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 */ 42typedef 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 57struct 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 */ 65typedef int so_upcall_t(struct socket *, void *, int); 66typedef void so_dtor_t(struct socket *); 67 68struct socket; 69 70enum socket_qstate { 71 SQ_NONE = 0, 72 SQ_INCOMP = 0x0800, /* on sol_incomp */ 73 SQ_COMP = 0x1000, /* on sol_comp */ 74}; 75 76/*- 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. 86 * (k) locked by KTLS workqueue mutex 87 */ 88TAILQ_HEAD(accept_queue, socket); 89struct socket { 90 struct mtx so_lock; 91 volatile u_int so_count; /* (b / refcount) */ 92 struct selinfo so_rdsel; /* (b/cr) for so_rcv/so_comp */ 93 struct selinfo so_wrsel; /* (b/cs) for so_snd */ 94 short so_type; /* (a) generic type, see socket.h */ 95 int so_options; /* (b) from socket call, see socket.h */ 96 short so_linger; /* time to linger close(2) */ 97 short so_state; /* (b) internal state flags SS_* */ 98 void *so_pcb; /* protocol control block */ 99 struct vnet *so_vnet; /* (a) network stack instance */ 100 struct protosw *so_proto; /* (a) protocol handle */ 101 short so_timeo; /* (g) connection timeout */ 102 u_short so_error; /* (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 */ 112 /* 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. 117 */ 118 int so_fibnum; /* routing domain for this socket */ 119 uint32_t so_user_cookie; 120 121 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 */ 123 union { 124 /* Regular (data flow) socket. */ 125 struct { 126 /* (cr, cs) Receive and send buffers. */ 127 struct sockbuf so_rcv, so_snd; 128 129 /* (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 */ 136 137 /* (k) Our place on KTLS RX work queue. */ 138 STAILQ_ENTRY(socket) so_ktls_rx_list; 139 }; 140 /* 141 * Listening socket, where accepts occur, is so_listen in all 142 * subsidiary sockets. If so_listen is NULL, socket is not 143 * related to an accept. For a listening socket itself 144 * sol_incomp queues partially completed connections, while 145 * sol_comp is a queue of connections ready to be accepted. 146 * If a connection is aborted and it has so_listen set, then 147 * it has to be pulled out of either sol_incomp or sol_comp. 148 * We allow connections to queue up based on current queue 149 * lengths and limit on number of queued connections for this 150 * socket. 151 */ 152 struct { 153 /* (e) queue of partial unaccepted connections */ 154 struct accept_queue sol_incomp; 155 /* (e) queue of complete unaccepted connections */ 156 struct accept_queue sol_comp; 157 u_int sol_qlen; /* (e) sol_comp length */ 158 u_int sol_incqlen; /* (e) sol_incomp length */ 159 u_int sol_qlimit; /* (e) queue limit */ 160 161 /* accept_filter(9) optional data */ 162 struct accept_filter *sol_accept_filter; 163 void *sol_accept_filter_arg; /* saved filter args */ 164 char *sol_accept_filter_str; /* saved user args */ 165 166 /* Optional upcall, for kernel socket. */ 167 so_upcall_t *sol_upcall; /* (e) */ 168 void *sol_upcallarg; /* (e) */ 169 170 /* Socket buffer parameters, to be copied to 171 * dataflow sockets, accepted from this one. */ 172 int sol_sbrcv_lowat; 173 int sol_sbsnd_lowat; 174 u_int sol_sbrcv_hiwat; 175 u_int sol_sbsnd_hiwat; 176 short sol_sbrcv_flags; 177 short sol_sbsnd_flags; 178 sbintime_t sol_sbrcv_timeo; 179 sbintime_t sol_sbsnd_timeo; 180 181 /* Information tracking listen queue overflows. */ 182 struct timeval sol_lastover; /* (e) */ 183 int sol_overcount; /* (e) */ 184 }; 185 }; 186}; 187#endif /* defined(_KERNEL) || defined(_WANT_SOCKET) */ 188 189/* 190 * Socket state bits. 191 * 192 * Historically, these bits were all kept in the so_state field. 