xref: /freebsd/sys/net/if_var.h (revision 7f32f0e2275b83b41a8a190e834d28cb2ab0ce96)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 1989, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the University nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  *
31  *	From: @(#)if.h	8.1 (Berkeley) 6/10/93
32  * $FreeBSD$
33  */
34 
35 #ifndef	_NET_IF_VAR_H_
36 #define	_NET_IF_VAR_H_
37 
38 /*
39  * Structures defining a network interface, providing a packet
40  * transport mechanism (ala level 0 of the PUP protocols).
41  *
42  * Each interface accepts output datagrams of a specified maximum
43  * length, and provides higher level routines with input datagrams
44  * received from its medium.
45  *
46  * Output occurs when the routine if_output is called, with three parameters:
47  *	(*ifp->if_output)(ifp, m, dst, rt)
48  * Here m is the mbuf chain to be sent and dst is the destination address.
49  * The output routine encapsulates the supplied datagram if necessary,
50  * and then transmits it on its medium.
51  *
52  * On input, each interface unwraps the data received by it, and either
53  * places it on the input queue of an internetwork datagram routine
54  * and posts the associated software interrupt, or passes the datagram to a raw
55  * packet input routine.
56  *
57  * Routines exist for locating interfaces by their addresses
58  * or for locating an interface on a certain network, as well as more general
59  * routing and gateway routines maintaining information used to locate
60  * interfaces.  These routines live in the files if.c and route.c
61  */
62 
63 struct	rtentry;		/* ifa_rtrequest */
64 struct	rt_addrinfo;		/* ifa_rtrequest */
65 struct	socket;
66 struct	carp_if;
67 struct	carp_softc;
68 struct  ifvlantrunk;
69 struct	route;			/* if_output */
70 struct	vnet;
71 struct	ifmedia;
72 struct	netmap_adapter;
73 struct	netdump_methods;
74 
75 #ifdef _KERNEL
76 #include <sys/_eventhandler.h>
77 #include <sys/mbuf.h>		/* ifqueue only? */
78 #include <sys/buf_ring.h>
79 #include <net/vnet.h>
80 #endif /* _KERNEL */
81 #include <sys/ck.h>
82 #include <sys/counter.h>
83 #include <sys/epoch.h>
84 #include <sys/lock.h>		/* XXX */
85 #include <sys/mutex.h>		/* struct ifqueue */
86 #include <sys/rwlock.h>		/* XXX */
87 #include <sys/sx.h>		/* XXX */
88 #include <sys/_task.h>		/* if_link_task */
89 #define	IF_DUNIT_NONE	-1
90 
91 #include <net/altq/if_altq.h>
92 
93 CK_STAILQ_HEAD(ifnethead, ifnet);	/* we use TAILQs so that the order of */
94 CK_STAILQ_HEAD(ifaddrhead, ifaddr);	/* instantiation is preserved in the list */
95 CK_STAILQ_HEAD(ifmultihead, ifmultiaddr);
96 CK_STAILQ_HEAD(ifgrouphead, ifg_group);
97 
98 #ifdef _KERNEL
99 VNET_DECLARE(struct pfil_head *, link_pfil_head);
100 #define	V_link_pfil_head	VNET(link_pfil_head)
101 #define	PFIL_ETHER_NAME		"ethernet"
102 
103 #define	HHOOK_IPSEC_INET	0
104 #define	HHOOK_IPSEC_INET6	1
105 #define	HHOOK_IPSEC_COUNT	2
106 VNET_DECLARE(struct hhook_head *, ipsec_hhh_in[HHOOK_IPSEC_COUNT]);
107 VNET_DECLARE(struct hhook_head *, ipsec_hhh_out[HHOOK_IPSEC_COUNT]);
108 #define	V_ipsec_hhh_in	VNET(ipsec_hhh_in)
109 #define	V_ipsec_hhh_out	VNET(ipsec_hhh_out)
110 extern epoch_t net_epoch_preempt;
111 extern epoch_t net_epoch;
112 #endif /* _KERNEL */
113 
114 typedef enum {
115 	IFCOUNTER_IPACKETS = 0,
116 	IFCOUNTER_IERRORS,
117 	IFCOUNTER_OPACKETS,
118 	IFCOUNTER_OERRORS,
119 	IFCOUNTER_COLLISIONS,
120 	IFCOUNTER_IBYTES,
121 	IFCOUNTER_OBYTES,
122 	IFCOUNTER_IMCASTS,
123 	IFCOUNTER_OMCASTS,
124 	IFCOUNTER_IQDROPS,
125 	IFCOUNTER_OQDROPS,
126 	IFCOUNTER_NOPROTO,
127 	IFCOUNTERS /* Array size. */
128 } ift_counter;
129 
130 typedef struct ifnet * if_t;
131 
132 typedef	void (*if_start_fn_t)(if_t);
133 typedef	int (*if_ioctl_fn_t)(if_t, u_long, caddr_t);
134 typedef	void (*if_init_fn_t)(void *);
135 typedef void (*if_qflush_fn_t)(if_t);
136 typedef int (*if_transmit_fn_t)(if_t, struct mbuf *);
137 typedef	uint64_t (*if_get_counter_t)(if_t, ift_counter);
138 
139 struct ifnet_hw_tsomax {
140 	u_int	tsomaxbytes;	/* TSO total burst length limit in bytes */
141 	u_int	tsomaxsegcount;	/* TSO maximum segment count */
142 	u_int	tsomaxsegsize;	/* TSO maximum segment size in bytes */
143 };
144 
145 /* Interface encap request types */
146 typedef enum {
147 	IFENCAP_LL = 1			/* pre-calculate link-layer header */
148 } ife_type;
149 
150 /*
151  * The structure below allows to request various pre-calculated L2/L3 headers
152  * for different media. Requests varies by type (rtype field).
