2 * Copyright (c) 1982, 1986, 1990, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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. 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.
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
29 * @(#)socketvar.h 8.3 (Berkeley) 2/19/95
34 #ifndef _SYS_SOCKETVAR_H_
35 #define _SYS_SOCKETVAR_H_
38 * Socket generation count type. Also used in xinpcb, xtcpcb, xunpcb.
40 typedef uint64_t so_gen_t;
42 #if defined(_KERNEL) || defined(_WANT_SOCKET)
43 #include <sys/queue.h> /* for TAILQ macros */
44 #include <sys/selinfo.h> /* for struct selinfo */
45 #include <sys/_lock.h>
46 #include <sys/_mutex.h>
49 #include <sys/sockbuf.h>
51 #include <sys/caprights.h>
52 #include <sys/sockopt.h>
58 * Kernel structure per socket.
59 * Contains send and receive buffer queues,
60 * handle on protocol and pointer to protocol
61 * private data and error information.
63 typedef int so_upcall_t(struct socket *, void *, int);
68 * Locking key to struct socket:
69 * (a) constant after allocation, no locking required.
70 * (b) locked by SOCK_LOCK(so).
71 * (cr) locked by SOCKBUF_LOCK(&so->so_rcv).
72 * (cs) locked by SOCKBUF_LOCK(&so->so_rcv).
73 * (e) locked by SOLISTEN_LOCK() of corresponding listening socket.
74 * (f) not locked since integer reads/writes are atomic.
75 * (g) used only as a sleep/wakeup address, no value.
76 * (h) locked by global mutex so_global_mtx.
78 TAILQ_HEAD(accept_queue, socket);
81 volatile u_int so_count; /* (b / refcount) */
82 struct selinfo so_rdsel; /* (b/cr) for so_rcv/so_comp */
83 struct selinfo so_wrsel; /* (b/cs) for so_snd */
84 short so_type; /* (a) generic type, see socket.h */
85 short so_options; /* (b) from socket call, see socket.h */
86 short so_linger; /* time to linger close(2) */
87 short so_state; /* (b) internal state flags SS_* */
88 void *so_pcb; /* protocol control block */
89 struct vnet *so_vnet; /* (a) network stack instance */
90 struct protosw *so_proto; /* (a) protocol handle */
91 short so_timeo; /* (g) connection timeout */
92 u_short so_error; /* (f) error affecting connection */
93 struct sigio *so_sigio; /* [sg] information for async I/O or
94 out of band data (SIGURG) */
95 struct ucred *so_cred; /* (a) user credentials */
96 struct label *so_label; /* (b) MAC label for socket */
97 /* NB: generation count must not be first. */
98 so_gen_t so_gencnt; /* (h) generation count */
99 void *so_emuldata; /* (b) private data for emulators */
100 struct osd osd; /* Object Specific extensions */
102 * so_fibnum, so_user_cookie and friends can be used to attach
103 * some user-specified metadata to a socket, which then can be
104 * used by the kernel for various actions.
105 * so_user_cookie is used by ipfw/dummynet.
107 int so_fibnum; /* routing domain for this socket */
108 uint32_t so_user_cookie;
110 int so_ts_clock; /* type of the clock used for timestamps */
111 uint32_t so_max_pacing_rate; /* (f) TX rate limit in bytes/s */
113 /* Regular (data flow) socket. */
115 /* (cr, cs) Receive and send buffers. */
116 struct sockbuf so_rcv, so_snd;
118 /* (e) Our place on accept queue. */
119 TAILQ_ENTRY(socket) so_list;
120 struct socket *so_listen; /* (b) */
123 SQ_INCOMP = 0x0800, /* on sol_incomp */
124 SQ_COMP = 0x1000, /* on sol_comp */
125 } so_qstate; /* (b) */
127 /* (b) cached MAC label for peer */
128 struct label *so_peerlabel;
129 u_long so_oobmark; /* chars to oob mark */
132 * Listening socket, where accepts occur, is so_listen in all
133 * subsidiary sockets. If so_listen is NULL, socket is not
134 * related to an accept. For a listening socket itself
135 * sol_incomp queues partially completed connections, while
136 * sol_comp is a queue of connections ready to be accepted.
