tcp_usrreq.c revision 1.62
1 /* $NetBSD: tcp_usrreq.c,v 1.62 2001/07/03 08:06:19 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_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 210 in_purgeif((struct ifnet *)control); 211 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 212 break; 213 #endif 214 #ifdef INET6 215 case PF_INET6: 216 in6_pcbpurgeif0(&tcb6, (struct ifnet *)control); 217 in6_purgeif((struct ifnet *)control); 218 in6_pcbpurgeif(&tcb6, (struct ifnet *)control); 219 break; 220 #endif 221 default: 222 return (EAFNOSUPPORT); 223 } 224 return (0); 225 } 226 227 s = splsoftnet(); 228 switch (family) { 229 #ifdef INET 230 case PF_INET: 231 inp = sotoinpcb(so); 232 #ifdef INET6 233 in6p = NULL; 234 #endif 235 break; 236 #endif 237 #ifdef INET6 238 case PF_INET6: 239 inp = NULL; 240 in6p = sotoin6pcb(so); 241 break; 242 #endif 243 default: 244 splx(s); 245 return EAFNOSUPPORT; 246 } 247 248 #ifdef DIAGNOSTIC 249 #ifdef INET6 250 if (inp && in6p) 251 panic("tcp_usrreq: both inp and in6p set to non-NULL"); 252 #endif 253 if (req != PRU_SEND && req != PRU_SENDOOB && control) 254 panic("tcp_usrreq: unexpected control mbuf"); 255 #endif 256 /* 257 * When a TCP is attached to a socket, then there will be 258 * a (struct inpcb) pointed at by the socket, and this 259 * structure will point at a subsidary (struct tcpcb). 260 */ 261 #ifndef INET6 262 if (inp == 0 && req != PRU_ATTACH) 263 #else 264 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 265 #endif 266 { 267 error = EINVAL; 268 goto release; 269 } 270 #ifdef INET 271 if (inp) { 272 tp = intotcpcb(inp); 273 /* WHAT IF TP IS 0? */ 274 #ifdef KPROF 275 tcp_acounts[tp->t_state][req]++; 276 #endif 277 ostate = tp->t_state; 278 } 279 #endif 280 #ifdef INET6 281 if (in6p) { 282 tp = in6totcpcb(in6p); 283 /* WHAT IF TP IS 0? */ 284 #ifdef KPROF 285 tcp_acounts[tp->t_state][req]++; 286 #endif 287 ostate = tp->t_state; 288 } 289 #endif 290 else 291 ostate = 0; 292 293 switch (req) { 294 295 /* 296 * TCP attaches to socket via PRU_ATTACH, reserving space, 297 * and an internet control block. 298 */ 299 case PRU_ATTACH: 300 #ifndef INET6 301 if (inp != 0) 302 #else 303 if (inp != 0 || in6p != 0) 304 #endif 305 { 306 error = EISCONN; 307 break; 308 } 309 error = tcp_attach(so); 310 if (error) 311 break; 312 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 313 so->so_linger = TCP_LINGERTIME; 314 tp = sototcpcb(so); 315 break; 316 317 /* 318 * PRU_DETACH detaches the TCP protocol from the socket. 319 */ 320 case PRU_DETACH: 321 tp = tcp_disconnect(tp); 322 break; 323 324 /* 325 * Give the socket an address. 326 */ 327 case PRU_BIND: 328 switch (family) { 329 #ifdef INET 330 case PF_INET: 331 error = in_pcbbind(inp, nam, p); 332 break; 333 #endif 334 #ifdef INET6 335 case PF_INET6: 336 error = in6_pcbbind(in6p, nam, p); 337 if (!error) { 338 /* mapped addr case */ 339 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 340 tp->t_family = AF_INET; 341 else 342 tp->t_family = AF_INET6; 343 } 344 break; 345 #endif 346 } 347 break; 348 349 /* 350 * Prepare to accept connections. 351 */ 352 case PRU_LISTEN: 353 #ifdef INET 354 if (inp && inp->inp_lport == 0) { 355 error = in_pcbbind(inp, (struct mbuf *)0, 356 (struct proc *)0); 357 if (error) 358 break; 359 } 360 #endif 361 #ifdef INET6 362 if (in6p && in6p->in6p_lport == 0) { 363 error = in6_pcbbind(in6p, (struct mbuf *)0, 364 (struct proc *)0); 365 if (error) 366 break; 367 } 368 #endif 369 tp->t_state = TCPS_LISTEN; 370 break; 371 372 /* 373 * Initiate connection to peer. 374 * Create a template for use in transmissions on this connection. 