socketvar.h revision 195423
150477Speter/*- 231706Spst * Copyright (c) 1982, 1986, 1990, 1993 326931Smsmith * The Regents of the University of California. All rights reserved. 41590Srgrimes * 5265420Simp * Redistribution and use in source and binary forms, with or without 6160750Syar * modification, are permitted provided that the following conditions 731706Spst * are met: 831706Spst * 1. Redistributions of source code must retain the above copyright 931706Spst * notice, this list of conditions and the following disclaimer. 1031706Spst * 2. Redistributions in binary form must reproduce the above copyright 11223328Sgavin * notice, this list of conditions and the following disclaimer in the 12223328Sgavin * documentation and/or other materials provided with the distribution. 1387824Smikeh * 4. Neither the name of the University nor the names of its contributors 141590Srgrimes * may be used to endorse or promote products derived from this software 15223328Sgavin * without specific prior written permission. 16223328Sgavin * 171590Srgrimes * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 18160750Syar * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 19160750Syar * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 20160750Syar * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 21160750Syar * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 22223328Sgavin * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 23275042Sbapt * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 2426931Smsmith * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 25201386Sed * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 26201386Sed * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 2731706Spst * SUCH DAMAGE. 2831706Spst * 2931706Spst * @(#)socketvar.h 8.3 (Berkeley) 2/19/95 3031706Spst * 3126931Smsmith * $FreeBSD: head/sys/sys/socketvar.h 195423 2009-07-07 09:43:44Z kib $ 321590Srgrimes */ 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/_sx.h> 42#include <sys/sockbuf.h> 43#include <sys/sockstate.h> 44#ifdef _KERNEL 45#include <sys/sockopt.h> 46#endif 47 48struct vnet; 49 50/* 51 * Kernel structure per socket. 52 * Contains send and receive buffer queues, 53 * handle on protocol and pointer to protocol 54 * private data and error information. 55 */ 56typedef u_quad_t so_gen_t; 57 58struct socket; 59 60/*- 61 * Locking key to struct socket: 62 * (a) constant after allocation, no locking required. 63 * (b) locked by SOCK_LOCK(so). 64 * (c) locked by SOCKBUF_LOCK(&so->so_rcv). 65 * (d) locked by SOCKBUF_LOCK(&so->so_snd). 66 * (e) locked by ACCEPT_LOCK(). 67 * (f) not locked since integer reads/writes are atomic. 68 * (g) used only as a sleep/wakeup address, no value. 69 * (h) locked by global mutex so_global_mtx. 70 */ 71struct socket { 72 int so_count; /* (b) reference count */ 73 short so_type; /* (a) generic type, see socket.h */ 74 short so_options; /* from socket call, see socket.h */ 75 short so_linger; /* time to linger while closing */ 76 short so_state; /* (b) internal state flags SS_* */ 77 int so_qstate; /* (e) internal state flags SQ_* */ 78 void *so_pcb; /* protocol control block */ 79 struct vnet *so_vnet; /* network stack instance */ 80 struct protosw *so_proto; /* (a) protocol handle */ 81/* 82 * Variables for connection queuing. 83 * Socket where accepts occur is so_head in all subsidiary sockets. 84 * If so_head is 0, socket is not related to an accept. 85 * For head socket so_incomp queues partially completed connections, 86 * while so_comp is a queue of connections ready to be accepted. 