xref: /freebsd/sys/sys/socketvar.h (revision f6379f7f)
1 /*-
2  * SPDX-License-Identifier: BSD-3-Clause
3  *
4  * Copyright (c) 1982, 1986, 1990, 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  *	@(#)socketvar.h	8.3 (Berkeley) 2/19/95
32  *
33  * $FreeBSD$
34  */
35 
36 #ifndef _SYS_SOCKETVAR_H_
37 #define _SYS_SOCKETVAR_H_
38 
39 /*
40  * Socket generation count type.  Also used in xinpcb, xtcpcb, xunpcb.
41  */
42 typedef uint64_t so_gen_t;
43 
44 #if defined(_KERNEL) || defined(_WANT_SOCKET)
45 #include <sys/queue.h>			/* for TAILQ macros */
46 #include <sys/selinfo.h>		/* for struct selinfo */
47 #include <sys/_lock.h>
48 #include <sys/_mutex.h>
49 #include <sys/osd.h>
50 #include <sys/_sx.h>
51 #include <sys/sockbuf.h>
52 #ifdef _KERNEL
53 #include <sys/caprights.h>
54 #include <sys/sockopt.h>
55 #endif
56 
57 struct vnet;
58 
59 /*
60  * Kernel structure per socket.
61  * Contains send and receive buffer queues,
62  * handle on protocol and pointer to protocol
63  * private data and error information.
64  */
65 typedef	int so_upcall_t(struct socket *, void *, int);
66 typedef	void so_dtor_t(struct socket *);
67 
68 struct socket;
69 
70 enum socket_qstate {
71 	SQ_NONE = 0,
72 	SQ_INCOMP = 0x0800,	/* on sol_incomp */
73 	SQ_COMP = 0x1000,	/* on sol_comp */
74 };
75 
76 /*-
77  * Locking key to struct socket:
78  * (a) constant after allocation, no locking required.
79  * (b) locked by SOCK_LOCK(so).
80  * (cr) locked by SOCK_RECVBUF_LOCK(so)
81  * (cs) locked by SOCK_SENDBUF_LOCK(so)
82  * (e) locked by SOLISTEN_LOCK() of corresponding listening socket.
83  * (f) not locked since integer reads/writes are atomic.
84  * (g) used only as a sleep/wakeup address, no value.
85  * (h) locked by global mutex so_global_mtx.
86  * (k) locked by KTLS workqueue mutex
87  */
88 TAILQ_HEAD(accept_queue, socket);
89 struct socket {
90 	struct mtx	so_lock;
91 	volatile u_int	so_count;	/* (b / refcount) */
92 	struct selinfo	so_rdsel;	/* (b/cr) for so_rcv/so_comp */
93 	struct selinfo	so_wrsel;	/* (b/cs) for so_snd */
94 	int	so_options;		/* (b) from socket call, see socket.h */
95 	short	so_type;		/* (a) generic type, see socket.h */
96 	short	so_state;		/* (b) internal state flags SS_* */
97 	void	*so_pcb;		/* protocol control block */
98 	struct	vnet *so_vnet;		/* (a) network stack instance */
99 	struct	protosw *so_proto;	/* (a) protocol handle */
100 	short	so_linger;		/* time to linger close(2) */
101 	short	so_timeo;		/* (g) connection timeout */
102 	u_short	so_error;		/* (f) error affecting connection */
103 	u_short so_rerror;		/* (f) error affecting connection */
104 	struct	sigio *so_sigio;	/* [sg] information for async I/O or
105 					   out of band data (SIGURG) */
106 	struct	ucred *so_cred;		/* (a) user credentials */
107 	struct	label *so_label;	/* (b) MAC label for socket */
108 	/* NB: generation count must not be first. */
109 	so_gen_t so_gencnt;		/* (h) generation count */
110 	void	*so_emuldata;		/* (b) private data for emulators */
111 	so_dtor_t *so_dtor;		/* (b) optional destructor */
112 	struct	osd	osd;		/* Object Specific extensions */
113 	/*
114 	 * so_fibnum, so_user_cookie and friends can be used to attach
115 	 * some user-specified metadata to a socket, which then can be
116 	 * used by the kernel for various actions.