153  *
154  * IFENCAP_LL type: pre-calculates link header based on address family
155  *   and destination lladdr.
156  *
157  *   Input data fields:
158  *     buf: pointer to destination buffer
159  *     bufsize: buffer size
160  *     flags: IFENCAP_FLAG_BROADCAST if destination is broadcast
161  *     family: address family defined by AF_ constant.
162  *     lladdr: pointer to link-layer address
163  *     lladdr_len: length of link-layer address
164  *     hdata: pointer to L3 header (optional, used for ARP requests).
165  *   Output data fields:
166  *     buf: encap data is stored here
167  *     bufsize: resulting encap length is stored here
168  *     lladdr_off: offset of link-layer address from encap hdr start
169  *     hdata: L3 header may be altered if necessary
170  */
171 
172 struct if_encap_req {
173 	u_char		*buf;		/* Destination buffer (w) */
174 	size_t		bufsize;	/* size of provided buffer (r) */
175 	ife_type	rtype;		/* request type (r) */
176 	uint32_t	flags;		/* Request flags (r) */
177 	int		family;		/* Address family AF_* (r) */
178 	int		lladdr_off;	/* offset from header start (w) */
179 	int		lladdr_len;	/* lladdr length (r) */
180 	char		*lladdr;	/* link-level address pointer (r) */
181 	char		*hdata;		/* Upper layer header data (rw) */
182 };
183 
184 #define	IFENCAP_FLAG_BROADCAST	0x02	/* Destination is broadcast */
185 
186 /*
187  * Network interface send tag support. The storage of "struct
188  * m_snd_tag" comes from the network driver and it is free to allocate
189  * as much additional space as it wants for its own use.
190  */
191 struct m_snd_tag;
192 
193 #define	IF_SND_TAG_TYPE_RATE_LIMIT 0
194 #define	IF_SND_TAG_TYPE_UNLIMITED 1
195 #define	IF_SND_TAG_TYPE_MAX 2
196 
197 struct if_snd_tag_alloc_header {
198 	uint32_t type;		/* send tag type, see IF_SND_TAG_XXX */
199 	uint32_t flowid;	/* mbuf hash value */
200 	uint32_t flowtype;	/* mbuf hash type */
201 };
202 
203 struct if_snd_tag_alloc_rate_limit {
204 	struct if_snd_tag_alloc_header hdr;
205 	uint64_t max_rate;	/* in bytes/s */
206 	uint32_t flags;		/* M_NOWAIT or M_WAITOK */
207 	uint32_t reserved;	/* alignment */
208 };
209 
210 struct if_snd_tag_rate_limit_params {
211 	uint64_t max_rate;	/* in bytes/s */
212 	uint32_t queue_level;	/* 0 (empty) .. 65535 (full) */
213 #define	IF_SND_QUEUE_LEVEL_MIN 0
214 #define	IF_SND_QUEUE_LEVEL_MAX 65535
215 	uint32_t flags;		/* M_NOWAIT or M_WAITOK */
216 };
217 
218 union if_snd_tag_alloc_params {
219 	struct if_snd_tag_alloc_header hdr;
220 	struct if_snd_tag_alloc_rate_limit rate_limit;
221 	struct if_snd_tag_alloc_rate_limit unlimited;
222 };
223 
224 union if_snd_tag_modify_params {
225 	struct if_snd_tag_rate_limit_params rate_limit;
226 	struct if_snd_tag_rate_limit_params unlimited;
227 };
228 
229 union if_snd_tag_query_params {
230 	struct if_snd_tag_rate_limit_params rate_limit;
231 	struct if_snd_tag_rate_limit_params unlimited;
232 };
233 
234 /* Query return flags */
235 #define RT_NOSUPPORT	  0x00000000	/* Not supported */
236 #define RT_IS_INDIRECT    0x00000001	/*
237 					 * Interface like a lagg, select
238 					 * the actual interface for
239 					 * capabilities.
240 					 */
241 #define RT_IS_SELECTABLE  0x00000002	/*
242 					 * No rate table, you select
243 					 * rates and the first
244 					 * number_of_rates are created.
