2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1982, 1986, 1990, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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.
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
31 * @(#)socketvar.h 8.3 (Berkeley) 2/19/95
36 #ifndef _SYS_SOCKETVAR_H_
37 #define _SYS_SOCKETVAR_H_
40 * Socket generation count type. Also used in xinpcb, xtcpcb, xunpcb.
42 typedef uint64_t so_gen_t;
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>
51 #include <sys/sockbuf.h>
53 #include <sys/caprights.h>
54 #include <sys/sockopt.h>
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.
65 typedef int so_upcall_t(struct socket *, void *, int);
66 typedef void so_dtor_t(struct socket *);
72 SQ_INCOMP = 0x0800, /* on sol_incomp */
73 SQ_COMP = 0x1000, /* on sol_comp */
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 SOCKBUF_LOCK(&so->so_rcv).
81 * (cs) locked by SOCKBUF_LOCK(&so->so_snd).
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.
87 TAILQ_HEAD(accept_queue, socket);
90 volatile u_int so_count; /* (b / refcount) */
91 struct selinfo so_rdsel; /* (b/cr) for so_rcv/so_comp */
92 struct selinfo so_wrsel; /* (b/cs) for so_snd */
93 short so_type; /* (a) generic type, see socket.h */
94 int so_options; /* (b) from socket call, see socket.h */
95 short so_linger; /* time to linger close(2) */
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_timeo; /* (g) connection timeout */
101 u_short so_error; /* (f) error affecting connection */
102 u_short so_rerror; /* (f) error affecting connection */
103 struct sigio *so_sigio; /* [sg] information for async I/O or
104 out of band data (SIGURG) */
105 struct ucred *so_cred; /* (a) user credentials */
106 struct label *so_label; /* (b) MAC label for socket */
107 /* NB: generation count must not be first. */
108 so_gen_t so_gencnt; /* (h) generation count */
109 void *so_emuldata; /* (b) private data for emulators */
110 so_dtor_t *so_dtor; /* (b) optional destructor */
111 struct osd osd; /* Object Specific extensions */
113 * so_fibnum, so_user_cookie and friends can be used to attach
114 * some user-specified metadata to a socket, which then can be
115 * used by the kernel for various actions.
116 * so_user_cookie is used by ipfw/dummynet.
118 int so_fibnum; /* routing domain for this socket */
119 uint32_t so_user_cookie;
121 int so_ts_clock; /* type of the clock used for timestamps */
122 uint32_t so_max_pacing_rate; /* (f) TX rate limit in bytes/s */
124 /* Regular (data flow) socket. */
126 /* (cr, cs) Receive and send buffers. */
127 struct sockbuf so_rcv, so_snd;
129 /* (e) Our place on accept queue. */
130 TAILQ_ENTRY(socket) so_list;
131 struct socket *so_listen; /* (b) */
132 enum socket_qstate so_qstate; /* (b) */
133 /* (b) cached MAC label for peer */
134 struct label *so_peerlabel;
135 u_long so_oobmark; /* chars to oob mark */
138 * Listening socket, where accepts occur, is so_listen in all
139 * subsidiary sockets. If so_listen is NULL, socket is not
140 * related to an accept. For a listening socket itself
141 * sol_incomp queues partially completed connections, while
142 * sol_comp is a queue of connections ready to be accepted.
143 * If a connection is aborted and it has so_listen set, then
144 * it has to be pulled out of either sol_incomp or sol_comp.