117 	 * so_user_cookie is used by ipfw/dummynet.
118 	 */
119 	int so_fibnum;		/* routing domain for this socket */
120 	uint32_t so_user_cookie;
121 
122 	int so_ts_clock;	/* type of the clock used for timestamps */
123 	uint32_t so_max_pacing_rate;	/* (f) TX rate limit in bytes/s */
124 
125 	/*
126 	 * Mutexes to prevent interleaving of socket I/O.  These have to be
127 	 * outside of the socket buffers in order to interlock with listen(2).
128 	 */
129 	struct sx so_snd_sx __aligned(CACHE_LINE_SIZE);
130 	struct mtx so_snd_mtx;
131 
132 	struct sx so_rcv_sx __aligned(CACHE_LINE_SIZE);
133 	struct mtx so_rcv_mtx;
134 
135 	union {
136 		/* Regular (data flow) socket. */
137 		struct {
138 			/* (cr, cs) Receive and send buffers. */
139 			struct sockbuf		so_rcv, so_snd;
140 
141 			/* (e) Our place on accept queue. */
142 			TAILQ_ENTRY(socket)	so_list;
143 			struct socket		*so_listen;	/* (b) */
144 			enum socket_qstate so_qstate;		/* (b) */
145 			/* (b) cached MAC label for peer */
146 			struct	label		*so_peerlabel;
147 			u_long	so_oobmark;	/* chars to oob mark */
148 
149 			/* (k) Our place on KTLS RX work queue. */
150 			STAILQ_ENTRY(socket)	so_ktls_rx_list;
151 		};
152 		/*
153 		 * Listening socket, where accepts occur, is so_listen in all
154 		 * subsidiary sockets.  If so_listen is NULL, socket is not
155 		 * related to an accept.  For a listening socket itself
156 		 * sol_incomp queues partially completed connections, while
157 		 * sol_comp is a queue of connections ready to be accepted.
158 		 * If a connection is aborted and it has so_listen set, then
159 		 * it has to be pulled out of either sol_incomp or sol_comp.
160 		 * We allow connections to queue up based on current queue
161 		 * lengths and limit on number of queued connections for this
162 		 * socket.
163 		 */
164 		struct {
165 			/* (e) queue of partial unaccepted connections */
166 			struct accept_queue	sol_incomp;
167 			/* (e) queue of complete unaccepted connections */
168 			struct accept_queue	sol_comp;
169 			u_int	sol_qlen;    /* (e) sol_comp length */
170 			u_int	sol_incqlen; /* (e) sol_incomp length */
171 			u_int	sol_qlimit;  /* (e) queue limit */
172 
173 			/* accept_filter(9) optional data */
174 			struct	accept_filter	*sol_accept_filter;
175 			void	*sol_accept_filter_arg;	/* saved filter args */
176 			char	*sol_accept_filter_str;	/* saved user args */
177 
178 			/* Optional upcall, for kernel socket. */
179 			so_upcall_t	*sol_upcall;	/* (e) */
180 			void		*sol_upcallarg;	/* (e) */
181 
182 			/* Socket buffer parameters, to be copied to
183 			 * dataflow sockets, accepted from this one. */
184 			int		sol_sbrcv_lowat;
185 			int		sol_sbsnd_lowat;
186 			u_int		sol_sbrcv_hiwat;
187 			u_int		sol_sbsnd_hiwat;
188 			short		sol_sbrcv_flags;
189 			short		sol_sbsnd_flags;
190 			sbintime_t	sol_sbrcv_timeo;
191 			sbintime_t	sol_sbsnd_timeo;
192 
193 			/* Information tracking listen queue overflows. */
194 			struct timeval	sol_lastover;	/* (e) */
195 			int		sol_overcount;	/* (e) */
196 		};
197 	};
198 };
199 #endif	/* defined(_KERNEL) || defined(_WANT_SOCKET) */
200 
201 /*
202  * Socket state bits.