245 					 */
246 #define RT_IS_FIXED_TABLE 0x00000004	/* A fixed table is attached */
247 #define RT_IS_UNUSABLE	  0x00000008	/* It is not usable for this */
248 
249 struct if_ratelimit_query_results {
250 	const uint64_t *rate_table;	/* Pointer to table if present */
251 	uint32_t flags;			/* Flags indicating results */
252 	uint32_t max_flows;		/* Max flows using, 0=unlimited */
253 	uint32_t number_of_rates;	/* How many unique rates can be created */
254 	uint32_t min_segment_burst;	/* The amount the adapter bursts at each send */
255 };
256 
257 typedef int (if_snd_tag_alloc_t)(struct ifnet *, union if_snd_tag_alloc_params *,
258     struct m_snd_tag **);
259 typedef int (if_snd_tag_modify_t)(struct m_snd_tag *, union if_snd_tag_modify_params *);
260 typedef int (if_snd_tag_query_t)(struct m_snd_tag *, union if_snd_tag_query_params *);
261 typedef void (if_snd_tag_free_t)(struct m_snd_tag *);
262 typedef void (if_ratelimit_query_t)(struct ifnet *,
263     struct if_ratelimit_query_results *);
264 
265 
266 /*
267  * Structure defining a network interface.
268  */
269 struct ifnet {
270 	/* General book keeping of interface lists. */
271 	CK_STAILQ_ENTRY(ifnet) if_link; 	/* all struct ifnets are chained (CK_) */
272 	LIST_ENTRY(ifnet) if_clones;	/* interfaces of a cloner */
273 	CK_STAILQ_HEAD(, ifg_list) if_groups; /* linked list of groups per if (CK_) */
274 					/* protected by if_addr_lock */
275 	u_char	if_alloctype;		/* if_type at time of allocation */
276 	uint8_t	if_numa_domain;		/* NUMA domain of device */
277 	/* Driver and protocol specific information that remains stable. */
278 	void	*if_softc;		/* pointer to driver state */
279 	void	*if_llsoftc;		/* link layer softc */
280 	void	*if_l2com;		/* pointer to protocol bits */
281 	const char *if_dname;		/* driver name */
282 	int	if_dunit;		/* unit or IF_DUNIT_NONE */
283 	u_short	if_index;		/* numeric abbreviation for this if  */
284 	short	if_index_reserved;	/* spare space to grow if_index */
285 	char	if_xname[IFNAMSIZ];	/* external name (name + unit) */
286 	char	*if_description;	/* interface description */
287 
288 	/* Variable fields that are touched by the stack and drivers. */
289 	int	if_flags;		/* up/down, broadcast, etc. */
290 	int	if_drv_flags;		/* driver-managed status flags */
291 	int	if_capabilities;	/* interface features & capabilities */
292 	int	if_capenable;		/* enabled features & capabilities */
293 	void	*if_linkmib;		/* link-type-specific MIB data */
294 	size_t	if_linkmiblen;		/* length of above data */
295 	u_int	if_refcount;		/* reference count */
296 
297 	/* These fields are shared with struct if_data. */
298 	uint8_t		if_type;	/* ethernet, tokenring, etc */
299 	uint8_t		if_addrlen;	/* media address length */
300 	uint8_t		if_hdrlen;	/* media header length */
301 	uint8_t		if_link_state;	/* current link state */
302 	uint32_t	if_mtu;		/* maximum transmission unit */
303 	uint32_t	if_metric;	/* routing metric (external only) */
304 	uint64_t	if_baudrate;	/* linespeed */
305 	uint64_t	if_hwassist;	/* HW offload capabilities, see IFCAP */
306 	time_t		if_epoch;	/* uptime at attach or stat reset */
307 	struct timeval	if_lastchange;	/* time of last administrative change */
308 
309 	struct  ifaltq if_snd;		/* output queue (includes altq) */
310 	struct	task if_linktask;	/* task for link change events */
311 
312 	/* Addresses of different protocol families assigned to this if. */
313 	struct mtx if_addr_lock;	/* lock to protect address lists */
314 		/*
315 		 * if_addrhead is the list of all addresses associated to
316 		 * an interface.
317 		 * Some code in the kernel assumes that first element
318 		 * of the list has type AF_LINK, and contains sockaddr_dl
319 		 * addresses which store the link-level address and the name
320 		 * of the interface.
321 		 * However, access to the AF_LINK address through this
322 		 * field is deprecated. Use if_addr or ifaddr_byindex() instead.