375 * Enter SYN_SENT state, and mark socket as connecting. 376 * Start keep-alive timer, and seed output sequence space. 377 * Send initial segment on connection. 378 */ 379 case PRU_CONNECT: 380 #ifdef INET 381 if (inp) { 382 if (inp->inp_lport == 0) { 383 error = in_pcbbind(inp, (struct mbuf *)0, 384 (struct proc *)0); 385 if (error) 386 break; 387 } 388 error = in_pcbconnect(inp, nam); 389 } 390 #endif 391 #ifdef INET6 392 if (in6p) { 393 if (in6p->in6p_lport == 0) { 394 error = in6_pcbbind(in6p, (struct mbuf *)0, 395 (struct proc *)0); 396 if (error) 397 break; 398 } 399 error = in6_pcbconnect(in6p, nam); 400 if (!error) { 401 /* mapped addr case */ 402 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 403 tp->t_family = AF_INET; 404 else 405 tp->t_family = AF_INET6; 406 } 407 } 408 #endif 409 if (error) 410 break; 411 tp->t_template = tcp_template(tp); 412 if (tp->t_template == 0) { 413 #ifdef INET 414 if (inp) 415 in_pcbdisconnect(inp); 416 #endif 417 #ifdef INET6 418 if (in6p) 419 in6_pcbdisconnect(in6p); 420 #endif 421 error = ENOBUFS; 422 break; 423 } 424 /* Compute window scaling to request. */ 425 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 426 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 427 tp->request_r_scale++; 428 soisconnecting(so); 429 tcpstat.tcps_connattempt++; 430 tp->t_state = TCPS_SYN_SENT; 431 TCP_TIMER_ARM(tp, TCPT_KEEP, TCPTV_KEEP_INIT); 432 tp->iss = tcp_new_iss(tp, 0); 433 tcp_sendseqinit(tp); 434 error = tcp_output(tp); 435 break; 436 437 /* 438 * Create a TCP connection between two sockets. 439 */ 440 case PRU_CONNECT2: 441 error = EOPNOTSUPP; 442 break; 443 444 /* 445 * Initiate disconnect from peer. 446 * If connection never passed embryonic stage, just drop; 447 * else if don't need to let data drain, then can just drop anyways, 448 * else have to begin TCP shutdown process: mark socket disconnecting, 449 * drain unread data, state switch to reflect user close, and 450 * send segment (e.g. FIN) to peer. Socket will be really disconnected 451 * when peer sends FIN and acks ours. 452 * 453 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 454 */ 455 case PRU_DISCONNECT: 456 tp = tcp_disconnect(tp); 457 break; 458 459 /* 460 * Accept a connection. Essentially all the work is 461 * done at higher levels; just return the address 462 * of the peer, storing through addr. 463 */ 464 case PRU_ACCEPT: 465 #ifdef INET 466 if (inp) 467 in_setpeeraddr(inp, nam); 468 #endif 469 #ifdef INET6 470 if (in6p) 471 in6_setpeeraddr(in6p, nam); 472 #endif 473 break; 474 475 /* 476 * Mark the connection as being incapable of further output. 477 */ 478 case PRU_SHUTDOWN: 479 socantsendmore(so); 480 tp = tcp_usrclosed(tp); 481 if (tp) 482 error = tcp_output(tp); 483 break; 484 485 /* 486 * After a receive, possibly send window update to peer. 487 */ 488 case PRU_RCVD: 489 /* 490 * soreceive() calls this function when a user receives 491 * ancillary data on a listening socket. We don't call 492 * tcp_output in such a case, since there is no header 493 * template for a listening socket and hence the kernel 494 * will panic. 