tcp_usrreq.c revision 1.54
1 /* $NetBSD: tcp_usrreq.c,v 1.54 2000/10/06 09:24:40 enami 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 #include "opt_ipsec.h" 106 107 #include <sys/param.h> 108 #include <sys/systm.h> 109 #include <sys/kernel.h> 110 #include <sys/malloc.h> 111 #include <sys/mbuf.h> 112 #include <sys/socket.h> 113 #include <sys/socketvar.h> 114 #include <sys/protosw.h> 115 #include <sys/errno.h> 116 #include <sys/stat.h> 117 #include <sys/proc.h> 118 #include <sys/ucred.h> 119 #include <sys/domain.h> 120 121 #include <uvm/uvm_extern.h> 122 #include <sys/sysctl.h> 123 124 #include <net/if.h> 125 #include <net/route.h> 126 127 #include <netinet/in.h> 128 #include <netinet/in_systm.h> 129 #include <netinet/in_var.h> 130 #include <netinet/ip.h> 131 #include <netinet/in_pcb.h> 132 #include <netinet/ip_var.h> 133 134 #ifdef INET6 135 #ifndef INET 136 #include <netinet/in.h> 137 #endif 138 #include <netinet/ip6.h> 139 #include <netinet6/in6_pcb.h> 140 #include <netinet6/ip6_var.h> 141 #endif 142 143 #include <netinet/tcp.h> 144 #include <netinet/tcp_fsm.h> 145 #include <netinet/tcp_seq.h> 146 #include <netinet/tcp_timer.h> 147 #include <netinet/tcp_var.h> 148 #include <netinet/tcpip.h> 149 #include <netinet/tcp_debug.h> 150 151 #include "opt_tcp_recvspace.h" 152 #include "opt_tcp_sendspace.h" 153 154 #ifdef IPSEC 155 #include <netinet6/ipsec.h> 156 #endif /*IPSEC*/ 157 158 /* 159 * TCP protocol interface to socket abstraction. 160 */ 161 extern char *tcpstates[]; 162 163 /* 164 * Process a TCP user request for TCP tb. If this is a send request 165 * then m is the mbuf chain of send data. If this is a timer expiration 166 * (called from the software clock routine), then timertype tells which timer. 167 */ 168 /*ARGSUSED*/ 169 int 170 tcp_usrreq(so, req, m, nam, control, p) 171 struct socket *so; 172 int req; 173 struct mbuf *m, *nam, *control; 174 struct proc *p; 175 { 176 struct inpcb *inp; 177 #ifdef INET6 178 struct in6pcb *in6p; 179 #endif 180 struct tcpcb *tp = NULL; 181 int s; 182 int error = 0; 183 int ostate; 184 int family; /* family of the socket */ 185 186 family = so->so_proto->pr_domain->dom_family; 187 188 if (req == PRU_CONTROL) { 189 switch (family) { 190 case PF_INET: 191 return (in_control(so, (long)m, (caddr_t)nam, 192 (struct ifnet *)control, p)); 193 #ifdef INET6 194 case PF_INET6: 195 return (in6_control(so, (long)m, (caddr_t)nam, 196 (struct ifnet *)control, p)); 197 #endif 198 default: 199 return EAFNOSUPPORT; 200 } 201 } 202 203 if (req == PRU_PURGEIF) { 204 switch (family) { 205 case PF_INET: 206 in_purgeif((struct ifnet *)control); 207 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 208 break; 209 #ifdef INET6 210 case PF_INET6: 211 in6_purgeif((struct ifnet *)control); 212 in6_pcbpurgeif(&tcb6, (struct ifnet *)control); 213 break; 214 #endif 215 default: 216 return (EAFNOSUPPORT); 217 } 218 return (0); 219 } 220 221 s = splsoftnet(); 222 switch (family) { 223 case PF_INET: 224 inp = sotoinpcb(so); 225 #ifdef INET6 226 in6p = NULL; 227 #endif 228 break; 229 #ifdef INET6 230 case PF_INET6: 231 inp = NULL; 232 in6p = sotoin6pcb(so); 233 break; 234 #endif 235 default: 236 splx(s); 237 return EAFNOSUPPORT; 238 } 239 240 #ifdef DIAGNOSTIC 241 if (req != PRU_SEND && req != PRU_SENDOOB && control) 242 panic("tcp_usrreq: unexpected control mbuf"); 243 #endif 244 /* 245 * When a TCP is attached to a socket, then there will be 246 * a (struct inpcb) pointed at by the socket, and this 247 * structure will point at a subsidary (struct tcpcb). 