87 * If a connection is aborted and it has so_head set, then 88 * it has to be pulled out of either so_incomp or so_comp. 89 * We allow connections to queue up based on current queue lengths 90 * and limit on number of queued connections for this socket. 91 */ 92 struct socket *so_head; /* (e) back pointer to listen socket */ 93 TAILQ_HEAD(, socket) so_incomp; /* (e) queue of partial unaccepted connections */ 94 TAILQ_HEAD(, socket) so_comp; /* (e) queue of complete unaccepted connections */ 95 TAILQ_ENTRY(socket) so_list; /* (e) list of unaccepted connections */ 96 u_short so_qlen; /* (e) number of unaccepted connections */ 97 u_short so_incqlen; /* (e) number of unaccepted incomplete 98 connections */ 99 u_short so_qlimit; /* (e) max number queued connections */ 100 short so_timeo; /* (g) connection timeout */ 101 u_short so_error; /* (f) error affecting connection */ 102 struct sigio *so_sigio; /* [sg] information for async I/O or 103 out of band data (SIGURG) */ 104 u_long so_oobmark; /* (c) chars to oob mark */ 105 TAILQ_HEAD(, aiocblist) so_aiojobq; /* AIO ops waiting on socket */ 106 107 struct sockbuf so_rcv, so_snd; 108 109 struct ucred *so_cred; /* (a) user credentials */ 110 struct label *so_label; /* (b) MAC label for socket */ 111 struct label *so_peerlabel; /* (b) cached MAC label for peer */ 112 /* NB: generation count must not be first. */ 113 so_gen_t so_gencnt; /* (h) generation count */ 114 void *so_emuldata; /* (b) private data for emulators */ 115 struct so_accf { 116 struct accept_filter *so_accept_filter; 117 void *so_accept_filter_arg; /* saved filter args */ 118 char *so_accept_filter_str; /* saved user args */ 119 } *so_accf; 120 int so_fibnum; /* routing domain for this socket */ 121}; 122 123/* 124 * Global accept mutex to serialize access to accept queues and 125 * fields associated with multiple sockets. This allows us to 126 * avoid defining a lock order between listen and accept sockets 127 * until such time as it proves to be a good idea. 128 */ 129extern struct mtx accept_mtx; 130#define ACCEPT_LOCK_ASSERT() mtx_assert(&accept_mtx, MA_OWNED) 131#define ACCEPT_UNLOCK_ASSERT() mtx_assert(&accept_mtx, MA_NOTOWNED) 132#define ACCEPT_LOCK() mtx_lock(&accept_mtx) 133#define ACCEPT_UNLOCK() mtx_unlock(&accept_mtx) 134 135/* 136 * Per-socket mutex: we reuse the receive socket buffer mutex for space 137 * efficiency. This decision should probably be revisited as we optimize 138 * locking for the socket code. 139 */ 140#define SOCK_MTX(_so) SOCKBUF_MTX(&(_so)->so_rcv) 141#define SOCK_LOCK(_so) SOCKBUF_LOCK(&(_so)->so_rcv) 142#define SOCK_OWNED(_so) SOCKBUF_OWNED(&(_so)->so_rcv) 143#define SOCK_UNLOCK(_so) SOCKBUF_UNLOCK(&(_so)->so_rcv) 144#define SOCK_LOCK_ASSERT(_so) SOCKBUF_LOCK_ASSERT(&(_so)->so_rcv) 145 146/* 147 * Socket state bits stored in so_qstate. 148 */ 149#define SQ_INCOMP 0x0800 /* unaccepted, incomplete connection */ 150#define SQ_COMP 0x1000 /* unaccepted, complete connection */ 151 152/* 153 * Externalized form of struct socket used by the sysctl(3) interface. 154 */ 155struct xsocket { 156 size_t xso_len; /* length of this structure */ 157 struct socket *xso_so; /* makes a convenient handle sometimes */ 158 short so_type; 159 short so_options; 160 short so_linger; 161 short so_state; 162 caddr_t so_pcb; /* another convenient handle */ 163 int xso_protocol; 164 int xso_family; 165 u_short so_qlen; 166 u_short so_incqlen; 167 u_short so_qlimit; 168 short so_timeo; 169 u_short so_error; 170 pid_t so_pgid; 171 u_long so_oobmark; 172 struct xsockbuf so_rcv, so_snd; 173 uid_t so_uid; /* XXX */ 174}; 175 176#ifdef _KERNEL 177 178/* 179 * Macros for sockets and socket buffering. 