137 * If a connection is aborted and it has so_listen set, then
138 * it has to be pulled out of either sol_incomp or sol_comp.
139 * We allow connections to queue up based on current queue
140 * lengths and limit on number of queued connections for this
144 /* (e) queue of partial unaccepted connections */
145 struct accept_queue sol_incomp;
146 /* (e) queue of complete unaccepted connections */
147 struct accept_queue sol_comp;
148 u_int sol_qlen; /* (e) sol_comp length */
149 u_int sol_incqlen; /* (e) sol_incomp length */
150 u_int sol_qlimit; /* (e) queue limit */
152 /* accept_filter(9) optional data */
153 struct accept_filter *sol_accept_filter;
154 void *sol_accept_filter_arg; /* saved filter args */
155 char *sol_accept_filter_str; /* saved user args */
157 /* Optional upcall, for kernel socket. */
158 so_upcall_t *sol_upcall; /* (e) */
159 void *sol_upcallarg; /* (e) */
161 /* Socket buffer parameters, to be copied to
162 * dataflow sockets, accepted from this one. */
165 u_int sol_sbrcv_hiwat;
166 u_int sol_sbsnd_hiwat;
167 short sol_sbrcv_flags;
168 short sol_sbsnd_flags;
169 sbintime_t sol_sbrcv_timeo;
170 sbintime_t sol_sbsnd_timeo;
174 #endif /* defined(_KERNEL) || defined(_WANT_SOCKET) */
179 * Historically, this bits were all kept in the so_state field. For
180 * locking reasons, they are now in multiple fields, as they are
181 * locked differently. so_state maintains basic socket state protected
182 * by the socket lock. so_qstate holds information about the socket
183 * accept queues. Each socket buffer also has a state field holding
184 * information relevant to that socket buffer (can't send, rcv). Many
185 * fields will be read without locks to improve performance and avoid
186 * lock order issues. However, this approach must be used with caution.
188 #define SS_NOFDREF 0x0001 /* no file table ref any more */
189 #define SS_ISCONNECTED 0x0002 /* socket connected to a peer */
190 #define SS_ISCONNECTING 0x0004 /* in process of connecting to peer */
191 #define SS_ISDISCONNECTING 0x0008 /* in process of disconnecting */
192 #define SS_NBIO 0x0100 /* non-blocking ops */
193 #define SS_ASYNC 0x0200 /* async i/o notify */
194 #define SS_ISCONFIRMING 0x0400 /* deciding to accept connection req */
195 #define SS_ISDISCONNECTED 0x2000 /* socket disconnected from peer */
198 * Protocols can mark a socket as SS_PROTOREF to indicate that, following
199 * pru_detach, they still want the socket to persist, and will free it
200 * themselves when they are done. Protocols should only ever call sofree()
201 * following setting this flag in pru_detach(), and never otherwise, as
202 * sofree() bypasses socket reference counting.
204 #define SS_PROTOREF 0x4000 /* strong protocol reference */
208 #define SOCK_MTX(so) &(so)->so_lock
209 #define SOCK_LOCK(so) mtx_lock(&(so)->so_lock)
210 #define SOCK_OWNED(so) mtx_owned(&(so)->so_lock)
211 #define SOCK_UNLOCK(so) mtx_unlock(&(so)->so_lock)
212 #define SOCK_LOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_OWNED)
213 #define SOCK_UNLOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_NOTOWNED)
215 #define SOLISTENING(sol) (((sol)->so_options & SO_ACCEPTCONN) != 0)
216 #define SOLISTEN_LOCK(sol) do { \
217 mtx_lock(&(sol)->so_lock); \
218 KASSERT(SOLISTENING(sol), \
219 ("%s: %p not listening", __func__, (sol))); \
221 #define SOLISTEN_TRYLOCK(sol) mtx_trylock(&(sol)->so_lock)
222 #define SOLISTEN_UNLOCK(sol) do { \
223 KASSERT(SOLISTENING(sol), \
224 ("%s: %p not listening", __func__, (sol))); \
225 mtx_unlock(&(sol)->so_lock); \
227 #define SOLISTEN_LOCK_ASSERT(sol) do { \
228 mtx_assert(&(sol)->so_lock, MA_OWNED); \
229 KASSERT(SOLISTENING(sol), \
230 ("%s: %p not listening", __func__, (sol))); \
234 * Macros for sockets and socket buffering.
