2 * INET An implementation of the TCP/IP protocol suite for the LINUX
3 * operating system. INET is implemented using the BSD Socket
4 * interface as the means of communication with the user level.
6 * Definitions for the Interfaces handler.
8 * Version: @(#)dev.h 1.0.10 08/12/93
11 * Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
12 * Corey Minyard <wf-rch!minyard@relay.EU.net>
13 * Donald J. Becker, <becker@cesdis.gsfc.nasa.gov>
14 * Alan Cox, <alan@lxorguk.ukuu.org.uk>
15 * Bjorn Ekwall. <bj0rn@blox.se>
16 * Pekka Riikonen <priikone@poseidon.pspt.fi>
18 * This program is free software; you can redistribute it and/or
19 * modify it under the terms of the GNU General Public License
20 * as published by the Free Software Foundation; either version
21 * 2 of the License, or (at your option) any later version.
23 * Moved to /usr/include/linux for NET3
25 #ifndef _LINUX_NETDEVICE_H
26 #define _LINUX_NETDEVICE_H
28 #include <linux/pm_qos.h>
29 #include <linux/timer.h>
30 #include <linux/bug.h>
31 #include <linux/delay.h>
32 #include <linux/atomic.h>
33 #include <linux/prefetch.h>
34 #include <asm/cache.h>
35 #include <asm/byteorder.h>
37 #include <linux/percpu.h>
38 #include <linux/rculist.h>
39 #include <linux/dmaengine.h>
40 #include <linux/workqueue.h>
41 #include <linux/dynamic_queue_limits.h>
43 #include <linux/ethtool.h>
44 #include <net/net_namespace.h>
47 #include <net/dcbnl.h>
49 #include <net/netprio_cgroup.h>
51 #include <linux/netdev_features.h>
52 #include <linux/neighbour.h>
53 #include <uapi/linux/netdevice.h>
54 #include <uapi/linux/if_bonding.h>
61 /* 802.15.4 specific */
64 void netdev_set_default_ethtool_ops(struct net_device
*dev
,
65 const struct ethtool_ops
*ops
);
67 /* Backlog congestion levels */
68 #define NET_RX_SUCCESS 0 /* keep 'em coming, baby */
69 #define NET_RX_DROP 1 /* packet dropped */
72 * Transmit return codes: transmit return codes originate from three different
75 * - qdisc return codes
76 * - driver transmit return codes
79 * Drivers are allowed to return any one of those in their hard_start_xmit()
80 * function. Real network devices commonly used with qdiscs should only return
81 * the driver transmit return codes though - when qdiscs are used, the actual
82 * transmission happens asynchronously, so the value is not propagated to
83 * higher layers. Virtual network devices transmit synchronously, in this case
84 * the driver transmit return codes are consumed by dev_queue_xmit(), all
85 * others are propagated to higher layers.
88 /* qdisc ->enqueue() return codes. */
89 #define NET_XMIT_SUCCESS 0x00
90 #define NET_XMIT_DROP 0x01 /* skb dropped */
91 #define NET_XMIT_CN 0x02 /* congestion notification */
92 #define NET_XMIT_POLICED 0x03 /* skb is shot by police */
93 #define NET_XMIT_MASK 0x0f /* qdisc flags in net/sch_generic.h */
95 /* NET_XMIT_CN is special. It does not guarantee that this packet is lost. It
96 * indicates that the device will soon be dropping packets, or already drops
97 * some packets of the same priority; prompting us to send less aggressively. */
98 #define net_xmit_eval(e) ((e) == NET_XMIT_CN ? 0 : (e))
99 #define net_xmit_errno(e) ((e) != NET_XMIT_CN ? -ENOBUFS : 0)
101 /* Driver transmit return codes */
102 #define NETDEV_TX_MASK 0xf0
105 __NETDEV_TX_MIN
= INT_MIN
, /* make sure enum is signed */
106 NETDEV_TX_OK
= 0x00, /* driver took care of packet */
107 NETDEV_TX_BUSY
= 0x10, /* driver tx path was busy*/
108 NETDEV_TX_LOCKED
= 0x20, /* driver tx lock was already taken */
110 typedef enum netdev_tx netdev_tx_t
;
113 * Current order: NETDEV_TX_MASK > NET_XMIT_MASK >= 0 is significant;
114 * hard_start_xmit() return < NET_XMIT_MASK means skb was consumed.
116 static inline bool dev_xmit_complete(int rc
)
119 * Positive cases with an skb consumed by a driver:
120 * - successful transmission (rc == NETDEV_TX_OK)
121 * - error while transmitting (rc < 0)
122 * - error while queueing to a different device (rc & NET_XMIT_MASK)
124 if (likely(rc
< NET_XMIT_MASK
))
131 * Compute the worst case header length according to the protocols
135 #if defined(CONFIG_WLAN) || IS_ENABLED(CONFIG_AX25)
136 # if defined(CONFIG_MAC80211_MESH)
137 # define LL_MAX_HEADER 128
139 # define LL_MAX_HEADER 96
142 # define LL_MAX_HEADER 32
145 #if !IS_ENABLED(CONFIG_NET_IPIP) && !IS_ENABLED(CONFIG_NET_IPGRE) && \
146 !IS_ENABLED(CONFIG_IPV6_SIT) && !IS_ENABLED(CONFIG_IPV6_TUNNEL)
147 #define MAX_HEADER LL_MAX_HEADER
149 #define MAX_HEADER (LL_MAX_HEADER + 48)
153 * Old network device statistics. Fields are native words
154 * (unsigned long) so they can be read and written atomically.
157 struct net_device_stats
{
158 unsigned long rx_packets
;
159 unsigned long tx_packets
;
160 unsigned long rx_bytes
;
161 unsigned long tx_bytes
;
162 unsigned long rx_errors
;
163 unsigned long tx_errors
;
164 unsigned long rx_dropped
;
165 unsigned long tx_dropped
;
166 unsigned long multicast
;
167 unsigned long collisions
;
168 unsigned long rx_length_errors
;
169 unsigned long rx_over_errors
;
170 unsigned long rx_crc_errors
;
171 unsigned long rx_frame_errors
;
172 unsigned long rx_fifo_errors
;
173 unsigned long rx_missed_errors
;
174 unsigned long tx_aborted_errors
;
175 unsigned long tx_carrier_errors
;
176 unsigned long tx_fifo_errors
;
177 unsigned long tx_heartbeat_errors
;
178 unsigned long tx_window_errors
;
179 unsigned long rx_compressed
;
180 unsigned long tx_compressed
;
184 #include <linux/cache.h>
185 #include <linux/skbuff.h>
188 #include <linux/static_key.h>
189 extern struct static_key rps_needed
;
196 struct netdev_hw_addr
{
197 struct list_head list
;
198 unsigned char addr
[MAX_ADDR_LEN
];
200 #define NETDEV_HW_ADDR_T_LAN 1
201 #define NETDEV_HW_ADDR_T_SAN 2
202 #define NETDEV_HW_ADDR_T_SLAVE 3
203 #define NETDEV_HW_ADDR_T_UNICAST 4
204 #define NETDEV_HW_ADDR_T_MULTICAST 5
209 struct rcu_head rcu_head
;
212 struct netdev_hw_addr_list
{
213 struct list_head list
;
217 #define netdev_hw_addr_list_count(l) ((l)->count)
218 #define netdev_hw_addr_list_empty(l) (netdev_hw_addr_list_count(l) == 0)
219 #define netdev_hw_addr_list_for_each(ha, l) \
220 list_for_each_entry(ha, &(l)->list, list)
222 #define netdev_uc_count(dev) netdev_hw_addr_list_count(&(dev)->uc)
223 #define netdev_uc_empty(dev) netdev_hw_addr_list_empty(&(dev)->uc)
224 #define netdev_for_each_uc_addr(ha, dev) \
225 netdev_hw_addr_list_for_each(ha, &(dev)->uc)
227 #define netdev_mc_count(dev) netdev_hw_addr_list_count(&(dev)->mc)
228 #define netdev_mc_empty(dev) netdev_hw_addr_list_empty(&(dev)->mc)
229 #define netdev_for_each_mc_addr(ha, dev) \
230 netdev_hw_addr_list_for_each(ha, &(dev)->mc)
237 /* cached hardware header; allow for machine alignment needs. */
238 #define HH_DATA_MOD 16
239 #define HH_DATA_OFF(__len) \
240 (HH_DATA_MOD - (((__len - 1) & (HH_DATA_MOD - 1)) + 1))
241 #define HH_DATA_ALIGN(__len) \
242 (((__len)+(HH_DATA_MOD-1))&~(HH_DATA_MOD - 1))
243 unsigned long hh_data
[HH_DATA_ALIGN(LL_MAX_HEADER
) / sizeof(long)];
246 /* Reserve HH_DATA_MOD byte aligned hard_header_len, but at least that much.
248 * dev->hard_header_len ? (dev->hard_header_len +
249 * (HH_DATA_MOD - 1)) & ~(HH_DATA_MOD - 1) : 0
251 * We could use other alignment values, but we must maintain the
252 * relationship HH alignment <= LL alignment.
254 #define LL_RESERVED_SPACE(dev) \
255 ((((dev)->hard_header_len+(dev)->needed_headroom)&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
256 #define LL_RESERVED_SPACE_EXTRA(dev,extra) \
257 ((((dev)->hard_header_len+(dev)->needed_headroom+(extra))&~(HH_DATA_MOD - 1)) + HH_DATA_MOD)
260 int (*create
) (struct sk_buff
*skb
, struct net_device
*dev
,
261 unsigned short type
, const void *daddr
,
262 const void *saddr
, unsigned int len
);
263 int (*parse
)(const struct sk_buff
*skb
, unsigned char *haddr
);
264 int (*cache
)(const struct neighbour
*neigh
, struct hh_cache
*hh
, __be16 type
);
265 void (*cache_update
)(struct hh_cache
*hh
,
266 const struct net_device
*dev
,
267 const unsigned char *haddr
);
270 /* These flag bits are private to the generic network queueing
271 * layer, they may not be explicitly referenced by any other
275 enum netdev_state_t
{
277 __LINK_STATE_PRESENT
,
278 __LINK_STATE_NOCARRIER
,
279 __LINK_STATE_LINKWATCH_PENDING
,
280 __LINK_STATE_DORMANT
,
285 * This structure holds at boot time configured netdevice settings. They
286 * are then used in the device probing.
288 struct netdev_boot_setup
{
292 #define NETDEV_BOOT_SETUP_MAX 8
294 int __init
netdev_boot_setup(char *str
);
297 * Structure for NAPI scheduling similar to tasklet but with weighting
300 /* The poll_list must only be managed by the entity which
301 * changes the state of the NAPI_STATE_SCHED bit. This means
302 * whoever atomically sets that bit can add this napi_struct
303 * to the per-cpu poll_list, and whoever clears that bit
304 * can remove from the list right before clearing the bit.
306 struct list_head poll_list
;
310 unsigned int gro_count
;
311 int (*poll
)(struct napi_struct
*, int);
312 #ifdef CONFIG_NETPOLL
313 spinlock_t poll_lock
;
316 struct net_device
*dev
;
317 struct sk_buff
*gro_list
;
319 struct hrtimer timer
;
320 struct list_head dev_list
;
321 struct hlist_node napi_hash_node
;
322 unsigned int napi_id
;
326 NAPI_STATE_SCHED
, /* Poll is scheduled */
327 NAPI_STATE_DISABLE
, /* Disable pending */
328 NAPI_STATE_NPSVC
, /* Netpoll - don't dequeue from poll_list */
329 NAPI_STATE_HASHED
, /* In NAPI hash */
339 typedef enum gro_result gro_result_t
;
342 * enum rx_handler_result - Possible return values for rx_handlers.
343 * @RX_HANDLER_CONSUMED: skb was consumed by rx_handler, do not process it
345 * @RX_HANDLER_ANOTHER: Do another round in receive path. This is indicated in
346 * case skb->dev was changed by rx_handler.
347 * @RX_HANDLER_EXACT: Force exact delivery, no wildcard.
348 * @RX_HANDLER_PASS: Do nothing, passe the skb as if no rx_handler was called.
350 * rx_handlers are functions called from inside __netif_receive_skb(), to do
351 * special processing of the skb, prior to delivery to protocol handlers.
353 * Currently, a net_device can only have a single rx_handler registered. Trying
354 * to register a second rx_handler will return -EBUSY.
356 * To register a rx_handler on a net_device, use netdev_rx_handler_register().
357 * To unregister a rx_handler on a net_device, use
358 * netdev_rx_handler_unregister().
360 * Upon return, rx_handler is expected to tell __netif_receive_skb() what to
363 * If the rx_handler consumed to skb in some way, it should return
364 * RX_HANDLER_CONSUMED. This is appropriate when the rx_handler arranged for
365 * the skb to be delivered in some other ways.
367 * If the rx_handler changed skb->dev, to divert the skb to another
368 * net_device, it should return RX_HANDLER_ANOTHER. The rx_handler for the
369 * new device will be called if it exists.
371 * If the rx_handler consider the skb should be ignored, it should return
372 * RX_HANDLER_EXACT. The skb will only be delivered to protocol handlers that
373 * are registered on exact device (ptype->dev == skb->dev).
375 * If the rx_handler didn't changed skb->dev, but want the skb to be normally
376 * delivered, it should return RX_HANDLER_PASS.
378 * A device without a registered rx_handler will behave as if rx_handler
379 * returned RX_HANDLER_PASS.
382 enum rx_handler_result
{
388 typedef enum rx_handler_result rx_handler_result_t
;
389 typedef rx_handler_result_t
rx_handler_func_t(struct sk_buff
**pskb
);
391 void __napi_schedule(struct napi_struct
*n
);
392 void __napi_schedule_irqoff(struct napi_struct
*n
);
394 static inline bool napi_disable_pending(struct napi_struct
*n
)
396 return test_bit(NAPI_STATE_DISABLE
, &n
->state
);
400 * napi_schedule_prep - check if napi can be scheduled
403 * Test if NAPI routine is already running, and if not mark
404 * it as running. This is used as a condition variable
405 * insure only one NAPI poll instance runs. We also make
406 * sure there is no pending NAPI disable.
408 static inline bool napi_schedule_prep(struct napi_struct
*n
)
410 return !napi_disable_pending(n
) &&
411 !test_and_set_bit(NAPI_STATE_SCHED
, &n
->state
);
415 * napi_schedule - schedule NAPI poll
418 * Schedule NAPI poll routine to be called if it is not already
421 static inline void napi_schedule(struct napi_struct
*n
)
423 if (napi_schedule_prep(n
))
428 * napi_schedule_irqoff - schedule NAPI poll
431 * Variant of napi_schedule(), assuming hard irqs are masked.
433 static inline void napi_schedule_irqoff(struct napi_struct
*n
)
435 if (napi_schedule_prep(n
))
436 __napi_schedule_irqoff(n
);
439 /* Try to reschedule poll. Called by dev->poll() after napi_complete(). */
440 static inline bool napi_reschedule(struct napi_struct
*napi
)
442 if (napi_schedule_prep(napi
)) {
443 __napi_schedule(napi
);
449 void __napi_complete(struct napi_struct
*n
);
450 void napi_complete_done(struct napi_struct
*n
, int work_done
);
452 * napi_complete - NAPI processing complete
455 * Mark NAPI processing as complete.
456 * Consider using napi_complete_done() instead.
458 static inline void napi_complete(struct napi_struct
*n
)
460 return napi_complete_done(n
, 0);
464 * napi_by_id - lookup a NAPI by napi_id
465 * @napi_id: hashed napi_id
467 * lookup @napi_id in napi_hash table
468 * must be called under rcu_read_lock()
470 struct napi_struct
*napi_by_id(unsigned int napi_id
);
473 * napi_hash_add - add a NAPI to global hashtable
474 * @napi: napi context
476 * generate a new napi_id and store a @napi under it in napi_hash
478 void napi_hash_add(struct napi_struct
*napi
);
481 * napi_hash_del - remove a NAPI from global table
482 * @napi: napi context
484 * Warning: caller must observe rcu grace period
485 * before freeing memory containing @napi
487 void napi_hash_del(struct napi_struct
*napi
);
490 * napi_disable - prevent NAPI from scheduling
493 * Stop NAPI from being scheduled on this context.
494 * Waits till any outstanding processing completes.
496 void napi_disable(struct napi_struct
*n
);
499 * napi_enable - enable NAPI scheduling
502 * Resume NAPI from being scheduled on this context.
503 * Must be paired with napi_disable.
505 static inline void napi_enable(struct napi_struct
*n
)
507 BUG_ON(!test_bit(NAPI_STATE_SCHED
, &n
->state
));
508 smp_mb__before_atomic();
509 clear_bit(NAPI_STATE_SCHED
, &n
->state
);
514 * napi_synchronize - wait until NAPI is not running
517 * Wait until NAPI is done being scheduled on this context.
518 * Waits till any outstanding processing completes but
519 * does not disable future activations.
521 static inline void napi_synchronize(const struct napi_struct
*n
)
523 while (test_bit(NAPI_STATE_SCHED
, &n
->state
))
527 # define napi_synchronize(n) barrier()
530 enum netdev_queue_state_t
{
531 __QUEUE_STATE_DRV_XOFF
,
532 __QUEUE_STATE_STACK_XOFF
,
533 __QUEUE_STATE_FROZEN
,
536 #define QUEUE_STATE_DRV_XOFF (1 << __QUEUE_STATE_DRV_XOFF)
537 #define QUEUE_STATE_STACK_XOFF (1 << __QUEUE_STATE_STACK_XOFF)
538 #define QUEUE_STATE_FROZEN (1 << __QUEUE_STATE_FROZEN)
540 #define QUEUE_STATE_ANY_XOFF (QUEUE_STATE_DRV_XOFF | QUEUE_STATE_STACK_XOFF)
541 #define QUEUE_STATE_ANY_XOFF_OR_FROZEN (QUEUE_STATE_ANY_XOFF | \
543 #define QUEUE_STATE_DRV_XOFF_OR_FROZEN (QUEUE_STATE_DRV_XOFF | \
547 * __QUEUE_STATE_DRV_XOFF is used by drivers to stop the transmit queue. The
548 * netif_tx_* functions below are used to manipulate this flag. The
549 * __QUEUE_STATE_STACK_XOFF flag is used by the stack to stop the transmit
550 * queue independently. The netif_xmit_*stopped functions below are called
551 * to check if the queue has been stopped by the driver or stack (either
552 * of the XOFF bits are set in the state). Drivers should not need to call
553 * netif_xmit*stopped functions, they should only be using netif_tx_*.