323 		 */
324 	struct	ifaddrhead if_addrhead;	/* linked list of addresses per if */
325 	struct	ifmultihead if_multiaddrs; /* multicast addresses configured */
326 	int	if_amcount;		/* number of all-multicast requests */
327 	struct	ifaddr	*if_addr;	/* pointer to link-level address */
328 	void	*if_hw_addr;		/* hardware link-level address */
329 	const u_int8_t *if_broadcastaddr; /* linklevel broadcast bytestring */
330 	struct	mtx if_afdata_lock;
331 	void	*if_afdata[AF_MAX];
332 	int	if_afdata_initialized;
333 
334 	/* Additional features hung off the interface. */
335 	u_int	if_fib;			/* interface FIB */
336 	struct	vnet *if_vnet;		/* pointer to network stack instance */
337 	struct	vnet *if_home_vnet;	/* where this ifnet originates from */
338 	struct  ifvlantrunk *if_vlantrunk; /* pointer to 802.1q data */
339 	struct	bpf_if *if_bpf;		/* packet filter structure */
340 	int	if_pcount;		/* number of promiscuous listeners */
341 	void	*if_bridge;		/* bridge glue */
342 	void	*if_lagg;		/* lagg glue */
343 	void	*if_pf_kif;		/* pf glue */
344 	struct	carp_if *if_carp;	/* carp interface structure */
345 	struct	label *if_label;	/* interface MAC label */
346 	struct	netmap_adapter *if_netmap; /* netmap(4) softc */
347 
348 	/* Various procedures of the layer2 encapsulation and drivers. */
349 	int	(*if_output)		/* output routine (enqueue) */
350 		(struct ifnet *, struct mbuf *, const struct sockaddr *,
351 		     struct route *);
352 	void	(*if_input)		/* input routine (from h/w driver) */
353 		(struct ifnet *, struct mbuf *);
354 	struct mbuf *(*if_bridge_input)(struct ifnet *, struct mbuf *);
355 	int	(*if_bridge_output)(struct ifnet *, struct mbuf *, struct sockaddr *,
356 		    struct rtentry *);
357 	void (*if_bridge_linkstate)(struct ifnet *ifp);
358 	if_start_fn_t	if_start;	/* initiate output routine */
359 	if_ioctl_fn_t	if_ioctl;	/* ioctl routine */
360 	if_init_fn_t	if_init;	/* Init routine */
361 	int	(*if_resolvemulti)	/* validate/resolve multicast */
362 		(struct ifnet *, struct sockaddr **, struct sockaddr *);
363 	if_qflush_fn_t	if_qflush;	/* flush any queue */
364 	if_transmit_fn_t if_transmit;   /* initiate output routine */
365 
366 	void	(*if_reassign)		/* reassign to vnet routine */
367 		(struct ifnet *, struct vnet *, char *);
368 	if_get_counter_t if_get_counter; /* get counter values */
369 	int	(*if_requestencap)	/* make link header from request */
370 		(struct ifnet *, struct if_encap_req *);
371 
372 	/* Statistics. */
373 	counter_u64_t	if_counters[IFCOUNTERS];
374 
375 	/* Stuff that's only temporary and doesn't belong here. */
376 
377 	/*
378 	 * Network adapter TSO limits:
379 	 * ===========================
380 	 *
381 	 * If the "if_hw_tsomax" field is zero the maximum segment
382 	 * length limit does not apply. If the "if_hw_tsomaxsegcount"
383 	 * or the "if_hw_tsomaxsegsize" field is zero the TSO segment
384 	 * count limit does not apply. If all three fields are zero,
385 	 * there is no TSO limit.
386 	 *
387 	 * NOTE: The TSO limits should reflect the values used in the
388 	 * BUSDMA tag a network adapter is using to load a mbuf chain
389 	 * for transmission. The TCP/IP network stack will subtract
390 	 * space for all linklevel and protocol level headers and
391 	 * ensure that the full mbuf chain passed to the network
392 	 * adapter fits within the given limits.
393 	 */
394 	u_int	if_hw_tsomax;		/* TSO maximum size in bytes */
395 	u_int	if_hw_tsomaxsegcount;	/* TSO maximum segment count */
396 	u_int	if_hw_tsomaxsegsize;	/* TSO maximum segment size in bytes */
397 
398 	/*
399 	 * Network adapter send tag support:
400 	 */
401 	if_snd_tag_alloc_t *if_snd_tag_alloc;
402 	if_snd_tag_modify_t *if_snd_tag_modify;
403 	if_snd_tag_query_t *if_snd_tag_query;
404 	if_snd_tag_free_t *if_snd_tag_free;
405 	if_ratelimit_query_t *if_ratelimit_query;
406 
407 	/* Ethernet PCP */
408 	uint8_t if_pcp;
409 
410 	/*
411 	 * Netdump hooks to be called while dumping.
412 	 */
413 	struct netdump_methods *if_netdump_methods;
414 	struct epoch_context	if_epoch_ctx;
415 
416 	/*
417 	 * Spare fields to be added before branching a stable branch, so
418 	 * that structure can be enhanced without changing the kernel
419 	 * binary interface.
420 	 */
421 	int	if_ispare[4];		/* general use */
422 };
423 
424 /* for compatibility with other BSDs */
425 #define	if_name(ifp)	((ifp)->if_xname)
426 
427 #define	IF_NODOM	255
428 /*
429  * Locks for address lists on the network interface.
430  */
431 #define	IF_ADDR_LOCK_INIT(if)	mtx_init(&(if)->if_addr_lock, "if_addr_lock", NULL, MTX_DEF)
432 #define	IF_ADDR_LOCK_DESTROY(if)	mtx_destroy(&(if)->if_addr_lock)
433 
434 #define	IF_ADDR_WLOCK(if)	mtx_lock(&(if)->if_addr_lock)
435 #define	IF_ADDR_WUNLOCK(if)	mtx_unlock(&(if)->if_addr_lock)
436 #define	IF_ADDR_LOCK_ASSERT(if)	MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(if)->if_addr_lock))
437 #define	IF_ADDR_WLOCK_ASSERT(if) mtx_assert(&(if)->if_addr_lock, MA_OWNED)
438 #define	NET_EPOCH_ENTER(et)	epoch_enter_preempt(net_epoch_preempt, &(et))
439 #define	NET_EPOCH_EXIT(et)	epoch_exit_preempt(net_epoch_preempt, &(et))
440 #define	NET_EPOCH_WAIT()	epoch_wait_preempt(net_epoch_preempt)
441 #define	NET_EPOCH_ASSERT()	MPASS(in_epoch(net_epoch_preempt))
442 
443 /*
444  * Function variations on locking macros intended to be used by loadable
445  * kernel modules in order to divorce them from the internals of address list
446  * locking.