203  *
204  * Historically, these bits were all kept in the so_state field.
205  * They are now split into separate, lock-specific fields.
206  * so_state maintains basic socket state protected by the socket lock.
207  * so_qstate holds information about the socket accept queues.
208  * Each socket buffer also has a state field holding information
209  * relevant to that socket buffer (can't send, rcv).
210  * Many fields will be read without locks to improve performance and avoid
211  * lock order issues.  However, this approach must be used with caution.
212  */
213 #define	SS_NOFDREF		0x0001	/* no file table ref any more */
214 #define	SS_ISCONNECTED		0x0002	/* socket connected to a peer */
215 #define	SS_ISCONNECTING		0x0004	/* in process of connecting to peer */
216 #define	SS_ISDISCONNECTING	0x0008	/* in process of disconnecting */
217 #define	SS_NBIO			0x0100	/* non-blocking ops */
218 #define	SS_ASYNC		0x0200	/* async i/o notify */
219 #define	SS_ISCONFIRMING		0x0400	/* deciding to accept connection req */
220 #define	SS_ISDISCONNECTED	0x2000	/* socket disconnected from peer */
221 
222 /*
223  * Protocols can mark a socket as SS_PROTOREF to indicate that, following
224  * pru_detach, they still want the socket to persist, and will free it
225  * themselves when they are done.  Protocols should only ever call sofree()
226  * following setting this flag in pru_detach(), and never otherwise, as
227  * sofree() bypasses socket reference counting.
228  */
229 #define	SS_PROTOREF		0x4000	/* strong protocol reference */
230 
231 #ifdef _KERNEL
232 
233 #define	SOCK_MTX(so)		(&(so)->so_lock)
234 #define	SOCK_LOCK(so)		mtx_lock(&(so)->so_lock)
235 #define	SOCK_OWNED(so)		mtx_owned(&(so)->so_lock)
236 #define	SOCK_UNLOCK(so)		mtx_unlock(&(so)->so_lock)
237 #define	SOCK_LOCK_ASSERT(so)	mtx_assert(&(so)->so_lock, MA_OWNED)
238 #define	SOCK_UNLOCK_ASSERT(so)	mtx_assert(&(so)->so_lock, MA_NOTOWNED)
239 
240 #define	SOLISTENING(sol)	(((sol)->so_options & SO_ACCEPTCONN) != 0)
241 #define	SOLISTEN_LOCK(sol)	do {					\
242 	mtx_lock(&(sol)->so_lock);					\
243 	KASSERT(SOLISTENING(sol),					\
244 	    ("%s: %p not listening", __func__, (sol)));			\
245 } while (0)
246 #define	SOLISTEN_TRYLOCK(sol)	mtx_trylock(&(sol)->so_lock)
247 #define	SOLISTEN_UNLOCK(sol)	do {					\
248 	KASSERT(SOLISTENING(sol),					\
249 	    ("%s: %p not listening", __func__, (sol)));			\
250 	mtx_unlock(&(sol)->so_lock);					\
251 } while (0)
252 #define	SOLISTEN_LOCK_ASSERT(sol)	do {				\
253 	mtx_assert(&(sol)->so_lock, MA_OWNED);				\
254 	KASSERT(SOLISTENING(sol),					\
255 	    ("%s: %p not listening", __func__, (sol)));			\
256 } while (0)
257 #define	SOLISTEN_UNLOCK_ASSERT(sol)	do {				\
258 	mtx_assert(&(sol)->so_lock, MA_NOTOWNED);			\
259 	KASSERT(SOLISTENING(sol),					\
260 	    ("%s: %p not listening", __func__, (sol)));			\
261 } while (0)
262 
263 /*
264  * Socket buffer locks.  These are strongly preferred over SOCKBUF_LOCK(sb)
265  * macros, as we are moving towards protocol specific socket buffers.
