tcp_usrreq.c revision 1.41
1 /* $NetBSD: tcp_usrreq.c,v 1.41 1999/07/02 12:45:32 itojun Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * 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 project 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 PROJECT 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 PROJECT 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 32 /*- 33 * Copyright (c) 1997, 1998 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 38 * Facility, NASA Ames Research Center. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69 /* 70 * Copyright (c) 1982, 1986, 1988, 1993, 1995 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. All advertising materials mentioning features or use of this software 82 * must display the following acknowledgement: 83 * This product includes software developed by the University of 84 * California, Berkeley and its contributors. 85 * 4. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 102 */ 103 104 #include "opt_inet.h" 105 106 #include <sys/param.h> 107 #include <sys/systm.h> 108 #include <sys/kernel.h> 109 #include <sys/malloc.h> 110 #include <sys/mbuf.h> 111 #include <sys/socket.h> 112 #include <sys/socketvar.h> 113 #include <sys/protosw.h> 114 #include <sys/errno.h> 115 #include <sys/stat.h> 116 #include <sys/proc.h> 117 #include <sys/ucred.h> 118 #include <sys/domain.h> 119 120 #include <vm/vm.h> 121 #include <sys/sysctl.h> 122 123 #include <net/if.h> 124 #include <net/route.h> 125 126 #include <netinet/in.h> 127 #include <netinet/in_systm.h> 128 #include <netinet/in_var.h> 129 #include <netinet/ip.h> 130 #include <netinet/in_pcb.h> 131 #include <netinet/ip_var.h> 132 133 #ifdef INET6 134 #ifndef INET 135 #include <netinet/in.h> 136 #endif 137 #include <netinet/ip6.h> 138 #include <netinet6/in6_pcb.h> 139 #include <netinet6/ip6_var.h> 140 #endif 141 142 #include <netinet/tcp.h> 143 #include <netinet/tcp_fsm.h> 144 #include <netinet/tcp_seq.h> 145 #include <netinet/tcp_timer.h> 146 #include <netinet/tcp_var.h> 147 #include <netinet/tcpip.h> 148 #include <netinet/tcp_debug.h> 149 150 #include "opt_tcp_recvspace.h" 151 #include "opt_tcp_sendspace.h" 152 153 #ifdef IPSEC 154 #include <netinet6/ipsec.h> 155 #endif /*IPSEC*/ 156 157 /* 158 * TCP protocol interface to socket abstraction. 159 */ 160 extern char *tcpstates[]; 161 162 /* 163 * Process a TCP user request for TCP tb. If this is a send request 164 * then m is the mbuf chain of send data. If this is a timer expiration 165 * (called from the software clock routine), then timertype tells which timer. 166 */ 167 /*ARGSUSED*/ 168 int 169 tcp_usrreq(so, req, m, nam, control, p) 170 struct socket *so; 171 int req; 172 struct mbuf *m, *nam, *control; 173 struct proc *p; 174 { 175 register struct inpcb *inp; 176 #ifdef INET6 177 register struct in6pcb *in6p; 178 #endif 179 register struct tcpcb *tp = NULL; 180 int s; 181 int error = 0; 182 int ostate; 183 int family; /* family of the socket */ 184 185 family = so->so_proto->pr_domain->dom_family; 186 187 if (req == PRU_CONTROL) { 188 switch (family) { 189 case PF_INET: 190 return (in_control(so, (long)m, (caddr_t)nam, 191 (struct ifnet *)control, p)); 192 #ifdef INET6 193 case PF_INET6: 194 return (in6_control(so, (long)m, (caddr_t)nam, 195 (struct ifnet *)control, p)); 196 #endif 197 default: 198 return EAFNOSUPPORT; 199 } 200 } 201 202 s = splsoftnet(); 203 switch (family) { 204 case PF_INET: 205 inp = sotoinpcb(so); 206 #ifdef INET6 207 in6p = NULL; 208 #endif 209 break; 210 #ifdef INET6 211 case PF_INET6: 212 inp = NULL; 213 in6p = sotoin6pcb(so); 214 break; 215 #endif 216 default: 217 splx(s); 218 return EAFNOSUPPORT; 219 } 220 221 #ifdef DIAGNOSTIC 222 if (req != PRU_SEND && req != PRU_SENDOOB && control) 223 panic("tcp_usrreq: unexpected control mbuf"); 224 #endif 225 /* 226 * When a TCP is attached to a socket, then there will be 227 * a (struct inpcb) pointed at by the socket, and this 228 * structure will point at a subsidary (struct tcpcb). 229 */ 230 #ifndef INET6 231 if (inp == 0 && req != PRU_ATTACH) 232 #else 233 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 234 #endif 235 { 236 error = EINVAL; 237 goto release; 238 } 239 if (inp) { 240 tp = intotcpcb(inp); 241 /* WHAT IF TP IS 0? */ 242 #ifdef KPROF 243 tcp_acounts[tp->t_state][req]++; 244 #endif 245 ostate = tp->t_state; 246 } 247 #ifdef INET6 248 else if (in6p) { 249 tp = in6totcpcb(in6p); 250 /* WHAT IF TP IS 0? */ 251 #ifdef KPROF 252 tcp_acounts[tp->t_state][req]++; 253 #endif 254 ostate = tp->t_state; 255 } 256 #endif 257 else 258 ostate = 0; 259 260 switch (req) { 261 262 /* 263 * TCP attaches to socket via PRU_ATTACH, reserving space, 264 * and an internet control block. 265 */ 266 case PRU_ATTACH: 267 #ifndef INET6 268 if (inp != 0) 269 #else 270 if (inp != 0 || in6p != 0) 271 #endif 272 { 273 error = EISCONN; 274 break; 275 } 276 error = tcp_attach(so); 277 if (error) 278 break; 279 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 280 so->so_linger = TCP_LINGERTIME; 281 tp = sototcpcb(so); 282 break; 283 284 /* 285 * PRU_DETACH detaches the TCP protocol from the socket. 286 */ 287 case PRU_DETACH: 288 tp = tcp_disconnect(tp); 289 break; 290 291 /* 292 * Give the socket an address. 293 */ 294 case PRU_BIND: 295 switch (family) { 296 case PF_INET: 297 error = in_pcbbind(inp, nam, p); 298 break; 299 #ifdef INET6 300 case PF_INET6: 301 error = in6_pcbbind(in6p, nam /*, p*/ ); 302 /* mapped addr case */ 303 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 304 tp->t_family = AF_INET; 305 break; 306 #endif 307 } 308 break; 309 310 /* 311 * Prepare to accept connections. 312 */ 313 case PRU_LISTEN: 314 if (inp && inp->inp_lport == 0) { 315 error = in_pcbbind(inp, (struct mbuf *)0, 316 (struct proc *)0); 317 if (error) 318 break; 319 } 320 #ifdef INET6 321 else if (in6p && in6p->in6p_lport == 0) { 322 error = in6_pcbbind(in6p, (struct mbuf *)0 /*, 323 (struct proc *)0 */ ); 324 if (error) 325 break; 326 } 327 #endif 328 tp->t_state = TCPS_LISTEN; 329 break; 330 331 /* 332 * Initiate connection to peer. 333 * Create a template for use in transmissions on this connection. 334 * Enter SYN_SENT state, and mark socket as connecting. 335 * Start keep-alive timer, and seed output sequence space. 336 * Send initial segment on connection. 