495 */ 496 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 497 (void) tcp_output(tp); 498 break; 499 500 /* 501 * Do a send by putting data in output queue and updating urgent 502 * marker if URG set. Possibly send more data. 503 */ 504 case PRU_SEND: 505 if (control && control->m_len) { 506 m_freem(control); 507 m_freem(m); 508 error = EINVAL; 509 break; 510 } 511 sbappend(&so->so_snd, m); 512 error = tcp_output(tp); 513 break; 514 515 /* 516 * Abort the TCP. 517 */ 518 case PRU_ABORT: 519 tp = tcp_drop(tp, ECONNABORTED); 520 break; 521 522 case PRU_SENSE: 523 /* 524 * stat: don't bother with a blocksize. 525 */ 526 splx(s); 527 return (0); 528 529 case PRU_RCVOOB: 530 if (control && control->m_len) { 531 m_freem(control); 532 m_freem(m); 533 error = EINVAL; 534 break; 535 } 536 if ((so->so_oobmark == 0 && 537 (so->so_state & SS_RCVATMARK) == 0) || 538 so->so_options & SO_OOBINLINE || 539 tp->t_oobflags & TCPOOB_HADDATA) { 540 error = EINVAL; 541 break; 542 } 543 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 544 error = EWOULDBLOCK; 545 break; 546 } 547 m->m_len = 1; 548 *mtod(m, caddr_t) = tp->t_iobc; 549 if (((long)nam & MSG_PEEK) == 0) 550 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 551 break; 552 553 case PRU_SENDOOB: 554 if (sbspace(&so->so_snd) < -512) { 555 m_freem(m); 556 error = ENOBUFS; 557 break; 558 } 559 /* 560 * According to RFC961 (Assigned Protocols), 561 * the urgent pointer points to the last octet 562 * of urgent data. We continue, however, 563 * to consider it to indicate the first octet 564 * of data past the urgent section. 565 * Otherwise, snd_up should be one lower. 566 */ 567 sbappend(&so->so_snd, m); 568 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 569 tp->t_force = 1; 570 error = tcp_output(tp); 571 tp->t_force = 0; 572 break; 573 574 case PRU_SOCKADDR: 575 #ifdef INET 576 if (inp) 577 in_setsockaddr(inp, nam); 578 #endif 579 #ifdef INET6 580 if (in6p) 581 in6_setsockaddr(in6p, nam); 582 #endif 583 break; 584 585 case PRU_PEERADDR: 586 #ifdef INET 587 if (inp) 588 in_setpeeraddr(inp, nam); 589 #endif 590 #ifdef INET6 591 if (in6p) 592 in6_setpeeraddr(in6p, nam); 593 #endif 594 break; 595 596 /* 597 * TCP slow timer went off; going through this 598 * routine for tracing's sake. 599 */ 600 case PRU_SLOWTIMO: 601 tp = tcp_timers(tp, (long)nam); 602 req |= (long)nam << 8; /* for debug's sake */ 603 break; 604 605 default: 606 panic("tcp_usrreq"); 607 } 608 if (tp && (so->so_options & SO_DEBUG)) 609 tcp_trace(TA_USER, ostate, tp, NULL, req); 610 611 release: 612 splx(s); 613 return (error); 614 } 615 616 int 617 tcp_ctloutput(op, so, level, optname, mp) 618 int op; 619 struct socket *so; 620 int level, optname; 621 struct mbuf **mp; 622 { 623 int error = 0, s; 624 struct inpcb *inp; 625 #ifdef INET6 626 struct in6pcb *in6p; 627 #endif 628 struct tcpcb *tp; 629 struct mbuf *m; 630 int i; 631 int family; /* family of the socket */ 632 633 family = so->so_proto->pr_domain->dom_family; 634 635 s = splsoftnet(); 636 switch (family) { 637 #ifdef INET 638 case PF_INET: 639 inp = sotoinpcb(so); 640 #ifdef INET6 641 in6p = NULL; 642 #endif 643 break; 644 #endif 645 #ifdef INET6 646 case PF_INET6: 647 inp = NULL; 648 in6p = sotoin6pcb(so); 649 break; 650 #endif 651 default: 652 splx(s); 653 return EAFNOSUPPORT; 654 } 655 #ifndef INET6 656 if (inp == NULL) 657 #else 658 if (inp == NULL && in6p == NULL) 659 #endif 660 { 661 splx(s); 662 if (op == PRCO_SETOPT && *mp) 663 (void) m_free(*mp); 664 return (ECONNRESET); 665 } 666 if (level != IPPROTO_TCP) { 667 switch (family) { 668 #ifdef INET 669 case PF_INET: 670 error = ip_ctloutput(op, so, level, optname, mp); 671 break; 672 #endif 673 #ifdef INET6 674 case PF_INET6: 