266  */
267 #define	SOCK_RECVBUF_MTX(so)						\
268 	(&(so)->so_rcv_mtx)
269 #define	SOCK_RECVBUF_LOCK(so)						\
270 	mtx_lock(SOCK_RECVBUF_MTX(so))
271 #define	SOCK_RECVBUF_UNLOCK(so)						\
272 	mtx_unlock(SOCK_RECVBUF_MTX(so))
273 #define	SOCK_RECVBUF_LOCK_ASSERT(so)					\
274 	mtx_assert(SOCK_RECVBUF_MTX(so), MA_OWNED)
275 #define	SOCK_RECVBUF_UNLOCK_ASSERT(so)					\
276 	mtx_assert(SOCK_RECVBUF_MTX(so), MA_NOTOWNED)
277 
278 #define	SOCK_SENDBUF_MTX(so)						\
279 	(&(so)->so_snd_mtx)
280 #define	SOCK_SENDBUF_LOCK(so)						\
281 	mtx_lock(SOCK_SENDBUF_MTX(so))
282 #define	SOCK_SENDBUF_UNLOCK(so)						\
283 	mtx_unlock(SOCK_SENDBUF_MTX(so))
284 #define	SOCK_SENDBUF_LOCK_ASSERT(so)					\
285 	mtx_assert(SOCK_SENDBUF_MTX(so), MA_OWNED)
286 #define	SOCK_SENDBUF_UNLOCK_ASSERT(so)					\
287 	mtx_assert(SOCK_SENDBUF_MTX(so), MA_NOTOWNED)
288 
289 #define	SOCK_BUF_LOCK(so, which)					\
290 	mtx_lock(soeventmtx(so, which))
291 #define	SOCK_BUF_UNLOCK(so, which)					\
292 	mtx_unlock(soeventmtx(so, which))
293 #define	SOCK_BUF_LOCK_ASSERT(so, which)					\
294 	mtx_assert(soeventmtx(so, which), MA_OWNED)
295 #define	SOCK_BUF_UNLOCK_ASSERT(so, which)				\
296 	mtx_assert(soeventmtx(so, which), MA_NOTOWNED)
297 
298 static inline struct sockbuf *
299 sobuf(struct socket *so, const sb_which which)
300 {
301 	return (which == SO_RCV ? &so->so_rcv : &so->so_snd);
302 }
303 
304 static inline struct mtx *
305 soeventmtx(struct socket *so, const sb_which which)
306 {
307 	return (which == SO_RCV ? SOCK_RECVBUF_MTX(so) : SOCK_SENDBUF_MTX(so));
308 }
309 
310 /*
311  * Macros for sockets and socket buffering.
312  */
313 
314 /*
315  * Flags to soiolock().
316  */
317 #define	SBL_WAIT	0x00000001	/* Wait if not immediately available. */
318 #define	SBL_NOINTR	0x00000002	/* Force non-interruptible sleep. */
319 #define	SBL_VALID	(SBL_WAIT | SBL_NOINTR)
320 
321 #define	SBLOCKWAIT(f)	(((f) & MSG_DONTWAIT) ? 0 : SBL_WAIT)
322 
323 #define	SOCK_IO_SEND_LOCK(so, flags)					\
324 	soiolock((so), &(so)->so_snd_sx, (flags))
325 #define	SOCK_IO_SEND_UNLOCK(so)						\
326 	soiounlock(&(so)->so_snd_sx)
327 #define	SOCK_IO_SEND_OWNED(so)	sx_xlocked(&(so)->so_snd_sx)
328 #define	SOCK_IO_RECV_LOCK(so, flags)					\
329 	soiolock((so), &(so)->so_rcv_sx, (flags))
330 #define	SOCK_IO_RECV_UNLOCK(so)						\
331 	soiounlock(&(so)->so_rcv_sx)
332 #define	SOCK_IO_RECV_OWNED(so)	sx_xlocked(&(so)->so_rcv_sx)
333 
334 /* do we have to send all at once on a socket? */
335 #define	sosendallatonce(so) \
336     ((so)->so_proto->pr_flags & PR_ATOMIC)
337 
338 /* can we read something from so? */
339 #define	soreadabledata(so) \
340 	(sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || \
341 	(so)->so_error || (so)->so_rerror)
342 #define	soreadable(so) \
343 	(soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE))
344 
345 /* can we write something to so? */
346 #define	sowriteable(so) \
347     ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
348 	(((so)->so_state&SS_ISCONNECTED) || \
349 	  ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
350      ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
351      (so)->so_error)
352 
353 /*
354  * soref()/sorele() ref-count the socket structure.