240 #define SBL_WAIT 0x00000001 /* Wait if not immediately available. */
241 #define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */
242 #define SBL_VALID (SBL_WAIT | SBL_NOINTR)
245 * Do we need to notify the other side when I/O is possible?
247 #define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \
248 SB_UPCALL | SB_AIO | SB_KNOTE)) != 0)
250 /* do we have to send all at once on a socket? */
251 #define sosendallatonce(so) \
252 ((so)->so_proto->pr_flags & PR_ATOMIC)
254 /* can we read something from so? */
255 #define soreadabledata(so) \
256 (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || (so)->so_error)
257 #define soreadable(so) \
258 (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE))
260 /* can we write something to so? */
261 #define sowriteable(so) \
262 ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \
263 (((so)->so_state&SS_ISCONNECTED) || \
264 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \
265 ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \
269 * soref()/sorele() ref-count the socket structure.
270 * soref() may be called without owning socket lock, but in that case a
271 * caller must own something that holds socket, and so_count must be not 0.
272 * Note that you must still explicitly close the socket, but the last ref
273 * count will free the structure.
275 #define soref(so) refcount_acquire(&(so)->so_count)
276 #define sorele(so) do { \
277 SOCK_LOCK_ASSERT(so); \
278 if (refcount_release(&(so)->so_count)) \
285 * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to
286 * avoid a non-atomic test-and-wakeup. However, sowakeup is
287 * responsible for releasing the lock if it is called. We unlock only
288 * if we don't call into sowakeup. If any code is introduced that
289 * directly invokes the underlying sowakeup() primitives, it must
290 * maintain the same semantics.
292 #define sorwakeup_locked(so) do { \
293 SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \
294 if (sb_notify(&(so)->so_rcv)) \
295 sowakeup((so), &(so)->so_rcv); \
297 SOCKBUF_UNLOCK(&(so)->so_rcv); \
300 #define sorwakeup(so) do { \
301 SOCKBUF_LOCK(&(so)->so_rcv); \
302 sorwakeup_locked(so); \
305 #define sowwakeup_locked(so) do { \
306 SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \
307 if (sb_notify(&(so)->so_snd)) \
308 sowakeup((so), &(so)->so_snd); \
310 SOCKBUF_UNLOCK(&(so)->so_snd); \
313 #define sowwakeup(so) do { \
314 SOCKBUF_LOCK(&(so)->so_snd); \
315 sowwakeup_locked(so); \
318 struct accept_filter {
321 (struct socket *so, void *arg, int waitflag);
322 void * (*accf_create)
323 (struct socket *so, char *arg);
326 SLIST_ENTRY(accept_filter) accf_next;
329 #ifdef MALLOC_DECLARE
330 MALLOC_DECLARE(M_ACCF);
331 MALLOC_DECLARE(M_PCB);
332 MALLOC_DECLARE(M_SONAME);
336 * Socket specific helper hook point identifiers
337 * Do not leave holes in the sequence, hook registration is a loop.
