_set_num_tc(struct net_device *dev, u8 num_tc);
2463
2464 static inline
netdev_get_num_tc(struct net_device * dev)2465 int netdev_get_num_tc(struct net_device *dev)
2466 {
2467 return dev->num_tc;
2468 }
2469
net_prefetch(void * p)2470 static inline void net_prefetch(void *p)
2471 {
2472 prefetch(p);
2473 #if L1_CACHE_BYTES < 128
2474 prefetch((u8 *)p + L1_CACHE_BYTES);
2475 #endif
2476 }
2477
net_prefetchw(void * p)2478 static inline void net_prefetchw(void *p)
2479 {
2480 prefetchw(p);
2481 #if L1_CACHE_BYTES < 128
2482 prefetchw((u8 *)p + L1_CACHE_BYTES);
2483 #endif
2484 }
2485
2486 void netdev_unbind_sb_channel(struct net_device *dev,
2487 struct net_device *sb_dev);
2488 int netdev_bind_sb_channel_queue(struct net_device *dev,
2489 struct net_device *sb_dev,
2490 u8 tc, u16 count, u16 offset);
2491 int netdev_set_sb_channel(struct net_device *dev, u16 channel);
netdev_get_sb_channel(struct net_device * dev)2492 static inline int netdev_get_sb_channel(struct net_device *dev)
2493 {
2494 return max_t(int, -dev->num_tc, 0);
2495 }
2496
2497 static inline
netdev_get_tx_queue(const struct net_device * dev,unsigned int index)2498 struct netdev_queue *netdev_get_tx_queue(const struct net_device *dev,
2499 unsigned int index)
2500 {
2501 DEBUG_NET_WARN_ON_ONCE(index >= dev->num_tx_queues);
2502 return &dev->_tx[index];
2503 }
2504
skb_get_tx_queue(const struct net_device * dev,const struct sk_buff * skb)2505 static inline struct netdev_queue *skb_get_tx_queue(const struct net_device *dev,
2506 const struct sk_buff *skb)
2507 {
2508 return netdev_get_tx_queue(dev, skb_get_queue_mapping(skb));
2509 }
2510
netdev_for_each_tx_queue(struct net_device * dev,void (* f)(struct net_device *,struct netdev_queue *,void *),void * arg)2511 static inline void netdev_for_each_tx_queue(struct net_device *dev,
2512 void (*f)(struct net_device *,
2513 struct netdev_queue *,
2514 void *),
2515 void *arg)
2516 {
2517 unsigned int i;
2518
2519 for (i = 0; i < dev->num_tx_queues; i++)
2520 f(dev, &dev->_tx[i], arg);
2521 }
2522
2523 #define netdev_lockdep_set_classes(dev) \
2524 { \
2525 static struct lock_class_key qdisc_tx_busylock_key; \
2526 static struct lock_class_key qdisc_xmit_lock_key; \
2527 static struct lock_class_key dev_addr_list_lock_key; \
2528 unsigned int i; \
2529 \
2530 (dev)->qdisc_tx_busylock = &qdisc_tx_busylock_key; \
2531 lockdep_set_class(&(dev)->addr_list_lock, \
2532 &dev_addr_list_lock_key); \
2533 for (i = 0; i < (dev)->num_tx_queues; i++) \
2534 lockdep_set_class(&(dev)->_tx[i]._xmit_lock, \
2535 &qdisc_xmit_lock_key); \
2536 }
2537
2538 u16 netdev_pick_tx(struct net_device *dev, struct sk_buff *skb,
2539 struct net_device *sb_dev);
2540 struct netdev_queue *netdev_core_pick_tx(struct net_device *dev,
2541 struct sk_buff *skb,
2542 struct net_device *sb_dev);
2543
2544 /* returns the headroom that the master device needs to take in account
2545 * when forwarding to this dev
2546 */
netdev_get_fwd_headroom(struct net_device * dev)2547 static inline unsigned netdev_get_fwd_headroom(struct net_device *dev)
2548 {
2549 return dev->priv_flags & IFF_PHONY_HEADROOM ? 0 : dev->needed_headroom;
2550 }
2551
netdev_set_rx_headroom(struct net_device * dev,int new_hr)2552 static inline void netdev_set_rx_headroom(struct net_device *dev, int new_hr)
2553 {
2554 if (dev->netdev_ops->ndo_set_rx_headroom)
2555 dev->netdev_ops->ndo_set_rx_headroom(dev, new_hr);
2556 }
2557
2558 /* set the device rx headroom to the dev's default */
netdev_reset_rx_headroom(struct net_device * dev)2559 static inline void netdev_reset_rx_headroom(struct net_device *dev)
2560 {
2561 netdev_set_rx_headroom(dev, -1);
2562 }
2563
netdev_get_ml_priv(struct net_device * dev,enum netdev_ml_priv_type type)2564 static inline void *netdev_get_ml_priv(struct net_device *dev,
2565 enum netdev_ml_priv_type type)
2566 {
2567 if (dev->ml_priv_type != type)
2568 return NULL;
2569
2570 return dev->ml_priv;
2571 }
2572
netdev_set_ml_priv(struct net_device * dev,void * ml_priv,enum netdev_ml_priv_type type)2573 static inline void netdev_set_ml_priv(struct net_device *dev,
2574 void *ml_priv,
2575 enum netdev_ml_priv_type type)
2576 {
2577 WARN(dev->ml_priv_type && dev->ml_priv_type != type,
2578 "Overwriting already set ml_priv_type (%u) with different ml_priv_type (%u)!\n",
2579 dev->ml_priv_type, type);
2580 WARN(!dev->ml_priv_type && dev->ml_priv,
2581 "Overwriting already set ml_priv and ml_priv_type is ML_PRIV_NONE!\n");
2582
2583 dev->ml_priv = ml_priv;
2584 dev->ml_priv_type = type;
2585 }
2586
2587 /*
2588 * Net namespace inlines
2589 */
2590 static inline
dev_net(const struct net_device * dev)2591 struct net *dev_net(const struct net_device *dev)
2592 {
2593 return read_pnet(&dev->nd_net);
2594 }
2595
2596 static inline
dev_net_set(struct net_device * dev,struct net * net)2597 void dev_net_set(struct net_device *dev, struct net *net)
2598 {
2599 write_pnet(&dev->nd_net, net);
2600 }
2601
2602 /**
2603 * netdev_priv - access network device private data
2604 * @dev: network device
2605 *
2606 * Get network device private data
2607 */
netdev_priv(const struct net_device * dev)2608 static inline void *netdev_priv(const struct net_device *dev)
2609 {
2610 return (char *)dev + ALIGN(sizeof(struct net_device), NETDEV_ALIGN);
2611 }
2612
2613 /* Set the sysfs physical device reference for the network logical device
2614 * if set prior to registration will cause a symlink during initialization.
2615 */
2616 #define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
2617
2618 /* Set the sysfs device type for the network logical device to allow
2619 * fine-grained identification of different network device types. For
2620 * example Ethernet, Wireless LAN, Bluetooth, WiMAX etc.
2621 */
2622 #define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
2623
2624 /* Default NAPI poll() weight
2625 * Device drivers are strongly advised to not use bigger value
2626 */
2627 #define NAPI_POLL_WEIGHT 64
2628
2629 void netif_napi_add_weight(struct net_device *dev, struct napi_struct *napi,
2630 int (*poll)(struct napi_struct *, int), int weight);
2631
2632 /**
2633 * netif_napi_add() - initialize a NAPI context
2634 * @dev: network device
2635 * @napi: NAPI context
2636 * @poll: polling function
2637 *
2638 * netif_napi_add() must be used to initialize a NAPI context prior to calling
2639 * *any* of the other NAPI-related functions.
2640 */
2641 static inline void
netif_napi_add(struct net_device * dev,struct napi_struct * napi,int (* poll)(struct napi_struct *,int))2642 netif_napi_add(struct net_device *dev, struct napi_struct *napi,
2643 int (*poll)(struct napi_struct *, int))
2644 {
2645 netif_napi_add_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
2646 }
2647
2648 static inline void
netif_napi_add_tx_weight(struct net_device * dev,struct napi_struct * napi,int (* poll)(struct napi_struct *,int),int weight)2649 netif_napi_add_tx_weight(struct net_device *dev,
2650 struct napi_struct *napi,
2651 int (*poll)(struct napi_struct *, int),
2652 int weight)
2653 {
2654 set_bit(NAPI_STATE_NO_BUSY_POLL, &napi->state);
2655 netif_napi_add_weight(dev, napi, poll, weight);
2656 }
2657
2658 /**
2659 * netif_napi_add_tx() - initialize a NAPI context to be used for Tx only
2660 * @dev: network device
2661 * @napi: NAPI context
2662 * @poll: polling function
2663 *
2664 * This variant of netif_napi_add() should be used from drivers using NAPI
2665 * to exclusively poll a TX queue.
2666 * This will avoid we add it into napi_hash[], thus polluting this hash table.
2667 */
netif_napi_add_tx(struct net_device * dev,struct napi_struct * napi,int (* poll)(struct napi_struct *,int))2668 static inline void netif_napi_add_tx(struct net_device *dev,
2669 struct napi_struct *napi,
2670 int (*poll)(struct napi_struct *, int))
2671 {
2672 netif_napi_add_tx_weight(dev, napi, poll, NAPI_POLL_WEIGHT);
2673 }
2674
2675 /**
2676 * __netif_napi_del - remove a NAPI context
2677 * @napi: NAPI context
2678 *
2679 * Warning: caller must observe RCU grace period before freeing memory
2680 * containing @napi. Drivers might want to call this helper to combine
2681 * all the needed RCU grace periods into a single one.
2682 */
2683 void __netif_napi_del(struct napi_struct *napi);
2684
2685 /**
2686 * netif_napi_del - remove a NAPI context
2687 * @napi: NAPI context
2688 *
2689 * netif_napi_del() removes a NAPI context from the network device NAPI list
2690 */
netif_napi_del(struct napi_struct * napi)2691 static inline void netif_napi_del(struct napi_struct *napi)
2692 {
2693 __netif_napi_del(napi);
2694 synchronize_net();
2695 }
2696
2697 struct packet_type {
2698 __be16 type; /* This is really htons(ether_type). */
2699 bool ignore_outgoing;
2700 struct net_device *dev; /* NULL is wildcarded here */
2701 netdevice_tracker dev_tracker;
2702 int (*func) (struct sk_buff *,
2703 struct net_device *,
2704 struct packet_type *,
2705 struct net_device *);
2706 void (*list_func) (struct list_head *,
2707 struct packet_type *,
2708 struct net_device *);
2709 bool (*id_match)(struct packet_type *ptype,
2710 struct sock *sk);
2711 struct net *af_packet_net;
2712 void *af_packet_priv;
2713 struct list_head list;
2714 };
2715
2716 struct offload_callbacks {
2717 struct sk_buff *(*gso_segment)(struct sk_buff *skb,
2718 netdev_features_t features);
2719 struct sk_buff *(*gro_receive)(struct list_head *head,
2720 struct sk_buff *skb);
2721 int (*gro_complete)(struct sk_buff *skb, int nhoff);
2722 };
2723
2724 struct packet_offload {
2725 __be16 type; /* This is really htons(ether_type). */
2726 u16 priority;
2727 struct offload_callbacks callbacks;
2728 struct list_head list;
2729 };
2730
2731 /* often modified stats are per-CPU, other are shared (netdev->stats) */
2732 struct pcpu_sw_netstats {
2733 u64_stats_t rx_packets;
2734 u64_stats_t rx_bytes;
2735 u64_stats_t tx_packets;
2736 u64_stats_t tx_bytes;
2737 struct u64_stats_sync syncp;
2738 } __aligned(4 * sizeof(u64));
2739
2740 struct pcpu_dstats {
2741 u64 rx_packets;
2742 u64 rx_bytes;
2743 u64 rx_drops;
2744 u64 tx_packets;
2745 u64 tx_bytes;
2746 u64 tx_drops;
2747 struct u64_stats_sync syncp;
2748 } __aligned(8 * sizeof(u64));
2749
2750 struct pcpu_lstats {
2751 u64_stats_t packets;
2752 u64_stats_t bytes;
2753 struct u64_stats_sync syncp;
2754 } __aligned(2 * sizeof(u64));
2755
2756 void dev_lstats_read(struct net_device *dev, u64 *packets, u64 *bytes);
2757
dev_sw_netstats_rx_add(struct net_device * dev,unsigned int len)2758 static inline void dev_sw_netstats_rx_add(struct net_device *dev, unsigned int len)
2759 {
2760 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2761
2762 u64_stats_update_begin(&tstats->syncp);
2763 u64_stats_add(&tstats->rx_bytes, len);
2764 u64_stats_inc(&tstats->rx_packets);
2765 u64_stats_update_end(&tstats->syncp);
2766 }
2767
dev_sw_netstats_tx_add(struct net_device * dev,unsigned int packets,unsigned int len)2768 static inline void dev_sw_netstats_tx_add(struct net_device *dev,
2769 unsigned int packets,
2770 unsigned int len)
2771 {
2772 struct pcpu_sw_netstats *tstats = this_cpu_ptr(dev->tstats);
2773
2774 u64_stats_update_begin(&tstats->syncp);
2775 u64_stats_add(&tstats->tx_bytes, len);
2776 u64_stats_add(&tstats->tx_packets, packets);
2777 u64_stats_update_end(&tstats->syncp);
2778 }
2779
dev_lstats_add(struct net_device * dev,unsigned int len)2780 static inline void dev_lstats_add(struct net_device *dev, unsigned int len)
2781 {
2782 struct pcpu_lstats *lstats = this_cpu_ptr(dev->lstats);
2783
2784 u64_stats_update_begin(&lstats->syncp);
2785 u64_stats_add(&lstats->bytes, len);
2786 u64_stats_inc(&lstats->packets);
2787 u64_stats_update_end(&lstats->syncp);
2788 }
2789
2790 #define __netdev_alloc_pcpu_stats(type, gfp) \
2791 ({ \
2792 typeof(type) __percpu *pcpu_stats = alloc_percpu_gfp(type, gfp);\
2793 if (pcpu_stats) { \
2794 int __cpu; \
2795 for_each_possible_cpu(__cpu) { \
2796 typeof(type) *stat; \
2797 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2798 u64_stats_init(&stat->syncp); \
2799 } \
2800 } \
2801 pcpu_stats; \
2802 })
2803
2804 #define netdev_alloc_pcpu_stats(type) \
2805 __netdev_alloc_pcpu_stats(type, GFP_KERNEL)
2806
2807 #define devm_netdev_alloc_pcpu_stats(dev, type) \
2808 ({ \
2809 typeof(type) __percpu *pcpu_stats = devm_alloc_percpu(dev, type);\
2810 if (pcpu_stats) { \
2811 int __cpu; \
2812 for_each_possible_cpu(__cpu) { \
2813 typeof(type) *stat; \
2814 stat = per_cpu_ptr(pcpu_stats, __cpu); \
2815 u64_stats_init(&stat->syncp); \
2816 } \
2817 } \
2818 pcpu_stats; \
2819 })
2820
2821 enum netdev_lag_tx_type {
2822 NETDEV_LAG_TX_TYPE_UNKNOWN,
2823 NETDEV_LAG_TX_TYPE_RANDOM,
2824 NETDEV_LAG_TX_TYPE_BROADCAST,
2825 NETDEV_LAG_TX_TYPE_ROUNDROBIN,
2826 NETDEV_LAG_TX_TYPE_ACTIVEBACKUP,
2827 NETDEV_LAG_TX_TYPE_HASH,
2828 };
2829
2830 enum netdev_lag_hash {
2831 NETDEV_LAG_HASH_NONE,
2832 NETDEV_LAG_HASH_L2,
2833 NETDEV_LAG_HASH_L34,
2834 NETDEV_LAG_HASH_L23,
2835 NETDEV_LAG_HASH_E23,
2836 NETDEV_LAG_HASH_E34,
2837 NETDEV_LAG_HASH_VLAN_SRCMAC,
2838 NETDEV_LAG_HASH_UNKNOWN,
2839 };
2840
2841 struct netdev_lag_upper_info {
2842 enum netdev_lag_tx_type tx_type;
2843 enum netdev_lag_hash hash_type;
2844 };
2845
2846 struct netdev_lag_lower_state_info {
2847 u8 link_up : 1,
2848 tx_enabled : 1;
2849 };
2850
2851 #include <linux/notifier.h>
2852
2853 /* netdevice notifier chain. Please remember to update netdev_cmd_to_name()
2854 * and the rtnetlink notification exclusion list in rtnetlink_event() when
2855 * adding new types.