556 struct netdev_queue
{
560 struct net_device
*dev
;
561 struct Qdisc __rcu
*qdisc
;
562 struct Qdisc
*qdisc_sleeping
;
566 #if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
572 spinlock_t _xmit_lock ____cacheline_aligned_in_smp
;
575 * please use this field instead of dev->trans_start
577 unsigned long trans_start
;
580 * Number of TX timeouts for this queue
581 * (/sys/class/net/DEV/Q/trans_timeout)
583 unsigned long trans_timeout
;
590 } ____cacheline_aligned_in_smp
;
592 static inline int netdev_queue_numa_node_read(const struct netdev_queue
*q
)
594 #if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
601 static inline void netdev_queue_numa_node_write(struct netdev_queue
*q
, int node
)
603 #if defined(CONFIG_XPS) && defined(CONFIG_NUMA)
610 * This structure holds an RPS map which can be of variable length. The
611 * map is an array of CPUs.
618 #define RPS_MAP_SIZE(_num) (sizeof(struct rps_map) + ((_num) * sizeof(u16)))
621 * The rps_dev_flow structure contains the mapping of a flow to a CPU, the
622 * tail pointer for that CPU's input queue at the time of last enqueue, and
623 * a hardware filter index.
625 struct rps_dev_flow
{
628 unsigned int last_qtail
;
630 #define RPS_NO_FILTER 0xffff
633 * The rps_dev_flow_table structure contains a table of flow mappings.
635 struct rps_dev_flow_table
{
638 struct rps_dev_flow flows
[0];
640 #define RPS_DEV_FLOW_TABLE_SIZE(_num) (sizeof(struct rps_dev_flow_table) + \
641 ((_num) * sizeof(struct rps_dev_flow)))
644 * The rps_sock_flow_table contains mappings of flows to the last CPU
645 * on which they were processed by the application (set in recvmsg).
646 * Each entry is a 32bit value. Upper part is the high order bits
647 * of flow hash, lower part is cpu number.
648 * rps_cpu_mask is used to partition the space, depending on number of
649 * possible cpus : rps_cpu_mask = roundup_pow_of_two(nr_cpu_ids) - 1
650 * For example, if 64 cpus are possible, rps_cpu_mask = 0x3f,
651 * meaning we use 32-6=26 bits for the hash.
653 struct rps_sock_flow_table
{
656 u32 ents
[0] ____cacheline_aligned_in_smp
;
658 #define RPS_SOCK_FLOW_TABLE_SIZE(_num) (offsetof(struct rps_sock_flow_table, ents[_num]))
660 #define RPS_NO_CPU 0xffff
662 extern u32 rps_cpu_mask
;
663 extern struct rps_sock_flow_table __rcu
*rps_sock_flow_table
;
665 static inline void rps_record_sock_flow(struct rps_sock_flow_table
*table
,
669 unsigned int index
= hash
& table
->mask
;
670 u32 val
= hash
& ~rps_cpu_mask
;
672 /* We only give a hint, preemption can change cpu under us */
673 val
|= raw_smp_processor_id();
675 if (table
->ents
[index
] != val
)
676 table
->ents
[index
] = val
;
680 #ifdef CONFIG_RFS_ACCEL
681 bool rps_may_expire_flow(struct net_device
*dev
, u16 rxq_index
, u32 flow_id
,
684 #endif /* CONFIG_RPS */
686 /* This structure contains an instance of an RX queue. */
687 struct netdev_rx_queue
{
689 struct rps_map __rcu
*rps_map
;
690 struct rps_dev_flow_table __rcu
*rps_flow_table
;
693 struct net_device
*dev
;
694 } ____cacheline_aligned_in_smp
;
697 * RX queue sysfs structures and functions.
699 struct rx_queue_attribute
{
700 struct attribute attr
;
701 ssize_t (*show
)(struct netdev_rx_queue
*queue
,
702 struct rx_queue_attribute
*attr
, char *buf
);
703 ssize_t (*store
)(struct netdev_rx_queue
*queue
,
704 struct rx_queue_attribute
*attr
, const char *buf
, size_t len
);
709 * This structure holds an XPS map which can be of variable length. The
710 * map is an array of queues.
714 unsigned int alloc_len
;
718 #define XPS_MAP_SIZE(_num) (sizeof(struct xps_map) + ((_num) * sizeof(u16)))
719 #define XPS_MIN_MAP_ALLOC ((L1_CACHE_BYTES - sizeof(struct xps_map)) \
723 * This structure holds all XPS maps for device. Maps are indexed by CPU.
725 struct xps_dev_maps
{
727 struct xps_map __rcu
*cpu_map
[0];
729 #define XPS_DEV_MAPS_SIZE (sizeof(struct xps_dev_maps) + \
730 (nr_cpu_ids * sizeof(struct xps_map *)))
731 #endif /* CONFIG_XPS */
733 #define TC_MAX_QUEUE 16
734 #define TC_BITMASK 15
735 /* HW offloaded queuing disciplines txq count and offset maps */
736 struct netdev_tc_txq
{
741 #if defined(CONFIG_FCOE) || defined(CONFIG_FCOE_MODULE)
743 * This structure is to hold information about the device
744 * configured to run FCoE protocol stack.
746 struct netdev_fcoe_hbainfo
{
747 char manufacturer
[64];
748 char serial_number
[64];
749 char hardware_version
[64];
750 char driver_version
[64];
751 char optionrom_version
[64];
752 char firmware_version
[64];
754 char model_description
[256];
758 #define MAX_PHYS_ITEM_ID_LEN 32
760 /* This structure holds a unique identifier to identify some
761 * physical item (port for example) used by a netdevice.
763 struct netdev_phys_item_id
{
764 unsigned char id
[MAX_PHYS_ITEM_ID_LEN
];
765 unsigned char id_len
;
768 typedef u16 (*select_queue_fallback_t
)(struct net_device
*dev
,
769 struct sk_buff
*skb
);
774 * This structure defines the management hooks for network devices.
775 * The following hooks can be defined; unless noted otherwise, they are
776 * optional and can be filled with a null pointer.
778 * int (*ndo_init)(struct net_device *dev);
779 * This function is called once when network device is registered.
780 * The network device can use this to any late stage initializaton
781 * or semantic validattion. It can fail with an error code which will
782 * be propogated back to register_netdev
784 * void (*ndo_uninit)(struct net_device *dev);
785 * This function is called when device is unregistered or when registration
786 * fails. It is not called if init fails.
788 * int (*ndo_open)(struct net_device *dev);
789 * This function is called when network device transistions to the up
792 * int (*ndo_stop)(struct net_device *dev);
793 * This function is called when network device transistions to the down
796 * netdev_tx_t (*ndo_start_xmit)(struct sk_buff *skb,
797 * struct net_device *dev);
798 * Called when a packet needs to be transmitted.
799 * Must return NETDEV_TX_OK , NETDEV_TX_BUSY.
800 * (can also return NETDEV_TX_LOCKED iff NETIF_F_LLTX)
801 * Required can not be NULL.
803 * u16 (*ndo_select_queue)(struct net_device *dev, struct sk_buff *skb,
804 * void *accel_priv, select_queue_fallback_t fallback);
805 * Called to decide which queue to when device supports multiple
808 * void (*ndo_change_rx_flags)(struct net_device *dev, int flags);
809 * This function is called to allow device receiver to make
810 * changes to configuration when multicast or promiscious is enabled.
812 * void (*ndo_set_rx_mode)(struct net_device *dev);
813 * This function is called device changes address list filtering.
814 * If driver handles unicast address filtering, it should set
815 * IFF_UNICAST_FLT to its priv_flags.
817 * int (*ndo_set_mac_address)(struct net_device *dev, void *addr);
818 * This function is called when the Media Access Control address
819 * needs to be changed. If this interface is not defined, the
820 * mac address can not be changed.
822 * int (*ndo_validate_addr)(struct net_device *dev);
823 * Test if Media Access Control address is valid for the device.
825 * int (*ndo_do_ioctl)(struct net_device *dev, struct ifreq *ifr, int cmd);
826 * Called when a user request an ioctl which can't be handled by
827 * the generic interface code. If not defined ioctl's return
828 * not supported error code.
830 * int (*ndo_set_config)(struct net_device *dev, struct ifmap *map);
831 * Used to set network devices bus interface parameters. This interface
832 * is retained for legacy reason, new devices should use the bus
833 * interface (PCI) for low level management.
835 * int (*ndo_change_mtu)(struct net_device *dev, int new_mtu);
836 * Called when a user wants to change the Maximum Transfer Unit
837 * of a device. If not defined, any request to change MTU will
838 * will return an error.
840 * void (*ndo_tx_timeout)(struct net_device *dev);
841 * Callback uses when the transmitter has not made any progress
842 * for dev->watchdog ticks.
844 * struct rtnl_link_stats64* (*ndo_get_stats64)(struct net_device *dev,
845 * struct rtnl_link_stats64 *storage);
846 * struct net_device_stats* (*ndo_get_stats)(struct net_device *dev);
847 * Called when a user wants to get the network device usage
848 * statistics. Drivers must do one of the following:
849 * 1. Define @ndo_get_stats64 to fill in a zero-initialised
850 * rtnl_link_stats64 structure passed by the caller.
851 * 2. Define @ndo_get_stats to update a net_device_stats structure
852 * (which should normally be dev->stats) and return a pointer to
853 * it. The structure may be changed asynchronously only if each
854 * field is written atomically.
855 * 3. Update dev->stats asynchronously and atomically, and define
858 * int (*ndo_vlan_rx_add_vid)(struct net_device *dev, __be16 proto, u16 vid);
859 * If device support VLAN filtering this function is called when a
860 * VLAN id is registered.
862 * int (*ndo_vlan_rx_kill_vid)(struct net_device *dev, __be16 proto, u16 vid);
863 * If device support VLAN filtering this function is called when a
864 * VLAN id is unregistered.
866 * void (*ndo_poll_controller)(struct net_device *dev);
868 * SR-IOV management functions.
869 * int (*ndo_set_vf_mac)(struct net_device *dev, int vf, u8* mac);
870 * int (*ndo_set_vf_vlan)(struct net_device *dev, int vf, u16 vlan, u8 qos);
871 * int (*ndo_set_vf_rate)(struct net_device *dev, int vf, int min_tx_rate,
873 * int (*ndo_set_vf_spoofchk)(struct net_device *dev, int vf, bool setting);
874 * int (*ndo_get_vf_config)(struct net_device *dev,
875 * int vf, struct ifla_vf_info *ivf);
876 * int (*ndo_set_vf_link_state)(struct net_device *dev, int vf, int link_state);
877 * int (*ndo_set_vf_port)(struct net_device *dev, int vf,
878 * struct nlattr *port[]);
879 * int (*ndo_get_vf_port)(struct net_device *dev, int vf, struct sk_buff *skb);
880 * int (*ndo_setup_tc)(struct net_device *dev, u8 tc)
881 * Called to setup 'tc' number of traffic classes in the net device. This
882 * is always called from the stack with the rtnl lock held and netif tx
883 * queues stopped. This allows the netdevice to perform queue management
886 * Fiber Channel over Ethernet (FCoE) offload functions.
887 * int (*ndo_fcoe_enable)(struct net_device *dev);
888 * Called when the FCoE protocol stack wants to start using LLD for FCoE
889 * so the underlying device can perform whatever needed configuration or
890 * initialization to support acceleration of FCoE traffic.
892 * int (*ndo_fcoe_disable)(struct net_device *dev);
893 * Called when the FCoE protocol stack wants to stop using LLD for FCoE
894 * so the underlying device can perform whatever needed clean-ups to
895 * stop supporting acceleration of FCoE traffic.
897 * int (*ndo_fcoe_ddp_setup)(struct net_device *dev, u16 xid,
898 * struct scatterlist *sgl, unsigned int sgc);
899 * Called when the FCoE Initiator wants to initialize an I/O that
900 * is a possible candidate for Direct Data Placement (DDP). The LLD can
901 * perform necessary setup and returns 1 to indicate the device is set up
902 * successfully to perform DDP on this I/O, otherwise this returns 0.
904 * int (*ndo_fcoe_ddp_done)(struct net_device *dev, u16 xid);
905 * Called when the FCoE Initiator/Target is done with the DDPed I/O as
906 * indicated by the FC exchange id 'xid', so the underlying device can
907 * clean up and reuse resources for later DDP requests.
909 * int (*ndo_fcoe_ddp_target)(struct net_device *dev, u16 xid,
910 * struct scatterlist *sgl, unsigned int sgc);
911 * Called when the FCoE Target wants to initialize an I/O that
912 * is a possible candidate for Direct Data Placement (DDP). The LLD can
913 * perform necessary setup and returns 1 to indicate the device is set up
914 * successfully to perform DDP on this I/O, otherwise this returns 0.
916 * int (*ndo_fcoe_get_hbainfo)(struct net_device *dev,
917 * struct netdev_fcoe_hbainfo *hbainfo);
918 * Called when the FCoE Protocol stack wants information on the underlying
919 * device. This information is utilized by the FCoE protocol stack to
920 * register attributes with Fiber Channel management service as per the
921 * FC-GS Fabric Device Management Information(FDMI) specification.
923 * int (*ndo_fcoe_get_wwn)(struct net_device *dev, u64 *wwn, int type);
924 * Called when the underlying device wants to override default World Wide
925 * Name (WWN) generation mechanism in FCoE protocol stack to pass its own
926 * World Wide Port Name (WWPN) or World Wide Node Name (WWNN) to the FCoE
927 * protocol stack to use.
930 * int (*ndo_rx_flow_steer)(struct net_device *dev, const struct sk_buff *skb,
931 * u16 rxq_index, u32 flow_id);
932 * Set hardware filter for RFS. rxq_index is the target queue index;
933 * flow_id is a flow ID to be passed to rps_may_expire_flow() later.
934 * Return the filter ID on success, or a negative error code.
936 * Slave management functions (for bridge, bonding, etc).
937 * int (*ndo_add_slave)(struct net_device *dev, struct net_device *slave_dev);
938 * Called to make another netdev an underling.
940 * int (*ndo_del_slave)(struct net_device *dev, struct net_device *slave_dev);
941 * Called to release previously enslaved netdev.
943 * Feature/offload setting functions.
944 * netdev_features_t (*ndo_fix_features)(struct net_device *dev,
945 * netdev_features_t features);
946 * Adjusts the requested feature flags according to device-specific
947 * constraints, and returns the resulting flags. Must not modify
950 * int (*ndo_set_features)(struct net_device *dev, netdev_features_t features);
951 * Called to update device configuration to new features. Passed
952 * feature set might be less than what was returned by ndo_fix_features()).
953 * Must return >0 or -errno if it changed dev->features itself.
955 * int (*ndo_fdb_add)(struct ndmsg *ndm, struct nlattr *tb[],
956 * struct net_device *dev,
957 * const unsigned char *addr, u16 vid, u16 flags)
958 * Adds an FDB entry to dev for addr.
959 * int (*ndo_fdb_del)(struct ndmsg *ndm, struct nlattr *tb[],
960 * struct net_device *dev,
961 * const unsigned char *addr, u16 vid)
962 * Deletes the FDB entry from dev coresponding to addr.
963 * int (*ndo_fdb_dump)(struct sk_buff *skb, struct netlink_callback *cb,
964 * struct net_device *dev, struct net_device *filter_dev,
966 * Used to add FDB entries to dump requests. Implementers should add
967 * entries to skb and update idx with the number of entries.
969 * int (*ndo_bridge_setlink)(struct net_device *dev, struct nlmsghdr *nlh)
970 * int (*ndo_bridge_getlink)(struct sk_buff *skb, u32 pid, u32 seq,
971 * struct net_device *dev, u32 filter_mask)
973 * int (*ndo_change_carrier)(struct net_device *dev, bool new_carrier);
974 * Called to change device carrier. Soft-devices (like dummy, team, etc)
975 * which do not represent real hardware may define this to allow their
976 * userspace components to manage their virtual carrier state. Devices
977 * that determine carrier state from physical hardware properties (eg
978 * network cables) or protocol-dependent mechanisms (eg
979 * USB_CDC_NOTIFY_NETWORK_CONNECTION) should NOT implement this function.
981 * int (*ndo_get_phys_port_id)(struct net_device *dev,
982 * struct netdev_phys_item_id *ppid);
983 * Called to get ID of physical port of this device. If driver does
984 * not implement this, it is assumed that the hw is not able to have
985 * multiple net devices on single physical port.
987 * void (*ndo_add_vxlan_port)(struct net_device *dev,
988 * sa_family_t sa_family, __be16 port);
989 * Called by vxlan to notiy a driver about the UDP port and socket
990 * address family that vxlan is listnening to. It is called only when
991 * a new port starts listening. The operation is protected by the
992 * vxlan_net->sock_lock.
994 * void (*ndo_del_vxlan_port)(struct net_device *dev,
995 * sa_family_t sa_family, __be16 port);
996 * Called by vxlan to notify the driver about a UDP port and socket
997 * address family that vxlan is not listening to anymore. The operation
998 * is protected by the vxlan_net->sock_lock.
1000 * void* (*ndo_dfwd_add_station)(struct net_device *pdev,
1001 * struct net_device *dev)
1002 * Called by upper layer devices to accelerate switching or other
1003 * station functionality into hardware. 'pdev is the lowerdev
1004 * to use for the offload and 'dev' is the net device that will
1005 * back the offload. Returns a pointer to the private structure
1006 * the upper layer will maintain.