145 * We allow connections to queue up based on current queue
146 * lengths and limit on number of queued connections for this
150 /* (e) queue of partial unaccepted connections */
151 struct accept_queue sol_incomp;
152 /* (e) queue of complete unaccepted connections */
153 struct accept_queue sol_comp;
154 u_int sol_qlen; /* (e) sol_comp length */
155 u_int sol_incqlen; /* (e) sol_incomp length */
156 u_int sol_qlimit; /* (e) queue limit */
158 /* accept_filter(9) optional data */
159 struct accept_filter *sol_accept_filter;
160 void *sol_accept_filter_arg; /* saved filter args */
161 char *sol_accept_filter_str; /* saved user args */
163 /* Optional upcall, for kernel socket. */
164 so_upcall_t *sol_upcall; /* (e) */
165 void *sol_upcallarg; /* (e) */
167 /* Socket buffer parameters, to be copied to
168 * dataflow sockets, accepted from this one. */
171 u_int sol_sbrcv_hiwat;
172 u_int sol_sbsnd_hiwat;
173 short sol_sbrcv_flags;
174 short sol_sbsnd_flags;
175 sbintime_t sol_sbrcv_timeo;
176 sbintime_t sol_sbsnd_timeo;
180 #endif /* defined(_KERNEL) || defined(_WANT_SOCKET) */
185 * Historically, this bits were all kept in the so_state field. For
186 * locking reasons, they are now in multiple fields, as they are
187 * locked differently. so_state maintains basic socket state protected
188 * by the socket lock. so_qstate holds information about the socket
189 * accept queues. Each socket buffer also has a state field holding
190 * information relevant to that socket buffer (can't send, rcv). Many
191 * fields will be read without locks to improve performance and avoid
192 * lock order issues. However, this approach must be used with caution.
194 #define SS_NOFDREF 0x0001 /* no file table ref any more */
195 #define SS_ISCONNECTED 0x0002 /* socket connected to a peer */
196 #define SS_ISCONNECTING 0x0004 /* in process of connecting to peer */
197 #define SS_ISDISCONNECTING 0x0008 /* in process of disconnecting */
198 #define SS_NBIO 0x0100 /* non-blocking ops */
199 #define SS_ASYNC 0x0200 /* async i/o notify */
200 #define SS_ISCONFIRMING 0x0400 /* deciding to accept connection req */
201 #define SS_ISDISCONNECTED 0x2000 /* socket disconnected from peer */
204 * Protocols can mark a socket as SS_PROTOREF to indicate that, following
205 * pru_detach, they still want the socket to persist, and will free it
206 * themselves when they are done. Protocols should only ever call sofree()
207 * following setting this flag in pru_detach(), and never otherwise, as
208 * sofree() bypasses socket reference counting.
210 #define SS_PROTOREF 0x4000 /* strong protocol reference */
214 #define SOCK_MTX(so) &(so)->so_lock
215 #define SOCK_LOCK(so) mtx_lock(&(so)->so_lock)
216 #define SOCK_OWNED(so) mtx_owned(&(so)->so_lock)
217 #define SOCK_UNLOCK(so) mtx_unlock(&(so)->so_lock)
218 #define SOCK_LOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_OWNED)
219 #define SOCK_UNLOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_NOTOWNED)
221 #define SOLISTENING(sol) (((sol)->so_options & SO_ACCEPTCONN) != 0)
222 #define SOLISTEN_LOCK(sol) do { \
223 mtx_lock(&(sol)->so_lock); \
224 KASSERT(SOLISTENING(sol), \
225 ("%s: %p not listening", __func__, (sol))); \
227 #define SOLISTEN_TRYLOCK(sol) mtx_trylock(&(sol)->so_lock)
228 #define SOLISTEN_UNLOCK(sol) do { \
229 KASSERT(SOLISTENING(sol), \
230 ("%s: %p not listening", __func__, (sol))); \
231 mtx_unlock(&(sol)->so_lock); \
233 #define SOLISTEN_LOCK_ASSERT(sol) do { \
234 mtx_assert(&(sol)->so_lock, MA_OWNED); \
235 KASSERT(SOLISTENING(sol), \
236 ("%s: %p not listening", __func__, (sol))); \
240 * Macros for sockets and socket buffering.
246 #define SBL_WAIT 0x00000001 /* Wait if not immediately available. */
247 #define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */
248 #define SBL_VALID (SBL_WAIT | SBL_NOINTR)
251 * Do we need to notify the other side when I/O is possible?
253 #define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \
254 SB_UPCALL | SB_AIO | SB_KNOTE)) != 0)
256 /* do we have to send all at once on a socket? */
257 #define sosendallatonce(so) \
258 ((so)->so_proto->pr_flags & PR_ATOMIC)
260 /* can we read something from so? */
261 #define soreadabledata(so) \
262 (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || \
263 (so)->so_error || (so)->so_rerror)
264 #define soreadable(so) \
265 (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE))
267 /* can we write something to so? */
268 #define sowriteable(so) \
269 ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
270 (((so)->so_state&SS_ISCONNECTED) || \
271 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
272 ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
276 * soref()/sorele() ref-count the socket structure.