193 * They are now split into separate, lock-specific fields. 194 * so_state maintains basic socket state protected by the socket lock. 195 * so_qstate holds information about the socket accept queues. 196 * Each socket buffer also has a state field holding information 197 * relevant to that socket buffer (can't send, rcv). 198 * Many fields will be read without locks to improve performance and avoid 199 * lock order issues. However, this approach must be used with caution. 200 */ 201#define SS_NOFDREF 0x0001 /* no file table ref any more */ 202#define SS_ISCONNECTED 0x0002 /* socket connected to a peer */ 203#define SS_ISCONNECTING 0x0004 /* in process of connecting to peer */ 204#define SS_ISDISCONNECTING 0x0008 /* in process of disconnecting */ 205#define SS_NBIO 0x0100 /* non-blocking ops */ 206#define SS_ASYNC 0x0200 /* async i/o notify */ 207#define SS_ISCONFIRMING 0x0400 /* deciding to accept connection req */ 208#define SS_ISDISCONNECTED 0x2000 /* socket disconnected from peer */ 209 210/* 211 * Protocols can mark a socket as SS_PROTOREF to indicate that, following 212 * pru_detach, they still want the socket to persist, and will free it 213 * themselves when they are done. Protocols should only ever call sofree() 214 * following setting this flag in pru_detach(), and never otherwise, as 215 * sofree() bypasses socket reference counting. 216 */ 217#define SS_PROTOREF 0x4000 /* strong protocol reference */ 218 219#ifdef _KERNEL 220 221#define SOCK_MTX(so) (&(so)->so_lock) 222#define SOCK_LOCK(so) mtx_lock(&(so)->so_lock) 223#define SOCK_OWNED(so) mtx_owned(&(so)->so_lock) 224#define SOCK_UNLOCK(so) mtx_unlock(&(so)->so_lock) 225#define SOCK_LOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_OWNED) 226#define SOCK_UNLOCK_ASSERT(so) mtx_assert(&(so)->so_lock, MA_NOTOWNED) 227 228#define SOLISTENING(sol) (((sol)->so_options & SO_ACCEPTCONN) != 0) 229#define SOLISTEN_LOCK(sol) do { \ 230 mtx_lock(&(sol)->so_lock); \ 231 KASSERT(SOLISTENING(sol), \ 232 ("%s: %p not listening", __func__, (sol))); \ 233} while (0) 234#define SOLISTEN_TRYLOCK(sol) mtx_trylock(&(sol)->so_lock) 235#define SOLISTEN_UNLOCK(sol) do { \ 236 KASSERT(SOLISTENING(sol), \ 237 ("%s: %p not listening", __func__, (sol))); \ 238 mtx_unlock(&(sol)->so_lock); \ 239} while (0) 240#define SOLISTEN_LOCK_ASSERT(sol) do { \ 241 mtx_assert(&(sol)->so_lock, MA_OWNED); \ 242 KASSERT(SOLISTENING(sol), \ 243 ("%s: %p not listening", __func__, (sol))); \ 244} while (0) 245 246/* 247 * Macros for sockets and socket buffering. 248 */ 249 250/* 251 * Flags to sblock(). 252 */ 253#define SBL_WAIT 0x00000001 /* Wait if not immediately available. */ 254#define SBL_NOINTR 0x00000002 /* Force non-interruptible sleep. */ 255#define SBL_VALID (SBL_WAIT | SBL_NOINTR) 256 257/* 258 * Do we need to notify the other side when I/O is possible? 