1007 * void (*ndo_dfwd_del_station)(struct net_device *pdev, void *priv)
1008 * Called by upper layer device to delete the station created
1009 * by 'ndo_dfwd_add_station'. 'pdev' is the net device backing
1010 * the station and priv is the structure returned by the add
1012 * netdev_tx_t (*ndo_dfwd_start_xmit)(struct sk_buff *skb,
1013 * struct net_device *dev,
1015 * Callback to use for xmit over the accelerated station. This
1016 * is used in place of ndo_start_xmit on accelerated net
1018 * netdev_features_t (*ndo_features_check) (struct sk_buff *skb,
1019 * struct net_device *dev
1020 * netdev_features_t features);
1021 * Called by core transmit path to determine if device is capable of
1022 * performing offload operations on a given packet. This is to give
1023 * the device an opportunity to implement any restrictions that cannot
1024 * be otherwise expressed by feature flags. The check is called with
1025 * the set of features that the stack has calculated and it returns
1026 * those the driver believes to be appropriate.
1028 * int (*ndo_switch_parent_id_get)(struct net_device *dev,
1029 * struct netdev_phys_item_id *psid);
1030 * Called to get an ID of the switch chip this port is part of.
1031 * If driver implements this, it indicates that it represents a port
1033 * int (*ndo_switch_port_stp_update)(struct net_device *dev, u8 state);
1034 * Called to notify switch device port of bridge port STP
1036 * int (*ndo_sw_parent_fib_ipv4_add)(struct net_device *dev, __be32 dst,
1037 * int dst_len, struct fib_info *fi,
1038 * u8 tos, u8 type, u32 tb_id);
1039 * Called to add/modify IPv4 route to switch device.
1040 * int (*ndo_sw_parent_fib_ipv4_del)(struct net_device *dev, __be32 dst,
1041 * int dst_len, struct fib_info *fi,
1042 * u8 tos, u8 type, u32 tb_id);
1043 * Called to delete IPv4 route from switch device.
1045 struct net_device_ops
{
1046 int (*ndo_init
)(struct net_device
*dev
);
1047 void (*ndo_uninit
)(struct net_device
*dev
);
1048 int (*ndo_open
)(struct net_device
*dev
);
1049 int (*ndo_stop
)(struct net_device
*dev
);
1050 netdev_tx_t (*ndo_start_xmit
) (struct sk_buff
*skb
,
1051 struct net_device
*dev
);
1052 u16 (*ndo_select_queue
)(struct net_device
*dev
,
1053 struct sk_buff
*skb
,
1055 select_queue_fallback_t fallback
);
1056 void (*ndo_change_rx_flags
)(struct net_device
*dev
,
1058 void (*ndo_set_rx_mode
)(struct net_device
*dev
);
1059 int (*ndo_set_mac_address
)(struct net_device
*dev
,
1061 int (*ndo_validate_addr
)(struct net_device
*dev
);
1062 int (*ndo_do_ioctl
)(struct net_device
*dev
,
1063 struct ifreq
*ifr
, int cmd
);
1064 int (*ndo_set_config
)(struct net_device
*dev
,
1066 int (*ndo_change_mtu
)(struct net_device
*dev
,
1068 int (*ndo_neigh_setup
)(struct net_device
*dev
,
1069 struct neigh_parms
*);
1070 void (*ndo_tx_timeout
) (struct net_device
*dev
);
1072 struct rtnl_link_stats64
* (*ndo_get_stats64
)(struct net_device
*dev
,
1073 struct rtnl_link_stats64
*storage
);
1074 struct net_device_stats
* (*ndo_get_stats
)(struct net_device
*dev
);
1076 int (*ndo_vlan_rx_add_vid
)(struct net_device
*dev
,
1077 __be16 proto
, u16 vid
);
1078 int (*ndo_vlan_rx_kill_vid
)(struct net_device
*dev
,
1079 __be16 proto
, u16 vid
);
1080 #ifdef CONFIG_NET_POLL_CONTROLLER
1081 void (*ndo_poll_controller
)(struct net_device
*dev
);
1082 int (*ndo_netpoll_setup
)(struct net_device
*dev
,
1083 struct netpoll_info
*info
);
1084 void (*ndo_netpoll_cleanup
)(struct net_device
*dev
);
1086 #ifdef CONFIG_NET_RX_BUSY_POLL
1087 int (*ndo_busy_poll
)(struct napi_struct
*dev
);
1089 int (*ndo_set_vf_mac
)(struct net_device
*dev
,
1090 int queue
, u8
*mac
);
1091 int (*ndo_set_vf_vlan
)(struct net_device
*dev
,
1092 int queue
, u16 vlan
, u8 qos
);
1093 int (*ndo_set_vf_rate
)(struct net_device
*dev
,
1094 int vf
, int min_tx_rate
,
1096 int (*ndo_set_vf_spoofchk
)(struct net_device
*dev
,
1097 int vf
, bool setting
);
1098 int (*ndo_get_vf_config
)(struct net_device
*dev
,
1100 struct ifla_vf_info
*ivf
);
1101 int (*ndo_set_vf_link_state
)(struct net_device
*dev
,
1102 int vf
, int link_state
);
1103 int (*ndo_set_vf_port
)(struct net_device
*dev
,
1105 struct nlattr
*port
[]);
1106 int (*ndo_get_vf_port
)(struct net_device
*dev
,
1107 int vf
, struct sk_buff
*skb
);
1108 int (*ndo_setup_tc
)(struct net_device
*dev
, u8 tc
);
1109 #if IS_ENABLED(CONFIG_FCOE)
1110 int (*ndo_fcoe_enable
)(struct net_device
*dev
);
1111 int (*ndo_fcoe_disable
)(struct net_device
*dev
);
1112 int (*ndo_fcoe_ddp_setup
)(struct net_device
*dev
,
1114 struct scatterlist
*sgl
,
1116 int (*ndo_fcoe_ddp_done
)(struct net_device
*dev
,
1118 int (*ndo_fcoe_ddp_target
)(struct net_device
*dev
,
1120 struct scatterlist
*sgl
,
1122 int (*ndo_fcoe_get_hbainfo
)(struct net_device
*dev
,
1123 struct netdev_fcoe_hbainfo
*hbainfo
);
1126 #if IS_ENABLED(CONFIG_LIBFCOE)
1127 #define NETDEV_FCOE_WWNN 0
1128 #define NETDEV_FCOE_WWPN 1
1129 int (*ndo_fcoe_get_wwn
)(struct net_device
*dev
,
1130 u64
*wwn
, int type
);
1133 #ifdef CONFIG_RFS_ACCEL
1134 int (*ndo_rx_flow_steer
)(struct net_device
*dev
,
1135 const struct sk_buff
*skb
,
1139 int (*ndo_add_slave
)(struct net_device
*dev
,
1140 struct net_device
*slave_dev
);
1141 int (*ndo_del_slave
)(struct net_device
*dev
,
1142 struct net_device
*slave_dev
);
1143 netdev_features_t (*ndo_fix_features
)(struct net_device
*dev
,
1144 netdev_features_t features
);
1145 int (*ndo_set_features
)(struct net_device
*dev
,
1146 netdev_features_t features
);
1147 int (*ndo_neigh_construct
)(struct neighbour
*n
);
1148 void (*ndo_neigh_destroy
)(struct neighbour
*n
);
1150 int (*ndo_fdb_add
)(struct ndmsg
*ndm
,
1151 struct nlattr
*tb
[],
1152 struct net_device
*dev
,
1153 const unsigned char *addr
,
1156 int (*ndo_fdb_del
)(struct ndmsg
*ndm
,
1157 struct nlattr
*tb
[],
1158 struct net_device
*dev
,
1159 const unsigned char *addr
,
1161 int (*ndo_fdb_dump
)(struct sk_buff
*skb
,
1162 struct netlink_callback
*cb
,
1163 struct net_device
*dev
,
1164 struct net_device
*filter_dev
,
1167 int (*ndo_bridge_setlink
)(struct net_device
*dev
,
1168 struct nlmsghdr
*nlh
,
1170 int (*ndo_bridge_getlink
)(struct sk_buff
*skb
,
1172 struct net_device
*dev
,
1174 int (*ndo_bridge_dellink
)(struct net_device
*dev
,
1175 struct nlmsghdr
*nlh
,
1177 int (*ndo_change_carrier
)(struct net_device
*dev
,
1179 int (*ndo_get_phys_port_id
)(struct net_device
*dev
,
1180 struct netdev_phys_item_id
*ppid
);
1181 void (*ndo_add_vxlan_port
)(struct net_device
*dev
,
1182 sa_family_t sa_family
,
1184 void (*ndo_del_vxlan_port
)(struct net_device
*dev
,
1185 sa_family_t sa_family
,
1188 void* (*ndo_dfwd_add_station
)(struct net_device
*pdev
,
1189 struct net_device
*dev
);
1190 void (*ndo_dfwd_del_station
)(struct net_device
*pdev
,
1193 netdev_tx_t (*ndo_dfwd_start_xmit
) (struct sk_buff
*skb
,
1194 struct net_device
*dev
,
1196 int (*ndo_get_lock_subclass
)(struct net_device
*dev
);
1197 netdev_features_t (*ndo_features_check
) (struct sk_buff
*skb
,
1198 struct net_device
*dev
,
1199 netdev_features_t features
);
1200 #ifdef CONFIG_NET_SWITCHDEV
1201 int (*ndo_switch_parent_id_get
)(struct net_device
*dev
,
1202 struct netdev_phys_item_id
*psid
);
1203 int (*ndo_switch_port_stp_update
)(struct net_device
*dev
,
1205 int (*ndo_switch_fib_ipv4_add
)(struct net_device
*dev
,
1208 struct fib_info
*fi
,
1211 int (*ndo_switch_fib_ipv4_del
)(struct net_device
*dev
,
1214 struct fib_info
*fi
,
1221 * enum net_device_priv_flags - &struct net_device priv_flags
1223 * These are the &struct net_device, they are only set internally
1224 * by drivers and used in the kernel. These flags are invisible to
1225 * userspace, this means that the order of these flags can change
1226 * during any kernel release.
1228 * You should have a pretty good reason to be extending these flags.
1230 * @IFF_802_1Q_VLAN: 802.1Q VLAN device
1231 * @IFF_EBRIDGE: Ethernet bridging device
1232 * @IFF_SLAVE_INACTIVE: bonding slave not the curr. active
1233 * @IFF_MASTER_8023AD: bonding master, 802.3ad
1234 * @IFF_MASTER_ALB: bonding master, balance-alb
1235 * @IFF_BONDING: bonding master or slave
1236 * @IFF_SLAVE_NEEDARP: need ARPs for validation
1237 * @IFF_ISATAP: ISATAP interface (RFC4214)
1238 * @IFF_MASTER_ARPMON: bonding master, ARP mon in use
1239 * @IFF_WAN_HDLC: WAN HDLC device
1240 * @IFF_XMIT_DST_RELEASE: dev_hard_start_xmit() is allowed to
1242 * @IFF_DONT_BRIDGE: disallow bridging this ether dev
1243 * @IFF_DISABLE_NETPOLL: disable netpoll at run-time
1244 * @IFF_MACVLAN_PORT: device used as macvlan port
1245 * @IFF_BRIDGE_PORT: device used as bridge port
1246 * @IFF_OVS_DATAPATH: device used as Open vSwitch datapath port
1247 * @IFF_TX_SKB_SHARING: The interface supports sharing skbs on transmit
1248 * @IFF_UNICAST_FLT: Supports unicast filtering
1249 * @IFF_TEAM_PORT: device used as team port
1250 * @IFF_SUPP_NOFCS: device supports sending custom FCS
1251 * @IFF_LIVE_ADDR_CHANGE: device supports hardware address
1252 * change when it's running
1253 * @IFF_MACVLAN: Macvlan device
1255 enum netdev_priv_flags
{
1256 IFF_802_1Q_VLAN
= 1<<0,
1258 IFF_SLAVE_INACTIVE
= 1<<2,
1259 IFF_MASTER_8023AD
= 1<<3,
1260 IFF_MASTER_ALB
= 1<<4,
1262 IFF_SLAVE_NEEDARP
= 1<<6,
1264 IFF_MASTER_ARPMON
= 1<<8,
1265 IFF_WAN_HDLC
= 1<<9,
1266 IFF_XMIT_DST_RELEASE
= 1<<10,
1267 IFF_DONT_BRIDGE
= 1<<11,
1268 IFF_DISABLE_NETPOLL
= 1<<12,
1269 IFF_MACVLAN_PORT
= 1<<13,
1270 IFF_BRIDGE_PORT
= 1<<14,
1271 IFF_OVS_DATAPATH
= 1<<15,
1272 IFF_TX_SKB_SHARING
= 1<<16,
1273 IFF_UNICAST_FLT
= 1<<17,
1274 IFF_TEAM_PORT
= 1<<18,
1275 IFF_SUPP_NOFCS
= 1<<19,
1276 IFF_LIVE_ADDR_CHANGE
= 1<<20,
1277 IFF_MACVLAN
= 1<<21,
1278 IFF_XMIT_DST_RELEASE_PERM
= 1<<22,
1279 IFF_IPVLAN_MASTER
= 1<<23,
1280 IFF_IPVLAN_SLAVE
= 1<<24,
1283 #define IFF_802_1Q_VLAN IFF_802_1Q_VLAN
1284 #define IFF_EBRIDGE IFF_EBRIDGE
1285 #define IFF_SLAVE_INACTIVE IFF_SLAVE_INACTIVE
1286 #define IFF_MASTER_8023AD IFF_MASTER_8023AD
1287 #define IFF_MASTER_ALB IFF_MASTER_ALB
1288 #define IFF_BONDING IFF_BONDING
1289 #define IFF_SLAVE_NEEDARP IFF_SLAVE_NEEDARP
1290 #define IFF_ISATAP IFF_ISATAP
1291 #define IFF_MASTER_ARPMON IFF_MASTER_ARPMON
1292 #define IFF_WAN_HDLC IFF_WAN_HDLC
1293 #define IFF_XMIT_DST_RELEASE IFF_XMIT_DST_RELEASE
1294 #define IFF_DONT_BRIDGE IFF_DONT_BRIDGE
1295 #define IFF_DISABLE_NETPOLL IFF_DISABLE_NETPOLL
1296 #define IFF_MACVLAN_PORT IFF_MACVLAN_PORT
1297 #define IFF_BRIDGE_PORT IFF_BRIDGE_PORT
1298 #define IFF_OVS_DATAPATH IFF_OVS_DATAPATH
1299 #define IFF_TX_SKB_SHARING IFF_TX_SKB_SHARING
1300 #define IFF_UNICAST_FLT IFF_UNICAST_FLT
1301 #define IFF_TEAM_PORT IFF_TEAM_PORT
1302 #define IFF_SUPP_NOFCS IFF_SUPP_NOFCS
1303 #define IFF_LIVE_ADDR_CHANGE IFF_LIVE_ADDR_CHANGE
1304 #define IFF_MACVLAN IFF_MACVLAN
1305 #define IFF_XMIT_DST_RELEASE_PERM IFF_XMIT_DST_RELEASE_PERM
1306 #define IFF_IPVLAN_MASTER IFF_IPVLAN_MASTER
1307 #define IFF_IPVLAN_SLAVE IFF_IPVLAN_SLAVE
1310 * struct net_device - The DEVICE structure.
1311 * Actually, this whole structure is a big mistake. It mixes I/O
1312 * data with strictly "high-level" data, and it has to know about
1313 * almost every data structure used in the INET module.
1315 * @name: This is the first field of the "visible" part of this structure
1316 * (i.e. as seen by users in the "Space.c" file). It is the name
1319 * @name_hlist: Device name hash chain, please keep it close to name[]
1320 * @ifalias: SNMP alias
1321 * @mem_end: Shared memory end
1322 * @mem_start: Shared memory start
1323 * @base_addr: Device I/O address
1324 * @irq: Device IRQ number
1326 * @state: Generic network queuing layer state, see netdev_state_t
1327 * @dev_list: The global list of network devices
1328 * @napi_list: List entry, that is used for polling napi devices
1329 * @unreg_list: List entry, that is used, when we are unregistering the
1330 * device, see the function unregister_netdev
1331 * @close_list: List entry, that is used, when we are closing the device
1333 * @adj_list: Directly linked devices, like slaves for bonding
1334 * @all_adj_list: All linked devices, *including* neighbours
1335 * @features: Currently active device features
1336 * @hw_features: User-changeable features
1338 * @wanted_features: User-requested features
1339 * @vlan_features: Mask of features inheritable by VLAN devices
1341 * @hw_enc_features: Mask of features inherited by encapsulating devices
1342 * This field indicates what encapsulation
1343 * offloads the hardware is capable of doing,
1344 * and drivers will need to set them appropriately.
1346 * @mpls_features: Mask of features inheritable by MPLS
1348 * @ifindex: interface index
1349 * @iflink: unique device identifier
1351 * @stats: Statistics struct, which was left as a legacy, use
1352 * rtnl_link_stats64 instead
1354 * @rx_dropped: Dropped packets by core network,
1355 * do not use this in drivers
1356 * @tx_dropped: Dropped packets by core network,
1357 * do not use this in drivers
1359 * @carrier_changes: Stats to monitor carrier on<->off transitions
1361 * @wireless_handlers: List of functions to handle Wireless Extensions,
1363 * see <net/iw_handler.h> for details.
1364 * @wireless_data: Instance data managed by the core of wireless extensions
1366 * @netdev_ops: Includes several pointers to callbacks,
1367 * if one wants to override the ndo_*() functions
1368 * @ethtool_ops: Management operations
1369 * @fwd_ops: Management operations
1370 * @header_ops: Includes callbacks for creating,parsing,caching,etc
1371 * of Layer 2 headers.
1373 * @flags: Interface flags (a la BSD)
1374 * @priv_flags: Like 'flags' but invisible to userspace,
1375 * see if.h for the definitions
1376 * @gflags: Global flags ( kept as legacy )
1377 * @padded: How much padding added by alloc_netdev()
1378 * @operstate: RFC2863 operstate
1379 * @link_mode: Mapping policy to operstate
1380 * @if_port: Selectable AUI, TP, ...