447  */
448 void	if_addr_rlock(struct ifnet *ifp);	/* if_addrhead */
449 void	if_addr_runlock(struct ifnet *ifp);	/* if_addrhead */
450 void	if_maddr_rlock(if_t ifp);	/* if_multiaddrs */
451 void	if_maddr_runlock(if_t ifp);	/* if_multiaddrs */
452 
453 #ifdef _KERNEL
454 /* interface link layer address change event */
455 typedef void (*iflladdr_event_handler_t)(void *, struct ifnet *);
456 EVENTHANDLER_DECLARE(iflladdr_event, iflladdr_event_handler_t);
457 /* interface address change event */
458 typedef void (*ifaddr_event_handler_t)(void *, struct ifnet *);
459 EVENTHANDLER_DECLARE(ifaddr_event, ifaddr_event_handler_t);
460 typedef void (*ifaddr_event_ext_handler_t)(void *, struct ifnet *,
461     struct ifaddr *, int);
462 EVENTHANDLER_DECLARE(ifaddr_event_ext, ifaddr_event_ext_handler_t);
463 #define	IFADDR_EVENT_ADD	0
464 #define	IFADDR_EVENT_DEL	1
465 /* new interface arrival event */
466 typedef void (*ifnet_arrival_event_handler_t)(void *, struct ifnet *);
467 EVENTHANDLER_DECLARE(ifnet_arrival_event, ifnet_arrival_event_handler_t);
468 /* interface departure event */
469 typedef void (*ifnet_departure_event_handler_t)(void *, struct ifnet *);
470 EVENTHANDLER_DECLARE(ifnet_departure_event, ifnet_departure_event_handler_t);
471 /* Interface link state change event */
472 typedef void (*ifnet_link_event_handler_t)(void *, struct ifnet *, int);
473 EVENTHANDLER_DECLARE(ifnet_link_event, ifnet_link_event_handler_t);
474 /* Interface up/down event */
475 #define IFNET_EVENT_UP		0
476 #define IFNET_EVENT_DOWN	1
477 #define IFNET_EVENT_PCP		2	/* priority code point, PCP */
478 
479 typedef void (*ifnet_event_fn)(void *, struct ifnet *ifp, int event);
480 EVENTHANDLER_DECLARE(ifnet_event, ifnet_event_fn);
481 
482 /*
483  * interface groups
484  */
485 struct ifg_group {
486 	char				 ifg_group[IFNAMSIZ];
487 	u_int				 ifg_refcnt;
488 	void				*ifg_pf_kif;
489 	CK_STAILQ_HEAD(, ifg_member)	 ifg_members; /* (CK_) */
490 	CK_STAILQ_ENTRY(ifg_group)		 ifg_next; /* (CK_) */
491 };
492 
493 struct ifg_member {
494 	CK_STAILQ_ENTRY(ifg_member)	 ifgm_next; /* (CK_) */
495 	struct ifnet		*ifgm_ifp;
496 };
497 
498 struct ifg_list {
499 	struct ifg_group	*ifgl_group;
500 	CK_STAILQ_ENTRY(ifg_list)	 ifgl_next; /* (CK_) */
501 };
502 
503 #ifdef _SYS_EVENTHANDLER_H_
504 /* group attach event */
505 typedef void (*group_attach_event_handler_t)(void *, struct ifg_group *);
506 EVENTHANDLER_DECLARE(group_attach_event, group_attach_event_handler_t);
507 /* group detach event */
508 typedef void (*group_detach_event_handler_t)(void *, struct ifg_group *);
509 EVENTHANDLER_DECLARE(group_detach_event, group_detach_event_handler_t);
510 /* group change event */
511 typedef void (*group_change_event_handler_t)(void *, const char *);
512 EVENTHANDLER_DECLARE(group_change_event, group_change_event_handler_t);
513 #endif /* _SYS_EVENTHANDLER_H_ */
514 
515 #define	IF_AFDATA_LOCK_INIT(ifp)	\
516 	mtx_init(&(ifp)->if_afdata_lock, "if_afdata", NULL, MTX_DEF)
517 
518 #define	IF_AFDATA_WLOCK(ifp)	mtx_lock(&(ifp)->if_afdata_lock)
519 #define	IF_AFDATA_WUNLOCK(ifp)	mtx_unlock(&(ifp)->if_afdata_lock)
520 #define	IF_AFDATA_LOCK(ifp)	IF_AFDATA_WLOCK(ifp)
521 #define	IF_AFDATA_UNLOCK(ifp)	IF_AFDATA_WUNLOCK(ifp)
522 #define	IF_AFDATA_TRYLOCK(ifp)	mtx_trylock(&(ifp)->if_afdata_lock)
523 #define	IF_AFDATA_DESTROY(ifp)	mtx_destroy(&(ifp)->if_afdata_lock)
524 
525 #define	IF_AFDATA_LOCK_ASSERT(ifp)	MPASS(in_epoch(net_epoch_preempt) || mtx_owned(&(ifp)->if_afdata_lock))
526 #define	IF_AFDATA_WLOCK_ASSERT(ifp)	mtx_assert(&(ifp)->if_afdata_lock, MA_OWNED)
527 #define	IF_AFDATA_UNLOCK_ASSERT(ifp)	mtx_assert(&(ifp)->if_afdata_lock, MA_NOTOWNED)
528 
529 /*
530  * 72 was chosen below because it is the size of a TCP/IP
531  * header (40) + the minimum mss (32).
