1 /* 2 * Copyright (c) 1982, 1986, 1988, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * Redistribution and use in source and binary forms, with or without 6 * modification, are permitted provided that the following conditions 7 * are met: 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. All advertising materials mentioning features or use of this software 14 * must display the following acknowledgement: 15 * This product includes software developed by the University of 16 * California, Berkeley and its contributors. 17 * 4. Neither the name of the University nor the names of its contributors 18 * may be used to endorse or promote products derived from this software 19 * without specific prior written permission. 20 * 21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 * 33 * From: @(#)tcp_usrreq.c 8.2 (Berkeley) 1/3/94 34 * $Id: tcp_usrreq.c,v 1.34 1997/09/14 03:10:40 peter Exp $ 35 */ 36 37 #include "opt_tcpdebug.h" 38 39 #include <sys/param.h> 40 #include <sys/systm.h> 41 #include <sys/kernel.h> 42 #include <sys/sysctl.h> 43 #include <sys/mbuf.h> 44 #include <sys/socket.h> 45 #include <sys/socketvar.h> 46 #include <sys/protosw.h> 47 48 #include <net/if.h> 49 #include <net/route.h> 50 51 #include <netinet/in.h> 52 #include <netinet/in_systm.h> 53 #include <netinet/in_pcb.h> 54 #include <netinet/in_var.h> 55 #include <netinet/ip_var.h> 56 #include <netinet/tcp.h> 57 #include <netinet/tcp_fsm.h> 58 #include <netinet/tcp_seq.h> 59 #include <netinet/tcp_timer.h> 60 #include <netinet/tcp_var.h> 61 #include <netinet/tcpip.h> 62 #ifdef TCPDEBUG 63 #include <netinet/tcp_debug.h> 64 #endif 65 66 /* 67 * TCP protocol interface to socket abstraction. 68 */ 69 extern char *tcpstates[]; /* XXX ??? */ 70 71 static int tcp_attach __P((struct socket *, struct proc *)); 72 static int tcp_connect __P((struct tcpcb *, struct sockaddr *, 73 struct proc *)); 74 static struct tcpcb * 75 tcp_disconnect __P((struct tcpcb *)); 76 static struct tcpcb * 77 tcp_usrclosed __P((struct tcpcb *)); 78 79 #ifdef TCPDEBUG 80 #define TCPDEBUG0 int ostate 81 #define TCPDEBUG1() ostate = tp ? tp->t_state : 0 82 #define TCPDEBUG2(req) if (tp && (so->so_options & SO_DEBUG)) \ 83 tcp_trace(TA_USER, ostate, tp, 0, req) 84 #else 85 #define TCPDEBUG0 86 #define TCPDEBUG1() 87 #define TCPDEBUG2(req) 88 #endif 89 90 /* 91 * TCP attaches to socket via pru_attach(), reserving space, 92 * and an internet control block. 93 */ 94 static int 95 tcp_usr_attach(struct socket *so, int proto, struct proc *p) 96 { 97 int s = splnet(); 98 int error; 99 struct inpcb *inp = sotoinpcb(so); 100 struct tcpcb *tp = 0; 101 TCPDEBUG0; 102 103 TCPDEBUG1(); 104 if (inp) { 105 error = EISCONN; 106 goto out; 107 } 108 109 error = tcp_attach(so, p); 110 if (error) 111 goto out; 112 113 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 114 so->so_linger = TCP_LINGERTIME * hz; 115 tp = sototcpcb(so); 116 out: 117 TCPDEBUG2(PRU_ATTACH); 118 splx(s); 119 return error; 120 } 121 122 /* 123 * pru_detach() detaches the TCP protocol from the socket. 