277 * soref() may be called without owning socket lock, but in that case a
278 * caller must own something that holds socket, and so_count must be not 0.
279 * Note that you must still explicitly close the socket, but the last ref
280 * count will free the structure.
282 #define soref(so) refcount_acquire(&(so)->so_count)
283 #define sorele(so) do { \
284 SOCK_LOCK_ASSERT(so); \
285 if (refcount_release(&(so)->so_count)) \
292 * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
293 * avoid a non-atomic test-and-wakeup. However, sowakeup is
294 * responsible for releasing the lock if it is called. We unlock only
295 * if we don't call into sowakeup. If any code is introduced that
296 * directly invokes the underlying sowakeup() primitives, it must
297 * maintain the same semantics.
299 #define sorwakeup_locked(so) do { \
300 SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \
301 if (sb_notify(&(so)->so_rcv)) \
302 sowakeup((so), &(so)->so_rcv); \
304 SOCKBUF_UNLOCK(&(so)->so_rcv); \
307 #define sorwakeup(so) do { \
308 SOCKBUF_LOCK(&(so)->so_rcv); \
309 sorwakeup_locked(so); \
312 #define sowwakeup_locked(so) do { \
313 SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \
314 if (sb_notify(&(so)->so_snd)) \
315 sowakeup((so), &(so)->so_snd); \
317 SOCKBUF_UNLOCK(&(so)->so_snd); \
320 #define sowwakeup(so) do { \
321 SOCKBUF_LOCK(&(so)->so_snd); \
322 sowwakeup_locked(so); \
325 struct accept_filter {
328 (struct socket *so, void *arg, int waitflag);
329 void * (*accf_create)
330 (struct socket *so, char *arg);
333 SLIST_ENTRY(accept_filter) accf_next;
336 #define ACCEPT_FILTER_DEFINE(modname, filtname, cb, create, destroy, ver) \
337 static struct accept_filter modname##_filter = { \
338 .accf_name = filtname, \
339 .accf_callback = cb, \
340 .accf_create = create, \
341 .accf_destroy = destroy, \
343 static moduledata_t modname##_mod = { \
344 .name = __XSTRING(modname), \
345 .evhand = accept_filt_generic_mod_event, \
346 .priv = &modname##_filter, \
348 DECLARE_MODULE(modname, modname##_mod, SI_SUB_DRIVERS, \
350 MODULE_VERSION(modname, ver)
352 #ifdef MALLOC_DECLARE
353 MALLOC_DECLARE(M_ACCF);
354 MALLOC_DECLARE(M_PCB);
355 MALLOC_DECLARE(M_SONAME);
359 * Socket specific helper hook point identifiers
360 * Do not leave holes in the sequence, hook registration is a loop.
362 #define HHOOK_SOCKET_OPT 0
363 #define HHOOK_SOCKET_CREATE 1
364 #define HHOOK_SOCKET_RCV 2
365 #define HHOOK_SOCKET_SND 3
366 #define HHOOK_FILT_SOREAD 4
367 #define HHOOK_FILT_SOWRITE 5
368 #define HHOOK_SOCKET_CLOSE 6
369 #define HHOOK_SOCKET_LAST HHOOK_SOCKET_CLOSE
371 struct socket_hhook_data {
374 void *hctx; /* hook point specific data*/
378 extern int maxsockets;
379 extern u_long sb_max;
380 extern so_gen_t so_gencnt;
390 /* 'which' values for socket upcalls. */
394 /* Return values for socket upcalls. */
396 #define SU_ISCONNECTED 1
399 * From uipc_socket and friends
401 int getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len);
402 int getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp,
403 struct file **fpp, u_int *fflagp, struct filecaps *havecaps);
404 void soabort(struct socket *so);
405 int soaccept(struct socket *so, struct sockaddr **nam);
406 void soaio_enqueue(struct task *task);
407 void soaio_rcv(void *context, int pending);
408 void soaio_snd(void *context, int pending);
409 int socheckuid(struct socket *so, uid_t uid);
410 int sobind(struct socket *so, struct sockaddr *nam, struct thread *td);
411 int sobindat(int fd, struct socket *so, struct sockaddr *nam,
413 int soclose(struct socket *so);
414 int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td);
415 int soconnectat(int fd, struct socket *so, struct sockaddr *nam,
417 int soconnect2(struct socket *so1, struct socket *so2);
418 int socreate(int dom, struct socket **aso, int type, int proto,