180 */ 181 182/* 183 * Flags to sblock(). 184 */ 185#define SBL_WAIT 0x00000001 /* Wait if not immediately available. */ 186#define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */ 187#define SBL_VALID (SBL_WAIT | SBL_NOINTR) 188 189/* 190 * Do we need to notify the other side when I/O is possible? 191 */ 192#define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \ 193 SB_UPCALL | SB_AIO | SB_KNOTE)) != 0) 194 195/* do we have to send all at once on a socket? */ 196#define sosendallatonce(so) \ 197 ((so)->so_proto->pr_flags & PR_ATOMIC) 198 199/* can we read something from so? */ 200#define soreadabledata(so) \ 201 ((so)->so_rcv.sb_cc >= (so)->so_rcv.sb_lowat || \ 202 !TAILQ_EMPTY(&(so)->so_comp) || (so)->so_error) 203#define soreadable(so) \ 204 (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE)) 205 206/* can we write something to so? */ 207#define sowriteable(so) \ 208 ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \ 209 (((so)->so_state&SS_ISCONNECTED) || \ 210 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \ 211 ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \ 212 (so)->so_error) 213 214/* 215 * soref()/sorele() ref-count the socket structure. Note that you must 216 * still explicitly close the socket, but the last ref count will free 217 * the structure. 218 */ 219#define soref(so) do { \ 220 SOCK_LOCK_ASSERT(so); \ 221 ++(so)->so_count; \ 222} while (0) 223 224#define sorele(so) do { \ 225 ACCEPT_LOCK_ASSERT(); \ 226 SOCK_LOCK_ASSERT(so); \ 227 if ((so)->so_count <= 0) \ 228 panic("sorele"); \ 229 if (--(so)->so_count == 0) \ 230 sofree(so); \ 231 else { \ 232 SOCK_UNLOCK(so); \ 233 ACCEPT_UNLOCK(); \ 234 } \ 235} while (0) 236 237#define sotryfree(so) do { \ 238 ACCEPT_LOCK_ASSERT(); \ 239 SOCK_LOCK_ASSERT(so); \ 240 if ((so)->so_count == 0) \ 241 sofree(so); \ 242 else { \ 243 SOCK_UNLOCK(so); \ 244 ACCEPT_UNLOCK(); \ 245 } \ 246} while(0) 247 248/* 249 * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to 250 * avoid a non-atomic test-and-wakeup. However, sowakeup is 251 * responsible for releasing the lock if it is called. We unlock only 252 * if we don't call into sowakeup. If any code is introduced that 253 * directly invokes the underlying sowakeup() primitives, it must 254 * maintain the same semantics. 255 */ 256#define sorwakeup_locked(so) do { \ 257 SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \ 258 if (sb_notify(&(so)->so_rcv)) \ 259 sowakeup((so), &(so)->so_rcv); \ 260 else \ 261 SOCKBUF_UNLOCK(&(so)->so_rcv); \ 262} while (0) 263 264#define sorwakeup(so) do { \ 265 SOCKBUF_LOCK(&(so)->so_rcv); \ 266 sorwakeup_locked(so); \ 267} while (0) 268 269#define sowwakeup_locked(so) do { \ 270 SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \ 271 if (sb_notify(&(so)->so_snd)) \ 272 sowakeup((so), &(so)->so_snd); \ 273 else \ 274 SOCKBUF_UNLOCK(&(so)->so_snd); \ 275} while (0) 276 277#define sowwakeup(so) do { \ 278 SOCKBUF_LOCK(&(so)->so_snd); \ 279 sowwakeup_locked(so); \ 280} while (0) 281 282struct accept_filter { 283 char accf_name[16]; 284 int (*accf_callback) 285 (struct socket *so, void *arg, int waitflag); 286 void * (*accf_create) 287 (struct socket *so, char *arg); 288 void (*accf_destroy) 289 (struct socket *so); 290 SLIST_ENTRY(accept_filter) accf_next; 291}; 292 293#ifdef MALLOC_DECLARE 294MALLOC_DECLARE(M_ACCF); 295MALLOC_DECLARE(M_PCB); 296MALLOC_DECLARE(M_SONAME); 297#endif 298 299extern int maxsockets; 300extern u_long sb_max; 301extern struct uma_zone *socket_zone; 302extern so_gen_t so_gencnt; 