259 */ 260#define sb_notify(sb) (((sb)->sb_flags & (SB_WAIT | SB_SEL | SB_ASYNC | \ 261 SB_UPCALL | SB_AIO | SB_KNOTE)) != 0) 262 263/* do we have to send all at once on a socket? */ 264#define sosendallatonce(so) \ 265 ((so)->so_proto->pr_flags & PR_ATOMIC) 266 267/* can we read something from so? */ 268#define soreadabledata(so) \ 269 (sbavail(&(so)->so_rcv) >= (so)->so_rcv.sb_lowat || (so)->so_error) 270#define soreadable(so) \ 271 (soreadabledata(so) || ((so)->so_rcv.sb_state & SBS_CANTRCVMORE)) 272 273/* can we write something to so? */ 274#define sowriteable(so) \ 275 ((sbspace(&(so)->so_snd) >= (so)->so_snd.sb_lowat && \ 276 (((so)->so_state&SS_ISCONNECTED) || \ 277 ((so)->so_proto->pr_flags&PR_CONNREQUIRED)==0)) || \ 278 ((so)->so_snd.sb_state & SBS_CANTSENDMORE) || \ 279 (so)->so_error) 280 281/* 282 * soref()/sorele() ref-count the socket structure. 283 * soref() may be called without owning socket lock, but in that case a 284 * caller must own something that holds socket, and so_count must be not 0. 285 * Note that you must still explicitly close the socket, but the last ref 286 * count will free the structure. 287 */ 288#define soref(so) refcount_acquire(&(so)->so_count) 289#define sorele(so) do { \ 290 SOCK_LOCK_ASSERT(so); \ 291 if (refcount_release(&(so)->so_count)) \ 292 sofree(so); \ 293 else \ 294 SOCK_UNLOCK(so); \ 295} while (0) 296 297/* 298 * In sorwakeup() and sowwakeup(), acquire the socket buffer lock to 299 * avoid a non-atomic test-and-wakeup. However, sowakeup is 300 * responsible for releasing the lock if it is called. We unlock only 301 * if we don't call into sowakeup. If any code is introduced that 302 * directly invokes the underlying sowakeup() primitives, it must 303 * maintain the same semantics. 304 */ 305#define sorwakeup_locked(so) do { \ 306 SOCKBUF_LOCK_ASSERT(&(so)->so_rcv); \ 307 if (sb_notify(&(so)->so_rcv)) \ 308 sowakeup((so), &(so)->so_rcv); \ 309 else \ 310 SOCKBUF_UNLOCK(&(so)->so_rcv); \ 311} while (0) 312 313#define sorwakeup(so) do { \ 314 SOCKBUF_LOCK(&(so)->so_rcv); \ 315 sorwakeup_locked(so); \ 316} while (0) 317 318#define sowwakeup_locked(so) do { \ 319 SOCKBUF_LOCK_ASSERT(&(so)->so_snd); \ 320 if (sb_notify(&(so)->so_snd)) \ 321 sowakeup((so), &(so)->so_snd); \ 322 else \ 323 SOCKBUF_UNLOCK(&(so)->so_snd); \ 324} while (0) 325 326#define sowwakeup(so) do { \ 327 SOCKBUF_LOCK(&(so)->so_snd); \ 328 sowwakeup_locked(so); \ 329} while (0) 330 331struct accept_filter { 332 char accf_name[16]; 333 int (*accf_callback) 334 (struct socket *so, void *arg, int waitflag); 335 void * (*accf_create) 336 (struct socket *so, char *arg); 337 void (*accf_destroy) 338 (struct socket *so); 339 SLIST_ENTRY(accept_filter) accf_next; 340}; 341 342#define ACCEPT_FILTER_DEFINE(modname, filtname, cb, create, destroy, ver) \ 343 static struct accept_filter modname##_filter = { \ 344 .accf_name = filtname, \ 345 .accf_callback = cb, \ 346 .accf_create = create, \ 347 .accf_destroy = destroy, \ 348 }; \ 349 static moduledata_t modname##_mod = { \ 350 .name = __XSTRING(modname), \ 351 .evhand = accept_filt_generic_mod_event, \ 352 .priv = &modname##_filter, \ 353 }; \ 354 DECLARE_MODULE(modname, modname##_mod, SI_SUB_DRIVERS, \ 355 SI_ORDER_MIDDLE); \ 356 MODULE_VERSION(modname, ver) 357 358#ifdef MALLOC_DECLARE 359MALLOC_DECLARE(M_ACCF); 360MALLOC_DECLARE(M_PCB); 361MALLOC_DECLARE(M_SONAME); 