532  */
533 #define	IF_MINMTU	72
534 #define	IF_MAXMTU	65535
535 
536 #define	TOEDEV(ifp)	((ifp)->if_llsoftc)
537 
538 /*
539  * The ifaddr structure contains information about one address
540  * of an interface.  They are maintained by the different address families,
541  * are allocated and attached when an address is set, and are linked
542  * together so all addresses for an interface can be located.
543  *
544  * NOTE: a 'struct ifaddr' is always at the beginning of a larger
545  * chunk of malloc'ed memory, where we store the three addresses
546  * (ifa_addr, ifa_dstaddr and ifa_netmask) referenced here.
547  */
548 struct ifaddr {
549 	struct	sockaddr *ifa_addr;	/* address of interface */
550 	struct	sockaddr *ifa_dstaddr;	/* other end of p-to-p link */
551 #define	ifa_broadaddr	ifa_dstaddr	/* broadcast address interface */
552 	struct	sockaddr *ifa_netmask;	/* used to determine subnet */
553 	struct	ifnet *ifa_ifp;		/* back-pointer to interface */
554 	struct	carp_softc *ifa_carp;	/* pointer to CARP data */
555 	CK_STAILQ_ENTRY(ifaddr) ifa_link;	/* queue macro glue */
556 	void	(*ifa_rtrequest)	/* check or clean routes (+ or -)'d */
557 		(int, struct rtentry *, struct rt_addrinfo *);
558 	u_short	ifa_flags;		/* mostly rt_flags for cloning */
559 #define	IFA_ROUTE	RTF_UP		/* route installed */
560 #define	IFA_RTSELF	RTF_HOST	/* loopback route to self installed */
561 	u_int	ifa_refcnt;		/* references to this structure */
562 
563 	counter_u64_t	ifa_ipackets;
564 	counter_u64_t	ifa_opackets;
565 	counter_u64_t	ifa_ibytes;
566 	counter_u64_t	ifa_obytes;
567 	struct	epoch_context	ifa_epoch_ctx;
568 };
569 
570 struct ifaddr *	ifa_alloc(size_t size, int flags);
571 void	ifa_free(struct ifaddr *ifa);
572 void	ifa_ref(struct ifaddr *ifa);
573 
574 /*
575  * Multicast address structure.  This is analogous to the ifaddr
576  * structure except that it keeps track of multicast addresses.
577  */
578 #define IFMA_F_ENQUEUED		0x1
579 struct ifmultiaddr {
580 	CK_STAILQ_ENTRY(ifmultiaddr) ifma_link; /* queue macro glue */
581 	struct	sockaddr *ifma_addr; 	/* address this membership is for */
582 	struct	sockaddr *ifma_lladdr;	/* link-layer translation, if any */
583 	struct	ifnet *ifma_ifp;	/* back-pointer to interface */
584 	u_int	ifma_refcount;		/* reference count */
585 	int	ifma_flags;
586 	void	*ifma_protospec;	/* protocol-specific state, if any */
587 	struct	ifmultiaddr *ifma_llifma; /* pointer to ifma for ifma_lladdr */
588 	struct	epoch_context	ifma_epoch_ctx;
589 };
590 
591 extern	struct rwlock ifnet_rwlock;
592 extern	struct sx ifnet_sxlock;
593 
594 #define	IFNET_WLOCK() do {						\
595 	sx_xlock(&ifnet_sxlock);					\
596 	rw_wlock(&ifnet_rwlock);					\
597 } while (0)
598 
599 #define	IFNET_WUNLOCK() do {						\
600 	rw_wunlock(&ifnet_rwlock);					\
601 	sx_xunlock(&ifnet_sxlock);					\
602 } while (0)
603 
604 /*
605  * To assert the ifnet lock, you must know not only whether it's for read or
606  * write, but also whether it was acquired with sleep support or not.
607  */
608 #define	IFNET_RLOCK_ASSERT()		sx_assert(&ifnet_sxlock, SA_SLOCKED)
609 #define	IFNET_WLOCK_ASSERT() do {					\
610 	sx_assert(&ifnet_sxlock, SA_XLOCKED);				\
611 	rw_assert(&ifnet_rwlock, RA_WLOCKED);				\
612 } while (0)
613 
614 #define	IFNET_RLOCK()		sx_slock(&ifnet_sxlock)
615 #define	IFNET_RUNLOCK()		sx_sunlock(&ifnet_sxlock)
616 
617 /*
618  * Look up an ifnet given its index; the _ref variant also acquires a
619  * reference that must be freed using if_rele().  It is almost always a bug
620  * to call ifnet_byindex() instead of ifnet_byindex_ref().