248 */ 249 #ifndef INET6 250 if (inp == 0 && req != PRU_ATTACH) 251 #else 252 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 253 #endif 254 { 255 error = EINVAL; 256 goto release; 257 } 258 if (inp) { 259 tp = intotcpcb(inp); 260 /* WHAT IF TP IS 0? */ 261 #ifdef KPROF 262 tcp_acounts[tp->t_state][req]++; 263 #endif 264 ostate = tp->t_state; 265 } 266 #ifdef INET6 267 else if (in6p) { 268 tp = in6totcpcb(in6p); 269 /* WHAT IF TP IS 0? */ 270 #ifdef KPROF 271 tcp_acounts[tp->t_state][req]++; 272 #endif 273 ostate = tp->t_state; 274 } 275 #endif 276 else 277 ostate = 0; 278 279 switch (req) { 280 281 /* 282 * TCP attaches to socket via PRU_ATTACH, reserving space, 283 * and an internet control block. 284 */ 285 case PRU_ATTACH: 286 #ifndef INET6 287 if (inp != 0) 288 #else 289 if (inp != 0 || in6p != 0) 290 #endif 291 { 292 error = EISCONN; 293 break; 294 } 295 error = tcp_attach(so); 296 if (error) 297 break; 298 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 299 so->so_linger = TCP_LINGERTIME; 300 tp = sototcpcb(so); 301 break; 302 303 /* 304 * PRU_DETACH detaches the TCP protocol from the socket. 305 */ 306 case PRU_DETACH: 307 tp = tcp_disconnect(tp); 308 break; 309 310 /* 311 * Give the socket an address. 312 */ 313 case PRU_BIND: 314 switch (family) { 315 case PF_INET: 316 error = in_pcbbind(inp, nam, p); 317 break; 318 #ifdef INET6 319 case PF_INET6: 320 error = in6_pcbbind(in6p, nam, p); 321 /* mapped addr case */ 322 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 323 tp->t_family = AF_INET; 324 break; 325 #endif 326 } 327 break; 328 329 /* 330 * Prepare to accept connections. 331 */ 332 case PRU_LISTEN: 333 if (inp && inp->inp_lport == 0) { 334 error = in_pcbbind(inp, (struct mbuf *)0, 335 (struct proc *)0); 336 if (error) 337 break; 338 } 339 #ifdef INET6 340 else if (in6p && in6p->in6p_lport == 0) { 341 error = in6_pcbbind(in6p, (struct mbuf *)0, 342 (struct proc *)0); 343 if (error) 344 break; 345 } 346 #endif 347 tp->t_state = TCPS_LISTEN; 348 break; 349 350 /* 351 * Initiate connection to peer. 352 * Create a template for use in transmissions on this connection. 353 * Enter SYN_SENT state, and mark socket as connecting. 354 * Start keep-alive timer, and seed output sequence space. 355 * Send initial segment on connection. 