124 * If the protocol state is non-embryonic, then can't 125 * do this directly: have to initiate a pru_disconnect(), 126 * which may finish later; embryonic TCB's can just 127 * be discarded here. 128 */ 129 static int 130 tcp_usr_detach(struct socket *so) 131 { 132 int s = splnet(); 133 int error = 0; 134 struct inpcb *inp = sotoinpcb(so); 135 struct tcpcb *tp; 136 TCPDEBUG0; 137 138 if (inp == 0) { 139 splx(s); 140 return EINVAL; /* XXX */ 141 } 142 tp = intotcpcb(inp); 143 TCPDEBUG1(); 144 tp = tcp_disconnect(tp); 145 146 TCPDEBUG2(PRU_DETACH); 147 splx(s); 148 return error; 149 } 150 151 #define COMMON_START() TCPDEBUG0; \ 152 do { \ 153 if (inp == 0) { \ 154 splx(s); \ 155 return EINVAL; \ 156 } \ 157 tp = intotcpcb(inp); \ 158 TCPDEBUG1(); \ 159 } while(0) 160 161 #define COMMON_END(req) out: TCPDEBUG2(req); splx(s); return error; goto out 162 163 164 /* 165 * Give the socket an address. 166 */ 167 static int 168 tcp_usr_bind(struct socket *so, struct sockaddr *nam, struct proc *p) 169 { 170 int s = splnet(); 171 int error = 0; 172 struct inpcb *inp = sotoinpcb(so); 173 struct tcpcb *tp; 174 struct sockaddr_in *sinp; 175 176 COMMON_START(); 177 178 /* 179 * Must check for multicast addresses and disallow binding 180 * to them. 181 */ 182 sinp = (struct sockaddr_in *)nam; 183 if (sinp->sin_family == AF_INET && 184 IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) { 185 error = EAFNOSUPPORT; 186 goto out; 187 } 188 error = in_pcbbind(inp, nam, p); 189 if (error) 190 goto out; 191 COMMON_END(PRU_BIND); 192 193 } 194 195 /* 196 * Prepare to accept connections. 197 */ 198 static int 199 tcp_usr_listen(struct socket *so, struct proc *p) 200 { 201 int s = splnet(); 202 int error = 0; 203 struct inpcb *inp = sotoinpcb(so); 204 struct tcpcb *tp; 205 206 COMMON_START(); 207 if (inp->inp_lport == 0) 208 error = in_pcbbind(inp, (struct sockaddr *)0, p); 209 if (error == 0) 210 tp->t_state = TCPS_LISTEN; 211 COMMON_END(PRU_LISTEN); 212 } 213 214 /* 215 * Initiate connection to peer. 216 * Create a template for use in transmissions on this connection. 217 * Enter SYN_SENT state, and mark socket as connecting. 218 * Start keep-alive timer, and seed output sequence space. 219 * Send initial segment on connection. 220 */ 221 static int 222 tcp_usr_connect(struct socket *so, struct sockaddr *nam, struct proc *p) 223 { 224 int s = splnet(); 225 int error = 0; 226 struct inpcb *inp = sotoinpcb(so); 227 struct tcpcb *tp; 228 struct sockaddr_in *sinp; 229 230 COMMON_START(); 231 232 /* 233 * Must disallow TCP ``connections'' to multicast addresses. 234 */ 235 sinp = (struct sockaddr_in *)nam; 236 if (sinp->sin_family == AF_INET 237 && IN_MULTICAST(ntohl(sinp->sin_addr.s_addr))) { 238 error = EAFNOSUPPORT; 239 goto out; 240 } 241 242 if ((error = tcp_connect(tp, nam, p)) != 0) 243 goto out; 244 error = tcp_output(tp); 245 COMMON_END(PRU_CONNECT); 246 } 247 248 /* 249 * Initiate disconnect from peer. 250 * If connection never passed embryonic stage, just drop; 251 * else if don't need to let data drain, then can just drop anyways, 252 * else have to begin TCP shutdown process: mark socket disconnecting, 253 * drain unread data, state switch to reflect user close, and 254 * send segment (e.g. FIN) to peer. Socket will be really disconnected 255 * when peer sends FIN and acks ours. 256 * 257 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 258 */ 259 static int 260 tcp_usr_disconnect(struct socket *so) 261 { 262 int s = splnet(); 263 int error = 0; 264 struct inpcb *inp = sotoinpcb(so); 265 struct tcpcb *tp; 266 267 COMMON_START(); 268 tp = tcp_disconnect(tp); 269 COMMON_END(PRU_DISCONNECT); 270 } 271 272 /* 273 * Accept a connection. Essentially all the work is 274 * done at higher levels; just return the address 275 * of the peer, storing through addr. 276 */ 277 static int 278 tcp_usr_accept(struct socket *so, struct sockaddr **nam) 279 { 280 int s = splnet(); 281 int error = 0; 282 struct inpcb *inp = sotoinpcb(so); 283 struct tcpcb *tp; 284 285 COMMON_START(); 286 in_setpeeraddr(so, nam); 287 COMMON_END(PRU_ACCEPT); 288 } 289 290 /* 291 * Mark the connection as being incapable of further output. 