xref: /dragonfly/sys/sys/socketvar.h (revision 92fc8b5c)
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  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)socketvar.h	8.3 (Berkeley) 2/19/95
34  * $FreeBSD: src/sys/sys/socketvar.h,v 1.46.2.10 2003/08/24 08:24:39 hsu Exp $
35  * $DragonFly: src/sys/sys/socketvar.h,v 1.35 2008/08/28 23:15:45 dillon Exp $
36  */
37 
38 #ifndef _SYS_SOCKETVAR_H_
39 #define _SYS_SOCKETVAR_H_
40 
41 #ifndef _SYS_TYPES_H_
42 #include <sys/types.h>
43 #endif
44 #ifndef _SYS_QUEUE_H_
45 #include <sys/queue.h>			/* for TAILQ macros */
46 #endif
47 #ifndef _SYS_EVENT_H_
48 #include <sys/event.h>			/* for struct kqinfo */
49 #endif
50 #ifndef _SYS_THREAD_H_
51 #include <sys/thread.h>			/* for struct lwkt_token */
52 #endif
53 #ifndef _SYS_SOCKBUF_H_
54 #include <sys/sockbuf.h>
55 #endif
56 
57 #if defined(_KERNEL) || defined(_KERNEL_STRUCTURES)
58 
59 struct accept_filter;
60 
61 /*
62  * Signaling socket buffers contain additional elements for locking
63  * and signaling conditions.  These are used primarily by sockets.
64  *
65  * WARNING: See partial clearing of fields in kern/uipc_socket.c
66  *	    sorflush() and sowflush().
67  */
68 struct signalsockbuf {
69 	struct sockbuf sb;
70 	struct kqinfo ssb_kq;	/* process selecting read/write */
71 	uint32_t ssb_flags;	/* flags, see below (use atomic ops) */
72 	u_int	ssb_timeo;	/* timeout for read/write */
73 	long	ssb_lowat;	/* low water mark */
74 	u_long	ssb_hiwat;	/* high water mark / max actual char count */
75 	u_long	ssb_mbmax;	/* max chars of mbufs to use */
76 	struct lwkt_token ssb_token; /* frontend/backend serializer */
77 };
78 
79 #define ssb_cc		sb.sb_cc	/* commonly used fields */
80 #define ssb_mb		sb.sb_mb	/* commonly used fields */
81 #define ssb_mbcnt	sb.sb_mbcnt	/* commonly used fields */
82 
83 #define	SSB_LOCK	0x0001		/* lock on data queue */
84 #define	SSB_WANT	0x0002		/* someone is waiting to lock */
85 #define	SSB_WAIT	0x0004		/* someone is waiting for data/space */
86 #define	SSB_ASYNC	0x0010		/* ASYNC I/O, need signals */
87 #define	SSB_UPCALL	0x0020		/* someone wants an upcall */
88 #define	SSB_NOINTR	0x0040		/* operations not interruptible */
89 /*#define SSB_AIO	0x0080*/	/* AIO operations queued */
90 #define SSB_KNOTE	0x0100		/* kernel note attached */
91 #define SSB_MEVENT	0x0200		/* need message event notification */
92 #define SSB_STOP	0x0400		/* backpressure indicator */
93 #define	SSB_AUTOSIZE	0x0800		/* automatically size socket buffer */
94 #define SSB_AUTOLOWAT	0x1000		/* automatically scale lowat */
95 #define SSB_WAKEUP	0x2000		/* wakeup event race */
96 
97 #define SSB_CLEAR_MASK	(SSB_ASYNC | SSB_UPCALL | SSB_STOP | \
98 			 SSB_AUTOSIZE | SSB_AUTOLOWAT)
99 
100 #define SSB_NOTIFY_MASK	(SSB_WAIT | SSB_ASYNC | SSB_UPCALL | \
101 			 SSB_KNOTE | SSB_MEVENT)
102 
103 /*
104  * Per-socket kernel structure.  Contains universal send and receive queues,
105  * protocol control handle, and error information.
