1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1982, 1986, 1990, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 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. 18 * 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 29 * SUCH DAMAGE. 30 * 31 * @(#)socketvar.h 8.3 (Berkeley) 2/19/95 32 * 33 * $FreeBSD$ 34 */ 35 36 #ifndef _SYS_SOCKETVAR_H_ 37 #define _SYS_SOCKETVAR_H_ 38 39 /* 40 * Socket generation count type. Also used in xinpcb, xtcpcb, xunpcb. 41 */ 42 typedef uint64_t so_gen_t; 43 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> 49 #include <sys/osd.h> 50 #include <sys/_sx.h> 51 #include <sys/sockbuf.h> 52 #ifdef _KERNEL 53 #include <sys/caprights.h> 54 #include <sys/sockopt.h> 55 #endif 56 57 struct vnet; 58 59 /* 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. 64 */ 65 typedef int so_upcall_t(struct socket *, void *, int); 66 67 struct socket; 68 69 /*- 70 * Locking key to struct socket: 71 * (a) constant after allocation, no locking required. 72 * (b) locked by SOCK_LOCK(so). 73 * (cr) locked by SOCKBUF_LOCK(&so->so_rcv). 74 * (cs) locked by SOCKBUF_LOCK(&so->so_snd). 75 * (e) locked by SOLISTEN_LOCK() of corresponding listening socket. 76 * (f) not locked since integer reads/writes are atomic. 77 * (g) used only as a sleep/wakeup address, no value. 78 * (h) locked by global mutex so_global_mtx. 79 */ 80 TAILQ_HEAD(accept_queue, socket); 81 struct socket { 82 struct mtx so_lock; 83 volatile u_int so_count; /* (b / refcount) */ 84 struct selinfo so_rdsel; /* (b/cr) for so_rcv/so_comp */ 85 struct selinfo so_wrsel; /* (b/cs) for so_snd */ 86 short so_type; /* (a) generic type, see socket.h */ 87 short so_options; /* (b) from socket call, see socket.h */ 88 short so_linger; /* time to linger close(2) */ 89 short so_state; /* (b) internal state flags SS_* */ 90 void *so_pcb; /* protocol control block */ 91 struct vnet *so_vnet; /* (a) network stack instance */ 92 struct protosw *so_proto; /* (a) protocol handle */ 93 short so_timeo; /* (g) connection timeout */ 94 u_short so_error; /* (f) error affecting connection */ 95 struct sigio *so_sigio; /* [sg] information for async I/O or 96 out of band data (SIGURG) */ 97 struct ucred *so_cred; /* (a) user credentials */ 98 struct label *so_label; /* (b) MAC label for socket */ 99 /* NB: generation count must not be first. */ 100 so_gen_t so_gencnt; /* (h) generation count */ 101 void *so_emuldata; /* (b) private data for emulators */ 102 struct osd osd; /* Object Specific extensions */ 103 /* 104 * so_fibnum, so_user_cookie and friends can be used to attach 105 * some user-specified metadata to a socket, which then can be 106 * used by the kernel for various actions. 107 * so_user_cookie is used by ipfw/dummynet. 108 */ 109 int so_fibnum; /* routing domain for this socket */ 110 uint32_t so_user_cookie; 111 112 int so_ts_clock; /* type of the clock used for timestamps */ 113 uint32_t so_max_pacing_rate; /* (f) TX rate limit in bytes/s */ 114 union { 115 /* Regular (data flow) socket. */ 116 struct { 117 /* (cr, cs) Receive and send buffers. */ 118 struct sockbuf so_rcv, so_snd; 119 120 /* (e) Our place on accept queue. */ 121 TAILQ_ENTRY(socket) so_list; 122 struct socket *so_listen; /* (b) */ 123 enum { 124 SQ_NONE = 0, 125 SQ_INCOMP = 0x0800, /* on sol_incomp */ 126 SQ_COMP = 0x1000, /* on sol_comp */ 127 } so_qstate; /* (b) */ 128 129 /* (b) cached MAC label for peer */ 130 struct label *so_peerlabel; 131 u_long so_oobmark; /* chars to oob mark */ 132 }; 133 /* 134 * Listening socket, where accepts occur, is so_listen in all 135 * subsidiary sockets. If so_listen is NULL, socket is not 136 * related to an accept. For a listening socket itself 137 * sol_incomp queues partially completed connections, while 138 * sol_comp is a queue of connections ready to be accepted. 