1382 * @mtu: Interface MTU value
1383 * @type: Interface hardware type
1384 * @hard_header_len: Hardware header length
1386 * @needed_headroom: Extra headroom the hardware may need, but not in all
1387 * cases can this be guaranteed
1388 * @needed_tailroom: Extra tailroom the hardware may need, but not in all
1389 * cases can this be guaranteed. Some cases also use
1390 * LL_MAX_HEADER instead to allocate the skb
1392 * interface address info:
1394 * @perm_addr: Permanent hw address
1395 * @addr_assign_type: Hw address assignment type
1396 * @addr_len: Hardware address length
1397 * @neigh_priv_len; Used in neigh_alloc(),
1398 * initialized only in atm/clip.c
1399 * @dev_id: Used to differentiate devices that share
1400 * the same link layer address
1401 * @dev_port: Used to differentiate devices that share
1403 * @addr_list_lock: XXX: need comments on this one
1404 * @uc: unicast mac addresses
1405 * @mc: multicast mac addresses
1406 * @dev_addrs: list of device hw addresses
1407 * @queues_kset: Group of all Kobjects in the Tx and RX queues
1408 * @uc_promisc: Counter, that indicates, that promiscuous mode
1409 * has been enabled due to the need to listen to
1410 * additional unicast addresses in a device that
1411 * does not implement ndo_set_rx_mode()
1412 * @promiscuity: Number of times, the NIC is told to work in
1413 * Promiscuous mode, if it becomes 0 the NIC will
1414 * exit from working in Promiscuous mode
1415 * @allmulti: Counter, enables or disables allmulticast mode
1417 * @vlan_info: VLAN info
1418 * @dsa_ptr: dsa specific data
1419 * @tipc_ptr: TIPC specific data
1420 * @atalk_ptr: AppleTalk link
1421 * @ip_ptr: IPv4 specific data
1422 * @dn_ptr: DECnet specific data
1423 * @ip6_ptr: IPv6 specific data
1424 * @ax25_ptr: AX.25 specific data
1425 * @ieee80211_ptr: IEEE 802.11 specific data, assign before registering
1427 * @last_rx: Time of last Rx
1428 * @dev_addr: Hw address (before bcast,
1429 * because most packets are unicast)
1431 * @_rx: Array of RX queues
1432 * @num_rx_queues: Number of RX queues
1433 * allocated at register_netdev() time
1434 * @real_num_rx_queues: Number of RX queues currently active in device
1436 * @rx_handler: handler for received packets
1437 * @rx_handler_data: XXX: need comments on this one
1438 * @ingress_queue: XXX: need comments on this one
1439 * @broadcast: hw bcast address
1441 * @_tx: Array of TX queues
1442 * @num_tx_queues: Number of TX queues allocated at alloc_netdev_mq() time
1443 * @real_num_tx_queues: Number of TX queues currently active in device
1444 * @qdisc: Root qdisc from userspace point of view
1445 * @tx_queue_len: Max frames per queue allowed
1446 * @tx_global_lock: XXX: need comments on this one
1448 * @xps_maps: XXX: need comments on this one
1450 * @rx_cpu_rmap: CPU reverse-mapping for RX completion interrupts,
1451 * indexed by RX queue number. Assigned by driver.
1452 * This must only be set if the ndo_rx_flow_steer
1453 * operation is defined
1455 * @trans_start: Time (in jiffies) of last Tx
1456 * @watchdog_timeo: Represents the timeout that is used by
1457 * the watchdog ( see dev_watchdog() )
1458 * @watchdog_timer: List of timers
1460 * @pcpu_refcnt: Number of references to this device
1461 * @todo_list: Delayed register/unregister
1462 * @index_hlist: Device index hash chain
1463 * @link_watch_list: XXX: need comments on this one
1465 * @reg_state: Register/unregister state machine
1466 * @dismantle: Device is going to be freed
1467 * @rtnl_link_state: This enum represents the phases of creating
1470 * @destructor: Called from unregister,
1471 * can be used to call free_netdev
1472 * @npinfo: XXX: need comments on this one
1473 * @nd_net: Network namespace this network device is inside
1475 * @ml_priv: Mid-layer private
1476 * @lstats: Loopback statistics
1477 * @tstats: Tunnel statistics
1478 * @dstats: Dummy statistics
1479 * @vstats: Virtual ethernet statistics
1484 * @dev: Class/net/name entry
1485 * @sysfs_groups: Space for optional device, statistics and wireless
1488 * @sysfs_rx_queue_group: Space for optional per-rx queue attributes
1489 * @rtnl_link_ops: Rtnl_link_ops
1491 * @gso_max_size: Maximum size of generic segmentation offload
1492 * @gso_max_segs: Maximum number of segments that can be passed to the
1494 * @gso_min_segs: Minimum number of segments that can be passed to the
1497 * @dcbnl_ops: Data Center Bridging netlink ops
1498 * @num_tc: Number of traffic classes in the net device
1499 * @tc_to_txq: XXX: need comments on this one
1500 * @prio_tc_map XXX: need comments on this one
1502 * @fcoe_ddp_xid: Max exchange id for FCoE LRO by ddp
1504 * @priomap: XXX: need comments on this one
1505 * @phydev: Physical device may attach itself
1506 * for hardware timestamping
1508 * @qdisc_tx_busylock: XXX: need comments on this one
1510 * @group: The group, that the device belongs to
1511 * @pm_qos_req: Power Management QoS object
1513 * FIXME: cleanup struct net_device such that network protocol info
1518 char name
[IFNAMSIZ
];
1519 struct hlist_node name_hlist
;
1522 * I/O specific fields
1523 * FIXME: Merge these and struct ifmap into one
1525 unsigned long mem_end
;
1526 unsigned long mem_start
;
1527 unsigned long base_addr
;
1531 * Some hardware also needs these fields (state,dev_list,
1532 * napi_list,unreg_list,close_list) but they are not
1533 * part of the usual set specified in Space.c.
1536 unsigned long state
;
1538 struct list_head dev_list
;
1539 struct list_head napi_list
;
1540 struct list_head unreg_list
;
1541 struct list_head close_list
;
1542 struct list_head ptype_all
;
1543 struct list_head ptype_specific
;
1546 struct list_head upper
;
1547 struct list_head lower
;
1551 struct list_head upper
;
1552 struct list_head lower
;
1555 netdev_features_t features
;
1556 netdev_features_t hw_features
;
1557 netdev_features_t wanted_features
;
1558 netdev_features_t vlan_features
;
1559 netdev_features_t hw_enc_features
;
1560 netdev_features_t mpls_features
;
1565 struct net_device_stats stats
;
1567 atomic_long_t rx_dropped
;
1568 atomic_long_t tx_dropped
;
1570 atomic_t carrier_changes
;
1572 #ifdef CONFIG_WIRELESS_EXT
1573 const struct iw_handler_def
* wireless_handlers
;
1574 struct iw_public_data
* wireless_data
;
1576 const struct net_device_ops
*netdev_ops
;
1577 const struct ethtool_ops
*ethtool_ops
;
1578 const struct forwarding_accel_ops
*fwd_ops
;
1580 const struct header_ops
*header_ops
;
1583 unsigned int priv_flags
;
1585 unsigned short gflags
;
1586 unsigned short padded
;
1588 unsigned char operstate
;
1589 unsigned char link_mode
;
1591 unsigned char if_port
;
1595 unsigned short type
;
1596 unsigned short hard_header_len
;
1598 unsigned short needed_headroom
;
1599 unsigned short needed_tailroom
;
1601 /* Interface address info. */
1602 unsigned char perm_addr
[MAX_ADDR_LEN
];
1603 unsigned char addr_assign_type
;
1604 unsigned char addr_len
;
1605 unsigned short neigh_priv_len
;
1606 unsigned short dev_id
;
1607 unsigned short dev_port
;
1608 spinlock_t addr_list_lock
;
1609 struct netdev_hw_addr_list uc
;
1610 struct netdev_hw_addr_list mc
;
1611 struct netdev_hw_addr_list dev_addrs
;
1614 struct kset
*queues_kset
;
1617 unsigned char name_assign_type
;
1620 unsigned int promiscuity
;
1621 unsigned int allmulti
;
1624 /* Protocol specific pointers */
1626 #if IS_ENABLED(CONFIG_VLAN_8021Q)
1627 struct vlan_info __rcu
*vlan_info
;
1629 #if IS_ENABLED(CONFIG_NET_DSA)
1630 struct dsa_switch_tree
*dsa_ptr
;
1632 #if IS_ENABLED(CONFIG_TIPC)
1633 struct tipc_bearer __rcu
*tipc_ptr
;
1636 struct in_device __rcu
*ip_ptr
;
1637 struct dn_dev __rcu
*dn_ptr
;
1638 struct inet6_dev __rcu
*ip6_ptr
;
1640 struct wireless_dev
*ieee80211_ptr
;
1641 struct wpan_dev
*ieee802154_ptr
;
1644 * Cache lines mostly used on receive path (including eth_type_trans())
1646 unsigned long last_rx
;
1648 /* Interface address info used in eth_type_trans() */
1649 unsigned char *dev_addr
;
1653 struct netdev_rx_queue
*_rx
;
1655 unsigned int num_rx_queues
;
1656 unsigned int real_num_rx_queues
;
1660 unsigned long gro_flush_timeout
;
1661 rx_handler_func_t __rcu
*rx_handler
;
1662 void __rcu
*rx_handler_data
;
1664 struct netdev_queue __rcu
*ingress_queue
;
1665 unsigned char broadcast
[MAX_ADDR_LEN
];
1669 * Cache lines mostly used on transmit path
1671 struct netdev_queue
*_tx ____cacheline_aligned_in_smp
;
1672 unsigned int num_tx_queues
;
1673 unsigned int real_num_tx_queues
;
1674 struct Qdisc
*qdisc
;
1675 unsigned long tx_queue_len
;
1676 spinlock_t tx_global_lock
;
1679 struct xps_dev_maps __rcu
*xps_maps
;
1681 #ifdef CONFIG_RFS_ACCEL
1682 struct cpu_rmap
*rx_cpu_rmap
;
1685 /* These may be needed for future network-power-down code. */
1688 * trans_start here is expensive for high speed devices on SMP,
1689 * please use netdev_queue->trans_start instead.
1691 unsigned long trans_start
;
1694 struct timer_list watchdog_timer
;
1696 int __percpu
*pcpu_refcnt
;
1697 struct list_head todo_list
;
1699 struct hlist_node index_hlist
;
1700 struct list_head link_watch_list
;
1702 enum { NETREG_UNINITIALIZED
=0,
1703 NETREG_REGISTERED
, /* completed register_netdevice */
1704 NETREG_UNREGISTERING
, /* called unregister_netdevice */
1705 NETREG_UNREGISTERED
, /* completed unregister todo */
1706 NETREG_RELEASED
, /* called free_netdev */
1707 NETREG_DUMMY
, /* dummy device for NAPI poll */
1713 RTNL_LINK_INITIALIZED
,
1714 RTNL_LINK_INITIALIZING
,
1715 } rtnl_link_state
:16;
1717 void (*destructor
)(struct net_device
*dev
);
1719 #ifdef CONFIG_NETPOLL
1720 struct netpoll_info __rcu
*npinfo
;
1723 #ifdef CONFIG_NET_NS
1727 /* mid-layer private */
1730 struct pcpu_lstats __percpu
*lstats
;
1731 struct pcpu_sw_netstats __percpu
*tstats
;
1732 struct pcpu_dstats __percpu
*dstats
;
1733 struct pcpu_vstats __percpu
*vstats
;
1736 struct garp_port __rcu
*garp_port
;
1737 struct mrp_port __rcu
*mrp_port
;
1740 const struct attribute_group
*sysfs_groups
[4];
1741 const struct attribute_group
*sysfs_rx_queue_group
;
1743 const struct rtnl_link_ops
*rtnl_link_ops
;
1745 /* for setting kernel sock attribute on TCP connection setup */
1746 #define GSO_MAX_SIZE 65536
1747 unsigned int gso_max_size
;
1748 #define GSO_MAX_SEGS 65535
1752 const struct dcbnl_rtnl_ops
*dcbnl_ops
;
1755 struct netdev_tc_txq tc_to_txq
[TC_MAX_QUEUE
];
1756 u8 prio_tc_map
[TC_BITMASK
+ 1];
1758 #if IS_ENABLED(CONFIG_FCOE)
1759 unsigned int fcoe_ddp_xid
;
1761 #if IS_ENABLED(CONFIG_CGROUP_NET_PRIO)
1762 struct netprio_map __rcu
*priomap
;
1764 struct phy_device
*phydev
;
1765 struct lock_class_key
*qdisc_tx_busylock
;
1767 struct pm_qos_request pm_qos_req
;
1769 #define to_net_dev(d) container_of(d, struct net_device, dev)
1771 #define NETDEV_ALIGN 32
1774 int netdev_get_prio_tc_map(const struct net_device
*dev
, u32 prio
)
1776 return dev
->prio_tc_map
[prio
& TC_BITMASK
];
1780 int netdev_set_prio_tc_map(struct net_device
*dev
, u8 prio
, u8 tc
)
1782 if (tc
>= dev
->num_tc
)
1785 dev
->prio_tc_map
[prio
& TC_BITMASK
] = tc
& TC_BITMASK
;
1790 void netdev_reset_tc(struct net_device
*dev
)
1793 memset(dev
->tc_to_txq
, 0, sizeof(dev
->tc_to_txq
));
1794 memset(dev
->prio_tc_map
, 0, sizeof(dev
->prio_tc_map
));
1798 int netdev_set_tc_queue(struct net_device
*dev
, u8 tc
, u16 count
, u16 offset
)
1800 if (tc
>= dev
->num_tc
)
1803 dev
->tc_to_txq
[tc
].count
= count
;
1804 dev
->tc_to_txq
[tc
].offset
= offset
;
1809 int netdev_set_num_tc(struct net_device
*dev
, u8 num_tc
)
1811 if (num_tc
> TC_MAX_QUEUE
)
1814 dev
->num_tc
= num_tc
;
1819 int netdev_get_num_tc(struct net_device
*dev
)
1825 struct netdev_queue
*netdev_get_tx_queue(const struct net_device
*dev
,
1828 return &dev
->_tx
[index
];
1831 static inline struct netdev_queue
*skb_get_tx_queue(const struct net_device
*dev
,
1832 const struct sk_buff
*skb
)
1834 return netdev_get_tx_queue(dev
, skb_get_queue_mapping(skb
));
1837 static inline void netdev_for_each_tx_queue(struct net_device
*dev
,
1838 void (*f
)(struct net_device
*,
1839 struct netdev_queue
*,
1845 for (i
= 0; i
< dev
->num_tx_queues
; i
++)
1846 f(dev
, &dev
->_tx
[i
], arg
);
1849 struct netdev_queue
*netdev_pick_tx(struct net_device
*dev
,
1850 struct sk_buff
*skb
,
1854 * Net namespace inlines
1857 struct net
*dev_net(const struct net_device
*dev
)
1859 return read_pnet(&dev
->nd_net
);
1863 void dev_net_set(struct net_device
*dev
, struct net
*net
)
1865 #ifdef CONFIG_NET_NS
1866 release_net(dev
->nd_net
);
1867 dev
->nd_net
= hold_net(net
);
1871 static inline bool netdev_uses_dsa(struct net_device
*dev
)
1873 #if IS_ENABLED(CONFIG_NET_DSA)
1874 if (dev
->dsa_ptr
!= NULL
)
1875 return dsa_uses_tagged_protocol(dev
->dsa_ptr
);
1881 * netdev_priv - access network device private data
1882 * @dev: network device
1884 * Get network device private data
1886 static inline void *netdev_priv(const struct net_device
*dev
)
1888 return (char *)dev
+ ALIGN(sizeof(struct net_device
), NETDEV_ALIGN
);
1891 /* Set the sysfs physical device reference for the network logical device
1892 * if set prior to registration will cause a symlink during initialization.
1894 #define SET_NETDEV_DEV(net, pdev) ((net)->dev.parent = (pdev))
1896 /* Set the sysfs device type for the network logical device to allow
1897 * fine-grained identification of different network device types. For
1898 * example Ethernet, Wirelss LAN, Bluetooth, WiMAX etc.
1900 #define SET_NETDEV_DEVTYPE(net, devtype) ((net)->dev.type = (devtype))
1902 /* Default NAPI poll() weight
1903 * Device drivers are strongly advised to not use bigger value
1905 #define NAPI_POLL_WEIGHT 64
1908 * netif_napi_add - initialize a napi context
1909 * @dev: network device
1910 * @napi: napi context
1911 * @poll: polling function
1912 * @weight: default weight
1914 * netif_napi_add() must be used to initialize a napi context prior to calling
1915 * *any* of the other napi related functions.