337 */ 338 case PRU_CONNECT: 339 if (inp) { 340 if (inp->inp_lport == 0) { 341 error = in_pcbbind(inp, (struct mbuf *)0, 342 (struct proc *)0); 343 if (error) 344 break; 345 } 346 error = in_pcbconnect(inp, nam); 347 } 348 #ifdef INET6 349 else if (in6p) { 350 if (in6p->in6p_lport == 0) { 351 error = in6_pcbbind(in6p, (struct mbuf *)0 /*, 352 (struct proc *)0 */ ); 353 if (error) 354 break; 355 } 356 error = in6_pcbconnect(in6p, nam); 357 /* mapped addr case */ 358 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 359 tp->t_family = AF_INET; 360 } 361 #endif 362 if (error) 363 break; 364 tp->t_template = tcp_template(tp); 365 if (tp->t_template == 0) { 366 if (inp) 367 in_pcbdisconnect(inp); 368 #ifdef INET6 369 else if (in6p) 370 in6_pcbdisconnect(in6p); 371 #endif 372 error = ENOBUFS; 373 break; 374 } 375 /* Compute window scaling to request. */ 376 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 377 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 378 tp->request_r_scale++; 379 soisconnecting(so); 380 tcpstat.tcps_connattempt++; 381 tp->t_state = TCPS_SYN_SENT; 382 TCP_TIMER_ARM(tp, TCPT_KEEP, TCPTV_KEEP_INIT); 383 tp->iss = tcp_new_iss(tp, sizeof(struct tcpcb), 0); 384 tcp_sendseqinit(tp); 385 error = tcp_output(tp); 386 break; 387 388 /* 389 * Create a TCP connection between two sockets. 390 */ 391 case PRU_CONNECT2: 392 error = EOPNOTSUPP; 393 break; 394 395 /* 396 * Initiate disconnect from peer. 397 * If connection never passed embryonic stage, just drop; 398 * else if don't need to let data drain, then can just drop anyways, 399 * else have to begin TCP shutdown process: mark socket disconnecting, 400 * drain unread data, state switch to reflect user close, and 401 * send segment (e.g. FIN) to peer. Socket will be really disconnected 402 * when peer sends FIN and acks ours. 403 * 404 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 405 */ 406 case PRU_DISCONNECT: 407 tp = tcp_disconnect(tp); 408 break; 409 410 /* 411 * Accept a connection. Essentially all the work is 412 * done at higher levels; just return the address 413 * of the peer, storing through addr. 414 */ 415 case PRU_ACCEPT: 416 if (inp) 417 in_setpeeraddr(inp, nam); 418 #ifdef INET6 419 else if (in6p) 420 in6_setpeeraddr(in6p, nam); 421 #endif 422 break; 423 424 /* 425 * Mark the connection as being incapable of further output. 426 */ 427 case PRU_SHUTDOWN: 428 socantsendmore(so); 429 tp = tcp_usrclosed(tp); 430 if (tp) 431 error = tcp_output(tp); 432 break; 433 434 /* 435 * After a receive, possibly send window update to peer. 436 */ 437 case PRU_RCVD: 438 (void) tcp_output(tp); 439 break; 440 441 /* 442 * Do a send by putting data in output queue and updating urgent 443 * marker if URG set. Possibly send more data. 444 */ 445 case PRU_SEND: 446 if (control && control->m_len) { 447 m_freem(control); 448 m_freem(m); 449 error = EINVAL; 450 break; 451 } 452 sbappend(&so->so_snd, m); 453 error = tcp_output(tp); 454 break; 455 456 /* 457 * Abort the TCP. 458 */ 459 case PRU_ABORT: 460 tp = tcp_drop(tp, ECONNABORTED); 461 break; 462 463 case PRU_SENSE: 464 /* 465 * stat: don't bother with a blocksize. 