139 * If a connection is aborted and it has so_listen set, then 140 * it has to be pulled out of either sol_incomp or sol_comp. 141 * We allow connections to queue up based on current queue 142 * lengths and limit on number of queued connections for this 143 * socket. 144 */ 145 struct { 146 /* (e) queue of partial unaccepted connections */ 147 struct accept_queue sol_incomp; 148 /* (e) queue of complete unaccepted connections */ 149 struct accept_queue sol_comp; 150 u_int sol_qlen; /* (e) sol_comp length */ 151 u_int sol_incqlen; /* (e) sol_incomp length */ 152 u_int sol_qlimit; /* (e) queue limit */ 153 154 /* accept_filter(9) optional data */ 155 struct accept_filter *sol_accept_filter; 156 void *sol_accept_filter_arg; /* saved filter args */ 157 char *sol_accept_filter_str; /* saved user args */ 158 159 /* Optional upcall, for kernel socket. */ 160 so_upcall_t *sol_upcall; /* (e) */ 161 void *sol_upcallarg; /* (e) */ 162 163 /* Socket buffer parameters, to be copied to 164 * dataflow sockets, accepted from this one. */ 165 int sol_sbrcv_lowat; 166 int sol_sbsnd_lowat; 167 u_int sol_sbrcv_hiwat; 168 u_int sol_sbsnd_hiwat; 169 short sol_sbrcv_flags; 170 short sol_sbsnd_flags; 171 sbintime_t sol_sbrcv_timeo; 172 sbintime_t sol_sbsnd_timeo; 173 }; 174 }; 175 }; 176 #endif /* defined(_KERNEL) || defined(_WANT_SOCKET) */ 177 178 /* 179 * Socket state bits. 180 * 181 * Historically, this bits were all kept in the so_state field. For 182 * locking reasons, they are now in multiple fields, as they are 183 * locked differently. so_state maintains basic socket state protected 184 * by the socket lock. so_qstate holds information about the socket 185 * accept queues. Each socket buffer also has a state field holding 186 * information relevant to that socket buffer (can't send, rcv). Many 187 * fields will be read without locks to improve performance and avoid 188 * lock order issues. However, this approach must be used with caution. 189 */ 190 #define SS_NOFDREF 0x0001 /* no file table ref any more */ 191 #define SS_ISCONNECTED 0x0002 /* socket connected to a peer */ 192 #define SS_ISCONNECTING 0x0004 /* in process of connecting to peer */ 193 #define SS_ISDISCONNECTING 0x0008 /* in process of disconnecting */ 194 #define SS_NBIO 0x0100 /* non-blocking ops */ 195 #define SS_ASYNC 0x0200 /* async i/o notify */ 196 #define SS_ISCONFIRMING 0x0400 /* deciding to accept connection req */ 197 #define SS_ISDISCONNECTED 0x2000 /* socket disconnected from peer */ 198 199 /* 200 * Protocols can mark a socket as SS_PROTOREF to indicate that, following 201 * pru_detach, they still want the socket to persist, and will free it 202 * themselves when they are done. Protocols should only ever call sofree() 203 * following setting this flag in pru_detach(), and never otherwise, as 204 * sofree() bypasses socket reference counting. 