2856 */
2857 enum netdev_cmd {
2858 NETDEV_UP = 1, /* For now you can't veto a device up/down */
2859 NETDEV_DOWN,
2860 NETDEV_REBOOT, /* Tell a protocol stack a network interface
2861 detected a hardware crash and restarted
2862 - we can use this eg to kick tcp sessions
2863 once done */
2864 NETDEV_CHANGE, /* Notify device state change */
2865 NETDEV_REGISTER,
2866 NETDEV_UNREGISTER,
2867 NETDEV_CHANGEMTU, /* notify after mtu change happened */
2868 NETDEV_CHANGEADDR, /* notify after the address change */
2869 NETDEV_PRE_CHANGEADDR, /* notify before the address change */
2870 NETDEV_GOING_DOWN,
2871 NETDEV_CHANGENAME,
2872 NETDEV_FEAT_CHANGE,
2873 NETDEV_BONDING_FAILOVER,
2874 NETDEV_PRE_UP,
2875 NETDEV_PRE_TYPE_CHANGE,
2876 NETDEV_POST_TYPE_CHANGE,
2877 NETDEV_POST_INIT,
2878 NETDEV_PRE_UNINIT,
2879 NETDEV_RELEASE,
2880 NETDEV_NOTIFY_PEERS,
2881 NETDEV_JOIN,
2882 NETDEV_CHANGEUPPER,
2883 NETDEV_RESEND_IGMP,
2884 NETDEV_PRECHANGEMTU, /* notify before mtu change happened */
2885 NETDEV_CHANGEINFODATA,
2886 NETDEV_BONDING_INFO,
2887 NETDEV_PRECHANGEUPPER,
2888 NETDEV_CHANGELOWERSTATE,
2889 NETDEV_UDP_TUNNEL_PUSH_INFO,
2890 NETDEV_UDP_TUNNEL_DROP_INFO,
2891 NETDEV_CHANGE_TX_QUEUE_LEN,
2892 NETDEV_CVLAN_FILTER_PUSH_INFO,
2893 NETDEV_CVLAN_FILTER_DROP_INFO,
2894 NETDEV_SVLAN_FILTER_PUSH_INFO,
2895 NETDEV_SVLAN_FILTER_DROP_INFO,
2896 NETDEV_OFFLOAD_XSTATS_ENABLE,
2897 NETDEV_OFFLOAD_XSTATS_DISABLE,
2898 NETDEV_OFFLOAD_XSTATS_REPORT_USED,
2899 NETDEV_OFFLOAD_XSTATS_REPORT_DELTA,
2900 NETDEV_XDP_FEAT_CHANGE,
2901 };
2902 const char *netdev_cmd_to_name(enum netdev_cmd cmd);
2903
2904 int register_netdevice_notifier(struct notifier_block *nb);
2905 int unregister_netdevice_notifier(struct notifier_block *nb);
2906 int register_netdevice_notifier_net(struct net *net, struct notifier_block *nb);
2907 int unregister_netdevice_notifier_net(struct net *net,
2908 struct notifier_block *nb);
2909 int register_netdevice_notifier_dev_net(struct net_device *dev,
2910 struct notifier_block *nb,
2911 struct netdev_net_notifier *nn);
2912 int unregister_netdevice_notifier_dev_net(struct net_device *dev,
2913 struct notifier_block *nb,
2914 struct netdev_net_notifier *nn);
2915
2916 struct netdev_notifier_info {
2917 struct net_device *dev;
2918 struct netlink_ext_ack *extack;
2919 };
2920
2921 struct netdev_notifier_info_ext {
2922 struct netdev_notifier_info info; /* must be first */
2923 union {
2924 u32 mtu;
2925 } ext;
2926 };
2927
2928 struct netdev_notifier_change_info {
2929 struct netdev_notifier_info info; /* must be first */
2930 unsigned int flags_changed;
2931 };
2932
2933 struct netdev_notifier_changeupper_info {
2934 struct netdev_notifier_info info; /* must be first */
2935 struct net_device *upper_dev; /* new upper dev */
2936 bool master; /* is upper dev master */
2937 bool linking; /* is the notification for link or unlink */
2938 void *upper_info; /* upper dev info */
2939 };
2940
2941 struct netdev_notifier_changelowerstate_info {
2942 struct netdev_notifier_info info; /* must be first */
2943 void *lower_state_info; /* is lower dev state */
2944 };
2945
2946 struct netdev_notifier_pre_changeaddr_info {
2947 struct netdev_notifier_info info; /* must be first */
2948 const unsigned char *dev_addr;
2949 };
2950
2951 enum netdev_offload_xstats_type {
2952 NETDEV_OFFLOAD_XSTATS_TYPE_L3 = 1,
2953 };
2954
2955 struct netdev_notifier_offload_xstats_info {
2956 struct netdev_notifier_info info; /* must be first */
2957 enum netdev_offload_xstats_type type;
2958
2959 union {
2960 /* NETDEV_OFFLOAD_XSTATS_REPORT_DELTA */
2961 struct netdev_notifier_offload_xstats_rd *report_delta;
2962 /* NETDEV_OFFLOAD_XSTATS_REPORT_USED */
2963 struct netdev_notifier_offload_xstats_ru *report_used;
2964 };
2965 };
2966
2967 int netdev_offload_xstats_enable(struct net_device *dev,
2968 enum netdev_offload_xstats_type type,
2969 struct netlink_ext_ack *extack);
2970 int netdev_offload_xstats_disable(struct net_device *dev,
2971 enum netdev_offload_xstats_type type);
2972 bool netdev_offload_xstats_enabled(const struct net_device *dev,
2973 enum netdev_offload_xstats_type type);
2974 int netdev_offload_xstats_get(struct net_device *dev,
2975 enum netdev_offload_xstats_type type,
2976 struct rtnl_hw_stats64 *stats, bool *used,
2977 struct netlink_ext_ack *extack);
2978 void
2979 netdev_offload_xstats_report_delta(struct netdev_notifier_offload_xstats_rd *rd,
2980 const struct rtnl_hw_stats64 *stats);
2981 void
2982 netdev_offload_xstats_report_used(struct netdev_notifier_offload_xstats_ru *ru);
2983 void netdev_offload_xstats_push_delta(struct net_device *dev,
2984 enum netdev_offload_xstats_type type,
2985 const struct rtnl_hw_stats64 *stats);
2986
netdev_notifier_info_init(struct netdev_notifier_info * info,struct net_device * dev)2987 static inline void netdev_notifier_info_init(struct netdev_notifier_info *info,
2988 struct net_device *dev)
2989 {
2990 info->dev = dev;
2991 info->extack = NULL;
2992 }
2993
2994 static inline struct net_device *
netdev_notifier_info_to_dev(const struct netdev_notifier_info * info)2995 netdev_notifier_info_to_dev(const struct netdev_notifier_info *info)
2996 {
2997 return info->dev;
2998 }
2999
3000 static inline struct netlink_ext_ack *
netdev_notifier_info_to_extack(const struct netdev_notifier_info * info)3001 netdev_notifier_info_to_extack(const struct netdev_notifier_info *info)
3002 {
3003 return info->extack;
3004 }
3005
3006 int call_netdevice_notifiers(unsigned long val, struct net_device *dev);
3007 int call_netdevice_notifiers_info(unsigned long val,
3008 struct netdev_notifier_info *info);
3009
3010 extern rwlock_t dev_base_lock; /* Device list lock */
3011
3012 #define for_each_netdev(net, d) \
3013 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
3014 #define for_each_netdev_reverse(net, d) \
3015 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
3016 #define for_each_netdev_rcu(net, d) \
3017 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
3018 #define for_each_netdev_safe(net, d, n) \
3019 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
3020 #define for_each_netdev_continue(net, d) \
3021 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
3022 #define for_each_netdev_continue_reverse(net, d) \
3023 list_for_each_entry_continue_reverse(d, &(net)->dev_base_head, \
3024 dev_list)
3025 #define for_each_netdev_continue_rcu(net, d) \
3026 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
3027 #define for_each_netdev_in_bond_rcu(bond, slave) \
3028 for_each_netdev_rcu(&init_net, slave) \
3029 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
3030 #define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
3031
3032 #define for_each_netdev_dump(net, d, ifindex) \
3033 xa_for_each_start(&(net)->dev_by_index, (ifindex), (d), (ifindex))
3034
next_net_device(struct net_device * dev)3035 static inline struct net_device *next_net_device(struct net_device *dev)
3036 {
3037 struct list_head *lh;
3038 struct net *net;
3039
3040 net = dev_net(dev);
3041 lh = dev->dev_list.next;
3042 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3043 }
3044
next_net_device_rcu(struct net_device * dev)3045 static inline struct net_device *next_net_device_rcu(struct net_device *dev)
3046 {
3047 struct list_head *lh;
3048 struct net *net;
3049
3050 net = dev_net(dev);
3051 lh = rcu_dereference(list_next_rcu(&dev->dev_list));
3052 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3053 }
3054
first_net_device(struct net * net)3055 static inline struct net_device *first_net_device(struct net *net)
3056 {
3057 return list_empty(&net->dev_base_head) ? NULL :
3058 net_device_entry(net->dev_base_head.next);
3059 }
3060
first_net_device_rcu(struct net * net)3061 static inline struct net_device *first_net_device_rcu(struct net *net)
3062 {
3063 struct list_head *lh = rcu_dereference(list_next_rcu(&net->dev_base_head));
3064
3065 return lh == &net->dev_base_head ? NULL : net_device_entry(lh);
3066 }
3067
3068 int netdev_boot_setup_check(struct net_device *dev);
3069 struct net_device *dev_getbyhwaddr_rcu(struct net *net, unsigned short type,
3070 const char *hwaddr);
3071 struct net_device *dev_getfirstbyhwtype(struct net *net, unsigned short type);
3072 void dev_add_pack(struct packet_type *pt);
3073 void dev_remove_pack(struct packet_type *pt);
3074 void __dev_remove_pack(struct packet_type *pt);
3075 void dev_add_offload(struct packet_offload *po);
3076 void dev_remove_offload(struct packet_offload *po);
3077
3078 int dev_get_iflink(const struct net_device *dev);
3079 int dev_fill_metadata_dst(struct net_device *dev, struct sk_buff *skb);
3080 int dev_fill_forward_path(const struct net_device *dev, const u8 *daddr,
3081 struct net_device_path_stack *stack);
3082 struct net_device *__dev_get_by_flags(struct net *net, unsigned short flags,
3083 unsigned short mask);
3084 struct net_device *dev_get_by_name(struct net *net, const char *name);
3085 struct net_device *dev_get_by_name_rcu(struct net *net, const char *name);
3086 struct net_device *__dev_get_by_name(struct net *net, const char *name);
3087 bool netdev_name_in_use(struct net *net, const char *name);
3088 int dev_alloc_name(struct net_device *dev, const char *name);
3089 int dev_open(struct net_device *dev, struct netlink_ext_ack *extack);
3090 void dev_close(struct net_device *dev);
3091 void dev_close_many(struct list_head *head, bool unlink);
3092 void dev_disable_lro(struct net_device *dev);
3093 int dev_loopback_xmit(struct net *net, struct sock *sk, struct sk_buff *newskb);
3094 u16 dev_pick_tx_zero(struct net_device *dev, struct sk_buff *skb,
3095 struct net_device *sb_dev);
3096 u16 dev_pick_tx_cpu_id(struct net_device *dev, struct sk_buff *skb,
3097 struct net_device *sb_dev);
3098
3099 int __dev_queue_xmit(struct sk_buff *skb, struct net_device *sb_dev);
3100 int __dev_direct_xmit(struct sk_buff *skb, u16 queue_id);
3101
dev_queue_xmit(struct sk_buff * skb)3102 static inline int dev_queue_xmit(struct sk_buff *skb)
3103 {
3104 return __dev_queue_xmit(skb, NULL);
3105 }
3106
dev_queue_xmit_accel(struct sk_buff * skb,struct net_device * sb_dev)3107 static inline int dev_queue_xmit_accel(struct sk_buff *skb,
3108 struct net_device *sb_dev)
3109 {
3110 return __dev_queue_xmit(skb, sb_dev);
3111 }
3112
dev_direct_xmit(struct sk_buff * skb,u16 queue_id)3113 static inline int dev_direct_xmit(struct sk_buff *skb, u16 queue_id)
3114 {
3115 int ret;
3116
3117 ret = __dev_direct_xmit(skb, queue_id);
3118 if (!dev_xmit_complete(ret))
3119 kfree_skb(skb);
3120 return ret;
3121 }
3122
3123 int register_netdevice(struct net_device *dev);
3124 void unregister_netdevice_queue(struct net_device *dev, struct list_head *head);
3125 void unregister_netdevice_many(struct list_head *head);
unregister_netdevice(struct net_device * dev)3126 static inline void unregister_netdevice(struct net_device *dev)
3127 {
3128 unregister_netdevice_queue(dev, NULL);
3129 }
3130
3131 int netdev_refcnt_read(const struct net_device *dev);
3132 void free_netdev(struct net_device *dev);
3133 void netdev_freemem(struct net_device *dev);
3134 int init_dummy_netdev(struct net_device *dev);
3135
3136 struct net_device *netdev_get_xmit_slave(struct net_device *dev,
3137 struct sk_buff *skb,
3138 bool all_slaves);
3139 struct net_device *netdev_sk_get_lowest_dev(struct net_device *dev,
3140 struct sock *sk);
3141 struct net_device *dev_get_by_index(struct net *net, int ifindex);
3142 struct net_device *__dev_get_by_index(struct net *net, int ifindex);
3143 struct net_device *netdev_get_by_index(struct net *net, int ifindex,
3144 netdevice_tracker *tracker, gfp_t gfp);
3145 struct net_device *netdev_get_by_name(struct net *net, const char *name,
3146 netdevice_tracker *tracker, gfp_t gfp);
3147 struct net_device *dev_get_by_index_rcu(struct net *net, int ifindex);
3148 struct net_device *dev_get_by_napi_id(unsigned int napi_id);
3149
dev_hard_header(struct sk_buff * skb,struct net_device * dev,unsigned short type,const void * daddr,const void * saddr,unsigned int len)3150 static inline int dev_hard_header(struct sk_buff *skb, struct net_device *dev,
3151 unsigned short type,
3152 const void *daddr, const void *saddr,
3153 unsigned int len)
3154 {
3155 if (!dev->header_ops || !dev->header_ops->create)
3156 return 0;
3157
3158 return dev->header_ops->create(skb, dev, type, daddr, saddr, len);
3159 }
3160
dev_parse_header(const struct sk_buff * skb,unsigned char * haddr)3161 static inline int dev_parse_header(const struct sk_buff *skb,
3162 unsigned char *haddr)
3163 {
3164 const struct net_device *dev = skb->dev;
3165
3166 if (!dev->header_ops || !dev->header_ops->parse)
3167 return 0;
3168 return dev->header_ops->parse(skb, haddr);
3169 }
3170
dev_parse_header_protocol(const struct sk_buff * skb)3171 static inline __be16 dev_parse_header_protocol(const struct sk_buff *skb)
3172 {
3173 const struct net_device *dev = skb->dev;
3174
3175 if (!dev->header_ops || !dev->header_ops->parse_protocol)
3176 return 0;
3177 return dev->header_ops->parse_protocol(skb);
3178 }
3179
3180 /* ll_header must have at least hard_header_len allocated */
dev_validate_header(const struct net_device * dev,char * ll_header,int len)3181 static inline bool dev_validate_header(const struct net_device *dev,
3182 char *ll_header, int len)
3183 {
3184 if (likely(len >= dev->hard_header_len))
3185 return true;
3186 if (len < dev->min_header_len)
3187 return false;
3188
3189 if (capable(CAP_SYS_RAWIO)) {
3190 memset(ll_header + len, 0, dev->hard_header_len - len);
3191 return true;
3192 }
3193
3194 if (dev->header_ops && dev->header_ops->validate)
3195 return dev->header_ops->validate(ll_header, len);
3196
3197 return false;
3198 }
3199
dev_has_header(const struct net_device * dev)3200 static inline bool dev_has_header(const struct net_device *dev)
3201 {
3202 return dev->header_ops && dev->header_ops->create;
3203 }
3204
3205 /*
3206 * Incoming packets are placed on per-CPU queues
3207 */
3208 struct softnet_data {
3209 struct list_head poll_list;
3210 struct sk_buff_head process_queue;
3211
3212 /* stats */
3213 unsigned int processed;
3214 unsigned int time_squeeze;
3215 #ifdef CONFIG_RPS
3216 struct softnet_data *rps_ipi_list;
3217 #endif
3218
3219 bool in_net_rx_action;
3220 bool in_napi_threaded_poll;
3221
3222 #ifdef CONFIG_NET_FLOW_LIMIT
3223 struct sd_flow_limit __rcu *flow_limit;
3224 #endif
3225 struct Qdisc *output_queue;
3226 struct Qdisc **output_queue_tailp;
3227 struct sk_buff *completion_queue;
3228 #ifdef CONFIG_XFRM_OFFLOAD
3229 struct sk_buff_head xfrm_backlog;
3230 #endif
3231 /* written and read only by owning cpu: */
3232 struct {
3233 u16 recursion;
3234 u8 more;
3235 #ifdef CONFIG_NET_EGRESS
3236 u8 skip_txqueue;
3237 #endif
3238 } xmit;
3239 #ifdef CONFIG_RPS
3240 /* input_queue_head should be written by cpu owning this struct,
3241 * and only read by other cpus. Worth using a cache line.