621  */
622 struct ifnet	*ifnet_byindex(u_short idx);
623 struct ifnet	*ifnet_byindex_locked(u_short idx);
624 struct ifnet	*ifnet_byindex_ref(u_short idx);
625 
626 /*
627  * Given the index, ifaddr_byindex() returns the one and only
628  * link-level ifaddr for the interface. You are not supposed to use
629  * it to traverse the list of addresses associated to the interface.
630  */
631 struct ifaddr	*ifaddr_byindex(u_short idx);
632 
633 VNET_DECLARE(struct ifnethead, ifnet);
634 VNET_DECLARE(struct ifgrouphead, ifg_head);
635 VNET_DECLARE(int, if_index);
636 VNET_DECLARE(struct ifnet *, loif);	/* first loopback interface */
637 
638 #define	V_ifnet		VNET(ifnet)
639 #define	V_ifg_head	VNET(ifg_head)
640 #define	V_if_index	VNET(if_index)
641 #define	V_loif		VNET(loif)
642 
643 #ifdef MCAST_VERBOSE
644 #define MCDPRINTF printf
645 #else
646 #define MCDPRINTF(...)
647 #endif
648 
649 int	if_addgroup(struct ifnet *, const char *);
650 int	if_delgroup(struct ifnet *, const char *);
651 int	if_addmulti(struct ifnet *, struct sockaddr *, struct ifmultiaddr **);
652 int	if_allmulti(struct ifnet *, int);
653 struct	ifnet* if_alloc(u_char);
654 struct	ifnet* if_alloc_dev(u_char, device_t dev);
655 struct	ifnet* if_alloc_domain(u_char, int numa_domain);
656 void	if_attach(struct ifnet *);
657 void	if_dead(struct ifnet *);
658 int	if_delmulti(struct ifnet *, struct sockaddr *);
659 void	if_delmulti_ifma(struct ifmultiaddr *);
660 void	if_delmulti_ifma_flags(struct ifmultiaddr *, int flags);
661 void	if_detach(struct ifnet *);
662 void	if_purgeaddrs(struct ifnet *);
663 void	if_delallmulti(struct ifnet *);
664 void	if_down(struct ifnet *);
665 struct ifmultiaddr *
666 	if_findmulti(struct ifnet *, const struct sockaddr *);
667 void	if_freemulti(struct ifmultiaddr *ifma);
668 void	if_free(struct ifnet *);
669 void	if_initname(struct ifnet *, const char *, int);
670 void	if_link_state_change(struct ifnet *, int);
671 int	if_printf(struct ifnet *, const char *, ...) __printflike(2, 3);
672 void	if_ref(struct ifnet *);
673 void	if_rele(struct ifnet *);
674 int	if_setlladdr(struct ifnet *, const u_char *, int);
675 int	if_tunnel_check_nesting(struct ifnet *, struct mbuf *, uint32_t, int);
676 void	if_up(struct ifnet *);
677 int	ifioctl(struct socket *, u_long, caddr_t, struct thread *);
678 int	ifpromisc(struct ifnet *, int);
679 struct	ifnet *ifunit(const char *);
680 struct	ifnet *ifunit_ref(const char *);
681 
682 int	ifa_add_loopback_route(struct ifaddr *, struct sockaddr *);
683 int	ifa_del_loopback_route(struct ifaddr *, struct sockaddr *);
684 int	ifa_switch_loopback_route(struct ifaddr *, struct sockaddr *);
685 
686 struct	ifaddr *ifa_ifwithaddr(const struct sockaddr *);
687 int		ifa_ifwithaddr_check(const struct sockaddr *);
688 struct	ifaddr *ifa_ifwithbroadaddr(const struct sockaddr *, int);
689 struct	ifaddr *ifa_ifwithdstaddr(const struct sockaddr *, int);
690 struct	ifaddr *ifa_ifwithnet(const struct sockaddr *, int, int);
691 struct	ifaddr *ifa_ifwithroute(int, const struct sockaddr *, struct sockaddr *,
692     u_int);
693 struct	ifaddr *ifaof_ifpforaddr(const struct sockaddr *, struct ifnet *);
694 int	ifa_preferred(struct ifaddr *, struct ifaddr *);
695 
696 int	if_simloop(struct ifnet *ifp, struct mbuf *m, int af, int hlen);
697 
698 typedef	void *if_com_alloc_t(u_char type, struct ifnet *ifp);
699 typedef	void if_com_free_t(void *com, u_char type);
700 void	if_register_com_alloc(u_char type, if_com_alloc_t *a, if_com_free_t *f);
701 void	if_deregister_com_alloc(u_char type);
702 void	if_data_copy(struct ifnet *, struct if_data *);
703 uint64_t if_get_counter_default(struct ifnet *, ift_counter);
704 void	if_inc_counter(struct ifnet *, ift_counter, int64_t);
705 
706 #define IF_LLADDR(ifp)							\
707     LLADDR((struct sockaddr_dl *)((ifp)->if_addr->ifa_addr))
708 
709 uint64_t if_setbaudrate(if_t ifp, uint64_t baudrate);
710 uint64_t if_getbaudrate(if_t ifp);