362#endif 363 364/* 365 * Socket specific helper hook point identifiers 366 * Do not leave holes in the sequence, hook registration is a loop. 367 */ 368#define HHOOK_SOCKET_OPT 0 369#define HHOOK_SOCKET_CREATE 1 370#define HHOOK_SOCKET_RCV 2 371#define HHOOK_SOCKET_SND 3 372#define HHOOK_FILT_SOREAD 4 373#define HHOOK_FILT_SOWRITE 5 374#define HHOOK_SOCKET_CLOSE 6 375#define HHOOK_SOCKET_LAST HHOOK_SOCKET_CLOSE 376 377struct socket_hhook_data { 378 struct socket *so; 379 struct mbuf *m; 380 void *hctx; /* hook point specific data*/ 381 int status; 382}; 383 384extern int maxsockets; 385extern u_long sb_max; 386extern so_gen_t so_gencnt; 387 388struct file; 389struct filecaps; 390struct filedesc; 391struct mbuf; 392struct sockaddr; 393struct ucred; 394struct uio; 395 396/* 'which' values for socket upcalls. */ 397#define SO_RCV 1 398#define SO_SND 2 399 400/* Return values for socket upcalls. */ 401#define SU_OK 0 402#define SU_ISCONNECTED 1 403 404/* 405 * From uipc_socket and friends 406 */ 407int getsockaddr(struct sockaddr **namp, const struct sockaddr *uaddr, 408 size_t len); 409int getsock_cap(struct thread *td, int fd, cap_rights_t *rightsp, 410 struct file **fpp, u_int *fflagp, struct filecaps *havecaps); 411void soabort(struct socket *so); 412int soaccept(struct socket *so, struct sockaddr **nam); 413void soaio_enqueue(struct task *task); 414void soaio_rcv(void *context, int pending); 415void soaio_snd(void *context, int pending); 416int socheckuid(struct socket *so, uid_t uid); 417int sobind(struct socket *so, struct sockaddr *nam, struct thread *td); 418int sobindat(int fd, struct socket *so, struct sockaddr *nam, 419 struct thread *td); 420int soclose(struct socket *so); 421int soconnect(struct socket *so, struct sockaddr *nam, struct thread *td); 422int soconnectat(int fd, struct socket *so, struct sockaddr *nam, 423 struct thread *td); 424int soconnect2(struct socket *so1, struct socket *so2); 425int socreate(int dom, struct socket **aso, int type, int proto, 426 struct ucred *cred, struct thread *td); 427int sodisconnect(struct socket *so); 428void sodtor_set(struct socket *, so_dtor_t *); 429struct sockaddr *sodupsockaddr(const struct sockaddr *sa, int mflags); 430void sofree(struct socket *so); 431void sohasoutofband(struct socket *so); 432int solisten(struct socket *so, int backlog, struct thread *td); 433void solisten_proto(struct socket *so, int backlog); 434int solisten_proto_check(struct socket *so); 435int solisten_dequeue(struct socket *, struct socket **, int); 436struct socket * 437 sonewconn(struct socket *head, int connstatus); 438struct socket * 439 sopeeloff(struct socket *); 440int sopoll(struct socket *so, int events, struct ucred *active_cred, 441 struct thread *td); 442int sopoll_generic(struct socket *so, int events, 443 struct ucred *active_cred, struct thread *td); 444int soreceive(struct socket *so, struct sockaddr **paddr, struct uio *uio, 445 struct mbuf **mp0, struct mbuf **controlp, int *flagsp); 446int soreceive_stream(struct socket *so, struct sockaddr **paddr, 447 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 448 int *flagsp); 449int soreceive_dgram(struct socket *so, struct sockaddr **paddr, 450 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 451 int *flagsp); 452int soreceive_generic(struct