3242 */
3243 unsigned int input_queue_head ____cacheline_aligned_in_smp;
3244
3245 /* Elements below can be accessed between CPUs for RPS/RFS */
3246 call_single_data_t csd ____cacheline_aligned_in_smp;
3247 struct softnet_data *rps_ipi_next;
3248 unsigned int cpu;
3249 unsigned int input_queue_tail;
3250 #endif
3251 unsigned int received_rps;
3252 unsigned int dropped;
3253 struct sk_buff_head input_pkt_queue;
3254 struct napi_struct backlog;
3255
3256 /* Another possibly contended cache line */
3257 spinlock_t defer_lock ____cacheline_aligned_in_smp;
3258 int defer_count;
3259 int defer_ipi_scheduled;
3260 struct sk_buff *defer_list;
3261 call_single_data_t defer_csd;
3262 };
3263
input_queue_head_incr(struct softnet_data * sd)3264 static inline void input_queue_head_incr(struct softnet_data *sd)
3265 {
3266 #ifdef CONFIG_RPS
3267 sd->input_queue_head++;
3268 #endif
3269 }
3270
input_queue_tail_incr_save(struct softnet_data * sd,unsigned int * qtail)3271 static inline void input_queue_tail_incr_save(struct softnet_data *sd,
3272 unsigned int *qtail)
3273 {
3274 #ifdef CONFIG_RPS
3275 *qtail = ++sd->input_queue_tail;
3276 #endif
3277 }
3278
3279 DECLARE_PER_CPU_ALIGNED(struct softnet_data, softnet_data);
3280
dev_recursion_level(void)3281 static inline int dev_recursion_level(void)
3282 {
3283 return this_cpu_read(softnet_data.xmit.recursion);
3284 }
3285
3286 #define XMIT_RECURSION_LIMIT 8
dev_xmit_recursion(void)3287 static inline bool dev_xmit_recursion(void)
3288 {
3289 return unlikely(__this_cpu_read(softnet_data.xmit.recursion) >
3290 XMIT_RECURSION_LIMIT);
3291 }
3292
dev_xmit_recursion_inc(void)3293 static inline void dev_xmit_recursion_inc(void)
3294 {
3295 __this_cpu_inc(softnet_data.xmit.recursion);
3296 }
3297
dev_xmit_recursion_dec(void)3298 static inline void dev_xmit_recursion_dec(void)
3299 {
3300 __this_cpu_dec(softnet_data.xmit.recursion);
3301 }
3302
3303 void __netif_schedule(struct Qdisc *q);
3304 void netif_schedule_queue(struct netdev_queue *txq);
3305
netif_tx_schedule_all(struct net_device * dev)3306 static inline void netif_tx_schedule_all(struct net_device *dev)
3307 {
3308 unsigned int i;
3309
3310 for (i = 0; i < dev->num_tx_queues; i++)
3311 netif_schedule_queue(netdev_get_tx_queue(dev, i));
3312 }
3313
netif_tx_start_queue(struct netdev_queue * dev_queue)3314 static __always_inline void netif_tx_start_queue(struct netdev_queue *dev_queue)
3315 {
3316 clear_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
3317 }
3318
3319 /**
3320 * netif_start_queue - allow transmit
3321 * @dev: network device
3322 *
3323 * Allow upper layers to call the device hard_start_xmit routine.
3324 */
netif_start_queue(struct net_device * dev)3325 static inline void netif_start_queue(struct net_device *dev)
3326 {
3327 netif_tx_start_queue(netdev_get_tx_queue(dev, 0));
3328 }
3329
netif_tx_start_all_queues(struct net_device * dev)3330 static inline void netif_tx_start_all_queues(struct net_device *dev)
3331 {
3332 unsigned int i;
3333
3334 for (i = 0; i < dev->num_tx_queues; i++) {
3335 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3336 netif_tx_start_queue(txq);
3337 }
3338 }
3339
3340 void netif_tx_wake_queue(struct netdev_queue *dev_queue);
3341
3342 /**
3343 * netif_wake_queue - restart transmit
3344 * @dev: network device
3345 *
3346 * Allow upper layers to call the device hard_start_xmit routine.
3347 * Used for flow control when transmit resources are available.
3348 */
netif_wake_queue(struct net_device * dev)3349 static inline void netif_wake_queue(struct net_device *dev)
3350 {
3351 netif_tx_wake_queue(netdev_get_tx_queue(dev, 0));
3352 }
3353
netif_tx_wake_all_queues(struct net_device * dev)3354 static inline void netif_tx_wake_all_queues(struct net_device *dev)
3355 {
3356 unsigned int i;
3357
3358 for (i = 0; i < dev->num_tx_queues; i++) {
3359 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
3360 netif_tx_wake_queue(txq);
3361 }
3362 }
3363
netif_tx_stop_queue(struct netdev_queue * dev_queue)3364 static __always_inline void netif_tx_stop_queue(struct netdev_queue *dev_queue)
3365 {
3366 /* Paired with READ_ONCE() from dev_watchdog() */
3367 WRITE_ONCE(dev_queue->trans_start, jiffies);
3368
3369 /* This barrier is paired with smp_mb() from dev_watchdog() */
3370 smp_mb__before_atomic();
3371
3372 /* Must be an atomic op see netif_txq_try_stop() */
3373 set_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
3374 }
3375
3376 /**
3377 * netif_stop_queue - stop transmitted packets
3378 * @dev: network device
3379 *
3380 * Stop upper layers calling the device hard_start_xmit routine.
3381 * Used for flow control when transmit resources are unavailable.
3382 */
netif_stop_queue(struct net_device * dev)3383 static inline void netif_stop_queue(struct net_device *dev)
3384 {
3385 netif_tx_stop_queue(netdev_get_tx_queue(dev, 0));
3386 }
3387
3388 void netif_tx_stop_all_queues(struct net_device *dev);
3389
netif_tx_queue_stopped(const struct netdev_queue * dev_queue)3390 static inline bool netif_tx_queue_stopped(const struct netdev_queue *dev_queue)
3391 {
3392 return test_bit(__QUEUE_STATE_DRV_XOFF, &dev_queue->state);
3393 }
3394
3395 /**
3396 * netif_queue_stopped - test if transmit queue is flowblocked
3397 * @dev: network device
3398 *
3399 * Test if transmit queue on device is currently unable to send.
3400 */
netif_queue_stopped(const struct net_device * dev)3401 static inline bool netif_queue_stopped(const struct net_device *dev)
3402 {
3403 return netif_tx_queue_stopped(netdev_get_tx_queue(dev, 0));
3404 }
3405
netif_xmit_stopped(const struct netdev_queue * dev_queue)3406 static inline bool netif_xmit_stopped(const struct netdev_queue *dev_queue)
3407 {
3408 return dev_queue->state & QUEUE_STATE_ANY_XOFF;
3409 }
3410
3411 static inline bool
netif_xmit_frozen_or_stopped(const struct netdev_queue * dev_queue)3412 netif_xmit_frozen_or_stopped(const struct netdev_queue *dev_queue)
3413 {
3414 return dev_queue->state & QUEUE_STATE_ANY_XOFF_OR_FROZEN;
3415 }
3416
3417 static inline bool
netif_xmit_frozen_or_drv_stopped(const struct netdev_queue * dev_queue)3418 netif_xmit_frozen_or_drv_stopped(const struct netdev_queue *dev_queue)
3419 {
3420 return dev_queue->state & QUEUE_STATE_DRV_XOFF_OR_FROZEN;
3421 }
3422
3423 /**
3424 * netdev_queue_set_dql_min_limit - set dql minimum limit
3425 * @dev_queue: pointer to transmit queue
3426 * @min_limit: dql minimum limit
3427 *
3428 * Forces xmit_more() to return true until the minimum threshold
3429 * defined by @min_limit is reached (or until the tx queue is
3430 * empty). Warning: to be use with care, misuse will impact the
3431 * latency.
3432 */
netdev_queue_set_dql_min_limit(struct netdev_queue * dev_queue,unsigned int min_limit)3433 static inline void netdev_queue_set_dql_min_limit(struct netdev_queue *dev_queue,
3434 unsigned int min_limit)
3435 {
3436 #ifdef CONFIG_BQL
3437 dev_queue->dql.min_limit = min_limit;
3438 #endif
3439 }
3440
3441 /**
3442 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
3443 * @dev_queue: pointer to transmit queue
3444 *
3445 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
3446 * to give appropriate hint to the CPU.
3447 */
netdev_txq_bql_enqueue_prefetchw(struct netdev_queue * dev_queue)3448 static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue *dev_queue)
3449 {
3450 #ifdef CONFIG_BQL
3451 prefetchw(&dev_queue->dql.num_queued);
3452 #endif
3453 }
3454
3455 /**
3456 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
3457 * @dev_queue: pointer to transmit queue
3458 *
3459 * BQL enabled drivers might use this helper in their TX completion path,
3460 * to give appropriate hint to the CPU.
3461 */
netdev_txq_bql_complete_prefetchw(struct netdev_queue * dev_queue)3462 static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue *dev_queue)
3463 {
3464 #ifdef CONFIG_BQL
3465 prefetchw(&dev_queue->dql.limit);
3466 #endif
3467 }
3468
3469 /**
3470 * netdev_tx_sent_queue - report the number of bytes queued to a given tx queue
3471 * @dev_queue: network device queue
3472 * @bytes: number of bytes queued to the device queue
3473 *
3474 * Report the number of bytes queued for sending/completion to the network
3475 * device hardware queue. @bytes should be a good approximation and should
3476 * exactly match netdev_completed_queue() @bytes.
3477 * This is typically called once per packet, from ndo_start_xmit().
3478 */
netdev_tx_sent_queue(struct netdev_queue * dev_queue,unsigned int bytes)3479 static inline void netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3480 unsigned int bytes)
3481 {
3482 #ifdef CONFIG_BQL
3483 dql_queued(&dev_queue->dql, bytes);
3484
3485 if (likely(dql_avail(&dev_queue->dql) >= 0))
3486 return;
3487
3488 /* Paired with READ_ONCE() from dev_watchdog() */
3489 WRITE_ONCE(dev_queue->trans_start, jiffies);
3490
3491 /* This barrier is paired with smp_mb() from dev_watchdog() */
3492 smp_mb__before_atomic();
3493
3494 set_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3495
3496 /*
3497 * The XOFF flag must be set before checking the dql_avail below,
3498 * because in netdev_tx_completed_queue we update the dql_completed
3499 * before checking the XOFF flag.
3500 */
3501 smp_mb();
3502
3503 /* check again in case another CPU has just made room avail */
3504 if (unlikely(dql_avail(&dev_queue->dql) >= 0))
3505 clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state);
3506 #endif
3507 }
3508
3509 /* Variant of netdev_tx_sent_queue() for drivers that are aware
3510 * that they should not test BQL status themselves.
3511 * We do want to change __QUEUE_STATE_STACK_XOFF only for the last
3512 * skb of a batch.
3513 * Returns true if the doorbell must be used to kick the NIC.
3514 */
__netdev_tx_sent_queue(struct netdev_queue * dev_queue,unsigned int bytes,bool xmit_more)3515 static inline bool __netdev_tx_sent_queue(struct netdev_queue *dev_queue,
3516 unsigned int bytes,
3517 bool xmit_more)
3518 {
3519 if (xmit_more) {
3520 #ifdef CONFIG_BQL
3521 dql_queued(&dev_queue->dql, bytes);
3522 #endif
3523 return netif_tx_queue_stopped(dev_queue);
3524 }
3525 netdev_tx_sent_queue(dev_queue, bytes);
3526 return true;
3527 }
3528
3529 /**
3530 * netdev_sent_queue - report the number of bytes queued to hardware
3531 * @dev: network device
3532 * @bytes: number of bytes queued to the hardware device queue
3533 *
3534 * Report the number of bytes queued for sending/completion to the network
3535 * device hardware queue#0. @bytes should be a good approximation and should
3536 * exactly match netdev_completed_queue() @bytes.
3537 * This is typically called once per packet, from ndo_start_xmit().
3538 */
netdev_sent_queue(struct net_device * dev,unsigned int bytes)3539 static inline void netdev_sent_queue(struct net_device *dev, unsigned int bytes)
3540 {
3541 netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes);
3542 }
3543
__netdev_sent_queue(struct net_device * dev,unsigned int bytes,bool xmit_more)3544 static inline bool __netdev_sent_queue(struct net_device *dev,
3545 unsigned int bytes,
3546 bool xmit_more)
3547 {
3548 return __netdev_tx_sent_queue(netdev_get_tx_queue(dev, 0), bytes,
3549 xmit_more);
3550 }
3551
3552 /**
3553 * netdev_tx_completed_queue - report number of packets/bytes at TX completion.
