xref: /NextBSD/sys/sys/socketvar.h (revision c21ffb8d6aca32c9584cfa072f309a5890a21aea)
1 /*-
2  * Copyright (c) 1982, 1986, 1990, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)socketvar.h	8.3 (Berkeley) 2/19/95
30  *
31  * $FreeBSD$
32  */
33 
34 #ifndef _SYS_SOCKETVAR_H_
35 #define _SYS_SOCKETVAR_H_
36 
37 #include <sys/queue.h>			/* for TAILQ macros */
38 #include <sys/selinfo.h>		/* for struct selinfo */
39 #include <sys/_lock.h>
40 #include <sys/_mutex.h>
41 #include <sys/osd.h>
42 #include <sys/_sx.h>
43 #include <sys/sockbuf.h>
44 #include <sys/sockstate.h>
45 #ifdef _KERNEL
46 #include <sys/caprights.h>
47 #include <sys/sockopt.h>
48 #endif
49 
50 struct vnet;
51 
52 /*
53  * Kernel structure per socket.
54  * Contains send and receive buffer queues,
55  * handle on protocol and pointer to protocol
56  * private data and error information.
57  */
58 typedef	u_quad_t so_gen_t;
59 
60 struct socket;
61 
62 /*-
63  * Locking key to struct socket:
64  * (a) constant after allocation, no locking required.
65  * (b) locked by SOCK_LOCK(so).
66  * (c) locked by SOCKBUF_LOCK(&so->so_rcv).
67  * (d) locked by SOCKBUF_LOCK(&so->so_snd).
68  * (e) locked by ACCEPT_LOCK().
69  * (f) not locked since integer reads/writes are atomic.
70  * (g) used only as a sleep/wakeup address, no value.
71  * (h) locked by global mutex so_global_mtx.
72  */
73 struct socket {
74 	int	so_count;		/* (b) reference count */
75 	short	so_type;		/* (a) generic type, see socket.h */
76 	short	so_options;		/* from socket call, see socket.h */
77 	short	so_linger;		/* time to linger while closing */
78 	short	so_state;		/* (b) internal state flags SS_* */
79 	int	so_qstate;		/* (e) internal state flags SQ_* */
80 	void	*so_pcb;		/* protocol control block */
81 	struct	vnet *so_vnet;		/* (a) network stack instance */
82 	struct	protosw *so_proto;	/* (a) protocol handle */
83 /*
84  * Variables for connection queuing.
85  * Socket where accepts occur is so_head in all subsidiary sockets.
86  * If so_head is 0, socket is not related to an accept.
87  * For head socket so_incomp queues partially completed connections,
88  * while so_comp is a queue of connections ready to be accepted.
89  * If a connection is aborted and it has so_head set, then
90  * it has to be pulled out of either so_incomp or so_comp.
91  * We allow connections to queue up based on current queue lengths
92  * and limit on number of queued connections for this socket.
93  */
94 	struct	socket *so_head;	/* (e) back pointer to listen socket */
95 	TAILQ_HEAD(, socket) so_incomp;	/* (e) queue of partial unaccepted connections */
96 	TAILQ_HEAD(, socket) so_comp;	/* (e) queue of complete unaccepted connections */
97 	TAILQ_ENTRY(socket) so_list;	/* (e) list of unaccepted connections */
98 	u_short	so_qlen;		/* (e) number of unaccepted connections */
99 	u_short	so_incqlen;		/* (e) number of unaccepted incomplete
100 					   connections */
101 	u_short	so_qlimit;		/* (e) max number queued connections */
102 	short	so_timeo;		/* (g) connection timeout */
103 	u_short	so_error;		/* (f) error affecting connection */
104 	struct	sigio *so_sigio;	/* [sg] information for async I/O or
105 					   out of band data (SIGURG) */
106 	u_long	so_oobmark;		/* (c) chars to oob mark */
107 	TAILQ_HEAD(, aiocblist) so_aiojobq; /* AIO ops waiting on socket */
108 
109 	struct sockbuf so_rcv, so_snd;
110 
111 	struct	ucred *so_cred;		/* (a) user credentials */
112 	struct	label *so_label;	/* (b) MAC label for socket */
113 	struct	label *so_peerlabel;	/* (b) cached MAC label for peer */
114 	/* NB: generation count must not be first. */
115 	so_gen_t so_gencnt;		/* (h) generation count */
116 	void	*so_emuldata;		/* (b) private data for emulators */
117  	struct so_accf {
118 		struct	accept_filter *so_accept_filter;
119 		void	*so_accept_filter_arg;	/* saved filter args */
120 		char	*so_accept_filter_str;	/* saved user args */
121 	} *so_accf;
122 	struct	osd	osd;		/* Object Specific extensions */
123 	/*
124 	 * so_fibnum, so_user_cookie and friends can be used to attach
125 	 * some user-specified metadata to a socket, which then can be
126 	 * used by the kernel for various actions.
