tcp_usrreq.c revision 1.57
1 /* $NetBSD: tcp_usrreq.c,v 1.57 2000/10/17 21:16:57 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 #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 #ifdef INET 191 case PF_INET: 192 return (in_control(so, (long)m, (caddr_t)nam, 193 (struct ifnet *)control, p)); 194 #endif 195 #ifdef INET6 196 case PF_INET6: 197 return (in6_control(so, (long)m, (caddr_t)nam, 198 (struct ifnet *)control, p)); 199 #endif 200 default: 201 return EAFNOSUPPORT; 202 } 203 } 204 205 if (req == PRU_PURGEIF) { 206 switch (family) { 207 #ifdef INET 208 case PF_INET: 209 in_purgeif((struct ifnet *)control); 210 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 211 break; 212 #endif 213 #ifdef INET6 214 case PF_INET6: 215 in6_purgeif((struct ifnet *)control); 216 in6_pcbpurgeif(&tcb6, (struct ifnet *)control); 217 break; 218 #endif 219 default: 220 return (EAFNOSUPPORT); 221 } 222 return (0); 223 } 224 225 s = splsoftnet(); 226 switch (family) { 227 #ifdef INET 228 case PF_INET: 229 inp = sotoinpcb(so); 230 #ifdef INET6 231 in6p = NULL; 232 #endif 233 break; 234 #endif 235 #ifdef INET6 236 case PF_INET6: 237 inp = NULL; 238 in6p = sotoin6pcb(so); 239 break; 240 #endif 241 default: 242 splx(s); 243 return EAFNOSUPPORT; 244 } 245 246 #ifdef DIAGNOSTIC 247 #ifdef INET6 248 if (inp && in6p) 249 panic("tcp_usrreq: both inp and in6p set to non-NULL"); 250 #endif 251 if (req != PRU_SEND && req != PRU_SENDOOB && control) 252 panic("tcp_usrreq: unexpected control mbuf"); 253 #endif 254 /* 255 * When a TCP is attached to a socket, then there will be 256 * a (struct inpcb) pointed at by the socket, and this 257 * structure will point at a subsidary (struct tcpcb). 258 */ 259 #ifndef INET6 260 if (inp == 0 && req != PRU_ATTACH) 261 #else 262 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 263 #endif 264 { 265 error = EINVAL; 266 goto release; 267 } 268 #ifdef INET 269 if (inp) { 270 tp = intotcpcb(inp); 271 /* WHAT IF TP IS 0? */ 272 #ifdef KPROF 273 tcp_acounts[tp->t_state][req]++; 274 #endif 275 ostate = tp->t_state; 276 } 277 #endif 278 #ifdef INET6 279 if (in6p) { 280 tp = in6totcpcb(in6p); 281 /* WHAT IF TP IS 0? */ 282 #ifdef KPROF 283 tcp_acounts[tp->t_state][req]++; 284 #endif 285 ostate = tp->t_state; 286 } 287 #endif 288 else 289 ostate = 0; 290 291 switch (req) { 292 293 /* 294 * TCP attaches to socket via PRU_ATTACH, reserving space, 295 * and an internet control block. 296 */ 297 case PRU_ATTACH: 298 #ifndef INET6 299 if (inp != 0) 300 #else 301 if (inp != 0 || in6p != 0) 302 #endif 303 { 304 error = EISCONN; 305 break; 306 } 307 error = tcp_attach(so); 308 if (error) 309 break; 310 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 311 so->so_linger = TCP_LINGERTIME; 312 tp = sototcpcb(so); 313 break; 314 315 /* 316 * PRU_DETACH detaches the TCP protocol from the socket. 317 */ 318 case PRU_DETACH: 319 tp = tcp_disconnect(tp); 320 break; 321 322 /* 323 * Give the socket an address. 324 */ 325 case PRU_BIND: 326 switch (family) { 327 #ifdef INET 328 case PF_INET: 329 error = in_pcbbind(inp, nam, p); 330 break; 331 #endif 332 #ifdef INET6 333 case PF_INET6: 334 error = in6_pcbbind(in6p, nam, p); 335 /* mapped addr case */ 336 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 337 tp->t_family = AF_INET; 338 break; 339 #endif 340 } 341 break; 342 343 /* 344 * Prepare to accept connections. 