3554 * @dev_queue: network device queue
3555 * @pkts: number of packets (currently ignored)
3556 * @bytes: number of bytes dequeued from the device queue
3557 *
3558 * Must be called at most once per TX completion round (and not per
3559 * individual packet), so that BQL can adjust its limits appropriately.
3560 */
netdev_tx_completed_queue(struct netdev_queue * dev_queue,unsigned int pkts,unsigned int bytes)3561 static inline void netdev_tx_completed_queue(struct netdev_queue *dev_queue,
3562 unsigned int pkts, unsigned int bytes)
3563 {
3564 #ifdef CONFIG_BQL
3565 if (unlikely(!bytes))
3566 return;
3567
3568 dql_completed(&dev_queue->dql, bytes);
3569
3570 /*
3571 * Without the memory barrier there is a small possiblity that
3572 * netdev_tx_sent_queue will miss the update and cause the queue to
3573 * be stopped forever
3574 */
3575 smp_mb(); /* NOTE: netdev_txq_completed_mb() assumes this exists */
3576
3577 if (unlikely(dql_avail(&dev_queue->dql) < 0))
3578 return;
3579
3580 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF, &dev_queue->state))
3581 netif_schedule_queue(dev_queue);
3582 #endif
3583 }
3584
3585 /**
3586 * netdev_completed_queue - report bytes and packets completed by device
3587 * @dev: network device
3588 * @pkts: actual number of packets sent over the medium
3589 * @bytes: actual number of bytes sent over the medium
3590 *
3591 * Report the number of bytes and packets transmitted by the network device
3592 * hardware queue over the physical medium, @bytes must exactly match the
3593 * @bytes amount passed to netdev_sent_queue()
3594 */
netdev_completed_queue(struct net_device * dev,unsigned int pkts,unsigned int bytes)3595 static inline void netdev_completed_queue(struct net_device *dev,
3596 unsigned int pkts, unsigned int bytes)
3597 {
3598 netdev_tx_completed_queue(netdev_get_tx_queue(dev, 0), pkts, bytes);
3599 }
3600
netdev_tx_reset_queue(struct netdev_queue * q)3601 static inline void netdev_tx_reset_queue(struct netdev_queue *q)
3602 {
3603 #ifdef CONFIG_BQL
3604 clear_bit(__QUEUE_STATE_STACK_XOFF, &q->state);
3605 dql_reset(&q->dql);
3606 #endif
3607 }
3608
3609 /**
3610 * netdev_reset_queue - reset the packets and bytes count of a network device
3611 * @dev_queue: network device
3612 *
3613 * Reset the bytes and packet count of a network device and clear the
3614 * software flow control OFF bit for this network device
3615 */
netdev_reset_queue(struct net_device * dev_queue)3616 static inline void netdev_reset_queue(struct net_device *dev_queue)
3617 {
3618 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue, 0));
3619 }
3620
3621 /**
3622 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
3623 * @dev: network device
3624 * @queue_index: given tx queue index
3625 *
3626 * Returns 0 if given tx queue index >= number of device tx queues,
3627 * otherwise returns the originally passed tx queue index.
3628 */
netdev_cap_txqueue(struct net_device * dev,u16 queue_index)3629 static inline u16 netdev_cap_txqueue(struct net_device *dev, u16 queue_index)
3630 {
3631 if (unlikely(queue_index >= dev->real_num_tx_queues)) {
3632 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
3633 dev->name, queue_index,
3634 dev->real_num_tx_queues);
3635 return 0;
3636 }
3637
3638 return queue_index;
3639 }
3640
3641 /**
3642 * netif_running - test if up
3643 * @dev: network device
3644 *
3645 * Test if the device has been brought up.
3646 */
netif_running(const struct net_device * dev)3647 static inline bool netif_running(const struct net_device *dev)
3648 {
3649 return test_bit(__LINK_STATE_START, &dev->state);
3650 }
3651
3652 /*
3653 * Routines to manage the subqueues on a device. We only need start,
3654 * stop, and a check if it's stopped. All other device management is
3655 * done at the overall netdevice level.
3656 * Also test the device if we're multiqueue.
3657 */
3658
3659 /**
3660 * netif_start_subqueue - allow sending packets on subqueue
3661 * @dev: network device
3662 * @queue_index: sub queue index
3663 *
3664 * Start individual transmit queue of a device with multiple transmit queues.
3665 */
netif_start_subqueue(struct net_device * dev,u16 queue_index)3666 static inline void netif_start_subqueue(struct net_device *dev, u16 queue_index)
3667 {
3668 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3669
3670 netif_tx_start_queue(txq);
3671 }
3672
3673 /**
3674 * netif_stop_subqueue - stop sending packets on subqueue
3675 * @dev: network device
3676 * @queue_index: sub queue index
3677 *
3678 * Stop individual transmit queue of a device with multiple transmit queues.
3679 */
netif_stop_subqueue(struct net_device * dev,u16 queue_index)3680 static inline void netif_stop_subqueue(struct net_device *dev, u16 queue_index)
3681 {
3682 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3683 netif_tx_stop_queue(txq);
3684 }
3685
3686 /**
3687 * __netif_subqueue_stopped - test status of subqueue
3688 * @dev: network device
3689 * @queue_index: sub queue index
3690 *
3691 * Check individual transmit queue of a device with multiple transmit queues.
3692 */
__netif_subqueue_stopped(const struct net_device * dev,u16 queue_index)3693 static inline bool __netif_subqueue_stopped(const struct net_device *dev,
3694 u16 queue_index)
3695 {
3696 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3697
3698 return netif_tx_queue_stopped(txq);
3699 }
3700
3701 /**
3702 * netif_subqueue_stopped - test status of subqueue
3703 * @dev: network device
3704 * @skb: sub queue buffer pointer
3705 *
3706 * Check individual transmit queue of a device with multiple transmit queues.
3707 */
netif_subqueue_stopped(const struct net_device * dev,struct sk_buff * skb)3708 static inline bool netif_subqueue_stopped(const struct net_device *dev,
3709 struct sk_buff *skb)
3710 {
3711 return __netif_subqueue_stopped(dev, skb_get_queue_mapping(skb));
3712 }
3713
3714 /**
3715 * netif_wake_subqueue - allow sending packets on subqueue
3716 * @dev: network device
3717 * @queue_index: sub queue index
3718 *
3719 * Resume individual transmit queue of a device with multiple transmit queues.
3720 */
netif_wake_subqueue(struct net_device * dev,u16 queue_index)3721 static inline void netif_wake_subqueue(struct net_device *dev, u16 queue_index)
3722 {
3723 struct netdev_queue *txq = netdev_get_tx_queue(dev, queue_index);
3724
3725 netif_tx_wake_queue(txq);
3726 }
3727
3728 #ifdef CONFIG_XPS
3729 int netif_set_xps_queue(struct net_device *dev, const struct cpumask *mask,
3730 u16 index);
3731 int __netif_set_xps_queue(struct net_device *dev, const unsigned long *mask,
3732 u16 index, enum xps_map_type type);
3733
3734 /**
3735 * netif_attr_test_mask - Test a CPU or Rx queue set in a mask
3736 * @j: CPU/Rx queue index
3737 * @mask: bitmask of all cpus/rx queues
3738 * @nr_bits: number of bits in the bitmask
3739 *
3740 * Test if a CPU or Rx queue index is set in a mask of all CPU/Rx queues.
3741 */
netif_attr_test_mask(unsigned long j,const unsigned long * mask,unsigned int nr_bits)3742 static inline bool netif_attr_test_mask(unsigned long j,
3743 const unsigned long *mask,
3744 unsigned int nr_bits)
3745 {
3746 cpu_max_bits_warn(j, nr_bits);
3747 return test_bit(j, mask);
3748 }
3749
3750 /**
3751 * netif_attr_test_online - Test for online CPU/Rx queue
3752 * @j: CPU/Rx queue index
3753 * @online_mask: bitmask for CPUs/Rx queues that are online
3754 * @nr_bits: number of bits in the bitmask
3755 *
3756 * Returns true if a CPU/Rx queue is online.
3757 */
netif_attr_test_online(unsigned long j,const unsigned long * online_mask,unsigned int nr_bits)3758 static inline bool netif_attr_test_online(unsigned long j,
3759 const unsigned long *online_mask,
3760 unsigned int nr_bits)
3761 {
3762 cpu_max_bits_warn(j, nr_bits);
3763
3764 if (online_mask)
3765 return test_bit(j, online_mask);
3766
3767 return (j < nr_bits);
3768 }
3769
3770 /**
3771 * netif_attrmask_next - get the next CPU/Rx queue in a cpu/Rx queues mask
3772 * @n: CPU/Rx queue index
3773 * @srcp: the cpumask/Rx queue mask pointer
3774 * @nr_bits: number of bits in the bitmask
3775 *
3776 * Returns >= nr_bits if no further CPUs/Rx queues set.
3777 */
netif_attrmask_next(int n,const unsigned long * srcp,unsigned int nr_bits)3778 static inline unsigned int netif_attrmask_next(int n, const unsigned long *srcp,
3779 unsigned int nr_bits)
3780 {
3781 /* -1 is a legal arg here. */
3782 if (n != -1)
3783 cpu_max_bits_warn(n, nr_bits);
3784
3785 if (srcp)
3786 return find_next_bit(srcp, nr_bits, n + 1);
3787
3788 return n + 1;
3789 }
3790
3791 /**
3792 * netif_attrmask_next_and - get the next CPU/Rx queue in \*src1p & \*src2p
3793 * @n: CPU/Rx queue index
3794 * @src1p: the first CPUs/Rx queues mask pointer
3795 * @src2p: the second CPUs/Rx queues mask pointer
3796 * @nr_bits: number of bits in the bitmask
3797 *
3798 * Returns >= nr_bits if no further CPUs/Rx queues set in both.
3799 */
netif_attrmask_next_and(int n,const unsigned long * src1p,const unsigned long * src2p,unsigned int nr_bits)3800 static inline int netif_attrmask_next_and(int n, const unsigned long *src1p,
3801 const unsigned long *src2p,
3802 unsigned int nr_bits)
3803 {
3804 /* -1 is a legal arg here. */
3805 if (n != -1)
3806 cpu_max_bits_warn(n, nr_bits);
3807
3808 if (src1p && src2p)
3809 return find_next_and_bit(src1p, src2p, nr_bits, n + 1);
3810 else if (src1p)
3811 return find_next_bit(src1p, nr_bits, n + 1);
3812 else if (src2p)
3813 return find_next_bit(src2p, nr_bits, n + 1);
3814
3815 return n + 1;
3816 }
3817 #else
netif_set_xps_queue(struct net_device * dev,const struct cpumask * mask,u16 index)3818 static inline int netif_set_xps_queue(struct net_device *dev,
3819 const struct cpumask *mask,
3820 u16 index)
3821 {
3822 return 0;
3823 }
3824
__netif_set_xps_queue(struct net_device * dev,const unsigned long * mask,u16 index,enum xps_map_type type)3825 static inline int __netif_set_xps_queue(struct net_device *dev,
3826 const unsigned long *mask,
3827 u16 index, enum xps_map_type type)
3828 {
3829 return 0;
3830 }
3831 #endif
3832
3833 /**
3834 * netif_is_multiqueue - test if device has multiple transmit queues
3835 * @dev: network device
3836 *
3837 * Check if device has multiple transmit queues
3838 */
netif_is_multiqueue(const struct net_device * dev)3839 static inline bool netif_is_multiqueue(const struct net_device *dev)
3840 {
3841 return dev->num_tx_queues > 1;
3842 }
3843
3844 int netif_set_real_num_tx_queues(struct net_device *dev, unsigned int txq);
3845
3846 #ifdef CONFIG_SYSFS
3847 int netif_set_real_num_rx_queues(struct net_device *dev, unsigned int rxq);
3848 #else
netif_set_real_num_rx_queues(struct net_device * dev,unsigned int rxqs)3849 static inline int netif_set_real_num_rx_queues(struct net_device *dev,
3850 unsigned int rxqs)
3851 {
3852 dev->real_num_rx_queues = rxqs;
3853 return 0;
3854 }
3855 #endif
3856 int netif_set_real_num_queues(struct net_device *dev,
3857 unsigned int txq, unsigned int rxq);
3858
3859 int netif_get_num_default_rss_queues(void);
3860
3861 void dev_kfree_skb_irq_reason(struct sk_buff *skb, enum skb_drop_reason reason);
3862 void dev_kfree_skb_any_reason(struct sk_buff *skb, enum skb_drop_reason reason);
3863
3864 /*
3865 * It is not allowed to call kfree_skb() or consume_skb() from hardware
3866 * interrupt context or with hardware interrupts being disabled.