127 	 * so_user_cookie is used by ipfw/dummynet.
128 	 */
129 	int so_fibnum;		/* routing domain for this socket */
130 	uint32_t so_user_cookie;
131 };
132 
133 /*
134  * Global accept mutex to serialize access to accept queues and
135  * fields associated with multiple sockets.  This allows us to
136  * avoid defining a lock order between listen and accept sockets
137  * until such time as it proves to be a good idea.
138  */
139 extern struct mtx accept_mtx;
140 #define	ACCEPT_LOCK_ASSERT()		mtx_assert(&accept_mtx, MA_OWNED)
141 #define	ACCEPT_UNLOCK_ASSERT()		mtx_assert(&accept_mtx, MA_NOTOWNED)
142 #define	ACCEPT_LOCK()			mtx_lock(&accept_mtx)
143 #define	ACCEPT_UNLOCK()			mtx_unlock(&accept_mtx)
144 
145 /*
146  * Per-socket mutex: we reuse the receive socket buffer mutex for space
147  * efficiency.  This decision should probably be revisited as we optimize
148  * locking for the socket code.
149  */
150 #define	SOCK_MTX(_so)			SOCKBUF_MTX(&(_so)->so_rcv)
151 #define	SOCK_LOCK(_so)			SOCKBUF_LOCK(&(_so)->so_rcv)
152 #define	SOCK_OWNED(_so)			SOCKBUF_OWNED(&(_so)->so_rcv)
153 #define	SOCK_UNLOCK(_so)		SOCKBUF_UNLOCK(&(_so)->so_rcv)
154 #define	SOCK_LOCK_ASSERT(_so)		SOCKBUF_LOCK_ASSERT(&(_so)->so_rcv)
155 
156 /*
157  * Socket state bits stored in so_qstate.
158  */
159 #define	SQ_INCOMP		0x0800	/* unaccepted, incomplete connection */
160 #define	SQ_COMP			0x1000	/* unaccepted, complete connection */
161 
162 /*
163  * Externalized form of struct socket used by the sysctl(3) interface.
164  */
165 struct xsocket {
166 	size_t	xso_len;	/* length of this structure */
167 	struct	socket *xso_so;	/* makes a convenient handle sometimes */
168 	short	so_type;
169 	short	so_options;
170 	short	so_linger;
171 	short	so_state;
172 	caddr_t	so_pcb;		/* another convenient handle */
173 	int	xso_protocol;
174 	int	xso_family;
175 	u_short	so_qlen;
176 	u_short	so_incqlen;
177 	u_short	so_qlimit;
178 	short	so_timeo;
179 	u_short	so_error;
180 	pid_t	so_pgid;
181 	u_long	so_oobmark;
182 	struct xsockbuf so_rcv, so_snd;
183 	uid_t	so_uid;		/* XXX */
184 };
185 
186 #ifdef _KERNEL
187 
188 /*
189  * Macros for sockets and socket buffering.
190  */
191 
192 /*
193  * Flags to sblock().
194  */
195 #define	SBL_WAIT	0x00000001	/* Wait if not immediately available. */
196 #define	SBL_NOINTR	0x00000002	/* Force non-interruptible sleep. */
197 #define	SBL_VALID	(SBL_WAIT | SBL_NOINTR)
198 
199 /*
200  * Do we need to notify the other side when I/O is possible?