292 */ 293 static int 294 tcp_usr_shutdown(struct socket *so) 295 { 296 int s = splnet(); 297 int error = 0; 298 struct inpcb *inp = sotoinpcb(so); 299 struct tcpcb *tp; 300 301 COMMON_START(); 302 socantsendmore(so); 303 tp = tcp_usrclosed(tp); 304 if (tp) 305 error = tcp_output(tp); 306 COMMON_END(PRU_SHUTDOWN); 307 } 308 309 /* 310 * After a receive, possibly send window update to peer. 311 */ 312 static int 313 tcp_usr_rcvd(struct socket *so, int flags) 314 { 315 int s = splnet(); 316 int error = 0; 317 struct inpcb *inp = sotoinpcb(so); 318 struct tcpcb *tp; 319 320 COMMON_START(); 321 tcp_output(tp); 322 COMMON_END(PRU_RCVD); 323 } 324 325 /* 326 * Do a send by putting data in output queue and updating urgent 327 * marker if URG set. Possibly send more data. 328 */ 329 static int 330 tcp_usr_send(struct socket *so, int flags, struct mbuf *m, 331 struct sockaddr *nam, struct mbuf *control, struct proc *p) 332 { 333 int s = splnet(); 334 int error = 0; 335 struct inpcb *inp = sotoinpcb(so); 336 struct tcpcb *tp; 337 338 COMMON_START(); 339 if (control && control->m_len) { 340 m_freem(control); /* XXX shouldn't caller do this??? */ 341 if (m) 342 m_freem(m); 343 return EINVAL; 344 } 345 346 if(!(flags & PRUS_OOB)) { 347 sbappend(&so->so_snd, m); 348 if (nam && tp->t_state < TCPS_SYN_SENT) { 349 /* 350 * Do implied connect if not yet connected, 351 * initialize window to default value, and 352 * initialize maxseg/maxopd using peer's cached 353 * MSS. 354 */ 355 error = tcp_connect(tp, nam, p); 356 if (error) 357 goto out; 358 tp->snd_wnd = TTCP_CLIENT_SND_WND; 359 tcp_mss(tp, -1); 360 } 361 362 if (flags & PRUS_EOF) { 363 /* 364 * Close the send side of the connection after 365 * the data is sent. 366 */ 367 socantsendmore(so); 368 tp = tcp_usrclosed(tp); 369 } 370 if (tp != NULL) 371 error = tcp_output(tp); 372 } else { 373 if (sbspace(&so->so_snd) < -512) { 374 m_freem(m); 375 error = ENOBUFS; 376 goto out; 377 } 378 /* 379 * According to RFC961 (Assigned Protocols), 380 * the urgent pointer points to the last octet 381 * of urgent data. We continue, however, 382 * to consider it to indicate the first octet 383 * of data past the urgent section. 384 * Otherwise, snd_up should be one lower. 385 */ 386 sbappend(&so->so_snd, m); 387 if (nam && tp->t_state < TCPS_SYN_SENT) { 388 /* 389 * Do implied connect if not yet connected, 390 * initialize window to default value, and 391 * initialize maxseg/maxopd using peer's cached 392 * MSS. 393 */ 394 error = tcp_connect(tp, nam, p); 395 if (error) 396 goto out; 397 tp->snd_wnd = TTCP_CLIENT_SND_WND; 398 tcp_mss(tp, -1); 399 } 400 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 401 tp->t_force = 1; 402 error = tcp_output(tp); 403 tp->t_force = 0; 404 } 405 COMMON_END((flags & PRUS_OOB) ? PRU_SENDOOB : 406 ((flags & PRUS_EOF) ? PRU_SEND_EOF : PRU_SEND)); 407 } 408 409 /* 410 * Abort the TCP. 411 */ 412 static int 413 tcp_usr_abort(struct socket *so) 414 { 415 int s = splnet(); 416 int error = 0; 417 struct inpcb *inp = sotoinpcb(so); 418 struct tcpcb *tp; 419 420 COMMON_START(); 421 tp = tcp_drop(tp, ECONNABORTED); 422 COMMON_END(PRU_ABORT); 423 } 424 425 /* 426 * Receive out-of-band data. 427 */ 428 static int 429 tcp_usr_rcvoob(struct socket *so, struct mbuf *m, int flags) 430 { 431 int s = splnet(); 432 int error = 0; 433 struct inpcb *inp = sotoinpcb(so); 434 struct tcpcb *tp; 435 436 COMMON_START(); 437 if ((so->so_oobmark == 0 && 438 (so->so_state & SS_RCVATMARK) == 0) || 439 so->so_options & SO_OOBINLINE || 440 tp->t_oobflags & TCPOOB_HADDATA) { 441 error = EINVAL; 442 goto out; 443 } 444 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 445 error = EWOULDBLOCK; 446 goto out; 447 } 448 m->m_len = 1; 449 *mtod(m, caddr_t) = tp->t_iobc; 450 if ((flags & MSG_PEEK) == 0) 451 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 452 COMMON_END(PRU_RCVOOB); 453 } 454 455 /* xxx - should be const */ 456 struct pr_usrreqs tcp_usrreqs = { 457 tcp_usr_abort, tcp_usr_accept, tcp_usr_attach, tcp_usr_bind, 458 tcp_usr_connect, pru_connect2_notsupp, in_control, tcp_usr_detach, 459 tcp_usr_disconnect, tcp_usr_listen, in_setpeeraddr, tcp_usr_rcvd, 460 tcp_usr_rcvoob, tcp_usr_send, pru_sense_null, tcp_usr_shutdown, 461 in_setsockaddr, sosend, soreceive, sopoll 462 }; 463 464 /* 465 * Common subroutine to open a TCP connection to remote host specified 466 * by struct sockaddr_in in mbuf *nam. Call in_pcbbind to assign a local 467 * port number if needed. Call in_pcbladdr to do the routing and to choose 468 * a local host address (interface). If there is an existing incarnation 469 * of the same connection in TIME-WAIT state and if the remote host was 470 * sending CC options and if the connection duration was < MSL, then 471 * truncate the previous TIME-WAIT state and proceed. 472 * Initialize connection parameters and enter SYN-SENT state. 473 */ 474 static int 475 tcp_connect(tp, nam, p) 476 register struct tcpcb *tp; 477 struct sockaddr *nam; 478 struct proc *p; 479 { 480 struct inpcb *inp = tp->t_inpcb, *oinp; 481 struct socket *so = inp->inp_socket; 482 struct tcpcb *otp; 483 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 484 struct sockaddr_in *ifaddr; 485 int error; 486 struct rmxp_tao *taop; 487 struct rmxp_tao tao_noncached; 488 489 if (inp->inp_lport == 0) { 490 error = in_pcbbind(inp, (struct sockaddr *)0, p); 491 if (error) 492 return error; 493 } 494 495 /* 496 * Cannot simply call in_pcbconnect, because there might be an 497 * earlier incarnation of this same connection still in 498 * TIME_WAIT state, creating an ADDRINUSE error. 499 */ 500 error = in_pcbladdr(inp, nam, &ifaddr); 501 if (error) 502 return error; 503 oinp = in_pcblookuphash(inp->inp_pcbinfo, 504 sin->sin_addr, sin->sin_port, 505 inp->inp_laddr.s_addr != INADDR_ANY ? inp->inp_laddr 506 : ifaddr->sin_addr, 507 inp->inp_lport, 0); 508 if (oinp) { 509 if (oinp != inp && (otp = intotcpcb(oinp)) != NULL && 510 otp->t_state == TCPS_TIME_WAIT && 511 otp->t_duration < TCPTV_MSL && 512 (otp->t_flags & TF_RCVD_CC)) 513 otp = tcp_close(otp); 514 else 515 return EADDRINUSE; 516 } 517 if (inp->inp_laddr.s_addr == INADDR_ANY) 518 inp->inp_laddr = ifaddr->sin_addr; 519 inp->inp_faddr = sin->sin_addr; 520 inp->inp_fport = sin->sin_port; 521 in_pcbrehash(inp); 522 523 tp->t_template = tcp_template(tp); 524 if (tp->t_template == 0) { 525 in_pcbdisconnect(inp); 526 return ENOBUFS; 527 } 528 529 /* Compute window scaling to request. */ 530 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 531 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 532 tp->request_r_scale++; 533 534 soisconnecting(so); 535 tcpstat.tcps_connattempt++; 536 tp->t_state = TCPS_SYN_SENT; 537 tp->t_timer[TCPT_KEEP] = tcp_keepinit; 538 tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2; 539 tcp_sendseqinit(tp); 540 541 /* 542 * Generate a CC value for this connection and 543 * check whether CC or CCnew should be used. 544 */ 545 if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) { 546 taop = &tao_noncached; 547 bzero(taop, sizeof(*taop)); 548 } 549 550 tp->cc_send = CC_INC(tcp_ccgen); 551 if (taop->tao_ccsent != 0 && 552 CC_GEQ(tp->cc_send, taop->tao_ccsent)) { 553 taop->tao_ccsent = tp->cc_send; 554 } else { 555 taop->tao_ccsent = 0; 556 tp->t_flags |= TF_SENDCCNEW; 557 } 558 559 return 0; 560 } 561 562 int 563 tcp_ctloutput(op, so, level, optname, mp, p) 564 int op; 565 struct socket *so; 566 int level, optname; 567 struct mbuf **mp; 568 struct proc *p; 569 { 570 int error = 0, s; 571 struct inpcb *inp; 572 register struct tcpcb *tp; 573 register struct mbuf *m; 574 register int i; 575 576 s = splnet(); 577 inp = sotoinpcb(so); 578 if (inp == NULL) { 579 splx(s); 580 if (op == PRCO_SETOPT && *mp) 581 (void) m_free(*mp); 582 return (ECONNRESET); 583 } 584 if (level != IPPROTO_TCP) { 585 error = ip_ctloutput(op, so, level, optname, mp, p); 586 splx(s); 587 return (error); 588 } 589 tp = intotcpcb(inp); 590 591 switch (op) { 592 593 case PRCO_SETOPT: 594 m = *mp; 595 switch (optname) { 596 597 case TCP_NODELAY: 598 if (m == NULL || m->m_len < sizeof (int)) 599 error = EINVAL; 600 else if (*mtod(m, int *)) 601 tp->t_flags |= TF_NODELAY; 602 else 603 tp->t_flags &= ~TF_NODELAY; 604 break; 605 606 case TCP_MAXSEG: 607 if (m && (i = *mtod(m, int *)) > 0 && i <= tp->t_maxseg) 608 tp->t_maxseg = i; 609 else 610 error = EINVAL; 611 break; 612 613 case TCP_NOOPT: 614 if (m == NULL || m->m_len < sizeof (int)) 615 error = EINVAL; 616 else if (*mtod(m, int *)) 617 tp->t_flags |= TF_NOOPT; 618 else 619 tp->t_flags &= ~TF_NOOPT; 620 break; 621 622 case TCP_NOPUSH: 623 if (m == NULL || m->m_len < sizeof (int)) 624 error = EINVAL; 625 else if (*mtod(m, int *)) 626 tp->t_flags |= TF_NOPUSH; 627 else 628 tp->t_flags &= ~TF_NOPUSH; 629 break; 630 631 default: 632 error = ENOPROTOOPT; 633 break; 634 } 635 if (m) 636 (void) m_free(m); 637 break; 638 639 case PRCO_GETOPT: 640 *mp = m = m_get(M_WAIT, MT_SOOPTS); 641 m->m_len = sizeof(int); 642 643 switch (optname) { 644 case TCP_NODELAY: 645 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 646 break; 647 case TCP_MAXSEG: 648 *mtod(m, int *) = tp->t_maxseg; 649 break; 650 case TCP_NOOPT: 651 *mtod(m, int *) = tp->t_flags & TF_NOOPT; 652 break; 653 case TCP_NOPUSH: 654 *mtod(m, int *) = tp->t_flags & TF_NOPUSH; 655 break; 656 default: 657 error = ENOPROTOOPT; 658 break; 659 } 660 break; 661 } 662 splx(s); 663 return (error); 664 } 665 666 /* 667 * tcp_sendspace and tcp_recvspace are the default send and receive window 668 * sizes, respectively. These are obsolescent (this information should 669 * be set by the route). 