355  * soref() may be called without owning socket lock, but in that case a
356  * caller must own something that holds socket, and so_count must be not 0.
357  * Note that you must still explicitly close the socket, but the last ref
358  * count will free the structure.
359  */
360 #define	soref(so)	refcount_acquire(&(so)->so_count)
361 #define	sorele(so) do {							\
362 	SOCK_UNLOCK_ASSERT(so);						\
363 	if (!refcount_release_if_not_last(&(so)->so_count)) {		\
364 		SOCK_LOCK(so);						\
365 		sorele_locked(so);					\
366 	}								\
367 } while (0)
368 
369 /*
370  * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
371  * avoid a non-atomic test-and-wakeup.  However, sowakeup is
372  * responsible for releasing the lock if it is called.  We unlock only
373  * if we don't call into sowakeup.  If any code is introduced that
374  * directly invokes the underlying sowakeup() primitives, it must
375  * maintain the same semantics.
376  */
377 #define	sorwakeup(so) do {						\
378 	SOCK_RECVBUF_LOCK(so);						\
379 	sorwakeup_locked(so);						\
380 } while (0)
381 
382 #define	sowwakeup(so) do {						\
383 	SOCK_SENDBUF_LOCK(so);						\
384 	sowwakeup_locked(so);						\
385 } while (0)
386 
387 struct accept_filter {
388 	char	accf_name[16];
389 	int	(*accf_callback)
390 		(struct socket *so, void *arg, int waitflag);
391 	void *	(*accf_create)
392 		(struct socket *so, char *arg);
393 	void	(*accf_destroy)
394 		(struct socket *so);
395 	SLIST_ENTRY(accept_filter) accf_next;
396 };
397 
398 #define	ACCEPT_FILTER_DEFINE(modname, filtname, cb, create, destroy, ver) \
399 	static struct accept_filter modname##_filter = {		\
400 		.accf_name = filtname,					\
401 		.accf_callback = cb,					\
402 		.accf_create = create,					\
403 		.accf_destroy = destroy,				\
404 	};								\
405 	static moduledata_t modname##_mod = {				\
406 		.name = __XSTRING(modname),				\
407 		.evhand = accept_filt_generic_mod_event,		\
408 		.priv = &modname##_filter,				\
409 	};								\
410 	DECLARE_MODULE(modname, modname##_mod, SI_SUB_DRIVERS,		\
411 	    SI_ORDER_MIDDLE);						\
412 	MODULE_VERSION(modname, ver)
413 
414 #ifdef MALLOC_DECLARE
415 MALLOC_DECLARE(M_ACCF);
416 MALLOC_DECLARE(M_PCB);
417 MALLOC_DECLARE(M_SONAME);
418 #endif
419 
420 /*
421  * Socket specific helper hook point identifiers
422  * Do not leave holes in the sequence, hook registration is a loop.