socket *so, struct sockaddr **paddr, 453 struct uio *uio, struct mbuf **mp0, struct mbuf **controlp, 454 int *flagsp); 455int soreserve(struct socket *so, u_long sndcc, u_long rcvcc); 456void sorflush(struct socket *so); 457int sosend(struct socket *so, struct sockaddr *addr, struct uio *uio, 458 struct mbuf *top, struct mbuf *control, int flags, 459 struct thread *td); 460int sosend_dgram(struct socket *so, struct sockaddr *addr, 461 struct uio *uio, struct mbuf *top, struct mbuf *control, 462 int flags, struct thread *td); 463int sosend_generic(struct socket *so, struct sockaddr *addr, 464 struct uio *uio, struct mbuf *top, struct mbuf *control, 465 int flags, struct thread *td); 466int soshutdown(struct socket *so, int how); 467void soupcall_clear(struct socket *, int); 468void soupcall_set(struct socket *, int, so_upcall_t, void *); 469void solisten_upcall_set(struct socket *, so_upcall_t, void *); 470void sowakeup(struct socket *so, struct sockbuf *sb); 471void sowakeup_aio(struct socket *so, struct sockbuf *sb); 472void solisten_wakeup(struct socket *); 473int selsocket(struct socket *so, int events, struct timeval *tv, 474 struct thread *td); 475void soisconnected(struct socket *so); 476void soisconnecting(struct socket *so); 477void soisdisconnected(struct socket *so); 478void soisdisconnecting(struct socket *so); 479void socantrcvmore(struct socket *so); 480void socantrcvmore_locked(struct socket *so); 481void socantsendmore(struct socket *so); 482void socantsendmore_locked(struct socket *so); 483 484/* 485 * Accept filter functions (duh). 486 */ 487int accept_filt_add(struct accept_filter *filt); 488int accept_filt_del(char *name); 489struct accept_filter *accept_filt_get(char *name); 490#ifdef ACCEPT_FILTER_MOD 491#ifdef SYSCTL_DECL 492SYSCTL_DECL(_net_inet_accf); 493#endif 494int accept_filt_generic_mod_event(module_t mod, int event, void *data); 495#endif 496 497#endif /* _KERNEL */ 498 499/* 500 * Structure to export socket from kernel to utilities, via sysctl(3). 501 */ 502struct xsocket { 503 ksize_t xso_len; /* length of this structure */ 504 kvaddr_t xso_so; /* kernel address of struct socket */ 505 kvaddr_t so_pcb; /* kernel address of struct inpcb */ 506 uint64_t so_oobmark; 507 int64_t so_spare64[8]; 508 int32_t xso_protocol; 509 int32_t xso_family; 510 uint32_t so_qlen; 511 uint32_t so_incqlen; 512 uint32_t so_qlimit; 513 pid_t so_pgid; 514 uid_t so_uid; 515 int32_t so_spare32[8]; 516 int16_t so_type; 517 int16_t so_options; 518 int16_t so_linger; 519 int16_t so_state; 520 int16_t so_timeo; 521 uint16_t so_error; 522 struct xsockbuf { 523 uint32_t sb_cc; 524 uint32_t sb_hiwat; 525 uint32_t sb_mbcnt; 526 uint32_t sb_mcnt; 527 uint32_t sb_ccnt; 528 uint32_t sb_mbmax; 529 int32_t sb_lowat; 530 int32_t sb_timeo; 531 int16_t sb_flags; 532 } so_rcv, so_snd; 533}; 534 535#ifdef _KERNEL 536void sotoxsocket(struct socket *so, struct xsocket *xso); 537void sbtoxsockbuf(struct sockbuf *sb, struct xsockbuf *xsb); 538#endif 539 540/* 541 * Socket buffer state bits. Exported via libprocstat(3). 542 */ 543#define SBS_CANTSENDMORE 0x0010 /* can't send more data to peer */ 544#define SBS_CANTRCVMORE 0x0020 /* can't receive more data from peer */ 545#define SBS_RCVATMARK 0x0040 /* at mark on input */ 546 547#endif /* !_SYS_SOCKETVAR_H_ */ 548