106  */
107 struct socket {
108 	short	so_type;		/* generic type, see socket.h */
109 	short	so_options;		/* from socket call, see socket.h */
110 	short	so_linger;		/* time to linger while closing */
111 	short	so_state;		/* internal state flags SS_*, below */
112 	void	*so_pcb;		/* protocol control block */
113 	struct	protosw *so_proto;	/* protocol handle */
114 	struct	socket *so_head;	/* back pointer to accept socket */
115 	lwkt_port_t so_port;		/* message port */
116 
117 	/*
118 	 * These fields are used to manage sockets capable of accepting
119 	 * new connections.
120 	 */
121 	TAILQ_HEAD(, socket) so_incomp;	/* in-progress, incomplete */
122 	TAILQ_HEAD(, socket) so_comp;	/* completed but not yet accepted */
123 	TAILQ_ENTRY(socket) so_list;	/* list of unaccepted connections */
124 	short	so_qlen;		/* so_comp count */
125 	short	so_incqlen;		/* so_incomp count */
126 	short	so_qlimit;		/* max number queued connections */
127 
128 	/*
129 	 * Misc socket support
130 	 */
131 	short	so_timeo;		/* connection timeout */
132 	u_short	so_error;		/* error affecting connection */
133 	struct  sigio *so_sigio;	/* information for async I/O or
134 					   out of band data (SIGURG) */
135 	u_long	so_oobmark;		/* chars to oob mark */
136 	TAILQ_HEAD(, aiocblist) so_aiojobq; /* AIO ops waiting on socket */
137 	struct signalsockbuf so_rcv;
138 	struct signalsockbuf so_snd;
139 
140 	void	(*so_upcall) (struct socket *, void *, int);
141 	void	*so_upcallarg;
142 	struct	ucred *so_cred;		/* user credentials */
143 	/* NB: generation count must not be first; easiest to make it last. */
144 	void	*so_emuldata;		/* private data for emulators */
145 	int	so_refs;		/* shutdown refs */
146 	struct	so_accf {
147 		struct	accept_filter *so_accept_filter;
148 		void	*so_accept_filter_arg;	/* saved filter args */
149 		char	*so_accept_filter_str;	/* saved user args */
150 	} *so_accf;
151 };
152 
153 #endif
154 
155 /*
156  * Socket state bits.
157  *
158  * NOTE: The following states are interlocked with so_refs:
159  *
160  *	SS_NOFDREF	so_refs while not set
161  *	(so_pcb)	so_refs while set
162  */
163 #define	SS_NOFDREF		0x0001	/* no file table ref any more */
164 #define	SS_ISCONNECTED		0x0002	/* socket connected to a peer */
165 #define	SS_ISCONNECTING		0x0004	/* in process of connecting to peer */
166 #define	SS_ISDISCONNECTING	0x0008	/* in process of disconnecting */
167 #define	SS_CANTSENDMORE		0x0010	/* can't send more data to peer */
168 #define	SS_CANTRCVMORE		0x0020	/* can't receive more data from peer */
169 #define	SS_RCVATMARK		0x0040	/* at mark on input */
170 
171 #define	SS_ASSERTINPROG		0x0100	/* sonewconn race debugging */
172 #define	SS_ASYNC		0x0200	/* async i/o notify */
173 #define	SS_ISCONFIRMING		0x0400	/* deciding to accept connection req */
174 
175 #define	SS_INCOMP		0x0800	/* unaccepted, incomplete connection */
176 #define	SS_COMP			0x1000	/* unaccepted, complete connection */
177 #define	SS_ISDISCONNECTED	0x2000	/* socket disconnected from peer */
178 
179 /*
180  * Externalized form of struct socket used by the sysctl(3) interface.