339 #define HHOOK_SOCKET_OPT 0
340 #define HHOOK_SOCKET_CREATE 1
341 #define HHOOK_SOCKET_RCV 2
342 #define HHOOK_SOCKET_SND 3
343 #define HHOOK_FILT_SOREAD 4
344 #define HHOOK_FILT_SOWRITE 5
345 #define HHOOK_SOCKET_CLOSE 6
346 #define HHOOK_SOCKET_LAST HHOOK_SOCKET_CLOSE
348 struct socket_hhook_data {
351 void *hctx; /* hook point specific data*/
355 extern int maxsockets;
356 extern u_long sb_max;
357 extern so_gen_t so_gencnt;
367 /* 'which' values for socket upcalls. */
371 /* Return values for socket upcalls. */
373 #define SU_ISCONNECTED 1
376 * From uipc_socket and friends
378 int getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len);
379 int getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp,
380 struct file **fpp, u_int *fflagp, struct filecaps *havecaps);
381 void soabort(struct socket *so);
382 int soaccept(struct socket *so, struct sockaddr **nam);
383 void soaio_enqueue(struct task *task);
384 void soaio_rcv(void *context, int pending);
385 void soaio_snd(void *context, int pending);
386 int socheckuid(struct socket *so, uid_t uid);
387 int sobind(struct socket *so, struct sockaddr *nam, struct thread *td);
388 int sobindat(int fd, struct socket *so, struct sockaddr *nam,
390 int soclose(struct socket *so);
391 int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td);
392 int soconnectat(int fd, struct socket *so, struct sockaddr *nam,
394 int soconnect2(struct socket *so1, struct socket *so2);
395 int socreate(int dom, struct socket **aso, int type, int proto,
396 struct ucred *cred, struct thread *td);
397 int sodisconnect(struct socket *so);
398 struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags);
399 void sofree(struct socket *so);
400 void sohasoutofband(struct socket *so);
401 int solisten(struct socket *so, int backlog, struct thread *td);
402 void solisten_proto(struct socket *so, int backlog);
403 int solisten_proto_check(struct socket *so);
404 int solisten_dequeue(struct socket *, struct socket **, int);
406 sonewconn(struct socket *head, int connstatus);
408 sopeeloff(struct socket *);
409 int sopoll(struct socket *so, int events, struct ucred *active_cred,
411 int sopoll_generic(struct socket *so, int events,
412 struct ucred *active_cred, struct thread *td);
413 int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio,
414 struct mbuf **mp0, struct mbuf **controlp, int *flagsp);
415 int soreceive_stream(struct socket *so, struct sockaddr **paddr,
416 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
418 int soreceive_dgram(struct socket *so, struct sockaddr **paddr,
419 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
421 int soreceive_generic(struct socket *so, struct sockaddr **paddr,
422 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp,
424 int soreserve(struct socket *so, u_long sndcc, u_long rcvcc);
425 void sorflush(struct socket *so);
426 int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio,
427 struct mbuf *top, struct mbuf *control, int flags,
429 int sosend_dgram(struct socket *so, struct sockaddr *addr,
430 struct uio *uio, struct mbuf *top, struct mbuf *control,
431 int flags, struct thread *td);
432 int sosend_generic(struct socket *so, struct sockaddr *addr,
433 struct uio *uio, struct mbuf *top, struct mbuf *control,
434 int flags, struct thread *td);
435 int soshutdown(struct socket *so, int how);
436 void soupcall_clear(struct socket *, int);
437 void soupcall_set(struct socket *, int, so_upcall_t, void *);
438 void solisten_upcall_set(struct socket *, so_upcall_t, void *);
439 void sowakeup(struct socket *so, struct sockbuf *sb);
440 void sowakeup_aio(struct socket *so, struct sockbuf *sb);
441 void solisten_wakeup(struct socket *);
442 int selsocket(struct socket *so, int events, struct timeval *tv,
444 void soisconnected(struct socket *so);
445 void soisconnecting(struct socket *so);
446 void soisdisconnected(struct socket *so);
447 void soisdisconnecting(struct socket *so);
448 void socantrcvmore(struct socket *so);
449 void socantrcvmore_locked(struct socket *so);
450 void socantsendmore(struct socket *so);
451 void socantsendmore_locked(struct socket *so);
454 * Accept filter functions (duh).
456 int accept_filt_add(struct accept_filter *filt);
457 int accept_filt_del(char *name);
458 struct accept_filter *accept_filt_get(char *name);
459 #ifdef ACCEPT_FILTER_MOD
461 SYSCTL_DECL(_net_inet_accf);
463 int accept_filt_generic_mod_event(module_t mod, int event, void *data);
469 * Structure to export socket from kernel to utilities, via sysctl(3).
472 size_t xso_len; /* length of this structure */
474 void *xso_so; /* kernel address of struct socket */
478 void *so_pcb; /* kernel address of struct inpcb */
482 int64_t so_spare64[8];
483 int32_t xso_protocol;
490 int32_t so_spare32[8];
511 void sotoxsocket(struct socket *so, struct xsocket *xso);
512 void sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb);
516 * Socket buffer state bits. Exported via libprocstat(3).
518 #define SBS_CANTSENDMORE 0x0010 /* can't send more data to peer */
519 #define SBS_CANTRCVMORE 0x0020 /* can't receive more data from peer */
520 #define SBS_RCVATMARK 0x0040 /* at mark on input */
522 #endif /* !_SYS_SOCKETVAR_H_ */