tcp_usrreq.c revision 1.166.4.1
1 /* $NetBSD: tcp_usrreq.c,v 1.166.4.1 2013/07/17 03:16:31 rmind 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, 2005, 2006 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 * This code is derived from software contributed to The NetBSD Foundation 40 * by Charles M. Hannum. 41 * This code is derived from software contributed to The NetBSD Foundation 42 * by Rui Paulo. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 54 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 55 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 56 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 57 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 58 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 59 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 60 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 61 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 62 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 63 * POSSIBILITY OF SUCH DAMAGE. 64 */ 65 66 /* 67 * Copyright (c) 1982, 1986, 1988, 1993, 1995 68 * The Regents of the University of California. All rights reserved. 69 * 70 * Redistribution and use in source and binary forms, with or without 71 * modification, are permitted provided that the following conditions 72 * are met: 73 * 1. Redistributions of source code must retain the above copyright 74 * notice, this list of conditions and the following disclaimer. 75 * 2. Redistributions in binary form must reproduce the above copyright 76 * notice, this list of conditions and the following disclaimer in the 77 * documentation and/or other materials provided with the distribution. 78 * 3. Neither the name of the University nor the names of its contributors 79 * may be used to endorse or promote products derived from this software 80 * without specific prior written permission. 81 * 82 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 83 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 84 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 85 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 86 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 87 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 88 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 89 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 90 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 91 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 92 * SUCH DAMAGE. 93 * 94 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 95 */ 96 97 #include <sys/cdefs.h> 98 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.166.4.1 2013/07/17 03:16:31 rmind Exp $"); 99 100 #include "opt_inet.h" 101 #include "opt_ipsec.h" 102 #include "opt_tcp_debug.h" 103 #include "opt_mbuftrace.h" 104 105 106 #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/domain.h> 118 #include <sys/sysctl.h> 119 #include <sys/kauth.h> 120 #include <sys/uidinfo.h> 121 122 #include <net/if.h> 123 #include <net/route.h> 124 125 #include <netinet/in.h> 126 #include <netinet/in_systm.h> 127 #include <netinet/in_var.h> 128 #include <netinet/ip.h> 129 #include <netinet/in_pcb.h> 130 #include <netinet/ip_var.h> 131 #include <netinet/in_offload.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 #include <netinet6/scope6_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/tcp_private.h> 149 #include <netinet/tcp_congctl.h> 150 #include <netinet/tcpip.h> 151 #include <netinet/tcp_debug.h> 152 #include <netinet/tcp_vtw.h> 153 154 #include "opt_tcp_space.h" 155 156 /* 157 * TCP protocol interface to socket abstraction. 158 */ 159 160 /* 161 * Process a TCP user request for TCP tb. If this is a send request 162 * then m is the mbuf chain of send data. If this is a timer expiration 163 * (called from the software clock routine), then timertype tells which timer. 