466 */ 467 splx(s); 468 return (0); 469 470 case PRU_RCVOOB: 471 if (control && control->m_len) { 472 m_freem(control); 473 m_freem(m); 474 error = EINVAL; 475 break; 476 } 477 if ((so->so_oobmark == 0 && 478 (so->so_state & SS_RCVATMARK) == 0) || 479 so->so_options & SO_OOBINLINE || 480 tp->t_oobflags & TCPOOB_HADDATA) { 481 error = EINVAL; 482 break; 483 } 484 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 485 error = EWOULDBLOCK; 486 break; 487 } 488 m->m_len = 1; 489 *mtod(m, caddr_t) = tp->t_iobc; 490 if (((long)nam & MSG_PEEK) == 0) 491 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 492 break; 493 494 case PRU_SENDOOB: 495 if (sbspace(&so->so_snd) < -512) { 496 m_freem(m); 497 error = ENOBUFS; 498 break; 499 } 500 /* 501 * According to RFC961 (Assigned Protocols), 502 * the urgent pointer points to the last octet 503 * of urgent data. We continue, however, 504 * to consider it to indicate the first octet 505 * of data past the urgent section. 506 * Otherwise, snd_up should be one lower. 507 */ 508 sbappend(&so->so_snd, m); 509 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 510 tp->t_force = 1; 511 error = tcp_output(tp); 512 tp->t_force = 0; 513 break; 514 515 case PRU_SOCKADDR: 516 if (inp) 517 in_setsockaddr(inp, nam); 518 #ifdef INET6 519 else if (in6p) 520 in6_setsockaddr(in6p, nam); 521 #endif 522 break; 523 524 case PRU_PEERADDR: 525 if (inp) 526 in_setpeeraddr(inp, nam); 527 #ifdef INET6 528 else if (in6p) 529 in6_setpeeraddr(in6p, nam); 530 #endif 531 break; 532 533 /* 534 * TCP slow timer went off; going through this 535 * routine for tracing's sake. 536 */ 537 case PRU_SLOWTIMO: 538 tp = tcp_timers(tp, (long)nam); 539 req |= (long)nam << 8; /* for debug's sake */ 540 break; 541 542 default: 543 panic("tcp_usrreq"); 544 } 545 if (tp && (so->so_options & SO_DEBUG)) 546 tcp_trace(TA_USER, ostate, tp, NULL, req); 547 548 release: 549 splx(s); 550 return (error); 551 } 552 553 int 554 tcp_ctloutput(op, so, level, optname, mp) 555 int op; 556 struct socket *so; 557 int level, optname; 558 struct mbuf **mp; 559 { 560 int error = 0, s; 561 struct inpcb *inp; 562 #ifdef INET6 563 register struct in6pcb *in6p; 564 #endif 565 register struct tcpcb *tp; 566 register struct mbuf *m; 567 register int i; 568 int family; /* family of the socket */ 569 570 family = so->so_proto->pr_domain->dom_family; 571 572 s = splsoftnet(); 573 switch (family) { 574 case PF_INET: 575 inp = sotoinpcb(so); 576 #ifdef INET6 577 in6p = NULL; 578 #endif 579 break; 580 #ifdef INET6 581 case PF_INET6: 582 inp = NULL; 583 in6p = sotoin6pcb(so); 584 break; 585 #endif 586 default: 587 splx(s); 588 return EAFNOSUPPORT; 589 } 590 #ifndef INET6 591 if (inp == NULL) 592 #else 593 if (inp == NULL && in6p == NULL) 594 #endif 595 { 596 splx(s); 597 if (op == PRCO_SETOPT && *mp) 598 (void) m_free(*mp); 599 return (ECONNRESET); 600 } 601 if (level != IPPROTO_TCP) { 602 switch (family) { 603 case PF_INET: 604 error = ip_ctloutput(op, so, level, optname, mp); 605 break; 606 #ifdef INET6 607 case PF_INET6: 608 error = ip6_ctloutput(op, so, level, optname, mp); 609 break; 610 #endif 611 } 612 splx(s); 613 return (error); 614 } 615 if (inp) 616 tp = intotcpcb(inp); 617 #ifdef INET6 618 else if (in6p) 619 tp = in6totcpcb(in6p); 620 #endif 621 else 622 tp = NULL; 623 624 switch (op) { 625 626 case PRCO_SETOPT: 627 m = *mp; 628 switch (optname) { 629 630 case TCP_NODELAY: 631 if (m == NULL || m->m_len < sizeof (int)) 632 error = EINVAL; 633 else if (*mtod(m, int *)) 634 tp->t_flags |= TF_NODELAY; 635 else 636 tp->t_flags &= ~TF_NODELAY; 637 break; 638 639 case TCP_MAXSEG: 640 if (m && (i = *mtod(m, int *)) > 0 && 641 i <= tp->t_peermss) 642 tp->t_peermss = i; /* limit on send size */ 643 else 644 error = EINVAL; 645 break; 646 647 default: 648 error = ENOPROTOOPT; 649 break; 650 } 651 if (m) 652 (void) m_free(m); 653 break; 654 655 case PRCO_GETOPT: 656 *mp = m = m_get(M_WAIT, MT_SOOPTS); 657 m->m_len = sizeof(int); 658 659 switch (optname) { 660 case TCP_NODELAY: 661 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 662 break; 663 case TCP_MAXSEG: 664 *mtod(m, int *) = tp->t_peermss; 665 break; 666 default: 667 error = ENOPROTOOPT; 668 break; 669 } 670 break; 671 } 672 splx(s); 673 return (error); 674 } 675 676 #ifndef TCP_SENDSPACE 677 #define TCP_SENDSPACE 1024*16; 678 #endif 679 int tcp_sendspace = TCP_SENDSPACE; 680 #ifndef TCP_RECVSPACE 681 #define TCP_RECVSPACE 1024*16; 682 #endif 683 int tcp_recvspace = TCP_RECVSPACE; 684 685 /* 686 * Attach TCP protocol to socket, allocating 687 * internet protocol control block, tcp control block, 688 * bufer space, and entering LISTEN state if to accept connections. 689 */ 690 int 691 tcp_attach(so) 692 struct socket *so; 693 { 694 register struct tcpcb *tp; 695 struct inpcb *inp; 696 #ifdef INET6 697 struct in6pcb *in6p; 698 #endif 699 int error; 700 int family; /* family of the socket */ 701 702 family = so->so_proto->pr_domain->dom_family; 703 704 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 705 error = soreserve(so, tcp_sendspace, tcp_recvspace); 706 if (error) 707 return (error); 708 } 709 switch (family) { 710 case PF_INET: 711 error = in_pcballoc(so, &tcbtable); 712 if (error) 713 return (error); 714 inp = sotoinpcb(so); 715 #ifdef INET6 716 in6p = NULL; 717 #endif 718 break; 719 #ifdef INET6 720 case PF_INET6: 721 error = in6_pcballoc(so, &tcb6); 722 if (error) 723 return (error); 724 inp = NULL; 725 in6p = sotoin6pcb(so); 726 break; 727 #endif 728 default: 729 return EAFNOSUPPORT; 730 } 731 #ifdef IPSEC 732 if (inp && (error = ipsec_init_policy(&inp->inp_sp)) != 0) { 733 in_pcbdetach(inp); 734 return (error); 735 } 736 #ifdef INET6 737 else if (in6p && (error = ipsec_init_policy(&in6p->in6p_sp)) != 0) { 738 in6_pcbdetach(in6p); 739 return (error); 740 } 741 #endif 742 #endif /*IPSEC*/ 743 if (inp) 744 tp = tcp_newtcpcb(family, (void *)inp); 745 #ifdef INET6 746 else if (in6p) 747 tp = tcp_newtcpcb(family, (void *)in6p); 748 #endif 749 else 750 tp = NULL; 751 752 if (tp == 0) { 753 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 754 755 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 756 if (inp) 757 in_pcbdetach(inp); 758 #ifdef INET6 759 else if (in6p) 760 in6_pcbdetach(in6p); 761 #endif 762 so->so_state |= nofd; 763 return (ENOBUFS); 764 } 765 tp->t_state = TCPS_CLOSED; 766 return (0); 767 } 768 769 /* 770 * Initiate (or continue) disconnect. 771 * If embryonic state, just send reset (once). 772 * If in ``let data drain'' option and linger null, just drop. 773 * Otherwise (hard), mark socket disconnecting and drop 774 * current input data; switch states based on user close, and 775 * send segment to peer (with FIN). 