3867 * (in_hardirq() || irqs_disabled())
3868 *
3869 * We provide four helpers that can be used in following contexts :
3870 *
3871 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
3872 * replacing kfree_skb(skb)
3873 *
3874 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
3875 * Typically used in place of consume_skb(skb) in TX completion path
3876 *
3877 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
3878 * replacing kfree_skb(skb)
3879 *
3880 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
3881 * and consumed a packet. Used in place of consume_skb(skb)
3882 */
dev_kfree_skb_irq(struct sk_buff * skb)3883 static inline void dev_kfree_skb_irq(struct sk_buff *skb)
3884 {
3885 dev_kfree_skb_irq_reason(skb, SKB_DROP_REASON_NOT_SPECIFIED);
3886 }
3887
dev_consume_skb_irq(struct sk_buff * skb)3888 static inline void dev_consume_skb_irq(struct sk_buff *skb)
3889 {
3890 dev_kfree_skb_irq_reason(skb, SKB_CONSUMED);
3891 }
3892
dev_kfree_skb_any(struct sk_buff * skb)3893 static inline void dev_kfree_skb_any(struct sk_buff *skb)
3894 {
3895 dev_kfree_skb_any_reason(skb, SKB_DROP_REASON_NOT_SPECIFIED);
3896 }
3897
dev_consume_skb_any(struct sk_buff * skb)3898 static inline void dev_consume_skb_any(struct sk_buff *skb)
3899 {
3900 dev_kfree_skb_any_reason(skb, SKB_CONSUMED);
3901 }
3902
3903 u32 bpf_prog_run_generic_xdp(struct sk_buff *skb, struct xdp_buff *xdp,
3904 struct bpf_prog *xdp_prog);
3905 void generic_xdp_tx(struct sk_buff *skb, struct bpf_prog *xdp_prog);
3906 int do_xdp_generic(struct bpf_prog *xdp_prog, struct sk_buff *skb);
3907 int netif_rx(struct sk_buff *skb);
3908 int __netif_rx(struct sk_buff *skb);
3909
3910 int netif_receive_skb(struct sk_buff *skb);
3911 int netif_receive_skb_core(struct sk_buff *skb);
3912 void netif_receive_skb_list_internal(struct list_head *head);
3913 void netif_receive_skb_list(struct list_head *head);
3914 gro_result_t napi_gro_receive(struct napi_struct *napi, struct sk_buff *skb);
3915 void napi_gro_flush(struct napi_struct *napi, bool flush_old);
3916 struct sk_buff *napi_get_frags(struct napi_struct *napi);
3917 void napi_get_frags_check(struct napi_struct *napi);
3918 gro_result_t napi_gro_frags(struct napi_struct *napi);
3919 struct packet_offload *gro_find_receive_by_type(__be16 type);
3920 struct packet_offload *gro_find_complete_by_type(__be16 type);
3921
napi_free_frags(struct napi_struct * napi)3922 static inline void napi_free_frags(struct napi_struct *napi)
3923 {
3924 kfree_skb(napi->skb);
3925 napi->skb = NULL;
3926 }
3927
3928 bool netdev_is_rx_handler_busy(struct net_device *dev);
3929 int netdev_rx_handler_register(struct net_device *dev,
3930 rx_handler_func_t *rx_handler,
3931 void *rx_handler_data);
3932 void netdev_rx_handler_unregister(struct net_device *dev);
3933
3934 bool dev_valid_name(const char *name);
is_socket_ioctl_cmd(unsigned int cmd)3935 static inline bool is_socket_ioctl_cmd(unsigned int cmd)
3936 {
3937 return _IOC_TYPE(cmd) == SOCK_IOC_TYPE;
3938 }
3939 int get_user_ifreq(struct ifreq *ifr, void __user **ifrdata, void __user *arg);
3940 int put_user_ifreq(struct ifreq *ifr, void __user *arg);
3941 int dev_ioctl(struct net *net, unsigned int cmd, struct ifreq *ifr,
3942 void __user *data, bool *need_copyout);
3943 int dev_ifconf(struct net *net, struct ifconf __user *ifc);
3944 int generic_hwtstamp_get_lower(struct net_device *dev,
3945 struct kernel_hwtstamp_config *kernel_cfg);
3946 int generic_hwtstamp_set_lower(struct net_device *dev,
3947 struct kernel_hwtstamp_config *kernel_cfg,
3948 struct netlink_ext_ack *extack);
3949 int dev_ethtool(struct net *net, struct ifreq *ifr, void __user *userdata);
3950 unsigned int dev_get_flags(const struct net_device *);
3951 int __dev_change_flags(struct net_device *dev, unsigned int flags,
3952 struct netlink_ext_ack *extack);
3953 int dev_change_flags(struct net_device *dev, unsigned int flags,
3954 struct netlink_ext_ack *extack);
3955 int dev_set_alias(struct net_device *, const char *, size_t);
3956 int dev_get_alias(const struct net_device *, char *, size_t);
3957 int __dev_change_net_namespace(struct net_device *dev, struct net *net,
3958 const char *pat, int new_ifindex);
3959 static inline
dev_change_net_namespace(struct net_device * dev,struct net * net,const char * pat)3960 int dev_change_net_namespace(struct net_device *dev, struct net *net,
3961 const char *pat)
3962 {
3963 return __dev_change_net_namespace(dev, net, pat, 0);
3964 }
3965 int __dev_set_mtu(struct net_device *, int);
3966 int dev_set_mtu(struct net_device *, int);
3967 int dev_pre_changeaddr_notify(struct net_device *dev, const char *addr,
3968 struct netlink_ext_ack *extack);
3969 int dev_set_mac_address(struct net_device *dev, struct sockaddr *sa,
3970 struct netlink_ext_ack *extack);
3971 int dev_set_mac_address_user(struct net_device *dev, struct sockaddr *sa,
3972 struct netlink_ext_ack *extack);
3973 int dev_get_mac_address(struct sockaddr *sa, struct net *net, char *dev_name);
3974 int dev_get_port_parent_id(struct net_device *dev,
3975 struct netdev_phys_item_id *ppid, bool recurse);
3976 bool netdev_port_same_parent_id(struct net_device *a, struct net_device *b);
3977 struct sk_buff *validate_xmit_skb_list(struct sk_buff *skb, struct net_device *dev, bool *again);
3978 struct sk_buff *dev_hard_start_xmit(struct sk_buff *skb, struct net_device *dev,
3979 struct netdev_queue *txq, int *ret);
3980
3981 int bpf_xdp_link_attach(const union bpf_attr *attr, struct bpf_prog *prog);
3982 u8 dev_xdp_prog_count(struct net_device *dev);
3983 u32 dev_xdp_prog_id(struct net_device *dev, enum bpf_xdp_mode mode);
3984
3985 int __dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
3986 int dev_forward_skb(struct net_device *dev, struct sk_buff *skb);
3987 int dev_forward_skb_nomtu(struct net_device *dev, struct sk_buff *skb);
3988 bool is_skb_forwardable(const struct net_device *dev,
3989 const struct sk_buff *skb);
3990
__is_skb_forwardable(const struct net_device * dev,const struct sk_buff * skb,const bool check_mtu)3991 static __always_inline bool __is_skb_forwardable(const struct net_device *dev,
3992 const struct sk_buff *skb,
3993 const bool check_mtu)
3994 {
3995 const u32 vlan_hdr_len = 4; /* VLAN_HLEN */
3996 unsigned int len;
3997
3998 if (!(dev->flags & IFF_UP))
3999 return false;
4000
4001 if (!check_mtu)
4002 return true;
4003
4004 len = dev->mtu + dev->hard_header_len + vlan_hdr_len;
4005 if (skb->len <= len)
4006 return true;
4007
4008 /* if TSO is enabled, we don't care about the length as the packet
4009 * could be forwarded without being segmented before
4010 */
4011 if (skb_is_gso(skb))
4012 return true;
4013
4014 return false;
4015 }
4016
4017 struct net_device_core_stats __percpu *netdev_core_stats_alloc(struct net_device *dev);
4018
dev_core_stats(struct net_device * dev)4019 static inline struct net_device_core_stats __percpu *dev_core_stats(struct net_device *dev)
4020 {
4021 /* This READ_ONCE() pairs with the write in netdev_core_stats_alloc() */
4022 struct net_device_core_stats __percpu *p = READ_ONCE(dev->core_stats);
4023
4024 if (likely(p))
4025 return p;
4026
4027 return netdev_core_stats_alloc(dev);
4028 }
4029
4030 #define DEV_CORE_STATS_INC(FIELD) \
4031 static inline void dev_core_stats_##FIELD##_inc(struct net_device *dev) \
4032 { \
4033 struct net_device_core_stats __percpu *p; \
4034 \
4035 p = dev_core_stats(dev); \
4036 if (p) \
4037 this_cpu_inc(p->FIELD); \
4038 }
4039 DEV_CORE_STATS_INC(rx_dropped)
DEV_CORE_STATS_INC(tx_dropped)4040 DEV_CORE_STATS_INC(tx_dropped)
4041 DEV_CORE_STATS_INC(rx_nohandler)
4042 DEV_CORE_STATS_INC(rx_otherhost_dropped)
4043
4044 static __always_inline int ____dev_forward_skb(struct net_device *dev,
4045 struct sk_buff *skb,
4046 const bool check_mtu)
4047 {
4048 if (skb_orphan_frags(skb, GFP_ATOMIC) ||
4049 unlikely(!__is_skb_forwardable(dev, skb, check_mtu))) {
4050 dev_core_stats_rx_dropped_inc(dev);
4051 kfree_skb(skb);
4052 return NET_RX_DROP;
4053 }
4054
4055 skb_scrub_packet(skb, !net_eq(dev_net(dev), dev_net(skb->dev)));
4056 skb->priority = 0;
4057 return 0;
4058 }
4059
4060 bool dev_nit_active(struct net_device *dev);
4061 void dev_queue_xmit_nit(struct sk_buff *skb, struct net_device *dev);
4062
__dev_put(struct net_device * dev)4063 static inline void __dev_put(struct net_device *dev)
4064 {
4065 if (dev) {
4066 #ifdef CONFIG_PCPU_DEV_REFCNT
4067 this_cpu_dec(*dev->pcpu_refcnt);
4068 #else
4069 refcount_dec(&dev->dev_refcnt);
4070 #endif
4071 }
4072 }
4073
__dev_hold(struct net_device * dev)4074 static inline void __dev_hold(struct net_device *dev)
4075 {
4076 if (dev) {
4077 #ifdef CONFIG_PCPU_DEV_REFCNT
4078 this_cpu_inc(*dev->pcpu_refcnt);
4079 #else
4080 refcount_inc(&dev->dev_refcnt);
4081 #endif
4082 }
4083 }
4084
__netdev_tracker_alloc(struct net_device * dev,netdevice_tracker * tracker,gfp_t gfp)4085 static inline void __netdev_tracker_alloc(struct net_device *dev,
4086 netdevice_tracker *tracker,
4087 gfp_t gfp)
4088 {
4089 #ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4090 ref_tracker_alloc(&dev->refcnt_tracker, tracker, gfp);
4091 #endif
4092 }
4093
4094 /* netdev_tracker_alloc() can upgrade a prior untracked reference
4095 * taken by dev_get_by_name()/dev_get_by_index() to a tracked one.
4096 */
netdev_tracker_alloc(struct net_device * dev,netdevice_tracker * tracker,gfp_t gfp)4097 static inline void netdev_tracker_alloc(struct net_device *dev,
4098 netdevice_tracker *tracker, gfp_t gfp)
4099 {
4100 #ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4101 refcount_dec(&dev->refcnt_tracker.no_tracker);
4102 __netdev_tracker_alloc(dev, tracker, gfp);
4103 #endif
4104 }
4105
netdev_tracker_free(struct net_device * dev,netdevice_tracker * tracker)4106 static inline void netdev_tracker_free(struct net_device *dev,
4107 netdevice_tracker *tracker)
4108 {
4109 #ifdef CONFIG_NET_DEV_REFCNT_TRACKER
4110 ref_tracker_free(&dev->refcnt_tracker, tracker);
4111 #endif
4112 }
4113
netdev_hold(struct net_device * dev,netdevice_tracker * tracker,gfp_t gfp)4114 static inline void netdev_hold(struct net_device *dev,
4115 netdevice_tracker *tracker, gfp_t gfp)
4116 {
4117 if (dev) {
4118 __dev_hold(dev);
4119 __netdev_tracker_alloc(dev, tracker, gfp);
4120 }
4121 }
4122
netdev_put(struct net_device * dev,netdevice_tracker * tracker)4123 static inline void netdev_put(struct net_device *dev,
4124 netdevice_tracker *tracker)
4125 {
4126 if (dev) {
4127 netdev_tracker_free(dev, tracker);
4128 __dev_put(dev);
4129 }
4130 }
4131
4132 /**
4133 * dev_hold - get reference to device
4134 * @dev: network device
4135 *
4136 * Hold reference to device to keep it from being freed.
4137 * Try using netdev_hold() instead.
4138 */
dev_hold(struct net_device * dev)4139 static inline void dev_hold(struct net_device *dev)
4140 {
4141 netdev_hold(dev, NULL, GFP_ATOMIC);
4142 }
4143
4144 /**
4145 * dev_put - release reference to device
4146 * @dev: network device
4147 *
4148 * Release reference to device to allow it to be freed.
4149 * Try using netdev_put() instead.
4150 */
dev_put(struct net_device * dev)4151 static inline void dev_put(struct net_device *dev)
4152 {
4153 netdev_put(dev, NULL);
4154 }
4155
netdev_ref_replace(struct net_device * odev,struct net_device * ndev,netdevice_tracker * tracker,gfp_t gfp)4156 static inline void netdev_ref_replace(struct net_device *odev,
4157 struct net_device *ndev,
4158 netdevice_tracker *tracker,
4159 gfp_t gfp)
4160 {
4161 if (odev)
4162 netdev_tracker_free(odev, tracker);
4163
4164 __dev_hold(ndev);
4165 __dev_put(odev);
4166
4167 if (ndev)
4168 __netdev_tracker_alloc(ndev, tracker, gfp);
4169 }
4170
4171 /* Carrier loss detection, dial on demand. The functions netif_carrier_on
4172 * and _off may be called from IRQ context, but it is caller
4173 * who is responsible for serialization of these calls.
4174 *
4175 * The name carrier is inappropriate, these functions should really be
4176 * called netif_lowerlayer_*() because they represent the state of any
4177 * kind of lower layer not just hardware media.
4178 */
4179 void linkwatch_fire_event(struct net_device *dev);
4180
4181 /**
4182 * netif_carrier_ok - test if carrier present
4183 * @dev: network device
4184 *
4185 * Check if carrier is present on device
4186 */
netif_carrier_ok(const struct net_device * dev)4187 static inline bool netif_carrier_ok(const struct net_device *dev)
4188 {
4189 return !test_bit(__LINK_STATE_NOCARRIER, &dev->state);
4190 }
4191
4192 unsigned long dev_trans_start(struct net_device *dev);
4193
4194 void __netdev_watchdog_up(struct net_device *dev);
4195
4196 void netif_carrier_on(struct net_device *dev);
4197 void netif_carrier_off(struct net_device *dev);
4198 void netif_carrier_event(struct net_device *dev);
4199
4200 /**
4201 * netif_dormant_on - mark device as dormant.
4202 * @dev: network device
4203 *
4204 * Mark device as dormant (as per RFC2863).
4205 *
4206 * The dormant state indicates that the relevant interface is not
4207 * actually in a condition to pass packets (i.e., it is not 'up') but is
4208 * in a "pending" state, waiting for some external event. For "on-
4209 * demand" interfaces, this new state identifies the situation where the
4210 * interface is waiting for events to place it in the up state.
4211 */
netif_dormant_on(struct net_device * dev)4212 static inline void netif_dormant_on(struct net_device *dev)
4213 {
4214 if (!test_and_set_bit(__LINK_STATE_DORMANT, &dev->state))
4215 linkwatch_fire_event(dev);
4216 }
4217
4218 /**
4219 * netif_dormant_off - set device as not dormant.
4220 * @dev: network device
4221 *
4222 * Device is not in dormant state.
4223 */
netif_dormant_off(struct net_device * dev)4224 static inline void netif_dormant_off(struct net_device *dev)
4225 {
4226 if (test_and_clear_bit(__LINK_STATE_DORMANT, &dev->state))
4227 linkwatch_fire_event(dev);
4228 }
4229
4230 /**
4231 * netif_dormant - test if device is dormant
4232 * @dev: network device
4233 *
4234 * Check if device is dormant.
4235 */
netif_dormant(const struct net_device * dev)4236 static inline bool netif_dormant(const struct net_device *dev)
4237 {
4238 return test_bit(__LINK_STATE_DORMANT, &dev->state);
4239 }
4240
4241
4242 /**
4243 * netif_testing_on - mark device as under test.
4244 * @dev: network device
4245 *
4246 * Mark device as under test (as per RFC2863).
4247 *
4248 * The testing state indicates that some test(s) must be performed on
4249 * the interface. After completion, of the test, the interface state
4250 * will change to up, dormant, or down, as appropriate.
4251 */
netif_testing_on(struct net_device * dev)4252 static inline void netif_testing_on(struct net_device *dev)
4253 {
4254 if (!test_and_set_bit(__LINK_STATE_TESTING, &dev->state))
4255 linkwatch_fire_event(dev);
4256 }
4257
4258 /**
4259 * netif_testing_off - set device as not under test.
4260 * @dev: network device
4261 *
4262 * Device is not in testing state.