1917 void netif_napi_add(struct net_device
*dev
, struct napi_struct
*napi
,
1918 int (*poll
)(struct napi_struct
*, int), int weight
);
1921 * netif_napi_del - remove a napi context
1922 * @napi: napi context
1924 * netif_napi_del() removes a napi context from the network device napi list
1926 void netif_napi_del(struct napi_struct
*napi
);
1928 struct napi_gro_cb
{
1929 /* Virtual address of skb_shinfo(skb)->frags[0].page + offset. */
1932 /* Length of frag0. */
1933 unsigned int frag0_len
;
1935 /* This indicates where we are processing relative to skb->data. */
1938 /* This is non-zero if the packet cannot be merged with the new skb. */
1941 /* Save the IP ID here and check when we get to the transport layer */
1944 /* Number of segments aggregated. */
1947 /* Start offset for remote checksum offload */
1948 u16 gro_remcsum_start
;
1950 /* jiffies when first packet was created/queued */
1953 /* Used in ipv6_gro_receive() and foo-over-udp */
1956 /* This is non-zero if the packet may be of the same flow. */
1959 /* Used in udp_gro_receive */
1962 /* GRO checksum is valid */
1965 /* Number of checksums via CHECKSUM_UNNECESSARY */
1970 #define NAPI_GRO_FREE 1
1971 #define NAPI_GRO_FREE_STOLEN_HEAD 2
1973 /* Used in foo-over-udp, set in udp[46]_gro_receive */
1978 /* used to support CHECKSUM_COMPLETE for tunneling protocols */
1981 /* used in skb_gro_receive() slow path */
1982 struct sk_buff
*last
;
1985 #define NAPI_GRO_CB(skb) ((struct napi_gro_cb *)(skb)->cb)
1987 struct packet_type
{
1988 __be16 type
; /* This is really htons(ether_type). */
1989 struct net_device
*dev
; /* NULL is wildcarded here */
1990 int (*func
) (struct sk_buff
*,
1991 struct net_device
*,
1992 struct packet_type
*,
1993 struct net_device
*);
1994 bool (*id_match
)(struct packet_type
*ptype
,
1996 void *af_packet_priv
;
1997 struct list_head list
;
2000 struct offload_callbacks
{
2001 struct sk_buff
*(*gso_segment
)(struct sk_buff
*skb
,
2002 netdev_features_t features
);
2003 struct sk_buff
**(*gro_receive
)(struct sk_buff
**head
,
2004 struct sk_buff
*skb
);
2005 int (*gro_complete
)(struct sk_buff
*skb
, int nhoff
);
2008 struct packet_offload
{
2009 __be16 type
; /* This is really htons(ether_type). */
2010 struct offload_callbacks callbacks
;
2011 struct list_head list
;
2016 struct udp_offload_callbacks
{
2017 struct sk_buff
**(*gro_receive
)(struct sk_buff
**head
,
2018 struct sk_buff
*skb
,
2019 struct udp_offload
*uoff
);
2020 int (*gro_complete
)(struct sk_buff
*skb
,
2022 struct udp_offload
*uoff
);
2025 struct udp_offload
{
2028 struct udp_offload_callbacks callbacks
;
2031 /* often modified stats are per cpu, other are shared (netdev->stats) */
2032 struct pcpu_sw_netstats
{
2037 struct u64_stats_sync syncp
;
2040 #define netdev_alloc_pcpu_stats(type) \
2042 typeof(type) __percpu *pcpu_stats = alloc_percpu(type); \
2045 for_each_possible_cpu(i) { \
2046 typeof(type) *stat; \
2047 stat = per_cpu_ptr(pcpu_stats, i); \
2048 u64_stats_init(&stat->syncp); \
2054 #include <linux/notifier.h>
2056 /* netdevice notifier chain. Please remember to update the rtnetlink
2057 * notification exclusion list in rtnetlink_event() when adding new
2060 #define NETDEV_UP 0x0001 /* For now you can't veto a device up/down */
2061 #define NETDEV_DOWN 0x0002
2062 #define NETDEV_REBOOT 0x0003 /* Tell a protocol stack a network interface
2063 detected a hardware crash and restarted
2064 - we can use this eg to kick tcp sessions
2066 #define NETDEV_CHANGE 0x0004 /* Notify device state change */
2067 #define NETDEV_REGISTER 0x0005
2068 #define NETDEV_UNREGISTER 0x0006
2069 #define NETDEV_CHANGEMTU 0x0007 /* notify after mtu change happened */
2070 #define NETDEV_CHANGEADDR 0x0008
2071 #define NETDEV_GOING_DOWN 0x0009
2072 #define NETDEV_CHANGENAME 0x000A
2073 #define NETDEV_FEAT_CHANGE 0x000B
2074 #define NETDEV_BONDING_FAILOVER 0x000C
2075 #define NETDEV_PRE_UP 0x000D
2076 #define NETDEV_PRE_TYPE_CHANGE 0x000E
2077 #define NETDEV_POST_TYPE_CHANGE 0x000F
2078 #define NETDEV_POST_INIT 0x0010
2079 #define NETDEV_UNREGISTER_FINAL 0x0011
2080 #define NETDEV_RELEASE 0x0012
2081 #define NETDEV_NOTIFY_PEERS 0x0013
2082 #define NETDEV_JOIN 0x0014
2083 #define NETDEV_CHANGEUPPER 0x0015
2084 #define NETDEV_RESEND_IGMP 0x0016
2085 #define NETDEV_PRECHANGEMTU 0x0017 /* notify before mtu change happened */
2086 #define NETDEV_CHANGEINFODATA 0x0018
2087 #define NETDEV_BONDING_INFO 0x0019
2089 int register_netdevice_notifier(struct notifier_block
*nb
);
2090 int unregister_netdevice_notifier(struct notifier_block
*nb
);
2092 struct netdev_notifier_info
{
2093 struct net_device
*dev
;
2096 struct netdev_notifier_change_info
{
2097 struct netdev_notifier_info info
; /* must be first */
2098 unsigned int flags_changed
;
2101 static inline void netdev_notifier_info_init(struct netdev_notifier_info
*info
,
2102 struct net_device
*dev
)
2107 static inline struct net_device
*
2108 netdev_notifier_info_to_dev(const struct netdev_notifier_info
*info
)
2113 int call_netdevice_notifiers(unsigned long val
, struct net_device
*dev
);
2116 extern rwlock_t dev_base_lock
; /* Device list lock */
2118 #define for_each_netdev(net, d) \
2119 list_for_each_entry(d, &(net)->dev_base_head, dev_list)
2120 #define for_each_netdev_reverse(net, d) \
2121 list_for_each_entry_reverse(d, &(net)->dev_base_head, dev_list)
2122 #define for_each_netdev_rcu(net, d) \
2123 list_for_each_entry_rcu(d, &(net)->dev_base_head, dev_list)
2124 #define for_each_netdev_safe(net, d, n) \
2125 list_for_each_entry_safe(d, n, &(net)->dev_base_head, dev_list)
2126 #define for_each_netdev_continue(net, d) \
2127 list_for_each_entry_continue(d, &(net)->dev_base_head, dev_list)
2128 #define for_each_netdev_continue_rcu(net, d) \
2129 list_for_each_entry_continue_rcu(d, &(net)->dev_base_head, dev_list)
2130 #define for_each_netdev_in_bond_rcu(bond, slave) \
2131 for_each_netdev_rcu(&init_net, slave) \
2132 if (netdev_master_upper_dev_get_rcu(slave) == (bond))
2133 #define net_device_entry(lh) list_entry(lh, struct net_device, dev_list)
2135 static inline struct net_device
*next_net_device(struct net_device
*dev
)
2137 struct list_head
*lh
;
2141 lh
= dev
->dev_list
.next
;
2142 return lh
== &net
->dev_base_head
? NULL
: net_device_entry(lh
);
2145 static inline struct net_device
*next_net_device_rcu(struct net_device
*dev
)
2147 struct list_head
*lh
;
2151 lh
= rcu_dereference(list_next_rcu(&dev
->dev_list
));
2152 return lh
== &net
->dev_base_head
? NULL
: net_device_entry(lh
);
2155 static inline struct net_device
*first_net_device(struct net
*net
)
2157 return list_empty(&net
->dev_base_head
) ? NULL
:
2158 net_device_entry(net
->dev_base_head
.next
);
2161 static inline struct net_device
*first_net_device_rcu(struct net
*net
)
2163 struct list_head
*lh
= rcu_dereference(list_next_rcu(&net
->dev_base_head
));
2165 return lh
== &net
->dev_base_head
? NULL
: net_device_entry(lh
);
2168 int netdev_boot_setup_check(struct net_device
*dev
);
2169 unsigned long netdev_boot_base(const char *prefix
, int unit
);
2170 struct net_device
*dev_getbyhwaddr_rcu(struct net
*net
, unsigned short type
,
2171 const char *hwaddr
);
2172 struct net_device
*dev_getfirstbyhwtype(struct net
*net
, unsigned short type
);
2173 struct net_device
*__dev_getfirstbyhwtype(struct net
*net
, unsigned short type
);
2174 void dev_add_pack(struct packet_type
*pt
);
2175 void dev_remove_pack(struct packet_type
*pt
);
2176 void __dev_remove_pack(struct packet_type
*pt
);
2177 void dev_add_offload(struct packet_offload
*po
);
2178 void dev_remove_offload(struct packet_offload
*po
);
2180 struct net_device
*__dev_get_by_flags(struct net
*net
, unsigned short flags
,
2181 unsigned short mask
);
2182 struct net_device
*dev_get_by_name(struct net
*net
, const char *name
);
2183 struct net_device
*dev_get_by_name_rcu(struct net
*net
, const char *name
);
2184 struct net_device
*__dev_get_by_name(struct net
*net
, const char *name
);
2185 int dev_alloc_name(struct net_device
*dev
, const char *name
);
2186 int dev_open(struct net_device
*dev
);
2187 int dev_close(struct net_device
*dev
);
2188 void dev_disable_lro(struct net_device
*dev
);
2189 int dev_loopback_xmit(struct sk_buff
*newskb
);
2190 int dev_queue_xmit(struct sk_buff
*skb
);
2191 int dev_queue_xmit_accel(struct sk_buff
*skb
, void *accel_priv
);
2192 int register_netdevice(struct net_device
*dev
);
2193 void unregister_netdevice_queue(struct net_device
*dev
, struct list_head
*head
);
2194 void unregister_netdevice_many(struct list_head
*head
);
2195 static inline void unregister_netdevice(struct net_device
*dev
)
2197 unregister_netdevice_queue(dev
, NULL
);
2200 int netdev_refcnt_read(const struct net_device
*dev
);
2201 void free_netdev(struct net_device
*dev
);
2202 void netdev_freemem(struct net_device
*dev
);
2203 void synchronize_net(void);
2204 int init_dummy_netdev(struct net_device
*dev
);
2206 struct net_device
*dev_get_by_index(struct net
*net
, int ifindex
);
2207 struct net_device
*__dev_get_by_index(struct net
*net
, int ifindex
);
2208 struct net_device
*dev_get_by_index_rcu(struct net
*net
, int ifindex
);
2209 int netdev_get_name(struct net
*net
, char *name
, int ifindex
);
2210 int dev_restart(struct net_device
*dev
);
2211 int skb_gro_receive(struct sk_buff
**head
, struct sk_buff
*skb
);
2213 static inline unsigned int skb_gro_offset(const struct sk_buff
*skb
)
2215 return NAPI_GRO_CB(skb
)->data_offset
;
2218 static inline unsigned int skb_gro_len(const struct sk_buff
*skb
)
2220 return skb
->len
- NAPI_GRO_CB(skb
)->data_offset
;
2223 static inline void skb_gro_pull(struct sk_buff
*skb
, unsigned int len
)
2225 NAPI_GRO_CB(skb
)->data_offset
+= len
;
2228 static inline void *skb_gro_header_fast(struct sk_buff
*skb
,
2229 unsigned int offset
)
2231 return NAPI_GRO_CB(skb
)->frag0
+ offset
;
2234 static inline int skb_gro_header_hard(struct sk_buff
*skb
, unsigned int hlen
)
2236 return NAPI_GRO_CB(skb
)->frag0_len
< hlen
;
2239 static inline void *skb_gro_header_slow(struct sk_buff
*skb
, unsigned int hlen
,
2240 unsigned int offset
)
2242 if (!pskb_may_pull(skb
, hlen
))
2245 NAPI_GRO_CB(skb
)->frag0
= NULL
;
2246 NAPI_GRO_CB(skb
)->frag0_len
= 0;
2247 return skb
->data
+ offset
;
2250 static inline void *skb_gro_network_header(struct sk_buff
*skb
)
2252 return (NAPI_GRO_CB(skb
)->frag0
?: skb
->data
) +
2253 skb_network_offset(skb
);
2256 static inline void skb_gro_postpull_rcsum(struct sk_buff
*skb
,
2257 const void *start
, unsigned int len
)
2259 if (NAPI_GRO_CB(skb
)->csum_valid
)
2260 NAPI_GRO_CB(skb
)->csum
= csum_sub(NAPI_GRO_CB(skb
)->csum
,
2261 csum_partial(start
, len
, 0));
2264 /* GRO checksum functions. These are logical equivalents of the normal
2265 * checksum functions (in skbuff.h) except that they operate on the GRO
2266 * offsets and fields in sk_buff.
2269 __sum16
__skb_gro_checksum_complete(struct sk_buff
*skb
);
2271 static inline bool skb_at_gro_remcsum_start(struct sk_buff
*skb
)
2273 return (NAPI_GRO_CB(skb
)->gro_remcsum_start
- skb_headroom(skb
) ==
2274 skb_gro_offset(skb
));
2277 static inline bool __skb_gro_checksum_validate_needed(struct sk_buff
*skb
,
2281 return ((skb
->ip_summed
!= CHECKSUM_PARTIAL
||
2282 skb_checksum_start_offset(skb
) <
2283 skb_gro_offset(skb
)) &&
2284 !skb_at_gro_remcsum_start(skb
) &&
2285 NAPI_GRO_CB(skb
)->csum_cnt
== 0 &&
2286 (!zero_okay
|| check
));
2289 static inline __sum16
__skb_gro_checksum_validate_complete(struct sk_buff
*skb
,
2292 if (NAPI_GRO_CB(skb
)->csum_valid
&&
2293 !csum_fold(csum_add(psum
, NAPI_GRO_CB(skb
)->csum
)))
2296 NAPI_GRO_CB(skb
)->csum
= psum
;
2298 return __skb_gro_checksum_complete(skb
);
2301 static inline void skb_gro_incr_csum_unnecessary(struct sk_buff
*skb
)
2303 if (NAPI_GRO_CB(skb
)->csum_cnt
> 0) {
2304 /* Consume a checksum from CHECKSUM_UNNECESSARY */
2305 NAPI_GRO_CB(skb
)->csum_cnt
--;
2307 /* Update skb for CHECKSUM_UNNECESSARY and csum_level when we
2308 * verified a new top level checksum or an encapsulated one
2309 * during GRO. This saves work if we fallback to normal path.
2311 __skb_incr_checksum_unnecessary(skb
);
2315 #define __skb_gro_checksum_validate(skb, proto, zero_okay, check, \
2318 __sum16 __ret = 0; \
2319 if (__skb_gro_checksum_validate_needed(skb, zero_okay, check)) \
2320 __ret = __skb_gro_checksum_validate_complete(skb, \
2321 compute_pseudo(skb, proto)); \
2323 __skb_mark_checksum_bad(skb); \
2325 skb_gro_incr_csum_unnecessary(skb); \
2329 #define skb_gro_checksum_validate(skb, proto, compute_pseudo) \
2330 __skb_gro_checksum_validate(skb, proto, false, 0, compute_pseudo)
2332 #define skb_gro_checksum_validate_zero_check(skb, proto, check, \
2334 __skb_gro_checksum_validate(skb, proto, true, check, compute_pseudo)
2336 #define skb_gro_checksum_simple_validate(skb) \
2337 __skb_gro_checksum_validate(skb, 0, false, 0, null_compute_pseudo)
2339 static inline bool __skb_gro_checksum_convert_check(struct sk_buff
*skb
)
2341 return (NAPI_GRO_CB(skb
)->csum_cnt
== 0 &&
2342 !NAPI_GRO_CB(skb
)->csum_valid
);
2345 static inline void __skb_gro_checksum_convert(struct sk_buff
*skb
,
2346 __sum16 check
, __wsum pseudo
)
2348 NAPI_GRO_CB(skb
)->csum
= ~pseudo
;
2349 NAPI_GRO_CB(skb
)->csum_valid
= 1;
2352 #define skb_gro_checksum_try_convert(skb, proto, check, compute_pseudo) \
2354 if (__skb_gro_checksum_convert_check(skb)) \
2355 __skb_gro_checksum_convert(skb, check, \
2356 compute_pseudo(skb, proto)); \
2359 struct gro_remcsum
{
2364 static inline void skb_gro_remcsum_init(struct gro_remcsum
*grc
)
2370 static inline void skb_gro_remcsum_process(struct sk_buff
*skb
, void *ptr
,
2371 int start
, int offset
,
2372 struct gro_remcsum
*grc
,
2377 BUG_ON(!NAPI_GRO_CB(skb
)->csum_valid
);
2380 NAPI_GRO_CB(skb
)->gro_remcsum_start
=
2381 ((unsigned char *)ptr
+ start
) - skb
->head
;
2385 delta
= remcsum_adjust(ptr
, NAPI_GRO_CB(skb
)->csum
, start
, offset
);
2387 /* Adjust skb->csum since we changed the packet */
2388 NAPI_GRO_CB(skb
)->csum
= csum_add(NAPI_GRO_CB(skb
)->csum
, delta
);
2390 grc
->offset
= (ptr
+ offset
) - (void *)skb
->head
;
2394 static inline void skb_gro_remcsum_cleanup(struct sk_buff
*skb
,
2395 struct gro_remcsum
*grc
)
2400 remcsum_unadjust((__sum16
*)(skb
->head
+ grc
->offset
), grc
->delta
);
2403 static inline int dev_hard_header(struct sk_buff
*skb
, struct net_device
*dev
,
2404 unsigned short type
,
2405 const void *daddr
, const void *saddr
,
2408 if (!dev
->header_ops
|| !dev
->header_ops
->create
)
2411 return dev
->header_ops
->create(skb
, dev
, type
, daddr
, saddr
, len
);
2414 static inline int dev_parse_header(const struct sk_buff
*skb
,
2415 unsigned char *haddr
)
2417 const struct net_device
*dev
= skb
->dev
;
2419 if (!dev
->header_ops
|| !dev
->header_ops
->parse
)
2421 return dev
->header_ops
->parse(skb
, haddr
);
2424 typedef int gifconf_func_t(struct net_device
* dev
, char __user
* bufptr
, int len
);
2425 int register_gifconf(unsigned int family
, gifconf_func_t
*gifconf
);
2426 static inline int unregister_gifconf(unsigned int family
)
2428 return register_gifconf(family
, NULL
);
2431 #ifdef CONFIG_NET_FLOW_LIMIT
2432 #define FLOW_LIMIT_HISTORY (1 << 7) /* must be ^2 and !overflow buckets */
2433 struct sd_flow_limit
{
2435 unsigned int num_buckets
;
2436 unsigned int history_head
;
2437 u16 history
[FLOW_LIMIT_HISTORY
];
2441 extern int netdev_flow_limit_table_len
;
2442 #endif /* CONFIG_NET_FLOW_LIMIT */
2445 * Incoming packets are placed on per-cpu queues
2447 struct softnet_data
{
2448 struct list_head poll_list
;
2449 struct sk_buff_head process_queue
;
2452 unsigned int processed
;
2453 unsigned int time_squeeze
;
2454 unsigned int cpu_collision
;
2455 unsigned int received_rps
;
2457 struct softnet_data
*rps_ipi_list
;
2459 #ifdef CONFIG_NET_FLOW_LIMIT
2460 struct sd_flow_limit __rcu
*flow_limit
;
2462 struct Qdisc
*output_queue
;
2463 struct Qdisc
**output_queue_tailp
;
2464 struct sk_buff
*completion_queue
;
2467 /* Elements below can be accessed between CPUs for RPS */
2468 struct call_single_data csd ____cacheline_aligned_in_smp
;
2469 struct softnet_data
*rps_ipi_next
;
2471 unsigned int input_queue_head
;
2472 unsigned int input_queue_tail
;
2474 unsigned int dropped
;
2475 struct sk_buff_head input_pkt_queue
;
2476 struct napi_struct backlog
;
2480 static inline void input_queue_head_incr(struct softnet_data
*sd
)
2483 sd
->input_queue_head
++;
2487 static inline void input_queue_tail_incr_save(struct softnet_data
*sd
,
2488 unsigned int *qtail
)
2491 *qtail
= ++sd
->input_queue_tail
;
2495 DECLARE_PER_CPU_ALIGNED(struct softnet_data
, softnet_data
);
2497 void __netif_schedule(struct Qdisc
*q
);
2498 void netif_schedule_queue(struct netdev_queue
*txq
);
2500 static inline void netif_tx_schedule_all(struct net_device
*dev
)
2504 for (i
= 0; i
< dev
->num_tx_queues
; i
++)
2505 netif_schedule_queue(netdev_get_tx_queue(dev
, i
));
2508 static inline void netif_tx_start_queue(struct netdev_queue
*dev_queue
)
2510 clear_bit(__QUEUE_STATE_DRV_XOFF
, &dev_queue
->state
);
2514 * netif_start_queue - allow transmit
2515 * @dev: network device
2517 * Allow upper layers to call the device hard_start_xmit routine.