776 */ 777 struct tcpcb * 778 tcp_disconnect(tp) 779 register struct tcpcb *tp; 780 { 781 struct socket *so; 782 783 if (tp->t_inpcb) 784 so = tp->t_inpcb->inp_socket; 785 #ifdef INET6 786 else if (tp->t_in6pcb) 787 so = tp->t_in6pcb->in6p_socket; 788 #endif 789 else 790 so = NULL; 791 792 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 793 tp = tcp_close(tp); 794 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 795 tp = tcp_drop(tp, 0); 796 else { 797 soisdisconnecting(so); 798 sbflush(&so->so_rcv); 799 tp = tcp_usrclosed(tp); 800 if (tp) 801 (void) tcp_output(tp); 802 } 803 return (tp); 804 } 805 806 /* 807 * User issued close, and wish to trail through shutdown states: 808 * if never received SYN, just forget it. If got a SYN from peer, 809 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 810 * If already got a FIN from peer, then almost done; go to LAST_ACK 811 * state. In all other cases, have already sent FIN to peer (e.g. 812 * after PRU_SHUTDOWN), and just have to play tedious game waiting 813 * for peer to send FIN or not respond to keep-alives, etc. 814 * We can let the user exit from the close as soon as the FIN is acked. 815 */ 816 struct tcpcb * 817 tcp_usrclosed(tp) 818 register struct tcpcb *tp; 819 { 820 821 switch (tp->t_state) { 822 823 case TCPS_CLOSED: 824 case TCPS_LISTEN: 825 case TCPS_SYN_SENT: 826 tp->t_state = TCPS_CLOSED; 827 tp = tcp_close(tp); 828 break; 829 830 case TCPS_SYN_RECEIVED: 831 case TCPS_ESTABLISHED: 832 tp->t_state = TCPS_FIN_WAIT_1; 833 break; 834 835 case TCPS_CLOSE_WAIT: 836 tp->t_state = TCPS_LAST_ACK; 837 break; 838 } 839 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 840 struct socket *so; 841 if (tp->t_inpcb) 842 so = tp->t_inpcb->inp_socket; 843 #ifdef INET6 844 else if (tp->t_in6pcb) 845 so = tp->t_in6pcb->in6p_socket; 846 #endif 847 else 848 so = NULL; 849 soisdisconnected(so); 850 /* 851 * If we are in FIN_WAIT_2, we arrived here because the 852 * application did a shutdown of the send side. Like the 853 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 854 * a full close, we start a timer to make sure sockets are 855 * not left in FIN_WAIT_2 forever. 856 */ 857 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tcp_maxidle > 0)) 858 TCP_TIMER_ARM(tp, TCPT_2MSL, tcp_maxidle); 859 } 860 return (tp); 861 } 862 863 static struct { 864 unsigned int valid : 1; 865 unsigned int rdonly : 1; 866 int *var; 867 int val; 868 } tcp_ctlvars[] = TCPCTL_VARIABLES; 869 870 /* 871 * Sysctl for tcp variables. 872 */ 873 int 874 tcp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 875 int *name; 876 u_int namelen; 877 void *oldp; 878 size_t *oldlenp; 879 void *newp; 880 size_t newlen; 881 { 882 /* All sysctl names at this level are terminal. */ 883 if (namelen != 1) 884 return (ENOTDIR); 885 886 if (name[0] < sizeof(tcp_ctlvars)/sizeof(tcp_ctlvars[0]) 887 && tcp_ctlvars[name[0]].valid) { 888 if (tcp_ctlvars[name[0]].rdonly) 889 return (sysctl_rdint(oldp, oldlenp, newp, 890 tcp_ctlvars[name[0]].val)); 891 else 892 return (sysctl_int(oldp, oldlenp, newp, newlen, 893 tcp_ctlvars[name[0]].var)); 894 } 895 896 return (ENOPROTOOPT); 897 } 898
Indexes created Sun Sep 21 20:09:37 GMT 2025