670 */ 671 u_long tcp_sendspace = 1024*16; 672 SYSCTL_INT(_net_inet_tcp, TCPCTL_SENDSPACE, sendspace, 673 CTLFLAG_RW, &tcp_sendspace , 0, ""); 674 u_long tcp_recvspace = 1024*16; 675 SYSCTL_INT(_net_inet_tcp, TCPCTL_RECVSPACE, recvspace, 676 CTLFLAG_RW, &tcp_recvspace , 0, ""); 677 678 /* 679 * Attach TCP protocol to socket, allocating 680 * internet protocol control block, tcp control block, 681 * bufer space, and entering LISTEN state if to accept connections. 682 */ 683 static int 684 tcp_attach(so, p) 685 struct socket *so; 686 struct proc *p; 687 { 688 register struct tcpcb *tp; 689 struct inpcb *inp; 690 int error; 691 692 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 693 error = soreserve(so, tcp_sendspace, tcp_recvspace); 694 if (error) 695 return (error); 696 } 697 error = in_pcballoc(so, &tcbinfo, p); 698 if (error) 699 return (error); 700 inp = sotoinpcb(so); 701 tp = tcp_newtcpcb(inp); 702 if (tp == 0) { 703 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 704 705 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 706 in_pcbdetach(inp); 707 so->so_state |= nofd; 708 return (ENOBUFS); 709 } 710 tp->t_state = TCPS_CLOSED; 711 return (0); 712 } 713 714 /* 715 * Initiate (or continue) disconnect. 716 * If embryonic state, just send reset (once). 717 * If in ``let data drain'' option and linger null, just drop. 718 * Otherwise (hard), mark socket disconnecting and drop 719 * current input data; switch states based on user close, and 720 * send segment to peer (with FIN). 721 */ 722 static struct tcpcb * 723 tcp_disconnect(tp) 724 register struct tcpcb *tp; 725 { 726 struct socket *so = tp->t_inpcb->inp_socket; 727 728 if (tp->t_state < TCPS_ESTABLISHED) 729 tp = tcp_close(tp); 730 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 731 tp = tcp_drop(tp, 0); 732 else { 733 soisdisconnecting(so); 734 sbflush(&so->so_rcv); 735 tp = tcp_usrclosed(tp); 736 if (tp) 737 (void) tcp_output(tp); 738 } 739 return (tp); 740 } 741 742 /* 743 * User issued close, and wish to trail through shutdown states: 744 * if never received SYN, just forget it. If got a SYN from peer, 745 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 746 * If already got a FIN from peer, then almost done; go to LAST_ACK 747 * state. In all other cases, have already sent FIN to peer (e.g. 748 * after PRU_SHUTDOWN), and just have to play tedious game waiting 749 * for peer to send FIN or not respond to keep-alives, etc. 750 * We can let the user exit from the close as soon as the FIN is acked. 751 */ 752 static struct tcpcb * 753 tcp_usrclosed(tp) 754 register struct tcpcb *tp; 755 { 756 757 switch (tp->t_state) { 758 759 case TCPS_CLOSED: 760 case TCPS_LISTEN: 761 tp->t_state = TCPS_CLOSED; 762 tp = tcp_close(tp); 763 break; 764 765 case TCPS_SYN_SENT: 766 case TCPS_SYN_RECEIVED: 767 tp->t_flags |= TF_NEEDFIN; 768 break; 769 770 case TCPS_ESTABLISHED: 771 tp->t_state = TCPS_FIN_WAIT_1; 772 break; 773 774 case TCPS_CLOSE_WAIT: 775 tp->t_state = TCPS_LAST_ACK; 776 break; 777 } 778 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 779 soisdisconnected(tp->t_inpcb->inp_socket); 780 /* To prevent the connection hanging in FIN_WAIT_2 forever. */ 781 if (tp->t_state == TCPS_FIN_WAIT_2) 782 tp->t_timer[TCPT_2MSL] = tcp_maxidle; 783 } 784 return (tp); 785 } 786 787