181  */
182 struct	xsocket {
183 	size_t	xso_len;	/* length of this structure */
184 	struct	socket *xso_so;	/* makes a convenient handle sometimes */
185 	short	so_type;
186 	short	so_options;
187 	short	so_linger;
188 	short	so_state;
189 	void	*so_pcb;		/* another convenient handle */
190 	int	xso_protocol;
191 	int	xso_family;
192 	short	so_qlen;
193 	short	so_incqlen;
194 	short	so_qlimit;
195 	short	so_timeo;
196 	u_short	so_error;
197 	pid_t	so_pgid;
198 	u_long	so_oobmark;
199 	struct	xsockbuf {
200 		u_long	sb_cc;
201 		u_long	sb_hiwat;
202 		u_long	sb_mbcnt;
203 		u_long	sb_mbmax;
204 		long	sb_lowat;
205 		u_int	sb_timeo;
206 		short	sb_flags;
207 	} so_rcv, so_snd;
208 	uid_t	so_uid;		/* XXX */
209 };
210 
211 /*
212  * Macros for sockets and socket buffering.
213  */
214 
215 #define	sosendallatonce(so) \
216     ((so)->so_proto->pr_flags & PR_ATOMIC)
217 
218 /* can we read something from so? */
219 #define	soreadable(so) \
220     ((so)->so_rcv.ssb_cc >= (so)->so_rcv.ssb_lowat || \
221 	((so)->so_state & SS_CANTRCVMORE) || \
222 	!TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error)
223 
224 /* can we write something to so? */
225 #define	sowriteable(so) \
226     ((ssb_space(&(so)->so_snd) >= (so)->so_snd.ssb_lowat && \
227 	(((so)->so_state&SS_ISCONNECTED) || \
228 	  ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
229      ((so)->so_state & SS_CANTSENDMORE) || \
230      (so)->so_error)
231 
232 /*
233  * Do we need to notify the other side when I/O is possible?
234  *
235  * NOTE: Interlock for ssb_wait/wakeup.  The protocol side will set
236  *	 SSB_WAKEUP asynchronously and this can race, so if it isn't
237  *	 set we have to go through the full-on notification check.
238  *	 If it is set but no waiting ever takes place it simply
239  *	 remains set.
240  */
241 #define ssb_notify(ssb)					\
242 	    (((ssb)->ssb_flags & SSB_NOTIFY_MASK) ||	\
243 	     ((ssb)->ssb_flags & SSB_WAKEUP) == 0)
244 
245 /* do we have to send all at once on a socket? */
246 
247 #ifdef _KERNEL
248 
249 /*
250  * How much space is there in a socket buffer (so->so_snd or so->so_rcv)?
251  * This is problematical if the fields are unsigned, as the space might
252  * still be negative (cc > hiwat or mbcnt > mbmax).  Should detect
253  * overflow and return 0.
254  *
255  * SSB_STOP ignores cc/hiwat and returns 0.  This is used by unix domain
256  * stream sockets to signal backpressure.
257  */
258 static __inline
259 long
260 ssb_space(struct signalsockbuf *ssb)
261 {
262 	long bleft;
263 	long mleft;
264 
265 	if (ssb->ssb_flags & SSB_STOP)
266 		return(0);
267 	bleft = ssb->ssb_hiwat - ssb->ssb_cc;
268 	mleft = ssb->ssb_mbmax - ssb->ssb_mbcnt;
269 	return((bleft < mleft) ? bleft : mleft);
270 }
271 
272 #endif
273 
274 #define ssb_append(ssb, m)						\
275 	sbappend(&(ssb)->sb, m)
276 
277 #define ssb_appendstream(ssb, m)					\
278 	sbappendstream(&(ssb)->sb, m)
279 
280 #define ssb_appendrecord(ssb, m)					\
281 	sbappendrecord(&(ssb)->sb, m)
282 
283 #define ssb_appendaddr(ssb, src, m, control)				\
284 	((ssb_space(ssb) <= 0) ? 0 : sbappendaddr(&(ssb)->sb, src, m, control))
285 
286 #define ssb_appendcontrol(ssb, m, control)				\
287 	((ssb_space(ssb) <= 0) ? 0 : sbappendcontrol(&(ssb)->sb, m, control))
288 
289 #define ssb_insert_knote(ssb, kn) {					\
290 	knote_insert(&(ssb)->ssb_kq.ki_note, kn);			\
291 	atomic_set_int(&(ssb)->ssb_flags, SSB_KNOTE);			\
292 }
293 
294 #define ssb_remove_knote(ssb, kn) {					\
295 	knote_remove(&(ssb)->ssb_kq.ki_note, kn);			\
296 	if (SLIST_EMPTY(&(ssb)->ssb_kq.ki_note))			\
297 		atomic_clear_int(&(ssb)->ssb_flags, SSB_KNOTE);		\
298 }
299 
300 #define	sorwakeup(so)						\
301 	do {							\
302 		if (ssb_notify(&(so)->so_rcv))			\
303 			sowakeup((so), &(so)->so_rcv);		\
304 	} while (0)
305 
306 #define	sowwakeup(so)						\
307 	do {							\
308 		if (ssb_notify(&(so)->so_snd))			\
309 			sowakeup((so), &(so)->so_snd);		\
310 	} while (0)
311 
312 #ifdef _KERNEL
313 
314 /*
315  * Argument structure for sosetopt et seq.  This is in the KERNEL
316  * section because it will never be visible to user code.