303 304struct mbuf; 305struct sockaddr; 306struct ucred; 307struct uio; 308 309/* 'which' values for socket upcalls. */ 310#define SO_RCV 1 311#define SO_SND 2 312 313/* Return values for socket upcalls. */ 314#define SU_OK 0 315#define SU_ISCONNECTED 1 316 317/* 318 * From uipc_socket and friends 319 */ 320int sockargs(struct mbuf **mp, caddr_t buf, int buflen, int type); 321int getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len); 322void soabort(struct socket *so); 323int soaccept(struct socket *so, struct sockaddr **nam); 324int socheckuid(struct socket *so, uid_t uid); 325int sobind(struct socket *so, struct sockaddr *nam, struct thread *td); 326int soclose(struct socket *so); 327int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td); 328int soconnect2(struct socket *so1, struct socket *so2); 329int socow_setup(struct mbuf *m0, struct uio *uio); 330int socreate(int dom, struct socket **aso, int type, int proto, 331 struct ucred *cred, struct thread *td); 332int sodisconnect(struct socket *so); 333struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags); 334void sofree(struct socket *so); 335void sohasoutofband(struct socket *so); 336int solisten(struct socket *so, int backlog, struct thread *td); 337void solisten_proto(struct socket *so, int backlog); 338int solisten_proto_check(struct socket *so); 339struct socket * 340 sonewconn(struct socket *head, int connstatus); 341 342 343int sopoll(struct socket *so, int events, struct ucred *active_cred, 344 struct thread *td); 345int sopoll_generic(struct socket *so, int events, 346 struct ucred *active_cred, struct thread *td); 347int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio, 348 struct mbuf **mp0, struct mbuf **controlp, int *flagsp); 349int soreceive_stream(struct socket *so, struct sockaddr **paddr, 350 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 351 int *flagsp); 352int soreceive_dgram(struct socket *so, struct sockaddr **paddr, 353 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 354 int *flagsp); 355int soreceive_generic(struct socket *so, struct sockaddr **paddr, 356 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 357 int *flagsp); 358int soreserve(struct socket *so, u_long sndcc, u_long rcvcc); 359void sorflush(struct socket *so); 360int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, 361 struct mbuf *top, struct mbuf *control, int flags, 362 struct thread *td); 363int sosend_dgram(struct socket *so, struct sockaddr *addr, 364 struct uio *uio, struct mbuf *top, struct mbuf *control, 365 int flags, struct thread *td); 366int sosend_generic(struct socket *so, struct sockaddr *addr, 367 struct uio *uio, struct mbuf *top, struct mbuf *control, 368 int flags, struct thread *td); 369int soshutdown(struct socket *so, int how); 370void sotoxsocket(struct socket *so, struct xsocket *xso); 371void soupcall_clear(struct socket *so, int which); 372void soupcall_set(struct socket *so, int which, 373 int (*func)(struct socket *, void *, int), void *arg); 374void sowakeup(struct socket *so, struct sockbuf *sb); 375int selsocket(struct socket *so, int events, struct timeval *tv, 376 struct thread *td); 377 378/* 379 * Accept filter functions (duh). 380 */ 381int accept_filt_add(struct accept_filter *filt); 382int accept_filt_del(char *name); 383struct accept_filter *accept_filt_get(char *name); 384#ifdef ACCEPT_FILTER_MOD 385#ifdef SYSCTL_DECL 386SYSCTL_DECL(_net_inet_accf); 387#endif 388int accept_filt_generic_mod_event(module_t mod, int event, void *data); 389#endif 390 391#endif /* _KERNEL */ 392 393#endif /* !_SYS_SOCKETVAR_H_ */ 394