419 struct ucred *cred, struct thread *td);
420 int sodisconnect(struct socket *so);
421 void sodtor_set(struct socket *, so_dtor_t *);
422 struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags);
423 void sofree(struct socket *so);
424 void sohasoutofband(struct socket *so);
425 int solisten(struct socket *so, int backlog, struct thread *td);
426 void solisten_proto(struct socket *so, int backlog);
427 int solisten_proto_check(struct socket *so);
428 int solisten_dequeue(struct socket *, struct socket **, int);
430 sonewconn(struct socket *head, int connstatus);
432 sopeeloff(struct socket *);
433 int sopoll(struct socket *so, int events, struct ucred *active_cred,
435 int sopoll_generic(struct socket *so, int events,
436 struct ucred *active_cred, struct thread *td);
437 int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio,
438 struct mbuf **mp0, struct mbuf **controlp, int *flagsp);
439 int soreceive_stream(struct socket *so, struct sockaddr **paddr,
440 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
442 int soreceive_dgram(struct socket *so, struct sockaddr **paddr,
443 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
445 int soreceive_generic(struct socket *so, struct sockaddr **paddr,
446 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
448 int soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
449 void sorflush(struct socket *so);
450 int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
451 struct mbuf *top, struct mbuf *control, int flags,
453 int sosend_dgram(struct socket *so, struct sockaddr *addr,
454 struct uio *uio, struct mbuf *top, struct mbuf *control,
455 int flags, struct thread *td);
456 int sosend_generic(struct socket *so, struct sockaddr *addr,
457 struct uio *uio, struct mbuf *top, struct mbuf *control,
458 int flags, struct thread *td);
459 int soshutdown(struct socket *so, int how);
460 void soupcall_clear(struct socket *, int);
461 void soupcall_set(struct socket *, int, so_upcall_t, void *);
462 void solisten_upcall_set(struct socket *, so_upcall_t, void *);
463 void sowakeup(struct socket *so, struct sockbuf *sb);
464 void sowakeup_aio(struct socket *so, struct sockbuf *sb);
465 void solisten_wakeup(struct socket *);
466 int selsocket(struct socket *so, int events, struct timeval *tv,
468 void soisconnected(struct socket *so);
469 void soisconnecting(struct socket *so);
470 void soisdisconnected(struct socket *so);
471 void soisdisconnecting(struct socket *so);
472 void socantrcvmore(struct socket *so);
473 void socantrcvmore_locked(struct socket *so);
474 void socantsendmore(struct socket *so);
475 void socantsendmore_locked(struct socket *so);
476 void soroverflow(struct socket *so);
477 void soroverflow_locked(struct socket *so);
480 * Accept filter functions (duh).
482 int accept_filt_add(struct accept_filter *filt);
483 int accept_filt_del(char *name);
484 struct accept_filter *accept_filt_get(char *name);
485 #ifdef ACCEPT_FILTER_MOD
487 SYSCTL_DECL(_net_inet_accf);
489 int accept_filt_generic_mod_event(module_t mod, int event, void *data);
495 * Structure to export socket from kernel to utilities, via sysctl(3).
498 ksize_t xso_len; /* length of this structure */
499 kvaddr_t xso_so; /* kernel address of struct socket */
500 kvaddr_t so_pcb; /* kernel address of struct inpcb */
502 int64_t so_spare64[8];
503 int32_t xso_protocol;
510 int32_t so_spare32[8];
531 void sotoxsocket(struct socket *so, struct xsocket *xso);
532 void sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb);
536 * Socket buffer state bits. Exported via libprocstat(3).
538 #define SBS_CANTSENDMORE 0x0010 /* can't send more data to peer */
539 #define SBS_CANTRCVMORE 0x0020 /* can't receive more data from peer */
540 #define SBS_RCVATMARK 0x0040 /* at mark on input */
542 #endif /* !_SYS_SOCKETVAR_H_ */