423  */
424 #define HHOOK_SOCKET_OPT		0
425 #define HHOOK_SOCKET_CREATE		1
426 #define HHOOK_SOCKET_RCV 		2
427 #define HHOOK_SOCKET_SND		3
428 #define HHOOK_FILT_SOREAD		4
429 #define HHOOK_FILT_SOWRITE		5
430 #define HHOOK_SOCKET_CLOSE		6
431 #define HHOOK_SOCKET_LAST		HHOOK_SOCKET_CLOSE
432 
433 struct socket_hhook_data {
434 	struct socket	*so;
435 	struct mbuf	*m;
436 	void		*hctx;		/* hook point specific data*/
437 	int		status;
438 };
439 
440 extern int	maxsockets;
441 extern u_long	sb_max;
442 extern so_gen_t so_gencnt;
443 
444 struct file;
445 struct filecaps;
446 struct filedesc;
447 struct mbuf;
448 struct sockaddr;
449 struct ucred;
450 struct uio;
451 
452 /* Return values for socket upcalls. */
453 #define	SU_OK		0
454 #define	SU_ISCONNECTED	1
455 
456 /*
457  * From uipc_socket and friends
458  */
459 int	getsockaddr(struct sockaddr **namp, const struct sockaddr *uaddr,
460 	    size_t len);
461 int	getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp,
462 	    struct file **fpp, u_int *fflagp, struct filecaps *havecaps);
463 void	soabort(struct socket *so);
464 int	soaccept(struct socket *so, struct sockaddr **nam);
465 void	soaio_enqueue(struct task *task);
466 void	soaio_rcv(void *context, int pending);
467 void	soaio_snd(void *context, int pending);
468 int	socheckuid(struct socket *so, uid_t uid);
469 int	sobind(struct socket *so, struct sockaddr *nam, struct thread *td);
470 int	sobindat(int fd, struct socket *so, struct sockaddr *nam,
471 	    struct thread *td);
472 int	soclose(struct socket *so);
473 int	soconnect(struct socket *so, struct sockaddr *nam, struct thread *td);
474 int	soconnectat(int fd, struct socket *so, struct sockaddr *nam,
475 	    struct thread *td);
476 int	soconnect2(struct socket *so1, struct socket *so2);
477 int	socreate(int dom, struct socket **aso, int type, int proto,
478 	    struct ucred *cred, struct thread *td);
479 int	sodisconnect(struct socket *so);
480 void	sodtor_set(struct socket *, so_dtor_t *);
481 struct	sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags);
482 void	sofree(struct socket *so);
483 void	sohasoutofband(struct socket *so);
484 int	solisten(struct socket *so, int backlog, struct thread *td);
485 void	solisten_proto(struct socket *so, int backlog);
486 void	solisten_proto_abort(struct socket *so);
487 int	solisten_proto_check(struct socket *so);
488 int	solisten_dequeue(struct socket *, struct socket **, int);
489 struct socket *
490 	sonewconn(struct socket *head, int connstatus);
491 struct socket *
492 	sopeeloff(struct socket *);
493 int	sopoll(struct socket *so, int events, struct ucred *active_cred,
494 	    struct thread *td);
495 int	sopoll_generic(struct socket *so, int events,
496 	    struct ucred *active_cred, struct thread *td);
497 int	soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio,
498 	    struct mbuf **mp0, struct mbuf **controlp, int *flagsp);
499 int	soreceive_stream(struct socket *so, struct sockaddr **paddr,
500 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
501 	    int *flagsp);
502 int	soreceive_dgram(struct socket *so, struct sockaddr **paddr,
503 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
504 	    int *flagsp);
505 int	soreceive_generic(struct socket *so, struct sockaddr **paddr,
506 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
507 	    int *flagsp);
508 void	sorele_locked(struct socket *so);
509 int	soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