2519 static inline void netif_start_queue(struct net_device
*dev
)
2521 netif_tx_start_queue(netdev_get_tx_queue(dev
, 0));
2524 static inline void netif_tx_start_all_queues(struct net_device
*dev
)
2528 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
2529 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
2530 netif_tx_start_queue(txq
);
2534 void netif_tx_wake_queue(struct netdev_queue
*dev_queue
);
2537 * netif_wake_queue - restart transmit
2538 * @dev: network device
2540 * Allow upper layers to call the device hard_start_xmit routine.
2541 * Used for flow control when transmit resources are available.
2543 static inline void netif_wake_queue(struct net_device
*dev
)
2545 netif_tx_wake_queue(netdev_get_tx_queue(dev
, 0));
2548 static inline void netif_tx_wake_all_queues(struct net_device
*dev
)
2552 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
2553 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
2554 netif_tx_wake_queue(txq
);
2558 static inline void netif_tx_stop_queue(struct netdev_queue
*dev_queue
)
2560 if (WARN_ON(!dev_queue
)) {
2561 pr_info("netif_stop_queue() cannot be called before register_netdev()\n");
2564 set_bit(__QUEUE_STATE_DRV_XOFF
, &dev_queue
->state
);
2568 * netif_stop_queue - stop transmitted packets
2569 * @dev: network device
2571 * Stop upper layers calling the device hard_start_xmit routine.
2572 * Used for flow control when transmit resources are unavailable.
2574 static inline void netif_stop_queue(struct net_device
*dev
)
2576 netif_tx_stop_queue(netdev_get_tx_queue(dev
, 0));
2579 static inline void netif_tx_stop_all_queues(struct net_device
*dev
)
2583 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
2584 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
2585 netif_tx_stop_queue(txq
);
2589 static inline bool netif_tx_queue_stopped(const struct netdev_queue
*dev_queue
)
2591 return test_bit(__QUEUE_STATE_DRV_XOFF
, &dev_queue
->state
);
2595 * netif_queue_stopped - test if transmit queue is flowblocked
2596 * @dev: network device
2598 * Test if transmit queue on device is currently unable to send.
2600 static inline bool netif_queue_stopped(const struct net_device
*dev
)
2602 return netif_tx_queue_stopped(netdev_get_tx_queue(dev
, 0));
2605 static inline bool netif_xmit_stopped(const struct netdev_queue
*dev_queue
)
2607 return dev_queue
->state
& QUEUE_STATE_ANY_XOFF
;
2611 netif_xmit_frozen_or_stopped(const struct netdev_queue
*dev_queue
)
2613 return dev_queue
->state
& QUEUE_STATE_ANY_XOFF_OR_FROZEN
;
2617 netif_xmit_frozen_or_drv_stopped(const struct netdev_queue
*dev_queue
)
2619 return dev_queue
->state
& QUEUE_STATE_DRV_XOFF_OR_FROZEN
;
2623 * netdev_txq_bql_enqueue_prefetchw - prefetch bql data for write
2624 * @dev_queue: pointer to transmit queue
2626 * BQL enabled drivers might use this helper in their ndo_start_xmit(),
2627 * to give appropriate hint to the cpu.
2629 static inline void netdev_txq_bql_enqueue_prefetchw(struct netdev_queue
*dev_queue
)
2632 prefetchw(&dev_queue
->dql
.num_queued
);
2637 * netdev_txq_bql_complete_prefetchw - prefetch bql data for write
2638 * @dev_queue: pointer to transmit queue
2640 * BQL enabled drivers might use this helper in their TX completion path,
2641 * to give appropriate hint to the cpu.
2643 static inline void netdev_txq_bql_complete_prefetchw(struct netdev_queue
*dev_queue
)
2646 prefetchw(&dev_queue
->dql
.limit
);
2650 static inline void netdev_tx_sent_queue(struct netdev_queue
*dev_queue
,
2654 dql_queued(&dev_queue
->dql
, bytes
);
2656 if (likely(dql_avail(&dev_queue
->dql
) >= 0))
2659 set_bit(__QUEUE_STATE_STACK_XOFF
, &dev_queue
->state
);
2662 * The XOFF flag must be set before checking the dql_avail below,
2663 * because in netdev_tx_completed_queue we update the dql_completed
2664 * before checking the XOFF flag.
2668 /* check again in case another CPU has just made room avail */
2669 if (unlikely(dql_avail(&dev_queue
->dql
) >= 0))
2670 clear_bit(__QUEUE_STATE_STACK_XOFF
, &dev_queue
->state
);
2675 * netdev_sent_queue - report the number of bytes queued to hardware
2676 * @dev: network device
2677 * @bytes: number of bytes queued to the hardware device queue
2679 * Report the number of bytes queued for sending/completion to the network
2680 * device hardware queue. @bytes should be a good approximation and should
2681 * exactly match netdev_completed_queue() @bytes
2683 static inline void netdev_sent_queue(struct net_device
*dev
, unsigned int bytes
)
2685 netdev_tx_sent_queue(netdev_get_tx_queue(dev
, 0), bytes
);
2688 static inline void netdev_tx_completed_queue(struct netdev_queue
*dev_queue
,
2689 unsigned int pkts
, unsigned int bytes
)
2692 if (unlikely(!bytes
))
2695 dql_completed(&dev_queue
->dql
, bytes
);
2698 * Without the memory barrier there is a small possiblity that
2699 * netdev_tx_sent_queue will miss the update and cause the queue to
2700 * be stopped forever
2704 if (dql_avail(&dev_queue
->dql
) < 0)
2707 if (test_and_clear_bit(__QUEUE_STATE_STACK_XOFF
, &dev_queue
->state
))
2708 netif_schedule_queue(dev_queue
);
2713 * netdev_completed_queue - report bytes and packets completed by device
2714 * @dev: network device
2715 * @pkts: actual number of packets sent over the medium
2716 * @bytes: actual number of bytes sent over the medium
2718 * Report the number of bytes and packets transmitted by the network device
2719 * hardware queue over the physical medium, @bytes must exactly match the
2720 * @bytes amount passed to netdev_sent_queue()
2722 static inline void netdev_completed_queue(struct net_device
*dev
,
2723 unsigned int pkts
, unsigned int bytes
)
2725 netdev_tx_completed_queue(netdev_get_tx_queue(dev
, 0), pkts
, bytes
);
2728 static inline void netdev_tx_reset_queue(struct netdev_queue
*q
)
2731 clear_bit(__QUEUE_STATE_STACK_XOFF
, &q
->state
);
2737 * netdev_reset_queue - reset the packets and bytes count of a network device
2738 * @dev_queue: network device
2740 * Reset the bytes and packet count of a network device and clear the
2741 * software flow control OFF bit for this network device
2743 static inline void netdev_reset_queue(struct net_device
*dev_queue
)
2745 netdev_tx_reset_queue(netdev_get_tx_queue(dev_queue
, 0));
2749 * netdev_cap_txqueue - check if selected tx queue exceeds device queues
2750 * @dev: network device
2751 * @queue_index: given tx queue index
2753 * Returns 0 if given tx queue index >= number of device tx queues,
2754 * otherwise returns the originally passed tx queue index.
2756 static inline u16
netdev_cap_txqueue(struct net_device
*dev
, u16 queue_index
)
2758 if (unlikely(queue_index
>= dev
->real_num_tx_queues
)) {
2759 net_warn_ratelimited("%s selects TX queue %d, but real number of TX queues is %d\n",
2760 dev
->name
, queue_index
,
2761 dev
->real_num_tx_queues
);
2769 * netif_running - test if up
2770 * @dev: network device
2772 * Test if the device has been brought up.
2774 static inline bool netif_running(const struct net_device
*dev
)
2776 return test_bit(__LINK_STATE_START
, &dev
->state
);
2780 * Routines to manage the subqueues on a device. We only need start
2781 * stop, and a check if it's stopped. All other device management is
2782 * done at the overall netdevice level.
2783 * Also test the device if we're multiqueue.
2787 * netif_start_subqueue - allow sending packets on subqueue
2788 * @dev: network device
2789 * @queue_index: sub queue index
2791 * Start individual transmit queue of a device with multiple transmit queues.
2793 static inline void netif_start_subqueue(struct net_device
*dev
, u16 queue_index
)
2795 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, queue_index
);
2797 netif_tx_start_queue(txq
);
2801 * netif_stop_subqueue - stop sending packets on subqueue
2802 * @dev: network device
2803 * @queue_index: sub queue index
2805 * Stop individual transmit queue of a device with multiple transmit queues.
2807 static inline void netif_stop_subqueue(struct net_device
*dev
, u16 queue_index
)
2809 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, queue_index
);
2810 netif_tx_stop_queue(txq
);
2814 * netif_subqueue_stopped - test status of subqueue
2815 * @dev: network device
2816 * @queue_index: sub queue index
2818 * Check individual transmit queue of a device with multiple transmit queues.
2820 static inline bool __netif_subqueue_stopped(const struct net_device
*dev
,
2823 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, queue_index
);
2825 return netif_tx_queue_stopped(txq
);
2828 static inline bool netif_subqueue_stopped(const struct net_device
*dev
,
2829 struct sk_buff
*skb
)
2831 return __netif_subqueue_stopped(dev
, skb_get_queue_mapping(skb
));
2834 void netif_wake_subqueue(struct net_device
*dev
, u16 queue_index
);
2837 int netif_set_xps_queue(struct net_device
*dev
, const struct cpumask
*mask
,
2840 static inline int netif_set_xps_queue(struct net_device
*dev
,
2841 const struct cpumask
*mask
,
2849 * Returns a Tx hash for the given packet when dev->real_num_tx_queues is used
2850 * as a distribution range limit for the returned value.
2852 static inline u16
skb_tx_hash(const struct net_device
*dev
,
2853 struct sk_buff
*skb
)
2855 return __skb_tx_hash(dev
, skb
, dev
->real_num_tx_queues
);
2859 * netif_is_multiqueue - test if device has multiple transmit queues
2860 * @dev: network device
2862 * Check if device has multiple transmit queues
2864 static inline bool netif_is_multiqueue(const struct net_device
*dev
)
2866 return dev
->num_tx_queues
> 1;
2869 int netif_set_real_num_tx_queues(struct net_device
*dev
, unsigned int txq
);
2872 int netif_set_real_num_rx_queues(struct net_device
*dev
, unsigned int rxq
);
2874 static inline int netif_set_real_num_rx_queues(struct net_device
*dev
,
2882 static inline unsigned int get_netdev_rx_queue_index(
2883 struct netdev_rx_queue
*queue
)
2885 struct net_device
*dev
= queue
->dev
;
2886 int index
= queue
- dev
->_rx
;
2888 BUG_ON(index
>= dev
->num_rx_queues
);
2893 #define DEFAULT_MAX_NUM_RSS_QUEUES (8)
2894 int netif_get_num_default_rss_queues(void);
2896 enum skb_free_reason
{
2897 SKB_REASON_CONSUMED
,
2901 void __dev_kfree_skb_irq(struct sk_buff
*skb
, enum skb_free_reason reason
);
2902 void __dev_kfree_skb_any(struct sk_buff
*skb
, enum skb_free_reason reason
);
2905 * It is not allowed to call kfree_skb() or consume_skb() from hardware
2906 * interrupt context or with hardware interrupts being disabled.
2907 * (in_irq() || irqs_disabled())
2909 * We provide four helpers that can be used in following contexts :
2911 * dev_kfree_skb_irq(skb) when caller drops a packet from irq context,
2912 * replacing kfree_skb(skb)
2914 * dev_consume_skb_irq(skb) when caller consumes a packet from irq context.
2915 * Typically used in place of consume_skb(skb) in TX completion path
2917 * dev_kfree_skb_any(skb) when caller doesn't know its current irq context,
2918 * replacing kfree_skb(skb)
2920 * dev_consume_skb_any(skb) when caller doesn't know its current irq context,
2921 * and consumed a packet. Used in place of consume_skb(skb)
2923 static inline void dev_kfree_skb_irq(struct sk_buff
*skb
)
2925 __dev_kfree_skb_irq(skb
, SKB_REASON_DROPPED
);
2928 static inline void dev_consume_skb_irq(struct sk_buff
*skb
)
2930 __dev_kfree_skb_irq(skb
, SKB_REASON_CONSUMED
);
2933 static inline void dev_kfree_skb_any(struct sk_buff
*skb
)
2935 __dev_kfree_skb_any(skb
, SKB_REASON_DROPPED
);
2938 static inline void dev_consume_skb_any(struct sk_buff
*skb
)
2940 __dev_kfree_skb_any(skb
, SKB_REASON_CONSUMED
);
2943 int netif_rx(struct sk_buff
*skb
);
2944 int netif_rx_ni(struct sk_buff
*skb
);
2945 int netif_receive_skb(struct sk_buff
*skb
);
2946 gro_result_t
napi_gro_receive(struct napi_struct
*napi
, struct sk_buff
*skb
);
2947 void napi_gro_flush(struct napi_struct
*napi
, bool flush_old
);
2948 struct sk_buff
*napi_get_frags(struct napi_struct
*napi
);
2949 gro_result_t
napi_gro_frags(struct napi_struct
*napi
);
2950 struct packet_offload
*gro_find_receive_by_type(__be16 type
);
2951 struct packet_offload
*gro_find_complete_by_type(__be16 type
);
2953 static inline void napi_free_frags(struct napi_struct
*napi
)
2955 kfree_skb(napi
->skb
);
2959 int netdev_rx_handler_register(struct net_device
*dev
,
2960 rx_handler_func_t
*rx_handler
,
2961 void *rx_handler_data
);
2962 void netdev_rx_handler_unregister(struct net_device
*dev
);
2964 bool dev_valid_name(const char *name
);
2965 int dev_ioctl(struct net
*net
, unsigned int cmd
, void __user
*);
2966 int dev_ethtool(struct net
*net
, struct ifreq
*);
2967 unsigned int dev_get_flags(const struct net_device
*);
2968 int __dev_change_flags(struct net_device
*, unsigned int flags
);
2969 int dev_change_flags(struct net_device
*, unsigned int);
2970 void __dev_notify_flags(struct net_device
*, unsigned int old_flags
,
2971 unsigned int gchanges
);
2972 int dev_change_name(struct net_device
*, const char *);
2973 int dev_set_alias(struct net_device
*, const char *, size_t);
2974 int dev_change_net_namespace(struct net_device
*, struct net
*, const char *);
2975 int dev_set_mtu(struct net_device
*, int);
2976 void dev_set_group(struct net_device
*, int);
2977 int dev_set_mac_address(struct net_device
*, struct sockaddr
*);
2978 int dev_change_carrier(struct net_device
*, bool new_carrier
);
2979 int dev_get_phys_port_id(struct net_device
*dev
,
2980 struct netdev_phys_item_id
*ppid
);
2981 struct sk_buff
*validate_xmit_skb_list(struct sk_buff
*skb
, struct net_device
*dev
);
2982 struct sk_buff
*dev_hard_start_xmit(struct sk_buff
*skb
, struct net_device
*dev
,
2983 struct netdev_queue
*txq
, int *ret
);
2984 int __dev_forward_skb(struct net_device
*dev
, struct sk_buff
*skb
);
2985 int dev_forward_skb(struct net_device
*dev
, struct sk_buff
*skb
);
2986 bool is_skb_forwardable(struct net_device
*dev
, struct sk_buff
*skb
);
2988 extern int netdev_budget
;
2990 /* Called by rtnetlink.c:rtnl_unlock() */
2991 void netdev_run_todo(void);
2994 * dev_put - release reference to device
2995 * @dev: network device
2997 * Release reference to device to allow it to be freed.
2999 static inline void dev_put(struct net_device
*dev
)
3001 this_cpu_dec(*dev
->pcpu_refcnt
);
3005 * dev_hold - get reference to device
3006 * @dev: network device
3008 * Hold reference to device to keep it from being freed.
3010 static inline void dev_hold(struct net_device
*dev
)
3012 this_cpu_inc(*dev
->pcpu_refcnt
);
3015 /* Carrier loss detection, dial on demand. The functions netif_carrier_on
3016 * and _off may be called from IRQ context, but it is caller
3017 * who is responsible for serialization of these calls.
3019 * The name carrier is inappropriate, these functions should really be
3020 * called netif_lowerlayer_*() because they represent the state of any
3021 * kind of lower layer not just hardware media.
3024 void linkwatch_init_dev(struct net_device
*dev
);
3025 void linkwatch_fire_event(struct net_device
*dev
);
3026 void linkwatch_forget_dev(struct net_device
*dev
);
3029 * netif_carrier_ok - test if carrier present
3030 * @dev: network device
3032 * Check if carrier is present on device
3034 static inline bool netif_carrier_ok(const struct net_device
*dev
)
3036 return !test_bit(__LINK_STATE_NOCARRIER
, &dev
->state
);
3039 unsigned long dev_trans_start(struct net_device
*dev
);
3041 void __netdev_watchdog_up(struct net_device
*dev
);
3043 void netif_carrier_on(struct net_device
*dev
);
3045 void netif_carrier_off(struct net_device
*dev
);
3048 * netif_dormant_on - mark device as dormant.