345 */ 346 case PRU_LISTEN: 347 #ifdef INET 348 if (inp && inp->inp_lport == 0) { 349 error = in_pcbbind(inp, (struct mbuf *)0, 350 (struct proc *)0); 351 if (error) 352 break; 353 } 354 #endif 355 #ifdef INET6 356 if (in6p && in6p->in6p_lport == 0) { 357 error = in6_pcbbind(in6p, (struct mbuf *)0, 358 (struct proc *)0); 359 if (error) 360 break; 361 } 362 #endif 363 tp->t_state = TCPS_LISTEN; 364 break; 365 366 /* 367 * Initiate connection to peer. 368 * Create a template for use in transmissions on this connection. 369 * Enter SYN_SENT state, and mark socket as connecting. 370 * Start keep-alive timer, and seed output sequence space. 371 * Send initial segment on connection. 372 */ 373 case PRU_CONNECT: 374 #ifdef INET 375 if (inp) { 376 if (inp->inp_lport == 0) { 377 error = in_pcbbind(inp, (struct mbuf *)0, 378 (struct proc *)0); 379 if (error) 380 break; 381 } 382 error = in_pcbconnect(inp, nam); 383 } 384 #endif 385 #ifdef INET6 386 if (in6p) { 387 if (in6p->in6p_lport == 0) { 388 error = in6_pcbbind(in6p, (struct mbuf *)0, 389 (struct proc *)0); 390 if (error) 391 break; 392 } 393 error = in6_pcbconnect(in6p, nam); 394 /* mapped addr case */ 395 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 396 tp->t_family = AF_INET; 397 } 398 #endif 399 if (error) 400 break; 401 tp->t_template = tcp_template(tp); 402 if (tp->t_template == 0) { 403 #ifdef INET 404 if (inp) 405 in_pcbdisconnect(inp); 406 #endif 407 #ifdef INET6 408 if (in6p) 409 in6_pcbdisconnect(in6p); 410 #endif 411 error = ENOBUFS; 412 break; 413 } 414 /* Compute window scaling to request. */ 415 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 416 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 417 tp->request_r_scale++; 418 soisconnecting(so); 419 tcpstat.tcps_connattempt++; 420 tp->t_state = TCPS_SYN_SENT; 421 TCP_TIMER_ARM(tp, TCPT_KEEP, TCPTV_KEEP_INIT); 422 tp->iss = tcp_new_iss(tp, sizeof(struct tcpcb), 0); 423 tcp_sendseqinit(tp); 424 error = tcp_output(tp); 425 break; 426 427 /* 428 * Create a TCP connection between two sockets. 429 */ 430 case PRU_CONNECT2: 431 error = EOPNOTSUPP; 432 break; 433 434 /* 435 * Initiate disconnect from peer. 436 * If connection never passed embryonic stage, just drop; 437 * else if don't need to let data drain, then can just drop anyways, 438 * else have to begin TCP shutdown process: mark socket disconnecting, 439 * drain unread data, state switch to reflect user close, and 440 * send segment (e.g. FIN) to peer. Socket will be really disconnected 441 * when peer sends FIN and acks ours. 442 * 443 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 444 */ 445 case PRU_DISCONNECT: 446 tp = tcp_disconnect(tp); 447 break; 448 449 /* 450 * Accept a connection. Essentially all the work is 451 * done at higher levels; just return the address 452 * of the peer, storing through addr. 453 */ 454 case PRU_ACCEPT: 455 #ifdef INET 456 if (inp) 457 in_setpeeraddr(inp, nam); 458 #endif 459 #ifdef INET6 460 if (in6p) 461 in6_setpeeraddr(in6p, nam); 462 #endif 463 break; 464 465 /* 466 * Mark the connection as being incapable of further output. 467 */ 468 case PRU_SHUTDOWN: 469 socantsendmore(so); 470 tp = tcp_usrclosed(tp); 471 if (tp) 472 error = tcp_output(tp); 473 break; 474 475 /* 476 * After a receive, possibly send window update to peer. 477 */ 478 case PRU_RCVD: 479 (void) tcp_output(tp); 480 break; 481 482 /* 483 * Do a send by putting data in output queue and updating urgent 484 * marker if URG set. Possibly send more data. 485 */ 486 case PRU_SEND: 487 if (control && control->m_len) { 488 m_freem(control); 489 m_freem(m); 490 error = EINVAL; 491 break; 492 } 493 sbappend(&so->so_snd, m); 494 error = tcp_output(tp); 495 break; 496 497 /* 498 * Abort the TCP. 499 */ 500 case PRU_ABORT: 501 tp = tcp_drop(tp, ECONNABORTED); 502 break; 503 504 case PRU_SENSE: 505 /* 506 * stat: don't bother with a blocksize. 