675 error = ip6_ctloutput(op, so, level, optname, mp); 676 break; 677 #endif 678 } 679 splx(s); 680 return (error); 681 } 682 if (inp) 683 tp = intotcpcb(inp); 684 #ifdef INET6 685 else if (in6p) 686 tp = in6totcpcb(in6p); 687 #endif 688 else 689 tp = NULL; 690 691 switch (op) { 692 693 case PRCO_SETOPT: 694 m = *mp; 695 switch (optname) { 696 697 case TCP_NODELAY: 698 if (m == NULL || m->m_len < sizeof (int)) 699 error = EINVAL; 700 else if (*mtod(m, int *)) 701 tp->t_flags |= TF_NODELAY; 702 else 703 tp->t_flags &= ~TF_NODELAY; 704 break; 705 706 case TCP_MAXSEG: 707 if (m && (i = *mtod(m, int *)) > 0 && 708 i <= tp->t_peermss) 709 tp->t_peermss = i; /* limit on send size */ 710 else 711 error = EINVAL; 712 break; 713 714 default: 715 error = ENOPROTOOPT; 716 break; 717 } 718 if (m) 719 (void) m_free(m); 720 break; 721 722 case PRCO_GETOPT: 723 *mp = m = m_get(M_WAIT, MT_SOOPTS); 724 m->m_len = sizeof(int); 725 726 switch (optname) { 727 case TCP_NODELAY: 728 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 729 break; 730 case TCP_MAXSEG: 731 *mtod(m, int *) = tp->t_peermss; 732 break; 733 default: 734 error = ENOPROTOOPT; 735 break; 736 } 737 break; 738 } 739 splx(s); 740 return (error); 741 } 742 743 #ifndef TCP_SENDSPACE 744 #define TCP_SENDSPACE 1024*16; 745 #endif 746 int tcp_sendspace = TCP_SENDSPACE; 747 #ifndef TCP_RECVSPACE 748 #define TCP_RECVSPACE 1024*16; 749 #endif 750 int tcp_recvspace = TCP_RECVSPACE; 751 752 /* 753 * Attach TCP protocol to socket, allocating 754 * internet protocol control block, tcp control block, 755 * bufer space, and entering LISTEN state if to accept connections. 756 */ 757 int 758 tcp_attach(so) 759 struct socket *so; 760 { 761 struct tcpcb *tp; 762 struct inpcb *inp; 763 #ifdef INET6 764 struct in6pcb *in6p; 765 #endif 766 int error; 767 int family; /* family of the socket */ 768 769 family = so->so_proto->pr_domain->dom_family; 770 771 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 772 error = soreserve(so, tcp_sendspace, tcp_recvspace); 773 if (error) 774 return (error); 775 } 776 switch (family) { 777 #ifdef INET 778 case PF_INET: 779 error = in_pcballoc(so, &tcbtable); 780 if (error) 781 return (error); 782 inp = sotoinpcb(so); 783 #ifdef INET6 784 in6p = NULL; 785 #endif 786 break; 787 #endif 788 #ifdef INET6 789 case PF_INET6: 790 error = in6_pcballoc(so, &tcb6); 791 if (error) 792 return (error); 793 inp = NULL; 794 in6p = sotoin6pcb(so); 795 break; 796 #endif 797 default: 798 return EAFNOSUPPORT; 799 } 800 #ifdef IPSEC 801 #ifdef INET 802 if (inp) { 803 error = ipsec_init_policy(so, &inp->inp_sp); 804 if (error != 0) { 805 in_pcbdetach(inp); 806 return (error); 807 } 808 } 809 #endif 810 #ifdef INET6 811 if (in6p) { 812 error = ipsec_init_policy(so, &in6p->in6p_sp); 813 if (error != 0) { 814 in6_pcbdetach(in6p); 815 return (error); 816 } 817 } 818 #endif 819 #endif /*IPSEC*/ 820 if (inp) 821 tp = tcp_newtcpcb(family, (void *)inp); 822 #ifdef INET6 823 else if (in6p) 824 tp = tcp_newtcpcb(family, (void *)in6p); 825 #endif 826 else 827 tp = NULL; 828 829 if (tp == 0) { 830 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 831 832 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 833 #ifdef INET 834 if (inp) 835 in_pcbdetach(inp); 836 #endif 837 #ifdef INET6 838 if (in6p) 839 in6_pcbdetach(in6p); 840 #endif 841 so->so_state |= nofd; 842 return (ENOBUFS); 843 } 844 tp->t_state = TCPS_CLOSED; 845 return (0); 846 } 847 848 /* 849 * Initiate (or continue) disconnect. 