3049 * @dev: network device
3051 * Mark device as dormant (as per RFC2863).
3053 * The dormant state indicates that the relevant interface is not
3054 * actually in a condition to pass packets (i.e., it is not 'up') but is
3055 * in a "pending" state, waiting for some external event. For "on-
3056 * demand" interfaces, this new state identifies the situation where the
3057 * interface is waiting for events to place it in the up state.
3060 static inline void netif_dormant_on(struct net_device
*dev
)
3062 if (!test_and_set_bit(__LINK_STATE_DORMANT
, &dev
->state
))
3063 linkwatch_fire_event(dev
);
3067 * netif_dormant_off - set device as not dormant.
3068 * @dev: network device
3070 * Device is not in dormant state.
3072 static inline void netif_dormant_off(struct net_device
*dev
)
3074 if (test_and_clear_bit(__LINK_STATE_DORMANT
, &dev
->state
))
3075 linkwatch_fire_event(dev
);
3079 * netif_dormant - test if carrier present
3080 * @dev: network device
3082 * Check if carrier is present on device
3084 static inline bool netif_dormant(const struct net_device
*dev
)
3086 return test_bit(__LINK_STATE_DORMANT
, &dev
->state
);
3091 * netif_oper_up - test if device is operational
3092 * @dev: network device
3094 * Check if carrier is operational
3096 static inline bool netif_oper_up(const struct net_device
*dev
)
3098 return (dev
->operstate
== IF_OPER_UP
||
3099 dev
->operstate
== IF_OPER_UNKNOWN
/* backward compat */);
3103 * netif_device_present - is device available or removed
3104 * @dev: network device
3106 * Check if device has not been removed from system.
3108 static inline bool netif_device_present(struct net_device
*dev
)
3110 return test_bit(__LINK_STATE_PRESENT
, &dev
->state
);
3113 void netif_device_detach(struct net_device
*dev
);
3115 void netif_device_attach(struct net_device
*dev
);
3118 * Network interface message level settings
3122 NETIF_MSG_DRV
= 0x0001,
3123 NETIF_MSG_PROBE
= 0x0002,
3124 NETIF_MSG_LINK
= 0x0004,
3125 NETIF_MSG_TIMER
= 0x0008,
3126 NETIF_MSG_IFDOWN
= 0x0010,
3127 NETIF_MSG_IFUP
= 0x0020,
3128 NETIF_MSG_RX_ERR
= 0x0040,
3129 NETIF_MSG_TX_ERR
= 0x0080,
3130 NETIF_MSG_TX_QUEUED
= 0x0100,
3131 NETIF_MSG_INTR
= 0x0200,
3132 NETIF_MSG_TX_DONE
= 0x0400,
3133 NETIF_MSG_RX_STATUS
= 0x0800,
3134 NETIF_MSG_PKTDATA
= 0x1000,
3135 NETIF_MSG_HW
= 0x2000,
3136 NETIF_MSG_WOL
= 0x4000,
3139 #define netif_msg_drv(p) ((p)->msg_enable & NETIF_MSG_DRV)
3140 #define netif_msg_probe(p) ((p)->msg_enable & NETIF_MSG_PROBE)
3141 #define netif_msg_link(p) ((p)->msg_enable & NETIF_MSG_LINK)
3142 #define netif_msg_timer(p) ((p)->msg_enable & NETIF_MSG_TIMER)
3143 #define netif_msg_ifdown(p) ((p)->msg_enable & NETIF_MSG_IFDOWN)
3144 #define netif_msg_ifup(p) ((p)->msg_enable & NETIF_MSG_IFUP)
3145 #define netif_msg_rx_err(p) ((p)->msg_enable & NETIF_MSG_RX_ERR)
3146 #define netif_msg_tx_err(p) ((p)->msg_enable & NETIF_MSG_TX_ERR)
3147 #define netif_msg_tx_queued(p) ((p)->msg_enable & NETIF_MSG_TX_QUEUED)
3148 #define netif_msg_intr(p) ((p)->msg_enable & NETIF_MSG_INTR)
3149 #define netif_msg_tx_done(p) ((p)->msg_enable & NETIF_MSG_TX_DONE)
3150 #define netif_msg_rx_status(p) ((p)->msg_enable & NETIF_MSG_RX_STATUS)
3151 #define netif_msg_pktdata(p) ((p)->msg_enable & NETIF_MSG_PKTDATA)
3152 #define netif_msg_hw(p) ((p)->msg_enable & NETIF_MSG_HW)
3153 #define netif_msg_wol(p) ((p)->msg_enable & NETIF_MSG_WOL)
3155 static inline u32
netif_msg_init(int debug_value
, int default_msg_enable_bits
)
3158 if (debug_value
< 0 || debug_value
>= (sizeof(u32
) * 8))
3159 return default_msg_enable_bits
;
3160 if (debug_value
== 0) /* no output */
3162 /* set low N bits */
3163 return (1 << debug_value
) - 1;
3166 static inline void __netif_tx_lock(struct netdev_queue
*txq
, int cpu
)
3168 spin_lock(&txq
->_xmit_lock
);
3169 txq
->xmit_lock_owner
= cpu
;
3172 static inline void __netif_tx_lock_bh(struct netdev_queue
*txq
)
3174 spin_lock_bh(&txq
->_xmit_lock
);
3175 txq
->xmit_lock_owner
= smp_processor_id();
3178 static inline bool __netif_tx_trylock(struct netdev_queue
*txq
)
3180 bool ok
= spin_trylock(&txq
->_xmit_lock
);
3182 txq
->xmit_lock_owner
= smp_processor_id();
3186 static inline void __netif_tx_unlock(struct netdev_queue
*txq
)
3188 txq
->xmit_lock_owner
= -1;
3189 spin_unlock(&txq
->_xmit_lock
);
3192 static inline void __netif_tx_unlock_bh(struct netdev_queue
*txq
)
3194 txq
->xmit_lock_owner
= -1;
3195 spin_unlock_bh(&txq
->_xmit_lock
);
3198 static inline void txq_trans_update(struct netdev_queue
*txq
)
3200 if (txq
->xmit_lock_owner
!= -1)
3201 txq
->trans_start
= jiffies
;
3205 * netif_tx_lock - grab network device transmit lock
3206 * @dev: network device
3208 * Get network device transmit lock
3210 static inline void netif_tx_lock(struct net_device
*dev
)
3215 spin_lock(&dev
->tx_global_lock
);
3216 cpu
= smp_processor_id();
3217 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
3218 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
3220 /* We are the only thread of execution doing a
3221 * freeze, but we have to grab the _xmit_lock in
3222 * order to synchronize with threads which are in
3223 * the ->hard_start_xmit() handler and already
3224 * checked the frozen bit.
3226 __netif_tx_lock(txq
, cpu
);
3227 set_bit(__QUEUE_STATE_FROZEN
, &txq
->state
);
3228 __netif_tx_unlock(txq
);
3232 static inline void netif_tx_lock_bh(struct net_device
*dev
)
3238 static inline void netif_tx_unlock(struct net_device
*dev
)
3242 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
3243 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
3245 /* No need to grab the _xmit_lock here. If the
3246 * queue is not stopped for another reason, we
3249 clear_bit(__QUEUE_STATE_FROZEN
, &txq
->state
);
3250 netif_schedule_queue(txq
);
3252 spin_unlock(&dev
->tx_global_lock
);
3255 static inline void netif_tx_unlock_bh(struct net_device
*dev
)
3257 netif_tx_unlock(dev
);
3261 #define HARD_TX_LOCK(dev, txq, cpu) { \
3262 if ((dev->features & NETIF_F_LLTX) == 0) { \
3263 __netif_tx_lock(txq, cpu); \
3267 #define HARD_TX_TRYLOCK(dev, txq) \
3268 (((dev->features & NETIF_F_LLTX) == 0) ? \
3269 __netif_tx_trylock(txq) : \
3272 #define HARD_TX_UNLOCK(dev, txq) { \
3273 if ((dev->features & NETIF_F_LLTX) == 0) { \
3274 __netif_tx_unlock(txq); \
3278 static inline void netif_tx_disable(struct net_device
*dev
)
3284 cpu
= smp_processor_id();
3285 for (i
= 0; i
< dev
->num_tx_queues
; i
++) {
3286 struct netdev_queue
*txq
= netdev_get_tx_queue(dev
, i
);
3288 __netif_tx_lock(txq
, cpu
);
3289 netif_tx_stop_queue(txq
);
3290 __netif_tx_unlock(txq
);
3295 static inline void netif_addr_lock(struct net_device
*dev
)
3297 spin_lock(&dev
->addr_list_lock
);
3300 static inline void netif_addr_lock_nested(struct net_device
*dev
)
3302 int subclass
= SINGLE_DEPTH_NESTING
;
3304 if (dev
->netdev_ops
->ndo_get_lock_subclass
)
3305 subclass
= dev
->netdev_ops
->ndo_get_lock_subclass(dev
);
3307 spin_lock_nested(&dev
->addr_list_lock
, subclass
);
3310 static inline void netif_addr_lock_bh(struct net_device
*dev
)
3312 spin_lock_bh(&dev
->addr_list_lock
);
3315 static inline void netif_addr_unlock(struct net_device
*dev
)
3317 spin_unlock(&dev
->addr_list_lock
);
3320 static inline void netif_addr_unlock_bh(struct net_device
*dev
)
3322 spin_unlock_bh(&dev
->addr_list_lock
);
3326 * dev_addrs walker. Should be used only for read access. Call with
3327 * rcu_read_lock held.
3329 #define for_each_dev_addr(dev, ha) \
3330 list_for_each_entry_rcu(ha, &dev->dev_addrs.list, list)
3332 /* These functions live elsewhere (drivers/net/net_init.c, but related) */
3334 void ether_setup(struct net_device
*dev
);
3336 /* Support for loadable net-drivers */
3337 struct net_device
*alloc_netdev_mqs(int sizeof_priv
, const char *name
,
3338 unsigned char name_assign_type
,
3339 void (*setup
)(struct net_device
*),
3340 unsigned int txqs
, unsigned int rxqs
);
3341 #define alloc_netdev(sizeof_priv, name, name_assign_type, setup) \
3342 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, 1, 1)
3344 #define alloc_netdev_mq(sizeof_priv, name, name_assign_type, setup, count) \
3345 alloc_netdev_mqs(sizeof_priv, name, name_assign_type, setup, count, \
3348 int register_netdev(struct net_device
*dev
);
3349 void unregister_netdev(struct net_device
*dev
);
3351 /* General hardware address lists handling functions */
3352 int __hw_addr_sync(struct netdev_hw_addr_list
*to_list
,
3353 struct netdev_hw_addr_list
*from_list
, int addr_len
);
3354 void __hw_addr_unsync(struct netdev_hw_addr_list
*to_list
,
3355 struct netdev_hw_addr_list
*from_list
, int addr_len
);
3356 int __hw_addr_sync_dev(struct netdev_hw_addr_list
*list
,
3357 struct net_device
*dev
,
3358 int (*sync
)(struct net_device
*, const unsigned char *),
3359 int (*unsync
)(struct net_device
*,
3360 const unsigned char *));
3361 void __hw_addr_unsync_dev(struct netdev_hw_addr_list
*list
,
3362 struct net_device
*dev
,
3363 int (*unsync
)(struct net_device
*,
3364 const unsigned char *));
3365 void __hw_addr_init(struct netdev_hw_addr_list
*list
);
3367 /* Functions used for device addresses handling */
3368 int dev_addr_add(struct net_device
*dev
, const unsigned char *addr
,
3369 unsigned char addr_type
);
3370 int dev_addr_del(struct net_device
*dev
, const unsigned char *addr
,
3371 unsigned char addr_type
);
3372 void dev_addr_flush(struct net_device
*dev
);
3373 int dev_addr_init(struct net_device
*dev
);
3375 /* Functions used for unicast addresses handling */
3376 int dev_uc_add(struct net_device
*dev
, const unsigned char *addr
);
3377 int dev_uc_add_excl(struct net_device
*dev
, const unsigned char *addr
);
3378 int dev_uc_del(struct net_device
*dev
, const unsigned char *addr
);
3379 int dev_uc_sync(struct net_device
*to
, struct net_device
*from
);
3380 int dev_uc_sync_multiple(struct net_device
*to
, struct net_device
*from
);
3381 void dev_uc_unsync(struct net_device
*to
, struct net_device
*from
);
3382 void dev_uc_flush(struct net_device
*dev
);
3383 void dev_uc_init(struct net_device
*dev
);
3386 * __dev_uc_sync - Synchonize device's unicast list
3387 * @dev: device to sync
3388 * @sync: function to call if address should be added
3389 * @unsync: function to call if address should be removed
3391 * Add newly added addresses to the interface, and release
3392 * addresses that have been deleted.
3394 static inline int __dev_uc_sync(struct net_device
*dev
,
3395 int (*sync
)(struct net_device
*,
3396 const unsigned char *),
3397 int (*unsync
)(struct net_device
*,
3398 const unsigned char *))
3400 return __hw_addr_sync_dev(&dev
->uc
, dev
, sync
, unsync
);
3404 * __dev_uc_unsync - Remove synchronized addresses from device
3405 * @dev: device to sync
3406 * @unsync: function to call if address should be removed
3408 * Remove all addresses that were added to the device by dev_uc_sync().
3410 static inline void __dev_uc_unsync(struct net_device
*dev
,
3411 int (*unsync
)(struct net_device
*,
3412 const unsigned char *))
3414 __hw_addr_unsync_dev(&dev
->uc
, dev
, unsync
);
3417 /* Functions used for multicast addresses handling */
3418 int dev_mc_add(struct net_device
*dev
, const unsigned char *addr
);
3419 int dev_mc_add_global(struct net_device
*dev
, const unsigned char *addr
);
3420 int dev_mc_add_excl(struct net_device
*dev
, const unsigned char *addr
);
3421 int dev_mc_del(struct net_device
*dev
, const unsigned char *addr
);
3422 int dev_mc_del_global(struct net_device
*dev
, const unsigned char *addr
);
3423 int dev_mc_sync(struct net_device
*to
, struct net_device
*from
);
3424 int dev_mc_sync_multiple(struct net_device
*to
, struct net_device
*from
);
3425 void dev_mc_unsync(struct net_device
*to
, struct net_device
*from
);
3426 void dev_mc_flush(struct net_device
*dev
);
3427 void dev_mc_init(struct net_device
*dev
);
3430 * __dev_mc_sync - Synchonize device's multicast list
3431 * @dev: device to sync
3432 * @sync: function to call if address should be added
3433 * @unsync: function to call if address should be removed
3435 * Add newly added addresses to the interface, and release
3436 * addresses that have been deleted.
3438 static inline int __dev_mc_sync(struct net_device
*dev
,
3439 int (*sync
)(struct net_device
*,
3440 const unsigned char *),
3441 int (*unsync
)(struct net_device
*,
3442 const unsigned char *))
3444 return __hw_addr_sync_dev(&dev
->mc
, dev
, sync
, unsync
);
3448 * __dev_mc_unsync - Remove synchronized addresses from device
3449 * @dev: device to sync
3450 * @unsync: function to call if address should be removed
3452 * Remove all addresses that were added to the device by dev_mc_sync().