201  */
202 #define	sb_notify(sb)	(((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \
203     SB_UPCALL | SB_AIO | SB_KNOTE)) != 0)
204 
205 /* do we have to send all at once on a socket? */
206 #define	sosendallatonce(so) \
207     ((so)->so_proto->pr_flags & PR_ATOMIC)
208 
209 /* can we read something from so? */
210 #define	soreadabledata(so) \
211     (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || \
212 	!TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error)
213 #define	soreadable(so) \
214 	(soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE))
215 
216 /* can we write something to so? */
217 #define	sowriteable(so) \
218     ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
219 	(((so)->so_state&SS_ISCONNECTED) || \
220 	  ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
221      ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
222      (so)->so_error)
223 
224 /*
225  * soref()/sorele() ref-count the socket structure.  Note that you must
226  * still explicitly close the socket, but the last ref count will free
227  * the structure.
228  */
229 #define	soref(so) do {							\
230 	SOCK_LOCK_ASSERT(so);						\
231 	++(so)->so_count;						\
232 } while (0)
233 
234 #define	sorele(so) do {							\
235 	ACCEPT_LOCK_ASSERT();						\
236 	SOCK_LOCK_ASSERT(so);						\
237 	if ((so)->so_count <= 0)					\
238 		panic("sorele");					\
239 	if (--(so)->so_count == 0)					\
240 		sofree(so);						\
241 	else {								\
242 		SOCK_UNLOCK(so);					\
243 		ACCEPT_UNLOCK();					\
244 	}								\
245 } while (0)
246 
247 /*
248  * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
249  * avoid a non-atomic test-and-wakeup.  However, sowakeup is
250  * responsible for releasing the lock if it is called.  We unlock only
251  * if we don't call into sowakeup.  If any code is introduced that
252  * directly invokes the underlying sowakeup() primitives, it must
253  * maintain the same semantics.
254  */
255 #define	sorwakeup_locked(so) do {					\
256 	SOCKBUF_LOCK_ASSERT(&(so)->so_rcv);				\
257 	if (sb_notify(&(so)->so_rcv))					\
258 		sowakeup((so), &(so)->so_rcv);	 			\
259 	else								\
260 		SOCKBUF_UNLOCK(&(so)->so_rcv);				\
261 } while (0)
262 
263 #define	sorwakeup(so) do {						\
264 	SOCKBUF_LOCK(&(so)->so_rcv);					\
265 	sorwakeup_locked(so);						\
266 } while (0)
267 
268 #define	sowwakeup_locked(so) do {					\
269 	SOCKBUF_LOCK_ASSERT(&(so)->so_snd);				\
270 	if (sb_notify(&(so)->so_snd))					\
271 		sowakeup((so), &(so)->so_snd); 				\
272 	else								\
273 		SOCKBUF_UNLOCK(&(so)->so_snd);				\
274 } while (0)
275 
276 #define	sowwakeup(so) do {						\
277 	SOCKBUF_LOCK(&(so)->so_snd);					\
278 	sowwakeup_locked(so);						\
279 } while (0)
280 
281 struct accept_filter {
282 	char	accf_name[16];
283 	int	(*accf_callback)
284 		(struct socket *so, void *arg, int waitflag);
285 	void *	(*accf_create)
286 		(struct socket *so, char *arg);
287 	void	(*accf_destroy)
288 		(struct socket *so);
289 	SLIST_ENTRY(accept_filter) accf_next;
290 };
291 
292 #ifdef MALLOC_DECLARE
293 MALLOC_DECLARE(M_ACCF);
294 MALLOC_DECLARE(M_PCB);
295 MALLOC_DECLARE(M_SONAME);
296 #endif
297 
298 /*
299  * Socket specific helper hook point identifiers
300  * Do not leave holes in the sequence, hook registration is a loop.