510 void	sorflush(struct socket *so);
511 int	sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
512 	    struct mbuf *top, struct mbuf *control, int flags,
513 	    struct thread *td);
514 int	sosend_dgram(struct socket *so, struct sockaddr *addr,
515 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
516 	    int flags, struct thread *td);
517 int	sosend_generic(struct socket *so, struct sockaddr *addr,
518 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
519 	    int flags, struct thread *td);
520 int	soshutdown(struct socket *so, int how);
521 void	soupcall_clear(struct socket *, sb_which);
522 void	soupcall_set(struct socket *, sb_which, so_upcall_t, void *);
523 void	solisten_upcall_set(struct socket *, so_upcall_t, void *);
524 void	sorwakeup_locked(struct socket *);
525 void	sowwakeup_locked(struct socket *);
526 void	sowakeup_aio(struct socket *, sb_which);
527 void	solisten_wakeup(struct socket *);
528 int	selsocket(struct socket *so, int events, struct timeval *tv,
529 	    struct thread *td);
530 void	soisconnected(struct socket *so);
531 void	soisconnecting(struct socket *so);
532 void	soisdisconnected(struct socket *so);
533 void	soisdisconnecting(struct socket *so);
534 void	socantrcvmore(struct socket *so);
535 void	socantrcvmore_locked(struct socket *so);
536 void	socantsendmore(struct socket *so);
537 void	socantsendmore_locked(struct socket *so);
538 void	soroverflow(struct socket *so);
539 void	soroverflow_locked(struct socket *so);
540 int	soiolock(struct socket *so, struct sx *sx, int flags);
541 void	soiounlock(struct sx *sx);
542 
543 /*
544  * Accept filter functions (duh).
545  */
546 int	accept_filt_add(struct accept_filter *filt);
547 int	accept_filt_del(char *name);
548 struct	accept_filter *accept_filt_get(char *name);
549 #ifdef ACCEPT_FILTER_MOD
550 #ifdef SYSCTL_DECL
551 SYSCTL_DECL(_net_inet_accf);
552 #endif
553 int	accept_filt_generic_mod_event(module_t mod, int event, void *data);
554 #endif
555 
556 #endif /* _KERNEL */
557 
558 /*
559  * Structure to export socket from kernel to utilities, via sysctl(3).
560  */
561 struct xsocket {
562 	ksize_t		xso_len;	/* length of this structure */
563 	kvaddr_t	xso_so;		/* kernel address of struct socket */
564 	kvaddr_t	so_pcb;		/* kernel address of struct inpcb */
565 	uint64_t	so_oobmark;
566 	int64_t		so_spare64[8];
567 	int32_t		xso_protocol;
568 	int32_t		xso_family;
569 	uint32_t	so_qlen;
570 	uint32_t	so_incqlen;
571 	uint32_t	so_qlimit;
572 	pid_t		so_pgid;
573 	uid_t		so_uid;
574 	int32_t		so_spare32[8];
575 	int16_t		so_type;
576 	int16_t		so_options;
577 	int16_t		so_linger;
578 	int16_t		so_state;
579 	int16_t		so_timeo;
580 	uint16_t	so_error;
581 	struct xsockbuf {
582 		uint32_t	sb_cc;
583 		uint32_t	sb_hiwat;
584 		uint32_t	sb_mbcnt;
585 		uint32_t	sb_spare0;	/* was sb_mcnt */
586 		uint32_t	sb_spare1;	/* was sb_ccnt */
587 		uint32_t	sb_mbmax;
588 		int32_t		sb_lowat;
589 		int32_t		sb_timeo;
590 		int16_t		sb_flags;
591 	} so_rcv, so_snd;
592 };
593 
594 #ifdef _KERNEL
595 void	sotoxsocket(struct socket *so, struct xsocket *xso);
596 void	sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb);
597 #endif
598 
599 /*
600  * Socket buffer state bits.  Exported via libprocstat(3).
601  */
602 #define	SBS_CANTSENDMORE	0x0010	/* can't send more data to peer */
603 #define	SBS_CANTRCVMORE		0x0020	/* can't receive more data from peer */
604 #define	SBS_RCVATMARK		0x0040	/* at mark on input */
605 
606 #endif /* !_SYS_SOCKETVAR_H_ */
607