507 */ 508 splx(s); 509 return (0); 510 511 case PRU_RCVOOB: 512 if (control && control->m_len) { 513 m_freem(control); 514 m_freem(m); 515 error = EINVAL; 516 break; 517 } 518 if ((so->so_oobmark == 0 && 519 (so->so_state & SS_RCVATMARK) == 0) || 520 so->so_options & SO_OOBINLINE || 521 tp->t_oobflags & TCPOOB_HADDATA) { 522 error = EINVAL; 523 break; 524 } 525 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 526 error = EWOULDBLOCK; 527 break; 528 } 529 m->m_len = 1; 530 *mtod(m, caddr_t) = tp->t_iobc; 531 if (((long)nam & MSG_PEEK) == 0) 532 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 533 break; 534 535 case PRU_SENDOOB: 536 if (sbspace(&so->so_snd) < -512) { 537 m_freem(m); 538 error = ENOBUFS; 539 break; 540 } 541 /* 542 * According to RFC961 (Assigned Protocols), 543 * the urgent pointer points to the last octet 544 * of urgent data. We continue, however, 545 * to consider it to indicate the first octet 546 * of data past the urgent section. 547 * Otherwise, snd_up should be one lower. 548 */ 549 sbappend(&so->so_snd, m); 550 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 551 tp->t_force = 1; 552 error = tcp_output(tp); 553 tp->t_force = 0; 554 break; 555 556 case PRU_SOCKADDR: 557 #ifdef INET 558 if (inp) 559 in_setsockaddr(inp, nam); 560 #endif 561 #ifdef INET6 562 if (in6p) 563 in6_setsockaddr(in6p, nam); 564 #endif 565 break; 566 567 case PRU_PEERADDR: 568 #ifdef INET 569 if (inp) 570 in_setpeeraddr(inp, nam); 571 #endif 572 #ifdef INET6 573 if (in6p) 574 in6_setpeeraddr(in6p, nam); 575 #endif 576 break; 577 578 /* 579 * TCP slow timer went off; going through this 580 * routine for tracing's sake. 581 */ 582 case PRU_SLOWTIMO: 583 tp = tcp_timers(tp, (long)nam); 584 req |= (long)nam << 8; /* for debug's sake */ 585 break; 586 587 default: 588 panic("tcp_usrreq"); 589 } 590 if (tp && (so->so_options & SO_DEBUG)) 591 tcp_trace(TA_USER, ostate, tp, NULL, req); 592 593 release: 594 splx(s); 595 return (error); 596 } 597 598 int 599 tcp_ctloutput(op, so, level, optname, mp) 600 int op; 601 struct socket *so; 602 int level, optname; 603 struct mbuf **mp; 604 { 605 int error = 0, s; 606 struct inpcb *inp; 607 #ifdef INET6 608 struct in6pcb *in6p; 609 #endif 610 struct tcpcb *tp; 611 struct mbuf *m; 612 int i; 613 int family; /* family of the socket */ 614 615 family = so->so_proto->pr_domain->dom_family; 616 617 s = splsoftnet(); 618 switch (family) { 619 #ifdef INET 620 case PF_INET: 621 inp = sotoinpcb(so); 622 #ifdef INET6 623 in6p = NULL; 624 #endif 625 break; 626 #endif 627 #ifdef INET6 628 case PF_INET6: 629 inp = NULL; 630 in6p = sotoin6pcb(so); 631 break; 632 #endif 633 default: 634 splx(s); 635 return EAFNOSUPPORT; 636 } 637 #ifndef INET6 638 if (inp == NULL) 639 #else 640 if (inp == NULL && in6p == NULL) 641 #endif 642 { 643 splx(s); 644 if (op == PRCO_SETOPT && *mp) 645 (void) m_free(*mp); 646 return (ECONNRESET); 647 } 648 if (level != IPPROTO_TCP) { 649 switch (family) { 650 #ifdef INET 651 case PF_INET: 652 error = ip_ctloutput(op, so, level, optname, mp); 653 break; 654 #endif 655 #ifdef INET6 656 case PF_INET6: 657 error = ip6_ctloutput(op, so, level, optname, mp); 658 break; 659 #endif 660 } 661 splx(s); 662 return (error); 663 } 664 if (inp) 665 tp = intotcpcb(inp); 666 #ifdef INET6 667 else if (in6p) 668 tp = in6totcpcb(in6p); 669 #endif 670 else 671 tp = NULL; 672 673 switch (op) { 674 675 case PRCO_SETOPT: 676 m = *mp; 677 switch (optname) { 678 679 case TCP_NODELAY: 680 if (m == NULL || m->m_len < sizeof (int)) 681 error = EINVAL; 682 else if (*mtod(m, int *)) 683 tp->t_flags |= TF_NODELAY; 684 else 685 tp->t_flags &= ~TF_NODELAY; 686 break; 687 688 case TCP_MAXSEG: 689 if (m && (i = *mtod(m, int *)) > 0 && 690 i <= tp->t_peermss) 691 tp->t_peermss = i; /* limit on send size */ 692 else 693 error = EINVAL; 694 break; 695 696 default: 697 error = ENOPROTOOPT; 698 break; 699 } 700 if (m) 701 (void) m_free(m); 702 break; 703 704 case PRCO_GETOPT: 705 *mp = m = m_get(M_WAIT, MT_SOOPTS); 706 m->m_len = sizeof(int); 707 708 switch (optname) { 709 case TCP_NODELAY: 710 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 711 break; 712 case TCP_MAXSEG: 713 *mtod(m, int *) = tp->t_peermss; 714 break; 715 default: 716 error = ENOPROTOOPT; 717 break; 718 } 719 break; 720 } 721 splx(s); 722 return (error); 723 } 724 725 #ifndef TCP_SENDSPACE 726 #define TCP_SENDSPACE 1024*16; 727 #endif 728 int tcp_sendspace = TCP_SENDSPACE; 729 #ifndef TCP_RECVSPACE 730 #define TCP_RECVSPACE 1024*16; 731 #endif 732 int tcp_recvspace = TCP_RECVSPACE; 733 734 /* 735 * Attach TCP protocol to socket, allocating 736 * internet protocol control block, tcp control block, 737 * bufer space, and entering LISTEN state if to accept connections. 738 */ 739 int 740 tcp_attach(so) 741 struct socket *so; 742 { 743 struct tcpcb *tp; 744 struct inpcb *inp; 745 #ifdef INET6 746 struct in6pcb *in6p; 747 #endif 748 int error; 749 int family; /* family of the socket */ 750 751 family = so->so_proto->pr_domain->dom_family; 752 753 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 754 error = soreserve(so, tcp_sendspace, tcp_recvspace); 755 if (error) 756 return (error); 757 } 758 switch (family) { 759 #ifdef INET 760 case PF_INET: 761 error = in_pcballoc(so, &tcbtable); 762 if (error) 763 return (error); 764 inp = sotoinpcb(so); 765 #ifdef INET6 766 in6p = NULL; 767 #endif 768 break; 769 #endif 770 #ifdef INET6 771 case PF_INET6: 772 error = in6_pcballoc(so, &tcb6); 773 if (error) 774 return (error); 775 inp = NULL; 776 in6p = sotoin6pcb(so); 777 break; 778 #endif 779 default: 780 return EAFNOSUPPORT; 781 } 782 #ifdef IPSEC 783 #ifdef INET 784 if (inp) { 785 error = ipsec_init_policy(so, &inp->inp_sp); 786 if (error != 0) { 787 in_pcbdetach(inp); 788 return (error); 789 } 790 } 791 #endif 792 #ifdef INET6 793 if (in6p) { 794 error = ipsec_init_policy(so, &in6p->in6p_sp); 795 if (error != 0) { 796 in6_pcbdetach(in6p); 797 return (error); 798 } 799 } 800 #endif 801 #endif /*IPSEC*/ 802 if (inp) 803 tp = tcp_newtcpcb(family, (void *)inp); 804 #ifdef INET6 805 else if (in6p) 806 tp = tcp_newtcpcb(family, (void *)in6p); 807 #endif 808 else 809 tp = NULL; 810 811 if (tp == 0) { 812 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 813 814 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 815 #ifdef INET 816 if (inp) 817 in_pcbdetach(inp); 818 #endif 819 #ifdef INET6 820 if (in6p) 821 in6_pcbdetach(in6p); 822 #endif 823 so->so_state |= nofd; 824 return (ENOBUFS); 825 } 826 tp->t_state = TCPS_CLOSED; 827 return (0); 828 } 829 830 /* 831 * Initiate (or continue) disconnect. 