317  */
318 enum sopt_dir { SOPT_GET, SOPT_SET };
319 struct sockopt {
320 	enum	sopt_dir sopt_dir; /* is this a get or a set? */
321 	int	sopt_level;	/* second arg of [gs]etsockopt */
322 	int	sopt_name;	/* third arg of [gs]etsockopt */
323 	void   *sopt_val;	/* fourth arg of [gs]etsockopt */
324 	size_t	sopt_valsize;	/* (almost) fifth arg of [gs]etsockopt */
325 	struct	thread *sopt_td; /* calling thread or null if kernel */
326 };
327 
328 struct accept_filter {
329 	char	accf_name[16];
330 	void	(*accf_callback)
331 		(struct socket *so, void *arg, int waitflag);
332 	void *	(*accf_create)
333 		(struct socket *so, char *arg);
334 	void	(*accf_destroy)
335 		(struct socket *so);
336 	SLIST_ENTRY(accept_filter) accf_next;	/* next on the list */
337 };
338 
339 #ifdef MALLOC_DECLARE
340 MALLOC_DECLARE(M_PCB);
341 MALLOC_DECLARE(M_SONAME);
342 MALLOC_DECLARE(M_ACCF);
343 #endif
344 
345 extern int	maxsockets;
346 extern u_long	sb_max;		/* nominal limit */
347 extern u_long	sb_max_adj;	/* actual limit used by sbreserve() */
348 
349 struct file;
350 struct filedesc;
351 struct mbuf;
352 struct rlimit;
353 struct sockaddr;
354 struct stat;
355 struct ucred;
356 struct uio;
357 struct knote;
358 struct sysmsg;
359 
360 /*
361  * File operations on sockets.
362  */
363 int	soo_read (struct file *fp, struct uio *uio, struct ucred *cred,
364 			int flags);
365 int	soo_write (struct file *fp, struct uio *uio, struct ucred *cred,
366 			int flags);
367 int	soo_close (struct file *fp);
368 int	soo_shutdown (struct file *fp, int how);
369 int	soo_ioctl (struct file *fp, u_long cmd, caddr_t data,
370 			struct ucred *cred, struct sysmsg *msg);
371 int	soo_stat (struct file *fp, struct stat *ub, struct ucred *cred);
372 int	sokqfilter (struct file *fp, struct knote *kn);
373 
374 /*
375  * From uipc_socket and friends
376  */
377 struct	sockaddr *dup_sockaddr (const struct sockaddr *sa);
378 int	getsockaddr (struct sockaddr **namp, caddr_t uaddr, size_t len);
379 
380 void	ssb_release (struct signalsockbuf *ssb, struct socket *so);
381 int	ssb_reserve (struct signalsockbuf *ssb, u_long cc, struct socket *so,
382 		   struct rlimit *rl);
383 void	ssbtoxsockbuf (struct signalsockbuf *sb, struct xsockbuf *xsb);
384 int	ssb_wait (struct signalsockbuf *sb);
385 int	_ssb_lock (struct signalsockbuf *sb);
386 
387 void	soabort (struct socket *so);
388 void	soaborta (struct socket *so);
389 void	soabort_oncpu (struct socket *so);
390 int	soaccept (struct socket *so, struct sockaddr **nam);
391 struct	socket *soalloc (int waitok);
392 int	sobind (struct socket *so, struct sockaddr *nam, struct thread *td);
393 void	socantrcvmore (struct socket *so);
394 void	socantsendmore (struct socket *so);
395 int	socket_wait (struct socket *so, struct timespec *ts, int *res);
396 int	soclose (struct socket *so, int fflags);
397 int	soconnect (struct socket *so, struct sockaddr *nam, struct thread *td);