356 */ 357 case PRU_CONNECT: 358 if (inp) { 359 if (inp->inp_lport == 0) { 360 error = in_pcbbind(inp, (struct mbuf *)0, 361 (struct proc *)0); 362 if (error) 363 break; 364 } 365 error = in_pcbconnect(inp, nam); 366 } 367 #ifdef INET6 368 else if (in6p) { 369 if (in6p->in6p_lport == 0) { 370 error = in6_pcbbind(in6p, (struct mbuf *)0, 371 (struct proc *)0); 372 if (error) 373 break; 374 } 375 error = in6_pcbconnect(in6p, nam); 376 /* mapped addr case */ 377 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 378 tp->t_family = AF_INET; 379 } 380 #endif 381 if (error) 382 break; 383 tp->t_template = tcp_template(tp); 384 if (tp->t_template == 0) { 385 if (inp) 386 in_pcbdisconnect(inp); 387 #ifdef INET6 388 else if (in6p) 389 in6_pcbdisconnect(in6p); 390 #endif 391 error = ENOBUFS; 392 break; 393 } 394 /* Compute window scaling to request. */ 395 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 396 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 397 tp->request_r_scale++; 398 soisconnecting(so); 399 tcpstat.tcps_connattempt++; 400 tp->t_state = TCPS_SYN_SENT; 401 TCP_TIMER_ARM(tp, TCPT_KEEP, TCPTV_KEEP_INIT); 402 tp->iss = tcp_new_iss(tp, sizeof(struct tcpcb), 0); 403 tcp_sendseqinit(tp); 404 error = tcp_output(tp); 405 break; 406 407 /* 408 * Create a TCP connection between two sockets. 409 */ 410 case PRU_CONNECT2: 411 error = EOPNOTSUPP; 412 break; 413 414 /* 415 * Initiate disconnect from peer. 416 * If connection never passed embryonic stage, just drop; 417 * else if don't need to let data drain, then can just drop anyways, 418 * else have to begin TCP shutdown process: mark socket disconnecting, 419 * drain unread data, state switch to reflect user close, and 420 * send segment (e.g. FIN) to peer. Socket will be really disconnected 421 * when peer sends FIN and acks ours. 422 * 423 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 424 */ 425 case PRU_DISCONNECT: 426 tp = tcp_disconnect(tp); 427 break; 428 429 /* 430 * Accept a connection. Essentially all the work is 431 * done at higher levels; just return the address 432 * of the peer, storing through addr. 433 */ 434 case PRU_ACCEPT: 435 if (inp) 436 in_setpeeraddr(inp, nam); 437 #ifdef INET6 438 else if (in6p) 439 in6_setpeeraddr(in6p, nam); 440 #endif 441 break; 442 443 /* 444 * Mark the connection as being incapable of further output. 445 */ 446 case PRU_SHUTDOWN: 447 socantsendmore(so); 448 tp = tcp_usrclosed(tp); 449 if (tp) 450 error = tcp_output(tp); 451 break; 452 453 /* 454 * After a receive, possibly send window update to peer. 455 */ 456 case PRU_RCVD: 457 (void) tcp_output(tp); 458 break; 459 460 /* 461 * Do a send by putting data in output queue and updating urgent 462 * marker if URG set. Possibly send more data. 463 */ 464 case PRU_SEND: 465 if (control && control->m_len) { 466 m_freem(control); 467 m_freem(m); 468 error = EINVAL; 469 break; 470 } 471 sbappend(&so->so_snd, m); 472 error = tcp_output(tp); 473 break; 474 475 /* 476 * Abort the TCP. 477 */ 478 case PRU_ABORT: 479 tp = tcp_drop(tp, ECONNABORTED); 480 break; 481 482 case PRU_SENSE: 483 /* 484 * stat: don't bother with a blocksize. 