832 * If embryonic state, just send reset (once). 833 * If in ``let data drain'' option and linger null, just drop. 834 * Otherwise (hard), mark socket disconnecting and drop 835 * current input data; switch states based on user close, and 836 * send segment to peer (with FIN). 837 */ 838 struct tcpcb * 839 tcp_disconnect(tp) 840 struct tcpcb *tp; 841 { 842 struct socket *so; 843 844 if (tp->t_inpcb) 845 so = tp->t_inpcb->inp_socket; 846 #ifdef INET6 847 else if (tp->t_in6pcb) 848 so = tp->t_in6pcb->in6p_socket; 849 #endif 850 else 851 so = NULL; 852 853 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 854 tp = tcp_close(tp); 855 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 856 tp = tcp_drop(tp, 0); 857 else { 858 soisdisconnecting(so); 859 sbflush(&so->so_rcv); 860 tp = tcp_usrclosed(tp); 861 if (tp) 862 (void) tcp_output(tp); 863 } 864 return (tp); 865 } 866 867 /* 868 * User issued close, and wish to trail through shutdown states: 869 * if never received SYN, just forget it. If got a SYN from peer, 870 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 871 * If already got a FIN from peer, then almost done; go to LAST_ACK 872 * state. In all other cases, have already sent FIN to peer (e.g. 873 * after PRU_SHUTDOWN), and just have to play tedious game waiting 874 * for peer to send FIN or not respond to keep-alives, etc. 875 * We can let the user exit from the close as soon as the FIN is acked. 876 */ 877 struct tcpcb * 878 tcp_usrclosed(tp) 879 struct tcpcb *tp; 880 { 881 882 switch (tp->t_state) { 883 884 case TCPS_CLOSED: 885 case TCPS_LISTEN: 886 case TCPS_SYN_SENT: 887 tp->t_state = TCPS_CLOSED; 888 tp = tcp_close(tp); 889 break; 890 891 case TCPS_SYN_RECEIVED: 892 case TCPS_ESTABLISHED: 893 tp->t_state = TCPS_FIN_WAIT_1; 894 break; 895 896 case TCPS_CLOSE_WAIT: 897 tp->t_state = TCPS_LAST_ACK; 898 break; 899 } 900 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 901 struct socket *so; 902 if (tp->t_inpcb) 903 so = tp->t_inpcb->inp_socket; 904 #ifdef INET6 905 else if (tp->t_in6pcb) 906 so = tp->t_in6pcb->in6p_socket; 907 #endif 908 else 909 so = NULL; 910 soisdisconnected(so); 911 /* 912 * If we are in FIN_WAIT_2, we arrived here because the 913 * application did a shutdown of the send side. Like the 914 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 915 * a full close, we start a timer to make sure sockets are 916 * not left in FIN_WAIT_2 forever. 917 */ 918 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tcp_maxidle > 0)) 919 TCP_TIMER_ARM(tp, TCPT_2MSL, tcp_maxidle); 920 } 921 return (tp); 922 } 923 924 static struct { 925 unsigned int valid : 1; 926 unsigned int rdonly : 1; 927 int *var; 928 int val; 929 } tcp_ctlvars[] = TCPCTL_VARIABLES; 930 931 /* 932 * Sysctl for tcp variables. 933 */ 934 int 935 tcp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 936 int *name; 937 u_int namelen; 938 void *oldp; 939 size_t *oldlenp; 940 void *newp; 941 size_t newlen; 942 { 943 944 /* All sysctl names at this level are terminal. */ 945 if (namelen != 1) 946 return (ENOTDIR); 947 948 if (name[0] < sizeof(tcp_ctlvars)/sizeof(tcp_ctlvars[0]) 949 && tcp_ctlvars[name[0]].valid) { 950 if (tcp_ctlvars[name[0]].rdonly) 951 return (sysctl_rdint(oldp, oldlenp, newp, 952 tcp_ctlvars[name[0]].val)); 953 else 954 return (sysctl_int(oldp, oldlenp, newp, newlen, 955 tcp_ctlvars[name[0]].var)); 956 } 957 958 return (ENOPROTOOPT); 959 } 960
Indexes created Sun Sep 21 18:09:52 GMT 2025