205 */ 206 #define SS_PROTOREF 0x4000 /* strong protocol reference */ 207 208 #ifdef _KERNEL 209 210 #define SOCK_MTX(so) &(so)->so_lock 211 #define SOCK_LOCK(so) mtx_lock(&(so)->so_lock) 212 #define SOCK_OWNED(so) mtx_owned(&(so)->so_lock) 213 #define SOCK_UNLOCK(so) mtx_unlock(&(so)->so_lock) 214 #define SOCK_LOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_OWNED) 215 #define SOCK_UNLOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_NOTOWNED) 216 217 #define SOLISTENING(sol) (((sol)->so_options & SO_ACCEPTCONN) != 0) 218 #define SOLISTEN_LOCK(sol) do { \ 219 mtx_lock(&(sol)->so_lock); \ 220 KASSERT(SOLISTENING(sol), \ 221 ("%s: %p not listening", __func__, (sol))); \ 222 } while (0) 223 #define SOLISTEN_TRYLOCK(sol) mtx_trylock(&(sol)->so_lock) 224 #define SOLISTEN_UNLOCK(sol) do { \ 225 KASSERT(SOLISTENING(sol), \ 226 ("%s: %p not listening", __func__, (sol))); \ 227 mtx_unlock(&(sol)->so_lock); \ 228 } while (0) 229 #define SOLISTEN_LOCK_ASSERT(sol) do { \ 230 mtx_assert(&(sol)->so_lock, MA_OWNED); \ 231 KASSERT(SOLISTENING(sol), \ 232 ("%s: %p not listening", __func__, (sol))); \ 233 } while (0) 234 235 /* 236 * Macros for sockets and socket buffering. 237 */ 238 239 /* 240 * Flags to sblock(). 241 */ 242 #define SBL_WAIT 0x00000001 /* Wait if not immediately available. */ 243 #define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */ 244 #define SBL_VALID (SBL_WAIT | SBL_NOINTR) 245 246 /* 247 * Do we need to notify the other side when I/O is possible? 248 */ 249 #define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \ 250 SB_UPCALL | SB_AIO | SB_KNOTE)) != 0) 251 252 /* do we have to send all at once on a socket? */ 253 #define sosendallatonce(so) \ 254 ((so)->so_proto->pr_flags & PR_ATOMIC) 255 256 /* can we read something from so? */ 257 #define soreadabledata(so) \ 258 (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || (so)->so_error) 259 #define soreadable(so) \ 260 (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE)) 261 262 /* can we write something to so? */ 263 #define sowriteable(so) \ 264 ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \ 265 (((so)->so_state&SS_ISCONNECTED) || \ 266 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \ 267 ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \ 268 (so)->so_error) 269 270 /* 271 * soref()/sorele() ref-count the socket structure. 272 * soref() may be called without owning socket lock, but in that case a 273 * caller must own something that holds socket, and so_count must be not 0. 274 * Note that you must still explicitly close the socket, but the last ref 275 * count will free the structure. 276 */ 277 #define soref(so) refcount_acquire(&(so)->so_count) 278 #define sorele(so) do { \ 279 SOCK_LOCK_ASSERT(so); \ 280 if (refcount_release(&(so)->so_count)) \ 281 sofree(so); \ 282 else \ 283 SOCK_UNLOCK(so); \ 284 } while (0) 285 286 /* 287 * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to 288 * avoid a non-atomic test-and-wakeup. However, sowakeup is 289 * responsible for releasing the lock if it is called. We unlock only 290 * if we don't call into sowakeup. If any code is introduced that 291 * directly invokes the underlying sowakeup() primitives, it must 292 * maintain the same semantics. 293 */ 294 #define sorwakeup_locked(so) do { \ 295 SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \ 296 if (sb_notify(&(so)->so_rcv)) \ 297 sowakeup((so), &(so)->so_rcv); \ 298 else \ 299 SOCKBUF_UNLOCK(&(so)->so_rcv); \ 300 } while (0) 301 302 #define sorwakeup(so) do { \ 303 SOCKBUF_LOCK(&(so)->so_rcv); \ 304 sorwakeup_locked(so); \ 305 } while (0) 306 307 #define sowwakeup_locked(so) do { \ 308 SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \ 309 if (sb_notify(&(so)->so_snd)) \ 310 sowakeup((so), &(so)->so_snd); \ 311 else \ 312 SOCKBUF_UNLOCK(&(so)->so_snd); \ 313 } while (0) 314 315 #define sowwakeup(so) do { \ 316 SOCKBUF_LOCK(&(so)->so_snd); \ 317 sowwakeup_locked(so); \ 