398 int	soconnect2 (struct socket *so1, struct socket *so2);
399 int	socreate (int dom, struct socket **aso, int type, int proto,
400 	    struct thread *td);
401 int	sodisconnect (struct socket *so);
402 void	sofree (struct socket *so);
403 int	sogetopt (struct socket *so, struct sockopt *sopt);
404 void	sohasoutofband (struct socket *so);
405 void	soisconnected (struct socket *so);
406 void	soisconnecting (struct socket *so);
407 void	soisdisconnected (struct socket *so);
408 void	soisdisconnecting (struct socket *so);
409 void	soisreconnected (struct socket *so);
410 void	soisreconnecting (struct socket *so);
411 void	sosetport (struct socket *so, struct lwkt_port *port);
412 int	solisten (struct socket *so, int backlog, struct thread *td);
413 struct socket *sonewconn (struct socket *head, int connstatus);
414 int	sooptcopyin (struct sockopt *sopt, void *buf, size_t len,
415 			 size_t minlen);
416 int	soopt_to_kbuf (struct sockopt *sopt, void *buf, size_t len,
417 			 size_t minlen);
418 int	sooptcopyout (struct sockopt *sopt, const void *buf, size_t len);
419 void	soopt_from_kbuf (struct sockopt *sopt, const void *buf, size_t len);
420 
421 /* XXX; prepare mbuf for (__FreeBSD__ < 3) routines. */
422 int	soopt_getm (struct sockopt *sopt, struct mbuf **mp);
423 int	soopt_mcopyin (struct sockopt *sopt, struct mbuf *m);
424 void	soopt_to_mbuf (struct sockopt *sopt, struct mbuf *m);
425 int	soopt_mcopyout (struct sockopt *sopt, struct mbuf *m);
426 int	soopt_from_mbuf (struct sockopt *sopt, struct mbuf *m);
427 
428 int	soreceive (struct socket *so, struct sockaddr **paddr,
429 		       struct uio *uio, struct sockbuf *sio,
430 		       struct mbuf **controlp, int *flagsp);
431 int	soreserve (struct socket *so, u_long sndcc, u_long rcvcc,
432 		   struct rlimit *rl);
433 void	sorflush (struct socket *so);
434 int	sosend (struct socket *so, struct sockaddr *addr, struct uio *uio,
435 		    struct mbuf *top, struct mbuf *control, int flags,
436 		    struct thread *td);
437 int	sosendudp (struct socket *so, struct sockaddr *addr, struct uio *uio,
438 		    struct mbuf *top, struct mbuf *control, int flags,
439 		    struct thread *td);
440 int	sosetopt (struct socket *so, struct sockopt *sopt);
441 int	soshutdown (struct socket *so, int how);
442 void	sotoxsocket (struct socket *so, struct xsocket *xso);
443 void	sowakeup (struct socket *so, struct signalsockbuf *sb);
444 
445 /* accept filter functions */
446 int	accept_filt_add (struct accept_filter *filt);
447 int	accept_filt_del (char *name);
448 struct accept_filter *	accept_filt_get (char *name);
449 #ifdef ACCEPT_FILTER_MOD
450 int accept_filt_generic_mod_event (module_t mod, int event, void *data);
451 SYSCTL_DECL(_net_inet_accf);
452 #endif /* ACCEPT_FILTER_MOD */
453 
454 #endif /* _KERNEL */
455 
456 #endif /* !_SYS_SOCKETVAR_H_ */
457