711 int if_setcapabilities(if_t ifp, int capabilities);
712 int if_setcapabilitiesbit(if_t ifp, int setbit, int clearbit);
713 int if_getcapabilities(if_t ifp);
714 int if_togglecapenable(if_t ifp, int togglecap);
715 int if_setcapenable(if_t ifp, int capenable);
716 int if_setcapenablebit(if_t ifp, int setcap, int clearcap);
717 int if_getcapenable(if_t ifp);
718 const char *if_getdname(if_t ifp);
719 int if_setdev(if_t ifp, void *dev);
720 int if_setdrvflagbits(if_t ifp, int if_setflags, int clear_flags);
721 int if_getdrvflags(if_t ifp);
722 int if_setdrvflags(if_t ifp, int flags);
723 int if_clearhwassist(if_t ifp);
724 int if_sethwassistbits(if_t ifp, int toset, int toclear);
725 int if_sethwassist(if_t ifp, int hwassist_bit);
726 int if_gethwassist(if_t ifp);
727 int if_setsoftc(if_t ifp, void *softc);
728 void *if_getsoftc(if_t ifp);
729 int if_setflags(if_t ifp, int flags);
730 int if_gethwaddr(if_t ifp, struct ifreq *);
731 int if_setmtu(if_t ifp, int mtu);
732 int if_getmtu(if_t ifp);
733 int if_getmtu_family(if_t ifp, int family);
734 int if_setflagbits(if_t ifp, int set, int clear);
735 int if_getflags(if_t ifp);
736 int if_sendq_empty(if_t ifp);
737 int if_setsendqready(if_t ifp);
738 int if_setsendqlen(if_t ifp, int tx_desc_count);
739 int if_sethwtsomax(if_t ifp, u_int if_hw_tsomax);
740 int if_sethwtsomaxsegcount(if_t ifp, u_int if_hw_tsomaxsegcount);
741 int if_sethwtsomaxsegsize(if_t ifp, u_int if_hw_tsomaxsegsize);
742 u_int if_gethwtsomax(if_t ifp);
743 u_int if_gethwtsomaxsegcount(if_t ifp);
744 u_int if_gethwtsomaxsegsize(if_t ifp);
745 int if_input(if_t ifp, struct mbuf* sendmp);
746 int if_sendq_prepend(if_t ifp, struct mbuf *m);
747 struct mbuf *if_dequeue(if_t ifp);
748 int if_setifheaderlen(if_t ifp, int len);
749 void if_setrcvif(struct mbuf *m, if_t ifp);
750 void if_setvtag(struct mbuf *m, u_int16_t tag);
751 u_int16_t if_getvtag(struct mbuf *m);
752 int if_vlantrunkinuse(if_t ifp);
753 caddr_t if_getlladdr(if_t ifp);
754 void *if_gethandle(u_char);
755 void if_bpfmtap(if_t ifp, struct mbuf *m);
756 void if_etherbpfmtap(if_t ifp, struct mbuf *m);
757 void if_vlancap(if_t ifp);
758 
759 int if_setupmultiaddr(if_t ifp, void *mta, int *cnt, int max);
760 int if_multiaddr_array(if_t ifp, void *mta, int *cnt, int max);
761 int if_multiaddr_count(if_t ifp, int max);
762 
763 int if_multi_apply(struct ifnet *ifp, int (*filter)(void *, struct ifmultiaddr *, int), void *arg);
764 int if_getamcount(if_t ifp);
765 struct ifaddr * if_getifaddr(if_t ifp);
766 
767 /* Functions */
768 void if_setinitfn(if_t ifp, void (*)(void *));
769 void if_setioctlfn(if_t ifp, int (*)(if_t, u_long, caddr_t));
770 void if_setstartfn(if_t ifp, void (*)(if_t));
771 void if_settransmitfn(if_t ifp, if_transmit_fn_t);
772 void if_setqflushfn(if_t ifp, if_qflush_fn_t);
773 void if_setgetcounterfn(if_t ifp, if_get_counter_t);
774 
775 /* Revisit the below. These are inline functions originally */
776 int drbr_inuse_drv(if_t ifp, struct buf_ring *br);
777 struct mbuf* drbr_dequeue_drv(if_t ifp, struct buf_ring *br);
778 int drbr_needs_enqueue_drv(if_t ifp, struct buf_ring *br);
779 int drbr_enqueue_drv(if_t ifp, struct buf_ring *br, struct mbuf *m);
780 
781 /* TSO */
782 void if_hw_tsomax_common(if_t ifp, struct ifnet_hw_tsomax *);
783 int if_hw_tsomax_update(if_t ifp, struct ifnet_hw_tsomax *);
784 
785 /* accessors for struct ifreq */
786 void *ifr_data_get_ptr(void *ifrp);
787 
788 int ifhwioctl(u_long, struct ifnet *, caddr_t, struct thread *);
789 
790 #ifdef DEVICE_POLLING
791 enum poll_cmd { POLL_ONLY, POLL_AND_CHECK_STATUS };
792 
793 typedef	int poll_handler_t(if_t ifp, enum poll_cmd cmd, int count);
794 int    ether_poll_register(poll_handler_t *h, if_t ifp);
795 int    ether_poll_deregister(if_t ifp);
796 #endif /* DEVICE_POLLING */
797 
798 #endif /* _KERNEL */
799 
800 #include <net/ifq.h>	/* XXXAO: temporary unconditional include */
801 
802 #endif /* !_NET_IF_VAR_H_ */
803