301  */
302 #define HHOOK_SOCKET_OPT		0
303 #define HHOOK_SOCKET_CREATE		1
304 #define HHOOK_SOCKET_RCV 		2
305 #define HHOOK_SOCKET_SND		3
306 #define HHOOK_FILT_SOREAD		4
307 #define HHOOK_FILT_SOWRITE		5
308 #define HHOOK_SOCKET_CLOSE		6
309 #define HHOOK_SOCKET_LAST		HHOOK_SOCKET_CLOSE
310 
311 struct socket_hhook_data {
312 	struct socket	*so;
313 	struct mbuf	*m;
314 	void		*hctx;		/* hook point specific data*/
315 	int		status;
316 };
317 
318 extern int	maxsockets;
319 extern u_long	sb_max;
320 extern so_gen_t so_gencnt;
321 
322 struct file;
323 struct filedesc;
324 struct mbuf;
325 struct sockaddr;
326 struct ucred;
327 struct uio;
328 
329 /* 'which' values for socket upcalls. */
330 #define	SO_RCV		1
331 #define	SO_SND		2
332 
333 /* Return values for socket upcalls. */
334 #define	SU_OK		0
335 #define	SU_ISCONNECTED	1
336 
337 /*
338  * From uipc_socket and friends
339  */
340 int	sockargs(struct mbuf **mp, caddr_t buf, int buflen, int type);
341 int	getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len);
342 int	getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp,
343 	    struct file **fpp, u_int *fflagp);
344 void	soabort(struct socket *so);
345 int	soaccept(struct socket *so, struct sockaddr **nam);
346 int	socheckuid(struct socket *so, uid_t uid);
347 int	sobind(struct socket *so, struct sockaddr *nam, struct thread *td);
348 int	sobindat(int fd, struct socket *so, struct sockaddr *nam,
349 	    struct thread *td);
350 int	soclose(struct socket *so);
351 int	soconnect(struct socket *so, struct sockaddr *nam, struct thread *td);
352 int	soconnectat(int fd, struct socket *so, struct sockaddr *nam,
353 	    struct thread *td);
354 int	soconnect2(struct socket *so1, struct socket *so2);
355 int	socreate(int dom, struct socket **aso, int type, int proto,
356 	    struct ucred *cred, struct thread *td);
357 int	sodisconnect(struct socket *so);
358 struct	sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags);
359 void	sofree(struct socket *so);
360 void	sohasoutofband(struct socket *so);
361 int	solisten(struct socket *so, int backlog, struct thread *td);
362 void	solisten_proto(struct socket *so, int backlog);
363 int	solisten_proto_check(struct socket *so);
364 struct socket *
365 	sonewconn(struct socket *head, int connstatus);
366 
367 
368 int	sopoll(struct socket *so, int events, struct ucred *active_cred,
369 	    struct thread *td);
370 int	sopoll_generic(struct socket *so, int events,
371 	    struct ucred *active_cred, struct thread *td);
372 int	soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio,
373 	    struct mbuf **mp0, struct mbuf **controlp, int *flagsp);
374 int	soreceive_stream(struct socket *so, struct sockaddr **paddr,
375 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
376 	    int *flagsp);
377 int	soreceive_dgram(struct socket *so, struct sockaddr **paddr,
378 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
379 	    int *flagsp);
380 int	soreceive_generic(struct socket *so, struct sockaddr **paddr,
381 	    struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
382 	    int *flagsp);
383 int	soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
384 void	sorflush(struct socket *so);
385 int	sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
386 	    struct mbuf *top, struct mbuf *control, int flags,
387 	    struct thread *td);
388 int	sosend_dgram(struct socket *so, struct sockaddr *addr,
389 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
390 	    int flags, struct thread *td);
391 int	sosend_generic(struct socket *so, struct sockaddr *addr,
392 	    struct uio *uio, struct mbuf *top, struct mbuf *control,
393 	    int flags, struct thread *td);
394 int	soshutdown(struct socket *so, int how);
395 void	sotoxsocket(struct socket *so, struct xsocket *xso);
396 void	soupcall_clear(struct socket *so, int which);
397 void	soupcall_set(struct socket *so, int which,
398 	    int (*func)(struct socket *, void *, int), void *arg);
399 void	sowakeup(struct socket *so, struct sockbuf *sb);
400 int	selsocket(struct socket *so, int events, struct timeval *tv,
401 	    struct thread *td);
402 
403 /*
404  * Accept filter functions (duh).
405  */
406 int	accept_filt_add(struct accept_filter *filt);
407 int	accept_filt_del(char *name);
408 struct	accept_filter *accept_filt_get(char *name);
409 #ifdef ACCEPT_FILTER_MOD
410 #ifdef SYSCTL_DECL
411 SYSCTL_DECL(_net_inet_accf);
412 #endif
413 int	accept_filt_generic_mod_event(module_t mod, int event, void *data);
414 #endif
415 
416 #endif /* _KERNEL */
417 
418 #endif /* !_SYS_SOCKETVAR_H_ */
419