4263 */
netif_testing_off(struct net_device * dev)4264 static inline void netif_testing_off(struct net_device *dev)
4265 {
4266 if (test_and_clear_bit(__LINK_STATE_TESTING, &dev->state))
4267 linkwatch_fire_event(dev);
4268 }
4269
4270 /**
4271 * netif_testing - test if device is under test
4272 * @dev: network device
4273 *
4274 * Check if device is under test
4275 */
netif_testing(const struct net_device * dev)4276 static inline bool netif_testing(const struct net_device *dev)
4277 {
4278 return test_bit(__LINK_STATE_TESTING, &dev->state);
4279 }
4280
4281
4282 /**
4283 * netif_oper_up - test if device is operational
4284 * @dev: network device
4285 *
4286 * Check if carrier is operational
4287 */
netif_oper_up(const struct net_device * dev)4288 static inline bool netif_oper_up(const struct net_device *dev)
4289 {
4290 return (dev->operstate == IF_OPER_UP ||
4291 dev->operstate == IF_OPER_UNKNOWN /* backward compat */);
4292 }
4293
4294 /**
4295 * netif_device_present - is device available or removed
4296 * @dev: network device
4297 *
4298 * Check if device has not been removed from system.
4299 */
netif_device_present(const struct net_device * dev)4300 static inline bool netif_device_present(const struct net_device *dev)
4301 {
4302 return test_bit(__LINK_STATE_PRESENT, &dev->state);
4303 }
4304
4305 void netif_device_detach(struct net_device *dev);
4306
4307 void netif_device_attach(struct net_device *dev);
4308
4309 /*
4310 * Network interface message level settings
4311 */
4312
4313 enum {
4314 NETIF_MSG_DRV_BIT,
4315 NETIF_MSG_PROBE_BIT,
4316 NETIF_MSG_LINK_BIT,
4317 NETIF_MSG_TIMER_BIT,
4318 NETIF_MSG_IFDOWN_BIT,
4319 NETIF_MSG_IFUP_BIT,
4320 NETIF_MSG_RX_ERR_BIT,
4321 NETIF_MSG_TX_ERR_BIT,
4322 NETIF_MSG_TX_QUEUED_BIT,
4323 NETIF_MSG_INTR_BIT,
4324 NETIF_MSG_TX_DONE_BIT,
4325 NETIF_MSG_RX_STATUS_BIT,
4326 NETIF_MSG_PKTDATA_BIT,
4327 NETIF_MSG_HW_BIT,
4328 NETIF_MSG_WOL_BIT,
4329
4330 /* When you add a new bit above, update netif_msg_class_names array
4331 * in net/ethtool/common.c
4332 */
4333 NETIF_MSG_CLASS_COUNT,
4334 };
4335 /* Both ethtool_ops interface and internal driver implementation use u32 */
4336 static_assert(NETIF_MSG_CLASS_COUNT <= 32);
4337
4338 #define __NETIF_MSG_BIT(bit) ((u32)1 << (bit))
4339 #define __NETIF_MSG(name) __NETIF_MSG_BIT(NETIF_MSG_ ## name ## _BIT)
4340
4341 #define NETIF_MSG_DRV __NETIF_MSG(DRV)
4342 #define NETIF_MSG_PROBE __NETIF_MSG(PROBE)
4343 #define NETIF_MSG_LINK __NETIF_MSG(LINK)
4344 #define NETIF_MSG_TIMER __NETIF_MSG(TIMER)
4345 #define NETIF_MSG_IFDOWN __NETIF_MSG(IFDOWN)
4346 #define NETIF_MSG_IFUP __NETIF_MSG(IFUP)
4347 #define NETIF_MSG_RX_ERR __NETIF_MSG(RX_ERR)
4348 #define NETIF_MSG_TX_ERR __NETIF_MSG(TX_ERR)
4349 #define NETIF_MSG_TX_QUEUED __NETIF_MSG(TX_QUEUED)
4350 #define NETIF_MSG_INTR __NETIF_MSG(INTR)
4351 #define NETIF_MSG_TX_DONE __NETIF_MSG(TX_DONE)
4352 #define NETIF_MSG_RX_STATUS __NETIF_MSG(RX_STATUS)
4353 #define NETIF_MSG_PKTDATA __NETIF_MSG(PKTDATA)
4354 #define NETIF_MSG_HW __NETIF_MSG(HW)
4355 #define NETIF_MSG_WOL __NETIF_MSG(WOL)
4356
4357 #define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
4358 #define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
4359 #define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
4360 #define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
4361 #define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
4362 #define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
4363 #define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
4364 #define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
4365 #define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
4366 #define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
4367 #define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
4368 #define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
4369 #define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
4370 #define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
4371 #define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
4372
netif_msg_init(int debug_value,int default_msg_enable_bits)4373 static inline u32 netif_msg_init(int debug_value, int default_msg_enable_bits)
4374 {
4375 /* use default */
4376 if (debug_value < 0 || debug_value >= (sizeof(u32) * 8))
4377 return default_msg_enable_bits;
4378 if (debug_value == 0) /* no output */
4379 return 0;
4380 /* set low N bits */
4381 return (1U << debug_value) - 1;
4382 }
4383
__netif_tx_lock(struct netdev_queue * txq,int cpu)4384 static inline void __netif_tx_lock(struct netdev_queue *txq, int cpu)
4385 {
4386 spin_lock(&txq->_xmit_lock);
4387 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4388 WRITE_ONCE(txq->xmit_lock_owner, cpu);
4389 }
4390
__netif_tx_acquire(struct netdev_queue * txq)4391 static inline bool __netif_tx_acquire(struct netdev_queue *txq)
4392 {
4393 __acquire(&txq->_xmit_lock);
4394 return true;
4395 }
4396
__netif_tx_release(struct netdev_queue * txq)4397 static inline void __netif_tx_release(struct netdev_queue *txq)
4398 {
4399 __release(&txq->_xmit_lock);
4400 }
4401
__netif_tx_lock_bh(struct netdev_queue * txq)4402 static inline void __netif_tx_lock_bh(struct netdev_queue *txq)
4403 {
4404 spin_lock_bh(&txq->_xmit_lock);
4405 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4406 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
4407 }
4408
__netif_tx_trylock(struct netdev_queue * txq)4409 static inline bool __netif_tx_trylock(struct netdev_queue *txq)
4410 {
4411 bool ok = spin_trylock(&txq->_xmit_lock);
4412
4413 if (likely(ok)) {
4414 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4415 WRITE_ONCE(txq->xmit_lock_owner, smp_processor_id());
4416 }
4417 return ok;
4418 }
4419
__netif_tx_unlock(struct netdev_queue * txq)4420 static inline void __netif_tx_unlock(struct netdev_queue *txq)
4421 {
4422 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4423 WRITE_ONCE(txq->xmit_lock_owner, -1);
4424 spin_unlock(&txq->_xmit_lock);
4425 }
4426
__netif_tx_unlock_bh(struct netdev_queue * txq)4427 static inline void __netif_tx_unlock_bh(struct netdev_queue *txq)
4428 {
4429 /* Pairs with READ_ONCE() in __dev_queue_xmit() */
4430 WRITE_ONCE(txq->xmit_lock_owner, -1);
4431 spin_unlock_bh(&txq->_xmit_lock);
4432 }
4433
4434 /*
4435 * txq->trans_start can be read locklessly from dev_watchdog()
4436 */
txq_trans_update(struct netdev_queue * txq)4437 static inline void txq_trans_update(struct netdev_queue *txq)
4438 {
4439 if (txq->xmit_lock_owner != -1)
4440 WRITE_ONCE(txq->trans_start, jiffies);
4441 }
4442
txq_trans_cond_update(struct netdev_queue * txq)4443 static inline void txq_trans_cond_update(struct netdev_queue *txq)
4444 {
4445 unsigned long now = jiffies;
4446
4447 if (READ_ONCE(txq->trans_start) != now)
4448 WRITE_ONCE(txq->trans_start, now);
4449 }
4450
4451 /* legacy drivers only, netdev_start_xmit() sets txq->trans_start */
netif_trans_update(struct net_device * dev)4452 static inline void netif_trans_update(struct net_device *dev)
4453 {
4454 struct netdev_queue *txq = netdev_get_tx_queue(dev, 0);
4455
4456 txq_trans_cond_update(txq);
4457 }
4458
4459 /**
4460 * netif_tx_lock - grab network device transmit lock
4461 * @dev: network device
4462 *
4463 * Get network device transmit lock
4464 */
4465 void netif_tx_lock(struct net_device *dev);
4466
netif_tx_lock_bh(struct net_device * dev)4467 static inline void netif_tx_lock_bh(struct net_device *dev)
4468 {
4469 local_bh_disable();
4470 netif_tx_lock(dev);
4471 }
4472
4473 void netif_tx_unlock(struct net_device *dev);
4474
netif_tx_unlock_bh(struct net_device * dev)4475 static inline void netif_tx_unlock_bh(struct net_device *dev)
4476 {
4477 netif_tx_unlock(dev);
4478 local_bh_enable();
4479 }
4480
4481 #define HARD_TX_LOCK(dev, txq, cpu) { \
4482 if ((dev->features & NETIF_F_LLTX) == 0) { \
4483 __netif_tx_lock(txq, cpu); \
4484 } else { \
4485 __netif_tx_acquire(txq); \
4486 } \
4487 }
4488
4489 #define HARD_TX_TRYLOCK(dev, txq) \
4490 (((dev->features & NETIF_F_LLTX) == 0) ? \
4491 __netif_tx_trylock(txq) : \
4492 __netif_tx_acquire(txq))
4493
4494 #define HARD_TX_UNLOCK(dev, txq) { \
4495 if ((dev->features & NETIF_F_LLTX) == 0) { \
4496 __netif_tx_unlock(txq); \
4497 } else { \
4498 __netif_tx_release(txq); \
4499 } \
4500 }
4501
netif_tx_disable(struct net_device * dev)4502 static inline void netif_tx_disable(struct net_device *dev)
4503 {
4504 unsigned int i;
4505 int cpu;
4506
4507 local_bh_disable();
4508 cpu = smp_processor_id();
4509 spin_lock(&dev->tx_global_lock);
4510 for (i = 0; i < dev->num_tx_queues; i++) {
4511 struct netdev_queue *txq = netdev_get_tx_queue(dev, i);
4512
4513 __netif_tx_lock(txq, cpu);
4514 netif_tx_stop_queue(txq);
4515 __netif_tx_unlock(txq);
4516 }
4517 spin_unlock(&dev->tx_global_lock);
4518 local_bh_enable();
4519 }
4520
netif_addr_lock(struct net_device * dev)4521 static inline void netif_addr_lock(struct net_device *dev)
4522 {
4523 unsigned char nest_level = 0;
4524
4525 #ifdef CONFIG_LOCKDEP
4526 nest_level = dev->nested_level;
4527 #endif
4528 spin_lock_nested(&dev->addr_list_lock, nest_level);
4529 }
4530
netif_addr_lock_bh(struct net_device * dev)4531 static inline void netif_addr_lock_bh(struct net_device *dev)
4532 {
4533 unsigned char nest_level = 0;
4534
4535 #ifdef CONFIG_LOCKDEP
4536 nest_level = dev->nested_level;
4537 #endif
4538 local_bh_disable();
4539 spin_lock_nested(&dev->addr_list_lock, nest_level);
4540 }
4541
netif_addr_unlock(struct net_device * dev)4542 static inline void netif_addr_unlock(struct net_device *dev)
4543 {
4544 spin_unlock(&dev->addr_list_lock);
4545 }
4546
netif_addr_unlock_bh(struct net_device * dev)4547 static inline void netif_addr_unlock_bh(struct net_device *dev)
4548 {
4549 spin_unlock_bh(&dev->addr_list_lock);
4550 }
4551
4552 /*
4553 * dev_addrs walker. Should be used only for read access. Call with
4554 * rcu_read_lock held.
4555 */
4556 #define for_each_dev_addr(dev, ha) \
4557 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
4558
4559 /* These functions live elsewhere (drivers/net/net_init.c, but related) */
4560
4561 void ether_setup(struct net_device *dev);
4562
4563 /* Support for loadable net-drivers */
4564 struct net_device *alloc_netdev_mqs(int sizeof_priv, const char *name,
4565 unsigned char name_assign_type,
4566 void (*setup)(struct net_device *),
4567 unsigned int txqs, unsigned int rxqs);
4568 #define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
4569 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
4570
4571 #define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
4572 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
4573 count)
4574
4575 int register_netdev(struct net_device *dev);
4576 void unregister_netdev(struct net_device *dev);
4577
4578 int devm_register_netdev(struct device *dev, struct net_device *ndev);
4579
4580 /* General hardware address lists handling functions */
4581 int __hw_addr_sync(struct netdev_hw_addr_list *to_list,
4582 struct netdev_hw_addr_list *from_list, int addr_len);
4583 void __hw_addr_unsync(struct netdev_hw_addr_list *to_list,
4584 struct netdev_hw_addr_list *from_list, int addr_len);
4585 int __hw_addr_sync_dev(struct netdev_hw_addr_list *list,
4586 struct net_device *dev,
4587 int (*sync)(struct net_device *, const unsigned char *),
4588 int (*unsync)(struct net_device *,
4589 const unsigned char *));
4590 int __hw_addr_ref_sync_dev(struct netdev_hw_addr_list *list,
4591 struct net_device *dev,
4592 int (*sync)(struct net_device *,
4593 const unsigned char *, int),
4594 int (*unsync)(struct net_device *,
4595 const unsigned char *, int));
4596 void __hw_addr_ref_unsync_dev(struct netdev_hw_addr_list *list,
4597 struct net_device *dev,
4598 int (*unsync)(struct net_device *,
4599 const unsigned char *, int));
4600 void __hw_addr_unsync_dev(struct netdev_hw_addr_list *list,
4601 struct net_device *dev,
4602 int (*unsync)(struct net_device *,
4603 const unsigned char *));
4604 void __hw_addr_init(struct netdev_hw_addr_list *list);
4605
4606 /* Functions used for device addresses handling */
4607 void dev_addr_mod(struct net_device *dev, unsigned int offset,
4608 const void *addr, size_t len);
4609
4610 static inline void
__dev_addr_set(struct net_device * dev,const void * addr,size_t len)4611 __dev_addr_set(struct net_device *dev, const void *addr, size_t len)
4612 {
4613 dev_addr_mod(dev, 0, addr, len);
4614 }
4615
dev_addr_set(struct net_device * dev,const u8 * addr)4616 static inline void dev_addr_set(struct net_device *dev, const u8 *addr)
4617 {
4618 __dev_addr_set(dev, addr, dev->addr_len);
4619 }
4620
4621 int dev_addr_add(struct net_device *dev, const unsigned char *addr,
4622 unsigned char addr_type);
4623 int dev_addr_del(struct net_device *dev, const unsigned char *addr,
4624 unsigned char addr_type);
4625
4626 /* Functions used for unicast addresses handling */
4627 int dev_uc_add(struct net_device *dev, const unsigned char *addr);
4628 int dev_uc_add_excl(struct net_device *dev, const unsigned char *addr);
4629 int dev_uc_del(struct net_device *dev, const unsigned char *addr);
4630 int dev_uc_sync(struct net_device *to, struct net_device *from);
4631 int dev_uc_sync_multiple(struct net_device *to, struct net_device *from);
4632 void dev_uc_unsync(struct net_device *to, struct net_device *from);
4633 void dev_uc_flush(struct net_device *dev);
4634 void dev_uc_init(struct net_device *dev);
4635
4636 /**
4637 * __dev_uc_sync - Synchonize device's unicast list
4638 * @dev: device to sync
4639 * @sync: function to call if address should be added
4640 * @unsync: function to call if address should be removed
4641 *
4642 * Add newly added addresses to the interface, and release
4643 * addresses that have been deleted.