485 */ 486 splx(s); 487 return (0); 488 489 case PRU_RCVOOB: 490 if (control && control->m_len) { 491 m_freem(control); 492 m_freem(m); 493 error = EINVAL; 494 break; 495 } 496 if ((so->so_oobmark == 0 && 497 (so->so_state & SS_RCVATMARK) == 0) || 498 so->so_options & SO_OOBINLINE || 499 tp->t_oobflags & TCPOOB_HADDATA) { 500 error = EINVAL; 501 break; 502 } 503 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 504 error = EWOULDBLOCK; 505 break; 506 } 507 m->m_len = 1; 508 *mtod(m, caddr_t) = tp->t_iobc; 509 if (((long)nam & MSG_PEEK) == 0) 510 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 511 break; 512 513 case PRU_SENDOOB: 514 if (sbspace(&so->so_snd) < -512) { 515 m_freem(m); 516 error = ENOBUFS; 517 break; 518 } 519 /* 520 * According to RFC961 (Assigned Protocols), 521 * the urgent pointer points to the last octet 522 * of urgent data. We continue, however, 523 * to consider it to indicate the first octet 524 * of data past the urgent section. 525 * Otherwise, snd_up should be one lower. 526 */ 527 sbappend(&so->so_snd, m); 528 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 529 tp->t_force = 1; 530 error = tcp_output(tp); 531 tp->t_force = 0; 532 break; 533 534 case PRU_SOCKADDR: 535 if (inp) 536 in_setsockaddr(inp, nam); 537 #ifdef INET6 538 else if (in6p) 539 in6_setsockaddr(in6p, nam); 540 #endif 541 break; 542 543 case PRU_PEERADDR: 544 if (inp) 545 in_setpeeraddr(inp, nam); 546 #ifdef INET6 547 else if (in6p) 548 in6_setpeeraddr(in6p, nam); 549 #endif 550 break; 551 552 /* 553 * TCP slow timer went off; going through this 554 * routine for tracing's sake. 555 */ 556 case PRU_SLOWTIMO: 557 tp = tcp_timers(tp, (long)nam); 558 req |= (long)nam << 8; /* for debug's sake */ 559 break; 560 561 default: 562 panic("tcp_usrreq"); 563 } 564 if (tp && (so->so_options & SO_DEBUG)) 565 tcp_trace(TA_USER, ostate, tp, NULL, req); 566 567 release: 568 splx(s); 569 return (error); 570 } 571 572 int 573 tcp_ctloutput(op, so, level, optname, mp) 574 int op; 575 struct socket *so; 576 int level, optname; 577 struct mbuf **mp; 578 { 579 int error = 0, s; 580 struct inpcb *inp; 581 #ifdef INET6 582 struct in6pcb *in6p; 583 #endif 584 struct tcpcb *tp; 585 struct mbuf *m; 586 int i; 587 int family; /* family of the socket */ 588 589 family = so->so_proto->pr_domain->dom_family; 590 591 s = splsoftnet(); 592 switch (family) { 593 case PF_INET: 594 inp = sotoinpcb(so); 595 #ifdef INET6 596 in6p = NULL; 597 #endif 598 break; 599 #ifdef INET6 600 case PF_INET6: 601 inp = NULL; 602 in6p = sotoin6pcb(so); 603 break; 604 #endif 605 default: 606 splx(s); 607 return EAFNOSUPPORT; 608 } 609 #ifndef INET6 610 if (inp == NULL) 611 #else 612 if (inp == NULL && in6p == NULL) 613 #endif 614 { 615 splx(s); 616 if (op == PRCO_SETOPT && *mp) 617 (void) m_free(*mp); 618 return (ECONNRESET); 619 } 620 if (level != IPPROTO_TCP) { 621 switch (family) { 622 case PF_INET: 623 error = ip_ctloutput(op, so, level, optname, mp); 624 break; 625 #ifdef INET6 626 case PF_INET6: 627 error = ip6_ctloutput(op, so, level, optname, mp); 628 break; 629 #endif 630 } 631 splx(s); 632 return (error); 633 } 634 if (inp) 635 tp = intotcpcb(inp); 636 #ifdef INET6 637 else if (in6p) 638 tp = in6totcpcb(in6p); 639 #endif 640 else 641 tp = NULL; 642 643 switch (op) { 644 645 case PRCO_SETOPT: 646 m = *mp; 647 switch (optname) { 648 649 case TCP_NODELAY: 650 if (m == NULL || m->m_len < sizeof (int)) 651 error = EINVAL; 652 else if (*mtod(m, int *)) 653 tp->t_flags |= TF_NODELAY; 654 else 655 tp->t_flags &= ~TF_NODELAY; 656 break; 657 658 case TCP_MAXSEG: 659 if (m && (i = *mtod(m, int *)) > 0 && 660 i <= tp->t_peermss) 661 tp->t_peermss = i; /* limit on send size */ 662 else 663 error = EINVAL; 664 break; 665 666 default: 667 error = ENOPROTOOPT; 668 break; 669 } 670 if (m) 671 (void) m_free(m); 672 break; 673 674 case PRCO_GETOPT: 675 *mp = m = m_get(M_WAIT, MT_SOOPTS); 676 m->m_len = sizeof(int); 677 678 switch (optname) { 679 case TCP_NODELAY: 680 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 681 break; 682 case TCP_MAXSEG: 683 *mtod(m, int *) = tp->t_peermss; 684 break; 685 default: 686 error = ENOPROTOOPT; 687 break; 688 } 689 break; 690 } 691 splx(s); 692 return (error); 693 } 694 695 #ifndef TCP_SENDSPACE 696 #define TCP_SENDSPACE 1024*16; 697 #endif 698 int tcp_sendspace = TCP_SENDSPACE; 699 #ifndef TCP_RECVSPACE 700 #define TCP_RECVSPACE 1024*16; 701 #endif 702 int tcp_recvspace = TCP_RECVSPACE; 703 704 /* 705 * Attach TCP protocol to socket, allocating 706 * internet protocol control block, tcp control block, 707 * bufer space, and entering LISTEN state if to accept connections. 708 */ 709 int 710 tcp_attach(so) 711 struct socket *so; 712 { 713 struct tcpcb *tp; 714 struct inpcb *inp; 715 #ifdef INET6 716 struct in6pcb *in6p; 717 #endif 718 int error; 719 int family; /* family of the socket */ 720 721 family = so->so_proto->pr_domain->dom_family; 722 723 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 724 error = soreserve(so, tcp_sendspace, tcp_recvspace); 725 if (error) 726 return (error); 727 } 728 switch (family) { 729 case PF_INET: 730 error = in_pcballoc(so, &tcbtable); 731 if (error) 732 return (error); 733 inp = sotoinpcb(so); 734 #ifdef INET6 735 in6p = NULL; 736 #endif 737 break; 738 #ifdef INET6 739 case PF_INET6: 740 error = in6_pcballoc(so, &tcb6); 741 if (error) 742 return (error); 743 inp = NULL; 744 in6p = sotoin6pcb(so); 745 break; 746 #endif 747 default: 748 return EAFNOSUPPORT; 749 } 750 #ifdef IPSEC 751 if (inp) { 752 error = ipsec_init_policy(so, &inp->inp_sp); 753 if (error != 0) { 754 in_pcbdetach(inp); 755 return (error); 756 } 757 } 758 #ifdef INET6 759 else if (in6p) { 760 error = ipsec_init_policy(so, &in6p->in6p_sp); 761 if (error != 0) { 762 in6_pcbdetach(in6p); 763 return (error); 764 } 765 } 766 #endif 767 #endif /*IPSEC*/ 768 if (inp) 769 tp = tcp_newtcpcb(family, (void *)inp); 770 #ifdef INET6 771 else if (in6p) 772 tp = tcp_newtcpcb(family, (void *)in6p); 773 #endif 774 else 775 tp = NULL; 776 777 if (tp == 0) { 778 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 779 780 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 781 if (inp) 782 in_pcbdetach(inp); 783 #ifdef INET6 784 else if (in6p) 785 in6_pcbdetach(in6p); 786 #endif 787 so->so_state |= nofd; 788 return (ENOBUFS); 789 } 790 tp->t_state = TCPS_CLOSED; 791 return (0); 792 } 793 794 /* 795 * Initiate (or continue) disconnect. 796 * If embryonic state, just send reset (once). 797 * If in ``let data drain'' option and linger null, just drop. 798 * Otherwise (hard), mark socket disconnecting and drop 799 * current input data; switch states based on user close, and 800 * send segment to peer (with FIN). 