3454 static inline void __dev_mc_unsync(struct net_device
*dev
,
3455 int (*unsync
)(struct net_device
*,
3456 const unsigned char *))
3458 __hw_addr_unsync_dev(&dev
->mc
, dev
, unsync
);
3461 /* Functions used for secondary unicast and multicast support */
3462 void dev_set_rx_mode(struct net_device
*dev
);
3463 void __dev_set_rx_mode(struct net_device
*dev
);
3464 int dev_set_promiscuity(struct net_device
*dev
, int inc
);
3465 int dev_set_allmulti(struct net_device
*dev
, int inc
);
3466 void netdev_state_change(struct net_device
*dev
);
3467 void netdev_notify_peers(struct net_device
*dev
);
3468 void netdev_features_change(struct net_device
*dev
);
3469 /* Load a device via the kmod */
3470 void dev_load(struct net
*net
, const char *name
);
3471 struct rtnl_link_stats64
*dev_get_stats(struct net_device
*dev
,
3472 struct rtnl_link_stats64
*storage
);
3473 void netdev_stats_to_stats64(struct rtnl_link_stats64
*stats64
,
3474 const struct net_device_stats
*netdev_stats
);
3476 extern int netdev_max_backlog
;
3477 extern int netdev_tstamp_prequeue
;
3478 extern int weight_p
;
3479 extern int bpf_jit_enable
;
3481 bool netdev_has_upper_dev(struct net_device
*dev
, struct net_device
*upper_dev
);
3482 struct net_device
*netdev_upper_get_next_dev_rcu(struct net_device
*dev
,
3483 struct list_head
**iter
);
3484 struct net_device
*netdev_all_upper_get_next_dev_rcu(struct net_device
*dev
,
3485 struct list_head
**iter
);
3487 /* iterate through upper list, must be called under RCU read lock */
3488 #define netdev_for_each_upper_dev_rcu(dev, updev, iter) \
3489 for (iter = &(dev)->adj_list.upper, \
3490 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)); \
3492 updev = netdev_upper_get_next_dev_rcu(dev, &(iter)))
3494 /* iterate through upper list, must be called under RCU read lock */
3495 #define netdev_for_each_all_upper_dev_rcu(dev, updev, iter) \
3496 for (iter = &(dev)->all_adj_list.upper, \
3497 updev = netdev_all_upper_get_next_dev_rcu(dev, &(iter)); \
3499 updev = netdev_all_upper_get_next_dev_rcu(dev, &(iter)))
3501 void *netdev_lower_get_next_private(struct net_device
*dev
,
3502 struct list_head
**iter
);
3503 void *netdev_lower_get_next_private_rcu(struct net_device
*dev
,
3504 struct list_head
**iter
);
3506 #define netdev_for_each_lower_private(dev, priv, iter) \
3507 for (iter = (dev)->adj_list.lower.next, \
3508 priv = netdev_lower_get_next_private(dev, &(iter)); \
3510 priv = netdev_lower_get_next_private(dev, &(iter)))
3512 #define netdev_for_each_lower_private_rcu(dev, priv, iter) \
3513 for (iter = &(dev)->adj_list.lower, \
3514 priv = netdev_lower_get_next_private_rcu(dev, &(iter)); \
3516 priv = netdev_lower_get_next_private_rcu(dev, &(iter)))
3518 void *netdev_lower_get_next(struct net_device
*dev
,
3519 struct list_head
**iter
);
3520 #define netdev_for_each_lower_dev(dev, ldev, iter) \
3521 for (iter = &(dev)->adj_list.lower, \
3522 ldev = netdev_lower_get_next(dev, &(iter)); \
3524 ldev = netdev_lower_get_next(dev, &(iter)))
3526 void *netdev_adjacent_get_private(struct list_head
*adj_list
);
3527 void *netdev_lower_get_first_private_rcu(struct net_device
*dev
);
3528 struct net_device
*netdev_master_upper_dev_get(struct net_device
*dev
);
3529 struct net_device
*netdev_master_upper_dev_get_rcu(struct net_device
*dev
);
3530 int netdev_upper_dev_link(struct net_device
*dev
, struct net_device
*upper_dev
);
3531 int netdev_master_upper_dev_link(struct net_device
*dev
,
3532 struct net_device
*upper_dev
);
3533 int netdev_master_upper_dev_link_private(struct net_device
*dev
,
3534 struct net_device
*upper_dev
,
3536 void netdev_upper_dev_unlink(struct net_device
*dev
,
3537 struct net_device
*upper_dev
);
3538 void netdev_adjacent_rename_links(struct net_device
*dev
, char *oldname
);
3539 void *netdev_lower_dev_get_private(struct net_device
*dev
,
3540 struct net_device
*lower_dev
);
3542 /* RSS keys are 40 or 52 bytes long */
3543 #define NETDEV_RSS_KEY_LEN 52
3544 extern u8 netdev_rss_key
[NETDEV_RSS_KEY_LEN
];
3545 void netdev_rss_key_fill(void *buffer
, size_t len
);
3547 int dev_get_nest_level(struct net_device
*dev
,
3548 bool (*type_check
)(struct net_device
*dev
));
3549 int skb_checksum_help(struct sk_buff
*skb
);
3550 struct sk_buff
*__skb_gso_segment(struct sk_buff
*skb
,
3551 netdev_features_t features
, bool tx_path
);
3552 struct sk_buff
*skb_mac_gso_segment(struct sk_buff
*skb
,
3553 netdev_features_t features
);
3555 struct netdev_bonding_info
{
3560 struct netdev_notifier_bonding_info
{
3561 struct netdev_notifier_info info
; /* must be first */
3562 struct netdev_bonding_info bonding_info
;
3565 void netdev_bonding_info_change(struct net_device
*dev
,
3566 struct netdev_bonding_info
*bonding_info
);
3569 struct sk_buff
*skb_gso_segment(struct sk_buff
*skb
, netdev_features_t features
)
3571 return __skb_gso_segment(skb
, features
, true);
3573 __be16
skb_network_protocol(struct sk_buff
*skb
, int *depth
);
3575 static inline bool can_checksum_protocol(netdev_features_t features
,
3578 return ((features
& NETIF_F_GEN_CSUM
) ||
3579 ((features
& NETIF_F_V4_CSUM
) &&
3580 protocol
== htons(ETH_P_IP
)) ||
3581 ((features
& NETIF_F_V6_CSUM
) &&
3582 protocol
== htons(ETH_P_IPV6
)) ||
3583 ((features
& NETIF_F_FCOE_CRC
) &&
3584 protocol
== htons(ETH_P_FCOE
)));
3588 void netdev_rx_csum_fault(struct net_device
*dev
);
3590 static inline void netdev_rx_csum_fault(struct net_device
*dev
)
3594 /* rx skb timestamps */
3595 void net_enable_timestamp(void);
3596 void net_disable_timestamp(void);
3598 #ifdef CONFIG_PROC_FS
3599 int __init
dev_proc_init(void);
3601 #define dev_proc_init() 0
3604 static inline netdev_tx_t
__netdev_start_xmit(const struct net_device_ops
*ops
,
3605 struct sk_buff
*skb
, struct net_device
*dev
,
3608 skb
->xmit_more
= more
? 1 : 0;
3609 return ops
->ndo_start_xmit(skb
, dev
);
3612 static inline netdev_tx_t
netdev_start_xmit(struct sk_buff
*skb
, struct net_device
*dev
,
3613 struct netdev_queue
*txq
, bool more
)
3615 const struct net_device_ops
*ops
= dev
->netdev_ops
;
3618 rc
= __netdev_start_xmit(ops
, skb
, dev
, more
);
3619 if (rc
== NETDEV_TX_OK
)
3620 txq_trans_update(txq
);
3625 int netdev_class_create_file_ns(struct class_attribute
*class_attr
,
3627 void netdev_class_remove_file_ns(struct class_attribute
*class_attr
,
3630 static inline int netdev_class_create_file(struct class_attribute
*class_attr
)
3632 return netdev_class_create_file_ns(class_attr
, NULL
);
3635 static inline void netdev_class_remove_file(struct class_attribute
*class_attr
)
3637 netdev_class_remove_file_ns(class_attr
, NULL
);
3640 extern struct kobj_ns_type_operations net_ns_type_operations
;
3642 const char *netdev_drivername(const struct net_device
*dev
);
3644 void linkwatch_run_queue(void);
3646 static inline netdev_features_t
netdev_intersect_features(netdev_features_t f1
,
3647 netdev_features_t f2
)
3649 if (f1
& NETIF_F_GEN_CSUM
)
3650 f1
|= (NETIF_F_ALL_CSUM
& ~NETIF_F_GEN_CSUM
);
3651 if (f2
& NETIF_F_GEN_CSUM
)
3652 f2
|= (NETIF_F_ALL_CSUM
& ~NETIF_F_GEN_CSUM
);
3654 if (f1
& NETIF_F_GEN_CSUM
)
3655 f1
&= ~(NETIF_F_ALL_CSUM
& ~NETIF_F_GEN_CSUM
);
3660 static inline netdev_features_t
netdev_get_wanted_features(
3661 struct net_device
*dev
)
3663 return (dev
->features
& ~dev
->hw_features
) | dev
->wanted_features
;
3665 netdev_features_t
netdev_increment_features(netdev_features_t all
,
3666 netdev_features_t one
, netdev_features_t mask
);
3668 /* Allow TSO being used on stacked device :
3669 * Performing the GSO segmentation before last device
3670 * is a performance improvement.
3672 static inline netdev_features_t
netdev_add_tso_features(netdev_features_t features
,
3673 netdev_features_t mask
)
3675 return netdev_increment_features(features
, NETIF_F_ALL_TSO
, mask
);
3678 int __netdev_update_features(struct net_device
*dev
);
3679 void netdev_update_features(struct net_device
*dev
);
3680 void netdev_change_features(struct net_device
*dev
);
3682 void netif_stacked_transfer_operstate(const struct net_device
*rootdev
,
3683 struct net_device
*dev
);
3685 netdev_features_t
netif_skb_features(struct sk_buff
*skb
);
3687 static inline bool net_gso_ok(netdev_features_t features
, int gso_type
)
3689 netdev_features_t feature
= gso_type
<< NETIF_F_GSO_SHIFT
;
3691 /* check flags correspondence */
3692 BUILD_BUG_ON(SKB_GSO_TCPV4
!= (NETIF_F_TSO
>> NETIF_F_GSO_SHIFT
));
3693 BUILD_BUG_ON(SKB_GSO_UDP
!= (NETIF_F_UFO
>> NETIF_F_GSO_SHIFT
));
3694 BUILD_BUG_ON(SKB_GSO_DODGY
!= (NETIF_F_GSO_ROBUST
>> NETIF_F_GSO_SHIFT
));
3695 BUILD_BUG_ON(SKB_GSO_TCP_ECN
!= (NETIF_F_TSO_ECN
>> NETIF_F_GSO_SHIFT
));
3696 BUILD_BUG_ON(SKB_GSO_TCPV6
!= (NETIF_F_TSO6
>> NETIF_F_GSO_SHIFT
));
3697 BUILD_BUG_ON(SKB_GSO_FCOE
!= (NETIF_F_FSO
>> NETIF_F_GSO_SHIFT
));
3698 BUILD_BUG_ON(SKB_GSO_GRE
!= (NETIF_F_GSO_GRE
>> NETIF_F_GSO_SHIFT
));
3699 BUILD_BUG_ON(SKB_GSO_GRE_CSUM
!= (NETIF_F_GSO_GRE_CSUM
>> NETIF_F_GSO_SHIFT
));
3700 BUILD_BUG_ON(SKB_GSO_IPIP
!= (NETIF_F_GSO_IPIP
>> NETIF_F_GSO_SHIFT
));
3701 BUILD_BUG_ON(SKB_GSO_SIT
!= (NETIF_F_GSO_SIT
>> NETIF_F_GSO_SHIFT
));
3702 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL
!= (NETIF_F_GSO_UDP_TUNNEL
>> NETIF_F_GSO_SHIFT
));
3703 BUILD_BUG_ON(SKB_GSO_UDP_TUNNEL_CSUM
!= (NETIF_F_GSO_UDP_TUNNEL_CSUM
>> NETIF_F_GSO_SHIFT
));
3704 BUILD_BUG_ON(SKB_GSO_TUNNEL_REMCSUM
!= (NETIF_F_GSO_TUNNEL_REMCSUM
>> NETIF_F_GSO_SHIFT
));
3706 return (features
& feature
) == feature
;
3709 static inline bool skb_gso_ok(struct sk_buff
*skb
, netdev_features_t features
)
3711 return net_gso_ok(features
, skb_shinfo(skb
)->gso_type
) &&
3712 (!skb_has_frag_list(skb
) || (features
& NETIF_F_FRAGLIST
));
3715 static inline bool netif_needs_gso(struct net_device
*dev
, struct sk_buff
*skb
,
3716 netdev_features_t features
)
3718 return skb_is_gso(skb
) && (!skb_gso_ok(skb
, features
) ||
3719 unlikely((skb
->ip_summed
!= CHECKSUM_PARTIAL
) &&
3720 (skb
->ip_summed
!= CHECKSUM_UNNECESSARY
)));
3723 static inline void netif_set_gso_max_size(struct net_device
*dev
,
3726 dev
->gso_max_size
= size
;
3729 static inline void skb_gso_error_unwind(struct sk_buff
*skb
, __be16 protocol
,
3730 int pulled_hlen
, u16 mac_offset
,
3733 skb
->protocol
= protocol
;
3734 skb
->encapsulation
= 1;
3735 skb_push(skb
, pulled_hlen
);
3736 skb_reset_transport_header(skb
);
3737 skb
->mac_header
= mac_offset
;
3738 skb
->network_header
= skb
->mac_header
+ mac_len
;
3739 skb
->mac_len
= mac_len
;
3742 static inline bool netif_is_macvlan(struct net_device
*dev
)
3744 return dev
->priv_flags
& IFF_MACVLAN
;
3747 static inline bool netif_is_macvlan_port(struct net_device
*dev
)
3749 return dev
->priv_flags
& IFF_MACVLAN_PORT
;
3752 static inline bool netif_is_ipvlan(struct net_device
*dev
)
3754 return dev
->priv_flags
& IFF_IPVLAN_SLAVE
;
3757 static inline bool netif_is_ipvlan_port(struct net_device
*dev
)
3759 return dev
->priv_flags
& IFF_IPVLAN_MASTER
;
3762 static inline bool netif_is_bond_master(struct net_device
*dev
)
3764 return dev
->flags
& IFF_MASTER
&& dev
->priv_flags
& IFF_BONDING
;
3767 static inline bool netif_is_bond_slave(struct net_device
*dev
)
3769 return dev
->flags
& IFF_SLAVE
&& dev
->priv_flags
& IFF_BONDING
;
3772 static inline bool netif_supports_nofcs(struct net_device
*dev
)
3774 return dev
->priv_flags
& IFF_SUPP_NOFCS
;
3777 /* This device needs to keep skb dst for qdisc enqueue or ndo_start_xmit() */
3778 static inline void netif_keep_dst(struct net_device
*dev
)
3780 dev
->priv_flags
&= ~(IFF_XMIT_DST_RELEASE
| IFF_XMIT_DST_RELEASE_PERM
);
3783 extern struct pernet_operations __net_initdata loopback_net_ops
;
3785 /* Logging, debugging and troubleshooting/diagnostic helpers. */
3787 /* netdev_printk helpers, similar to dev_printk */
3789 static inline const char *netdev_name(const struct net_device
*dev
)
3791 if (!dev
->name
[0] || strchr(dev
->name
, '%'))
3792 return "(unnamed net_device)";
3796 static inline const char *netdev_reg_state(const struct net_device
*dev
)
3798 switch (dev
->reg_state
) {
3799 case NETREG_UNINITIALIZED
: return " (uninitialized)";
3800 case NETREG_REGISTERED
: return "";
3801 case NETREG_UNREGISTERING
: return " (unregistering)";
3802 case NETREG_UNREGISTERED
: return " (unregistered)";
3803 case NETREG_RELEASED
: return " (released)";
3804 case NETREG_DUMMY
: return " (dummy)";
3807 WARN_ONCE(1, "%s: unknown reg_state %d\n", dev
->name
, dev
->reg_state
);
3808 return " (unknown)";
3812 void netdev_printk(const char *level
, const struct net_device
*dev
,
3813 const char *format
, ...);
3815 void netdev_emerg(const struct net_device
*dev
, const char *format
, ...);
3817 void netdev_alert(const struct net_device
*dev
, const char *format
, ...);
3819 void netdev_crit(const struct net_device
*dev
, const char *format
, ...);
3821 void netdev_err(const struct net_device
*dev
, const char *format
, ...);
3823 void netdev_warn(const struct net_device
*dev
, const char *format
, ...);
3825 void netdev_notice(const struct net_device
*dev
, const char *format
, ...);
3827 void netdev_info(const struct net_device
*dev
, const char *format
, ...);
3829 #define MODULE_ALIAS_NETDEV(device) \
3830 MODULE_ALIAS("netdev-" device)
3832 #if defined(CONFIG_DYNAMIC_DEBUG)
3833 #define netdev_dbg(__dev, format, args...) \
3835 dynamic_netdev_dbg(__dev, format, ##args); \
3837 #elif defined(DEBUG)
3838 #define netdev_dbg(__dev, format, args...) \
3839 netdev_printk(KERN_DEBUG, __dev, format, ##args)
3841 #define netdev_dbg(__dev, format, args...) \
3844 netdev_printk(KERN_DEBUG, __dev, format, ##args); \
3848 #if defined(VERBOSE_DEBUG)
3849 #define netdev_vdbg netdev_dbg
3852 #define netdev_vdbg(dev, format, args...) \
3855 netdev_printk(KERN_DEBUG, dev, format, ##args); \
3861 * netdev_WARN() acts like dev_printk(), but with the key difference
3862 * of using a WARN/WARN_ON to get the message out, including the
3863 * file/line information and a backtrace.
3865 #define netdev_WARN(dev, format, args...) \
3866 WARN(1, "netdevice: %s%s\n" format, netdev_name(dev), \
3867 netdev_reg_state(dev), ##args)
3869 /* netif printk helpers, similar to netdev_printk */
3871 #define netif_printk(priv, type, level, dev, fmt, args...) \
3873 if (netif_msg_##type(priv)) \
3874 netdev_printk(level, (dev), fmt, ##args); \
3877 #define netif_level(level, priv, type, dev, fmt, args...) \
3879 if (netif_msg_##type(priv)) \
3880 netdev_##level(dev, fmt, ##args); \
3883 #define netif_emerg(priv, type, dev, fmt, args...) \
3884 netif_level(emerg, priv, type, dev, fmt, ##args)
3885 #define netif_alert(priv, type, dev, fmt, args...) \
3886 netif_level(alert, priv, type, dev, fmt, ##args)
3887 #define netif_crit(priv, type, dev, fmt, args...) \
3888 netif_level(crit, priv, type, dev, fmt, ##args)
3889 #define netif_err(priv, type, dev, fmt, args...) \
3890 netif_level(err, priv, type, dev, fmt, ##args)
3891 #define netif_warn(priv, type, dev, fmt, args...) \
3892 netif_level(warn, priv, type, dev, fmt, ##args)
3893 #define netif_notice(priv, type, dev, fmt, args...) \
3894 netif_level(notice, priv, type, dev, fmt, ##args)
3895 #define netif_info(priv, type, dev, fmt, args...) \
3896 netif_level(info, priv, type, dev, fmt, ##args)
3898 #if defined(CONFIG_DYNAMIC_DEBUG)
3899 #define netif_dbg(priv, type, netdev, format, args...) \
3901 if (netif_msg_##type(priv)) \
3902 dynamic_netdev_dbg(netdev, format, ##args); \
3904 #elif defined(DEBUG)
3905 #define netif_dbg(priv, type, dev, format, args...) \
3906 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args)
3908 #define netif_dbg(priv, type, dev, format, args...) \
3911 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
3916 #if defined(VERBOSE_DEBUG)
3917 #define netif_vdbg netif_dbg
3919 #define netif_vdbg(priv, type, dev, format, args...) \
3922 netif_printk(priv, type, KERN_DEBUG, dev, format, ##args); \
3928 * The list of packet types we will receive (as opposed to discard)
3929 * and the routines to invoke.
3931 * Why 16. Because with 16 the only overlap we get on a hash of the
3932 * low nibble of the protocol value is RARP/SNAP/X.25.
3934 * NOTE: That is no longer true with the addition of VLAN tags. Not
3935 * sure which should go first, but I bet it won't make much
3936 * difference if we are running VLANs. The good news is that
3937 * this protocol won't be in the list unless compiled in, so
3938 * the average user (w/out VLANs) will not be adversely affected.
3954 #define PTYPE_HASH_SIZE (16)
3955 #define PTYPE_HASH_MASK (PTYPE_HASH_SIZE - 1)
3957 #endif /* _LINUX_NETDEVICE_H */