4644 */
__dev_uc_sync(struct net_device * dev,int (* sync)(struct net_device *,const unsigned char *),int (* unsync)(struct net_device *,const unsigned char *))4645 static inline int __dev_uc_sync(struct net_device *dev,
4646 int (*sync)(struct net_device *,
4647 const unsigned char *),
4648 int (*unsync)(struct net_device *,
4649 const unsigned char *))
4650 {
4651 return __hw_addr_sync_dev(&dev->uc, dev, sync, unsync);
4652 }
4653
4654 /**
4655 * __dev_uc_unsync - Remove synchronized addresses from device
4656 * @dev: device to sync
4657 * @unsync: function to call if address should be removed
4658 *
4659 * Remove all addresses that were added to the device by dev_uc_sync().
4660 */
__dev_uc_unsync(struct net_device * dev,int (* unsync)(struct net_device *,const unsigned char *))4661 static inline void __dev_uc_unsync(struct net_device *dev,
4662 int (*unsync)(struct net_device *,
4663 const unsigned char *))
4664 {
4665 __hw_addr_unsync_dev(&dev->uc, dev, unsync);
4666 }
4667
4668 /* Functions used for multicast addresses handling */
4669 int dev_mc_add(struct net_device *dev, const unsigned char *addr);
4670 int dev_mc_add_global(struct net_device *dev, const unsigned char *addr);
4671 int dev_mc_add_excl(struct net_device *dev, const unsigned char *addr);
4672 int dev_mc_del(struct net_device *dev, const unsigned char *addr);
4673 int dev_mc_del_global(struct net_device *dev, const unsigned char *addr);
4674 int dev_mc_sync(struct net_device *to, struct net_device *from);
4675 int dev_mc_sync_multiple(struct net_device *to, struct net_device *from);
4676 void dev_mc_unsync(struct net_device *to, struct net_device *from);
4677 void dev_mc_flush(struct net_device *dev);
4678 void dev_mc_init(struct net_device *dev);
4679
4680 /**
4681 * __dev_mc_sync - Synchonize device's multicast list
4682 * @dev: device to sync
4683 * @sync: function to call if address should be added
4684 * @unsync: function to call if address should be removed
4685 *
4686 * Add newly added addresses to the interface, and release
4687 * addresses that have been deleted.
4688 */
__dev_mc_sync(struct net_device * dev,int (* sync)(struct net_device *,const unsigned char *),int (* unsync)(struct net_device *,const unsigned char *))4689 static inline int __dev_mc_sync(struct net_device *dev,
4690 int (*sync)(struct net_device *,
4691 const unsigned char *),
4692 int (*unsync)(struct net_device *,
4693 const unsigned char *))
4694 {
4695 return __hw_addr_sync_dev(&dev->mc, dev, sync, unsync);
4696 }
4697
4698 /**
4699 * __dev_mc_unsync - Remove synchronized addresses from device
4700 * @dev: device to sync
4701 * @unsync: function to call if address should be removed
4702 *
4703 * Remove all addresses that were added to the device by dev_mc_sync().
4704 */
__dev_mc_unsync(struct net_device * dev,int (* unsync)(struct net_device *,const unsigned char *))4705 static inline void __dev_mc_unsync(struct net_device *dev,
4706 int (*unsync)(struct net_device *,
4707 const unsigned char *))
4708 {
4709 __hw_addr_unsync_dev(&dev->mc, dev, unsync);
4710 }
4711
4712 /* Functions used for secondary unicast and multicast support */
4713 void dev_set_rx_mode(struct net_device *dev);
4714 int dev_set_promiscuity(struct net_device *dev, int inc);
4715 int dev_set_allmulti(struct net_device *dev, int inc);
4716 void netdev_state_change(struct net_device *dev);
4717 void __netdev_notify_peers(struct net_device *dev);
4718 void netdev_notify_peers(struct net_device *dev);
4719 void netdev_features_change(struct net_device *dev);
4720 /* Load a device via the kmod */
4721 void dev_load(struct net *net, const char *name);
4722 struct rtnl_link_stats64 *dev_get_stats(struct net_device *dev,
4723 struct rtnl_link_stats64 *storage);
4724 void netdev_stats_to_stats64(struct rtnl_link_stats64 *stats64,
4725 const struct net_device_stats *netdev_stats);
4726 void dev_fetch_sw_netstats(struct rtnl_link_stats64 *s,
4727 const struct pcpu_sw_netstats __percpu *netstats);
4728 void dev_get_tstats64(struct net_device *dev, struct rtnl_link_stats64 *s);
4729
4730 extern int netdev_max_backlog;
4731 extern int dev_rx_weight;
4732 extern int dev_tx_weight;
4733 extern int gro_normal_batch;
4734
4735 enum {
4736 NESTED_SYNC_IMM_BIT,
4737 NESTED_SYNC_TODO_BIT,
4738 };
4739
4740 #define __NESTED_SYNC_BIT(bit) ((u32)1 << (bit))
4741 #define __NESTED_SYNC(name) __NESTED_SYNC_BIT(NESTED_SYNC_ ## name ## _BIT)
4742
4743 #define NESTED_SYNC_IMM __NESTED_SYNC(IMM)
4744 #define NESTED_SYNC_TODO __NESTED_SYNC(TODO)
4745
4746 struct netdev_nested_priv {
4747 unsigned char flags;
4748 void *data;
4749 };
4750
4751 bool netdev_has_upper_dev(struct net_device *dev, struct net_device *upper_dev);
4752 struct net_device *netdev_upper_get_next_dev_rcu(struct net_device *dev,
4753 struct list_head **iter);
4754
4755 /* iterate through upper list, must be called under RCU read lock */
4756 #define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
4757 for (iter = &(dev)->adj_list.upper, \
4758 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
4759 updev; \
4760 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
4761
4762 int netdev_walk_all_upper_dev_rcu(struct net_device *dev,
4763 int (*fn)(struct net_device *upper_dev,
4764 struct netdev_nested_priv *priv),
4765 struct netdev_nested_priv *priv);
4766
4767 bool netdev_has_upper_dev_all_rcu(struct net_device *dev,
4768 struct net_device *upper_dev);
4769
4770 bool netdev_has_any_upper_dev(struct net_device *dev);
4771
4772 void *netdev_lower_get_next_private(struct net_device *dev,
4773 struct list_head **iter);
4774 void *netdev_lower_get_next_private_rcu(struct net_device *dev,
4775 struct list_head **iter);
4776
4777 #define netdev_for_each_lower_private(dev, priv, iter) \
4778 for (iter = (dev)->adj_list.lower.next, \
4779 priv = netdev_lower_get_next_private(dev, &(iter)); \
4780 priv; \
4781 priv = netdev_lower_get_next_private(dev, &(iter)))
4782
4783 #define netdev_for_each_lower_private_rcu(dev, priv, iter) \
4784 for (iter = &(dev)->adj_list.lower, \
4785 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
4786 priv; \
4787 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
4788
4789 void *netdev_lower_get_next(struct net_device *dev,
4790 struct list_head **iter);
4791
4792 #define netdev_for_each_lower_dev(dev, ldev, iter) \
4793 for (iter = (dev)->adj_list.lower.next, \
4794 ldev = netdev_lower_get_next(dev, &(iter)); \
4795 ldev; \
4796 ldev = netdev_lower_get_next(dev, &(iter)))
4797
4798 struct net_device *netdev_next_lower_dev_rcu(struct net_device *dev,
4799 struct list_head **iter);
4800 int netdev_walk_all_lower_dev(struct net_device *dev,
4801 int (*fn)(struct net_device *lower_dev,
4802 struct netdev_nested_priv *priv),
4803 struct netdev_nested_priv *priv);
4804 int netdev_walk_all_lower_dev_rcu(struct net_device *dev,
4805 int (*fn)(struct net_device *lower_dev,
4806 struct netdev_nested_priv *priv),
4807 struct netdev_nested_priv *priv);
4808
4809 void *netdev_adjacent_get_private(struct list_head *adj_list);
4810 void *netdev_lower_get_first_private_rcu(struct net_device *dev);
4811 struct net_device *netdev_master_upper_dev_get(struct net_device *dev);
4812 struct net_device *netdev_master_upper_dev_get_rcu(struct net_device *dev);
4813 int netdev_upper_dev_link(struct net_device *dev, struct net_device *upper_dev,
4814 struct netlink_ext_ack *extack);
4815 int netdev_master_upper_dev_link(struct net_device *dev,
4816 struct net_device *upper_dev,
4817 void *upper_priv, void *upper_info,
4818 struct netlink_ext_ack *extack);
4819 void netdev_upper_dev_unlink(struct net_device *dev,
4820 struct net_device *upper_dev);
4821 int netdev_adjacent_change_prepare(struct net_device *old_dev,
4822 struct net_device *new_dev,
4823 struct net_device *dev,
4824 struct netlink_ext_ack *extack);
4825 void netdev_adjacent_change_commit(struct net_device *old_dev,
4826 struct net_device *new_dev,
4827 struct net_device *dev);
4828 void netdev_adjacent_change_abort(struct net_device *old_dev,
4829 struct net_device *new_dev,
4830 struct net_device *dev);
4831 void netdev_adjacent_rename_links(struct net_device *dev, char *oldname);
4832 void *netdev_lower_dev_get_private(struct net_device *dev,
4833 struct net_device *lower_dev);
4834 void netdev_lower_state_changed(struct net_device *lower_dev,
4835 void *lower_state_info);
4836
4837 /* RSS keys are 40 or 52 bytes long */
4838 #define NETDEV_RSS_KEY_LEN 52
4839 extern u8 netdev_rss_key[NETDEV_RSS_KEY_LEN] __read_mostly;
4840 void netdev_rss_key_fill(void *buffer, size_t len);
4841
4842 int skb_checksum_help(struct sk_buff *skb);
4843 int skb_crc32c_csum_help(struct sk_buff *skb);
4844 int skb_csum_hwoffload_help(struct sk_buff *skb,
4845 const netdev_features_t features);
4846
4847 struct netdev_bonding_info {
4848 ifslave slave;
4849 ifbond master;
4850 };
4851
4852 struct netdev_notifier_bonding_info {
4853 struct netdev_notifier_info info; /* must be first */
4854 struct netdev_bonding_info bonding_info;
4855 };
4856
4857 void netdev_bonding_info_change(struct net_device *dev,
4858 struct netdev_bonding_info *bonding_info);
4859
4860 #if IS_ENABLED(CONFIG_ETHTOOL_NETLINK)
4861 void ethtool_notify(struct net_device *dev, unsigned int cmd, const void *data);
4862 #else
ethtool_notify(struct net_device * dev,unsigned int cmd,const void * data)4863 static inline void ethtool_notify(struct net_device *dev, unsigned int cmd,
4864 const void *data)
4865 {
4866 }
4867 #endif
4868
4869 __be16 skb_network_protocol(struct sk_buff *skb, int *depth);
4870
can_checksum_protocol(netdev_features_t features,__be16 protocol)4871 static inline bool can_checksum_protocol(netdev_features_t features,
4872 __be16 protocol)
4873 {
4874 if (protocol == htons(ETH_P_FCOE))
4875 return !!(features & NETIF_F_FCOE_CRC);
4876
4877 /* Assume this is an IP checksum (not SCTP CRC) */
4878
4879 if (features & NETIF_F_HW_CSUM) {
4880 /* Can checksum everything */
4881 return true;
4882 }
4883
4884 switch (protocol) {
4885 case htons(ETH_P_IP):
4886 return !!(features & NETIF_F_IP_CSUM);
4887 case htons(ETH_P_IPV6):
4888 return !!(features & NETIF_F_IPV6_CSUM);
4889 default:
4890 return false;
4891 }
4892 }
4893
4894 #ifdef CONFIG_BUG
4895 void netdev_rx_csum_fault(struct net_device *dev, struct sk_buff *skb);
4896 #else
netdev_rx_csum_fault(struct net_device * dev,struct sk_buff * skb)4897 static inline void netdev_rx_csum_fault(struct net_device *dev,
4898 struct sk_buff *skb)
4899 {
4900 }
4901 #endif
4902 /* rx skb timestamps */
4903 void net_enable_timestamp(void);
4904 void net_disable_timestamp(void);
4905
netdev_get_tstamp(struct net_device * dev,const struct skb_shared_hwtstamps * hwtstamps,bool cycles)4906 static inline ktime_t netdev_get_tstamp(struct net_device *dev,
4907 const struct skb_shared_hwtstamps *hwtstamps,
4908 bool cycles)
4909 {
4910 const struct net_device_ops *ops = dev->netdev_ops;
4911
4912 if (ops->ndo_get_tstamp)
4913 return ops->ndo_get_tstamp(dev, hwtstamps, cycles);
4914
4915 return hwtstamps->hwtstamp;
4916 }
4917
__netdev_start_xmit(const struct net_device_ops * ops,struct sk_buff * skb,struct net_device * dev,bool more)4918 static inline netdev_tx_t __netdev_start_xmit(const struct net_device_ops *ops,
4919 struct sk_buff *skb, struct net_device *dev,
4920 bool more)
4921 {
4922 __this_cpu_write(softnet_data.xmit.more, more);
4923 return ops->ndo_start_xmit(skb, dev);
4924 }
4925
netdev_xmit_more(void)4926 static inline bool netdev_xmit_more(void)
4927 {
4928 return __this_cpu_read(softnet_data.xmit.more);
4929 }
4930
netdev_start_xmit(struct sk_buff * skb,struct net_device * dev,struct netdev_queue * txq,bool more)4931 static inline netdev_tx_t netdev_start_xmit(struct sk_buff *skb, struct net_device *dev,
4932 struct netdev_queue *txq, bool more)
4933 {
4934 const struct net_device_ops *ops = dev->netdev_ops;
4935 netdev_tx_t rc;
4936
4937 rc = __netdev_start_xmit(ops, skb, dev, more);
4938 if (rc == NETDEV_TX_OK)
4939 txq_trans_update(txq);
4940
4941 return rc;
4942 }
4943
4944 int netdev_class_create_file_ns(const struct class_attribute *class_attr,
4945 const void *ns);
4946 void netdev_class_remove_file_ns(const struct class_attribute *class_attr,
4947 const void *ns);
4948
4949 extern const struct kobj_ns_type_operations net_ns_type_operations;
4950
4951 const char *netdev_drivername(const struct net_device *dev);
4952
netdev_intersect_features(netdev_features_t f1,netdev_features_t f2)4953 static inline netdev_features_t netdev_intersect_features(netdev_features_t f1,
4954 netdev_features_t f2)
4955 {
4956 if ((f1 ^ f2) & NETIF_F_HW_CSUM) {
4957 if (f1 & NETIF_F_HW_CSUM)
4958 f1 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
4959 else
4960 f2 |= (NETIF_F_IP_CSUM|NETIF_F_IPV6_CSUM);
4961 }
4962
4963 return f1 & f2;
4964 }
4965
netdev_get_wanted_features(struct net_device * dev)4966 static inline netdev_features_t netdev_get_wanted_features(
4967 struct net_device *dev)
4968 {
4969 return (dev->features & ~dev->hw_features) | dev->wanted_features;
4970 }
4971 netdev_features_t netdev_increment_features(netdev_features_t all,
4972 netdev_features_t one, netdev_features_t mask);
4973
4974 /* Allow TSO being used on stacked device :
4975 * Performing the GSO segmentation before last device
4976 * is a performance improvement.