850 * If embryonic state, just send reset (once). 851 * If in ``let data drain'' option and linger null, just drop. 852 * Otherwise (hard), mark socket disconnecting and drop 853 * current input data; switch states based on user close, and 854 * send segment to peer (with FIN). 855 */ 856 struct tcpcb * 857 tcp_disconnect(tp) 858 struct tcpcb *tp; 859 { 860 struct socket *so; 861 862 if (tp->t_inpcb) 863 so = tp->t_inpcb->inp_socket; 864 #ifdef INET6 865 else if (tp->t_in6pcb) 866 so = tp->t_in6pcb->in6p_socket; 867 #endif 868 else 869 so = NULL; 870 871 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 872 tp = tcp_close(tp); 873 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 874 tp = tcp_drop(tp, 0); 875 else { 876 soisdisconnecting(so); 877 sbflush(&so->so_rcv); 878 tp = tcp_usrclosed(tp); 879 if (tp) 880 (void) tcp_output(tp); 881 } 882 return (tp); 883 } 884 885 /* 886 * User issued close, and wish to trail through shutdown states: 887 * if never received SYN, just forget it. If got a SYN from peer, 888 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 889 * If already got a FIN from peer, then almost done; go to LAST_ACK 890 * state. In all other cases, have already sent FIN to peer (e.g. 891 * after PRU_SHUTDOWN), and just have to play tedious game waiting 892 * for peer to send FIN or not respond to keep-alives, etc. 893 * We can let the user exit from the close as soon as the FIN is acked. 894 */ 895 struct tcpcb * 896 tcp_usrclosed(tp) 897 struct tcpcb *tp; 898 { 899 900 switch (tp->t_state) { 901 902 case TCPS_CLOSED: 903 case TCPS_LISTEN: 904 case TCPS_SYN_SENT: 905 tp->t_state = TCPS_CLOSED; 906 tp = tcp_close(tp); 907 break; 908 909 case TCPS_SYN_RECEIVED: 910 case TCPS_ESTABLISHED: 911 tp->t_state = TCPS_FIN_WAIT_1; 912 break; 913 914 case TCPS_CLOSE_WAIT: 915 tp->t_state = TCPS_LAST_ACK; 916 break; 917 } 918 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 919 struct socket *so; 920 if (tp->t_inpcb) 921 so = tp->t_inpcb->inp_socket; 922 #ifdef INET6 923 else if (tp->t_in6pcb) 924 so = tp->t_in6pcb->in6p_socket; 925 #endif 926 else 927 so = NULL; 928 soisdisconnected(so); 929 /* 930 * If we are in FIN_WAIT_2, we arrived here because the 931 * application did a shutdown of the send side. Like the 932 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 933 * a full close, we start a timer to make sure sockets are 934 * not left in FIN_WAIT_2 forever. 935 */ 936 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tcp_maxidle > 0)) 937 TCP_TIMER_ARM(tp, TCPT_2MSL, tcp_maxidle); 938 } 939 return (tp); 940 } 941 942 static const struct { 943 unsigned int valid : 1; 944 unsigned int rdonly : 1; 945 int *var; 946 int val; 947 } tcp_ctlvars[] = TCPCTL_VARIABLES; 948 949 /* 950 * Sysctl for tcp variables. 951 */ 952 int 953 tcp_sysctl(name, namelen, oldp, oldlenp, newp, newlen) 954 int *name; 955 u_int namelen; 956 void *oldp; 957 size_t *oldlenp; 958 void *newp; 959 size_t newlen; 960 { 961 962 /* All sysctl names at this level are terminal. */ 963 if (namelen != 1) 964 return (ENOTDIR); 965 966 if (name[0] < sizeof(tcp_ctlvars)/sizeof(tcp_ctlvars[0]) 967 && tcp_ctlvars[name[0]].valid) { 968 if (tcp_ctlvars[name[0]].rdonly) 969 return (sysctl_rdint(oldp, oldlenp, newp, 970 tcp_ctlvars[name[0]].val)); 971 else 972 return (sysctl_int(oldp, oldlenp, newp, newlen, 973 tcp_ctlvars[name[0]].var)); 974 } 975 976 return (ENOPROTOOPT); 977 } 978
Indexes created Sun Sep 21 16:09:51 GMT 2025