801 */ 802 struct tcpcb * 803 tcp_disconnect(tp) 804 struct tcpcb *tp; 805 { 806 struct socket *so; 807 808 if (tp->t_inpcb) 809 so = tp->t_inpcb->inp_socket; 810 #ifdef INET6 811 else if (tp->t_in6pcb) 812 so = tp->t_in6pcb->in6p_socket; 813 #endif 814 else 815 so = NULL; 816 817 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 818 tp = tcp_close(tp); 819 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 820 tp = tcp_drop(tp, 0); 821 else { 822 soisdisconnecting(so); 823 sbflush(&so->so_rcv); 824 tp = tcp_usrclosed(tp); 825 if (tp) 826 (void) tcp_output(tp); 827 } 828 return (tp); 829 } 830 831 /* 832 * User issued close, and wish to trail through shutdown states: 833 * if never received SYN, just forget it. If got a SYN from peer, 834 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 835 * If already got a FIN from peer, then almost done; go to LAST_ACK 836 * state. In all other cases, have already sent FIN to peer (e.g. 837 * after PRU_SHUTDOWN), and just have to play tedious game waiting 838 * for peer to send FIN or not respond to keep-alives, etc. 839 * We can let the user exit from the close as soon as the FIN is acked. 840 */ 841 struct tcpcb * 842 tcp_usrclosed(tp) 843 struct tcpcb *tp; 844 { 845 846 switch (tp->t_state) { 847 848 case TCPS_CLOSED: 849 case TCPS_LISTEN: 850 case TCPS_SYN_SENT: 851 tp->t_state = TCPS_CLOSED; 852 tp = tcp_close(tp); 853 break; 854 855 case TCPS_SYN_RECEIVED: 856 case TCPS_ESTABLISHED: 857 tp->t_state = TCPS_FIN_WAIT_1; 858 break; 859 860 case TCPS_CLOSE_WAIT: 861 tp->t_state = TCPS_LAST_ACK; 862 break; 863 } 864 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 865 struct socket *so; 866 if (tp->t_inpcb) 867 so = tp->t_inpcb->inp_socket; 868 #ifdef INET6 869 else if (tp->t_in6pcb) 870 so = tp->t_in6pcb->in6p_socket; 871 #endif 872 else 873 so = NULL; 874 soisdisconnected(so); 875 /* 876 * If we are in FIN_WAIT_2, we arrived here because the 877 * application did a shutdown of the send side. Like the 878 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 879 * a full close, we start a timer to make sure sockets are 880 * not left in FIN_WAIT_2 forever. 881 */ 882 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tcp_maxidle > 0)) 883 TCP_TIMER_ARM(tp, TCPT_2MSL, tcp_maxidle); 884 } 885 return (tp); 886 } 887 888 static struct { 889 unsigned int valid : 1; 890 unsigned int rdonly : 1; 891 int *var; 892 int val; 893 } tcp_ctlvars[] = TCPCTL_VARIABLES; 894 895 /* 896 * Sysctl for tcp variables. 897 */ 898 int 899 tcp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 900 int *name; 901 u_int namelen; 902 void *oldp; 903 size_t *oldlenp; 904 void *newp; 905 size_t newlen; 906 { 907 908 /* All sysctl names at this level are terminal. */ 909 if (namelen != 1) 910 return (ENOTDIR); 911 912 if (name[0] < sizeof(tcp_ctlvars)/sizeof(tcp_ctlvars[0]) 913 && tcp_ctlvars[name[0]].valid) { 914 if (tcp_ctlvars[name[0]].rdonly) 915 return (sysctl_rdint(oldp, oldlenp, newp, 916 tcp_ctlvars[name[0]].val)); 917 else 918 return (sysctl_int(oldp, oldlenp, newp, newlen, 919 tcp_ctlvars[name[0]].var)); 920 } 921 922 return (ENOPROTOOPT); 923 } 924
Indexes created Sun Sep 21 20:09:37 GMT 2025