4977 */
netdev_add_tso_features(netdev_features_t features,netdev_features_t mask)4978 static inline netdev_features_t netdev_add_tso_features(netdev_features_t features,
4979 netdev_features_t mask)
4980 {
4981 return netdev_increment_features(features, NETIF_F_ALL_TSO, mask);
4982 }
4983
4984 int __netdev_update_features(struct net_device *dev);
4985 void netdev_update_features(struct net_device *dev);
4986 void netdev_change_features(struct net_device *dev);
4987
4988 void netif_stacked_transfer_operstate(const struct net_device *rootdev,
4989 struct net_device *dev);
4990
4991 netdev_features_t passthru_features_check(struct sk_buff *skb,
4992 struct net_device *dev,
4993 netdev_features_t features);
4994 netdev_features_t netif_skb_features(struct sk_buff *skb);
4995 void skb_warn_bad_offload(const struct sk_buff *skb);
4996
net_gso_ok(netdev_features_t features,int gso_type)4997 static inline bool net_gso_ok(netdev_features_t features, int gso_type)
4998 {
4999 netdev_features_t feature = (netdev_features_t)gso_type << NETIF_F_GSO_SHIFT;
5000
5001 /* check flags correspondence */
5002 BUILD_BUG_ON(SKB_GSO_TCPV4 != (NETIF_F_TSO >> NETIF_F_GSO_SHIFT));
5003 BUILD_BUG_ON(SKB_GSO_DODGY != (NETIF_F_GSO_ROBUST >> NETIF_F_GSO_SHIFT));
5004 BUILD_BUG_ON(SKB_GSO_TCP_ECN != (NETIF_F_TSO_ECN >> NETIF_F_GSO_SHIFT));
5005 BUILD_BUG_ON(SKB_GSO_TCP_FIXEDID != (NETIF_F_TSO_MANGLEID >> NETIF_F_GSO_SHIFT));
5006 BUILD_BUG_ON(SKB_GSO_TCPV6 != (NETIF_F_TSO6 >> NETIF_F_GSO_SHIFT));
5007 BUILD_BUG_ON(SKB_GSO_FCOE != (NETIF_F_FSO >> NETIF_F_GSO_SHIFT));
5008 BUILD_BUG_ON(SKB_GSO_GRE != (NETIF_F_GSO_GRE >> NETIF_F_GSO_SHIFT));
5009 BUILD_BUG_ON(SKB_GSO_GRE_CSUM != (NETIF_F_GSO_GRE_CSUM >> NETIF_F_GSO_SHIFT));
5010 BUILD_BUG_ON(SKB_GSO_IPXIP4 != (NETIF_F_GSO_IPXIP4 >> NETIF_F_GSO_SHIFT));
5011 BUILD_BUG_ON(SKB_GSO_IPXIP6 != (NETIF_F_GSO_IPXIP6 >> NETIF_F_GSO_SHIFT));
5012 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL != (NETIF_F_GSO_UDP_TUNNEL >> NETIF_F_GSO_SHIFT));
5013 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM != (NETIF_F_GSO_UDP_TUNNEL_CSUM >> NETIF_F_GSO_SHIFT));
5014 BUILD_BUG_ON(SKB_GSO_PARTIAL != (NETIF_F_GSO_PARTIAL >> NETIF_F_GSO_SHIFT));
5015 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM != (NETIF_F_GSO_TUNNEL_REMCSUM >> NETIF_F_GSO_SHIFT));
5016 BUILD_BUG_ON(SKB_GSO_SCTP != (NETIF_F_GSO_SCTP >> NETIF_F_GSO_SHIFT));
5017 BUILD_BUG_ON(SKB_GSO_ESP != (NETIF_F_GSO_ESP >> NETIF_F_GSO_SHIFT));
5018 BUILD_BUG_ON(SKB_GSO_UDP != (NETIF_F_GSO_UDP >> NETIF_F_GSO_SHIFT));
5019 BUILD_BUG_ON(SKB_GSO_UDP_L4 != (NETIF_F_GSO_UDP_L4 >> NETIF_F_GSO_SHIFT));
5020 BUILD_BUG_ON(SKB_GSO_FRAGLIST != (NETIF_F_GSO_FRAGLIST >> NETIF_F_GSO_SHIFT));
5021
5022 return (features & feature) == feature;
5023 }
5024
skb_gso_ok(struct sk_buff * skb,netdev_features_t features)5025 static inline bool skb_gso_ok(struct sk_buff *skb, netdev_features_t features)
5026 {
5027 return net_gso_ok(features, skb_shinfo(skb)->gso_type) &&
5028 (!skb_has_frag_list(skb) || (features & NETIF_F_FRAGLIST));
5029 }
5030
netif_needs_gso(struct sk_buff * skb,netdev_features_t features)5031 static inline bool netif_needs_gso(struct sk_buff *skb,
5032 netdev_features_t features)
5033 {
5034 return skb_is_gso(skb) && (!skb_gso_ok(skb, features) ||
5035 unlikely((skb->ip_summed != CHECKSUM_PARTIAL) &&
5036 (skb->ip_summed != CHECKSUM_UNNECESSARY)));
5037 }
5038
5039 void netif_set_tso_max_size(struct net_device *dev, unsigned int size);
5040 void netif_set_tso_max_segs(struct net_device *dev, unsigned int segs);
5041 void netif_inherit_tso_max(struct net_device *to,
5042 const struct net_device *from);
5043
5044 static inline unsigned int
netif_get_gro_max_size(const struct net_device * dev,const struct sk_buff * skb)5045 netif_get_gro_max_size(const struct net_device *dev, const struct sk_buff *skb)
5046 {
5047 /* pairs with WRITE_ONCE() in netif_set_gro(_ipv4)_max_size() */
5048 return skb->protocol == htons(ETH_P_IPV6) ?
5049 READ_ONCE(dev->gro_max_size) :
5050 READ_ONCE(dev->gro_ipv4_max_size);
5051 }
5052
5053 static inline unsigned int
netif_get_gso_max_size(const struct net_device * dev,const struct sk_buff * skb)5054 netif_get_gso_max_size(const struct net_device *dev, const struct sk_buff *skb)
5055 {
5056 /* pairs with WRITE_ONCE() in netif_set_gso(_ipv4)_max_size() */
5057 return skb->protocol == htons(ETH_P_IPV6) ?
5058 READ_ONCE(dev->gso_max_size) :
5059 READ_ONCE(dev->gso_ipv4_max_size);
5060 }
5061
netif_is_macsec(const struct net_device * dev)5062 static inline bool netif_is_macsec(const struct net_device *dev)
5063 {
5064 return dev->priv_flags & IFF_MACSEC;
5065 }
5066
netif_is_macvlan(const struct net_device * dev)5067 static inline bool netif_is_macvlan(const struct net_device *dev)
5068 {
5069 return dev->priv_flags & IFF_MACVLAN;
5070 }
5071
netif_is_macvlan_port(const struct net_device * dev)5072 static inline bool netif_is_macvlan_port(const struct net_device *dev)
5073 {
5074 return dev->priv_flags & IFF_MACVLAN_PORT;
5075 }
5076
netif_is_bond_master(const struct net_device * dev)5077 static inline bool netif_is_bond_master(const struct net_device *dev)
5078 {
5079 return dev->flags & IFF_MASTER && dev->priv_flags & IFF_BONDING;
5080 }
5081
netif_is_bond_slave(const struct net_device * dev)5082 static inline bool netif_is_bond_slave(const struct net_device *dev)
5083 {
5084 return dev->flags & IFF_SLAVE && dev->priv_flags & IFF_BONDING;
5085 }
5086
netif_supports_nofcs(struct net_device * dev)5087 static inline bool netif_supports_nofcs(struct net_device *dev)
5088 {
5089 return dev->priv_flags & IFF_SUPP_NOFCS;
5090 }
5091
netif_has_l3_rx_handler(const struct net_device * dev)5092 static inline bool netif_has_l3_rx_handler(const struct net_device *dev)
5093 {
5094 return dev->priv_flags & IFF_L3MDEV_RX_HANDLER;
5095 }
5096
netif_is_l3_master(const struct net_device * dev)5097 static inline bool netif_is_l3_master(const struct net_device *dev)
5098 {
5099 return dev->priv_flags & IFF_L3MDEV_MASTER;
5100 }
5101
netif_is_l3_slave(const struct net_device * dev)5102 static inline bool netif_is_l3_slave(const struct net_device *dev)
5103 {
5104 return dev->priv_flags & IFF_L3MDEV_SLAVE;
5105 }
5106
dev_sdif(const struct net_device * dev)5107 static inline int dev_sdif(const struct net_device *dev)
5108 {
5109 #ifdef CONFIG_NET_L3_MASTER_DEV
5110 if (netif_is_l3_slave(dev))
5111 return dev->ifindex;
5112 #endif
5113 return 0;
5114 }
5115
netif_is_bridge_master(const struct net_device * dev)5116 static inline bool netif_is_bridge_master(const struct net_device *dev)
5117 {
5118 return dev->priv_flags & IFF_EBRIDGE;
5119 }
5120
netif_is_bridge_port(const struct net_device * dev)5121 static inline bool netif_is_bridge_port(const struct net_device *dev)
5122 {
5123 return dev->priv_flags & IFF_BRIDGE_PORT;
5124 }
5125
netif_is_ovs_master(const struct net_device * dev)5126 static inline bool netif_is_ovs_master(const struct net_device *dev)
5127 {
5128 return dev->priv_flags & IFF_OPENVSWITCH;
5129 }
5130
netif_is_ovs_port(const struct net_device * dev)5131 static inline bool netif_is_ovs_port(const struct net_device *dev)
5132 {
5133 return dev->priv_flags & IFF_OVS_DATAPATH;
5134 }
5135
netif_is_any_bridge_master(const struct net_device * dev)5136 static inline bool netif_is_any_bridge_master(const struct net_device *dev)
5137 {
5138 return netif_is_bridge_master(dev) || netif_is_ovs_master(dev);
5139 }
5140
netif_is_any_bridge_port(const struct net_device * dev)5141 static inline bool netif_is_any_bridge_port(const struct net_device *dev)
5142 {
5143 return netif_is_bridge_port(dev) || netif_is_ovs_port(dev);
5144 }
5145
netif_is_team_master(const struct net_device * dev)5146 static inline bool netif_is_team_master(const struct net_device *dev)
5147 {
5148 return dev->priv_flags & IFF_TEAM;
5149 }
5150
netif_is_team_port(const struct net_device * dev)5151 static inline bool netif_is_team_port(const struct net_device *dev)
5152 {
5153 return dev->priv_flags & IFF_TEAM_PORT;
5154 }
5155
netif_is_lag_master(const struct net_device * dev)5156 static inline bool netif_is_lag_master(const struct net_device *dev)
5157 {
5158 return netif_is_bond_master(dev) || netif_is_team_master(dev);
5159 }
5160
netif_is_lag_port(const struct net_device * dev)5161 static inline bool netif_is_lag_port(const struct net_device *dev)
5162 {
5163 return netif_is_bond_slave(dev) || netif_is_team_port(dev);
5164 }
5165
netif_is_rxfh_configured(const struct net_device * dev)5166 static inline bool netif_is_rxfh_configured(const struct net_device *dev)
5167 {
5168 return dev->priv_flags & IFF_RXFH_CONFIGURED;
5169 }
5170
netif_is_failover(const struct net_device * dev)5171 static inline bool netif_is_failover(const struct net_device *dev)
5172 {
5173 return dev->priv_flags & IFF_FAILOVER;
5174 }
5175
netif_is_failover_slave(const struct net_device * dev)5176 static inline bool netif_is_failover_slave(const struct net_device *dev)
5177 {
5178 return dev->priv_flags & IFF_FAILOVER_SLAVE;
5179 }
5180
5181 /* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
netif_keep_dst(struct net_device * dev)5182 static inline void netif_keep_dst(struct net_device *dev)
5183 {
5184 dev->priv_flags &= ~(IFF_XMIT_DST_RELEASE | IFF_XMIT_DST_RELEASE_PERM);
5185 }
5186
5187 /* return true if dev can't cope with mtu frames that need vlan tag insertion */
netif_reduces_vlan_mtu(struct net_device * dev)5188 static inline bool netif_reduces_vlan_mtu(struct net_device *dev)
5189 {
5190 /* TODO: reserve and use an additional IFF bit, if we get more users */
5191 return netif_is_macsec(dev);
5192 }
5193
5194 extern struct pernet_operations __net_initdata loopback_net_ops;
5195
5196 /* Logging, debugging and troubleshooting/diagnostic helpers. */
5197
5198 /* netdev_printk helpers, similar to dev_printk */
5199
netdev_name(const struct net_device * dev)5200 static inline const char *netdev_name(const struct net_device *dev)
5201 {
5202 if (!dev->name[0] || strchr(dev->name, '%'))
5203 return "(unnamed net_device)";
5204 return dev->name;
5205 }
5206
netdev_reg_state(const struct net_device * dev)5207 static inline const char *netdev_reg_state(const struct net_device *dev)
5208 {
5209 switch (dev->reg_state) {
5210 case NETREG_UNINITIALIZED: return " (uninitialized)";
5211 case NETREG_REGISTERED: return "";
5212 case NETREG_UNREGISTERING: return " (unregistering)";
5213 case NETREG_UNREGISTERED: return " (unregistered)";
5214 case NETREG_RELEASED: return " (released)";
5215 case NETREG_DUMMY: return " (dummy)";
5216 }
5217
5218 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev->name, dev->reg_state);
5219 return " (unknown)";
5220 }
5221
5222 #define MODULE_ALIAS_NETDEV(device) \
5223 MODULE_ALIAS("netdev-" device)
5224
5225 /*
5226 * netdev_WARN() acts like dev_printk(), but with the key difference
5227 * of using a WARN/WARN_ON to get the message out, including the
5228 * file/line information and a backtrace.
5229 */
5230 #define netdev_WARN(dev, format, args...) \
5231 WARN(1, "netdevice: %s%s: " format, netdev_name(dev), \
5232 netdev_reg_state(dev), ##args)
5233
5234 #define netdev_WARN_ONCE(dev, format, args...) \
5235 WARN_ONCE(1, "netdevice: %s%s: " format, netdev_name(dev), \
5236 netdev_reg_state(dev), ##args)
5237
5238 /*
5239 * The list of packet types we will receive (as opposed to discard)
5240 * and the routines to invoke.
5241 *
5242 * Why 16. Because with 16 the only overlap we get on a hash of the
5243 * low nibble of the protocol value is RARP/SNAP/X.25.
5244 *
5245 * 0800 IP
5246 * 0001 802.3
5247 * 0002 AX.25
5248 * 0004 802.2
5249 * 8035 RARP
5250 * 0005 SNAP
5251 * 0805 X.25
5252 * 0806 ARP
5253 * 8137 IPX
5254 * 0009 Localtalk
5255 * 86DD IPv6
5256 */
5257 #define PTYPE_HASH_SIZE (16)
5258 #define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
5259
5260 extern struct list_head ptype_all __read_mostly;
5261 extern struct list_head ptype_base[PTYPE_HASH_SIZE] __read_mostly;
5262
5263 extern struct net_device *blackhole_netdev;
5264
5265 /* Note: Avoid these macros in fast path, prefer per-cpu or per-queue counters. */
5266 #define DEV_STATS_INC(DEV, FIELD) atomic_long_inc(&(DEV)->stats.__##FIELD)
5267 #define DEV_STATS_ADD(DEV, FIELD, VAL) \
5268 atomic_long_add((VAL), &(DEV)->stats.__##FIELD)
5269 #define DEV_STATS_READ(DEV, FIELD) atomic_long_read(&(DEV)->stats.__##FIELD)
5270
5271 #endif /* _LINUX_NETDEVICE_H */
5272