318 } while (0) 319 320 struct accept_filter { 321 char accf_name[16]; 322 int (*accf_callback) 323 (struct socket *so, void *arg, int waitflag); 324 void * (*accf_create) 325 (struct socket *so, char *arg); 326 void (*accf_destroy) 327 (struct socket *so); 328 SLIST_ENTRY(accept_filter) accf_next; 329 }; 330 331 #ifdef MALLOC_DECLARE 332 MALLOC_DECLARE(M_ACCF); 333 MALLOC_DECLARE(M_PCB); 334 MALLOC_DECLARE(M_SONAME); 335 #endif 336 337 /* 338 * Socket specific helper hook point identifiers 339 * Do not leave holes in the sequence, hook registration is a loop. 340 */ 341 #define HHOOK_SOCKET_OPT 0 342 #define HHOOK_SOCKET_CREATE 1 343 #define HHOOK_SOCKET_RCV 2 344 #define HHOOK_SOCKET_SND 3 345 #define HHOOK_FILT_SOREAD 4 346 #define HHOOK_FILT_SOWRITE 5 347 #define HHOOK_SOCKET_CLOSE 6 348 #define HHOOK_SOCKET_LAST HHOOK_SOCKET_CLOSE 349 350 struct socket_hhook_data { 351 struct socket *so; 352 struct mbuf *m; 353 void *hctx; /* hook point specific data*/ 354 int status; 355 }; 356 357 extern int maxsockets; 358 extern u_long sb_max; 359 extern so_gen_t so_gencnt; 360 361 struct file; 362 struct filecaps; 363 struct filedesc; 364 struct mbuf; 365 struct sockaddr; 366 struct ucred; 367 struct uio; 368 369 /* 'which' values for socket upcalls. */ 370 #define SO_RCV 1 371 #define SO_SND 2 372 373 /* Return values for socket upcalls. */ 374 #define SU_OK 0 375 #define SU_ISCONNECTED 1 376 377 /* 378 * From uipc_socket and friends 379 */ 380 int getsockaddr(struct sockaddr **namp, caddr_t uaddr, size_t len); 381 int getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp, 382 struct file **fpp, u_int *fflagp, struct filecaps *havecaps); 383 void soabort(struct socket *so); 384 int soaccept(struct socket *so, struct sockaddr **nam); 385 void soaio_enqueue(struct task *task); 386 void soaio_rcv(void *context, int pending); 387 void soaio_snd(void *context, int pending); 388 int socheckuid(struct socket *so, uid_t uid); 389 int sobind(struct socket *so, struct sockaddr *nam, struct thread *td); 390 int sobindat(int fd, struct socket *so, struct sockaddr *nam, 391 struct thread *td); 392 int soclose(struct socket *so); 393 int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td); 394 int soconnectat(int fd, struct socket *so, struct sockaddr *nam, 395 struct thread *td); 396 int soconnect2(struct socket *so1, struct socket *so2); 397 int socreate(int dom, struct socket **aso, int type, int proto, 398 struct ucred *cred, struct thread *td); 399 int sodisconnect(struct socket *so); 400 struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags); 401 void sofree(struct socket *so); 402 void sohasoutofband(struct socket *so); 403 int solisten(struct socket *so, int backlog, struct thread *td); 404 void solisten_proto(struct socket *so, int backlog); 405 int solisten_proto_check(struct socket *so); 406 int solisten_dequeue(struct socket *, struct socket **, int); 407 struct socket * 408 sonewconn(struct socket *head, int connstatus); 409 struct socket * 410 sopeeloff(struct socket *); 411 int sopoll(struct socket *so, int events, struct ucred *active_cred, 412 struct thread *td); 413 int sopoll_generic(struct socket *so, int events, 414 struct ucred *active_cred, struct thread *td); 415 int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio, 416 struct mbuf **mp0, struct mbuf **controlp, int *flagsp); 417 int soreceive_stream(struct socket *so, struct sockaddr **paddr, 418 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 419 int *flagsp); 420 int soreceive_dgram(struct socket *so, struct sockaddr **paddr, 421 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 422 int *flagsp); 423 int soreceive_generic(struct socket *so, struct sockaddr **paddr, 424 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 425 int *flagsp); 426 int soreserve(struct socket *so, u_long sndcc, u_long rcvcc); 427 void sorflush(struct socket *so); 428 int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, 429 struct mbuf *top, struct mbuf *control, int flags, 430 struct thread *td); 431 int sosend_dgram(struct socket *so, struct sockaddr *addr, 432 struct uio *uio, struct mbuf *top, struct mbuf *control, 433 int flags, struct thread *td); 434 int sosend_generic(struct socket *so, struct sockaddr *addr, 435 struct uio *uio, struct mbuf *top, struct mbuf *control, 436 int flags, struct thread *td); 437 int soshutdown(struct socket *so, int how); 438 void soupcall_clear(struct socket *, int); 439 void soupcall_set(struct socket *, int, so_upcall_t, void *); 440 void solisten_upcall_set(struct socket *, so_upcall_t, void *); 441 void sowakeup(struct socket *so, struct sockbuf *sb); 442 void sowakeup_aio(struct socket *so, struct sockbuf *sb); 443 void solisten_wakeup(struct socket *); 444 int selsocket(struct socket *so, int events, struct timeval *tv, 445 struct thread *td); 446 void soisconnected(struct socket *so); 447 void soisconnecting(struct socket *so); 448 void soisdisconnected(struct socket *so); 449 void soisdisconnecting(struct socket *so); 450 void socantrcvmore(struct socket *so); 451 void socantrcvmore_locked(struct socket *so); 452 void socantsendmore(struct socket *so); 453 void socantsendmore_locked(struct socket *so); 454 455 /* 456 * Accept filter functions (duh). 457 */ 458 int accept_filt_add(struct accept_filter *filt); 459 int accept_filt_del(char *name); 460 struct accept_filter *accept_filt_get(char *name); 461 #ifdef ACCEPT_FILTER_MOD 462 #ifdef SYSCTL_DECL 463 SYSCTL_DECL(_net_inet_accf); 464 #endif 465 int accept_filt_generic_mod_event(module_t mod, int event, void *data); 466 #endif 467 468 #endif /* _KERNEL */ 469 470 /* 471 * Structure to export socket from kernel to utilities, via sysctl(3). 472 */ 473 struct xsocket { 474 size_t xso_len; /* length of this structure */ 475 union { 476 void *xso_so; /* kernel address of struct socket */ 477 int64_t ph_so; 478 }; 479 union { 480 void *so_pcb; /* kernel address of struct inpcb */ 481 int64_t ph_pcb; 482 }; 483 uint64_t so_oobmark; 484 int64_t so_spare64[8]; 485 int32_t xso_protocol; 486 int32_t xso_family; 487 uint32_t so_qlen; 488 uint32_t so_incqlen; 489 uint32_t so_qlimit; 490 pid_t so_pgid; 491 uid_t so_uid; 492 int32_t so_spare32[8]; 493 int16_t so_type; 494 int16_t so_options; 495 int16_t so_linger; 496 int16_t so_state; 497 int16_t so_timeo; 498 uint16_t so_error; 499 struct xsockbuf { 500 uint32_t sb_cc; 501 uint32_t sb_hiwat; 502 uint32_t sb_mbcnt; 503 uint32_t sb_mcnt; 504 uint32_t sb_ccnt; 505 uint32_t sb_mbmax; 506 int32_t sb_lowat; 507 int32_t sb_timeo; 508 int16_t sb_flags; 509 } so_rcv, so_snd; 510 }; 511 512 #ifdef _KERNEL 513 void sotoxsocket(struct socket *so, struct xsocket *xso); 514 void sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb); 515 #endif 516 517 /* 518 * Socket buffer state bits. Exported via libprocstat(3). 519 */ 520 #define SBS_CANTSENDMORE 0x0010 /* can't send more data to peer */ 521 #define SBS_CANTRCVMORE 0x0020 /* can't receive more data from peer */ 522 #define SBS_RCVATMARK 0x0040 /* at mark on input */ 523 524 #endif /* !_SYS_SOCKETVAR_H_ */ 525