164 */ 165 int 166 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 167 struct mbuf *control, struct lwp *l) 168 { 169 inpcb_t *inp; 170 #ifdef INET6 171 struct in6pcb *in6p; 172 #endif 173 struct tcpcb *tp = NULL; 174 int s; 175 int error = 0; 176 #ifdef TCP_DEBUG 177 int ostate = 0; 178 #endif 179 int family; /* family of the socket */ 180 181 family = so->so_proto->pr_domain->dom_family; 182 183 if (req == PRU_CONTROL) { 184 switch (family) { 185 #ifdef INET 186 case PF_INET: 187 return (in_control(so, (long)m, (void *)nam, 188 (struct ifnet *)control, l)); 189 #endif 190 #ifdef INET6 191 case PF_INET6: 192 return (in6_control(so, (long)m, (void *)nam, 193 (struct ifnet *)control, l)); 194 #endif 195 default: 196 return EAFNOSUPPORT; 197 } 198 } 199 200 s = splsoftnet(); 201 202 if (req == PRU_PURGEIF) { 203 mutex_enter(softnet_lock); 204 switch (family) { 205 #ifdef INET 206 case PF_INET: 207 inpcb_purgeif0(tcbtable, (struct ifnet *)control); 208 in_purgeif((struct ifnet *)control); 209 inpcb_purgeif(tcbtable, (struct ifnet *)control); 210 break; 211 #endif 212 #ifdef INET6 213 case PF_INET6: 214 in6_pcbpurgeif0(tcbtable, (struct ifnet *)control); 215 in6_purgeif((struct ifnet *)control); 216 in6_pcbpurgeif(tcbtable, (struct ifnet *)control); 217 break; 218 #endif 219 default: 220 mutex_exit(softnet_lock); 221 splx(s); 222 return (EAFNOSUPPORT); 223 } 224 mutex_exit(softnet_lock); 225 splx(s); 226 return (0); 227 } 228 229 KASSERT(req == PRU_ATTACH || solocked(so)); 230 231 if (req == PRU_ATTACH) 232 sosetlock(so); 233 234 switch (family) { 235 #ifdef INET 236 case PF_INET: 237 inp = sotoinpcb(so); 238 #ifdef INET6 239 in6p = NULL; 240 #endif 241 break; 242 #endif 243 #ifdef INET6 244 case PF_INET6: 245 inp = NULL; 246 in6p = sotoin6pcb(so); 247 break; 248 #endif 249 default: 250 splx(s); 251 return EAFNOSUPPORT; 252 } 253 254 #ifdef DIAGNOSTIC 255 #ifdef INET6 256 if (inp && in6p) 257 panic("tcp_usrreq: both inp and in6p set to non-NULL"); 258 #endif 259 if (req != PRU_SEND && req != PRU_SENDOOB && control) 260 panic("tcp_usrreq: unexpected control mbuf"); 261 #endif 262 /* 263 * When a TCP is attached to a socket, then there will be 264 * a (struct inpcb) pointed at by the socket, and this 265 * structure will point at a subsidary (struct tcpcb). 266 */ 267 if ((inp == 0 268 #ifdef INET6 269 && in6p == 0 270 #endif 271 ) && (req != PRU_ATTACH && req != PRU_SENSE)) 272 { 273 error = EINVAL; 274 goto release; 275 } 276 #ifdef INET 277 if (inp) { 278 tp = intotcpcb(inp); 279 /* WHAT IF TP IS 0? */ 280 #ifdef KPROF 281 tcp_acounts[tp->t_state][req]++; 282 #endif 283 #ifdef TCP_DEBUG 284 ostate = tp->t_state; 285 #endif 286 } 287 #endif 288 #ifdef INET6 289 if (in6p) { 290 tp = in6totcpcb(in6p); 291 /* WHAT IF TP IS 0? */ 292 #ifdef KPROF 293 tcp_acounts[tp->t_state][req]++; 294 #endif 295 #ifdef TCP_DEBUG 296 ostate = tp->t_state; 297 #endif 298 } 299 #endif 300 301 switch (req) { 302 303 /* 304 * TCP attaches to socket via PRU_ATTACH, reserving space, 305 * and an internet control block. 306 */ 307 case PRU_ATTACH: 308 #ifndef INET6 309 if (inp != 0) 310 #else 311 if (inp != 0 || in6p != 0) 312 #endif 313 { 314 error = EISCONN; 315 break; 316 } 317 error = tcp_attach(so); 318 if (error) 319 break; 320 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 321 so->so_linger = TCP_LINGERTIME; 322 tp = sototcpcb(so); 323 break; 324 325 /* 326 * PRU_DETACH detaches the TCP protocol from the socket. 327 */ 328 case PRU_DETACH: 329 tp = tcp_disconnect(tp); 330 break; 331 332 /* 333 * Give the socket an address. 334 */ 335 case PRU_BIND: 336 switch (family) { 337 #ifdef INET 338 case PF_INET: 339 error = inpcb_bind(inp, nam, l); 340 break; 341 #endif 342 #ifdef INET6 343 case PF_INET6: 344 error = in6_pcbbind(in6p, nam, l); 345 if (!error) { 346 /* mapped addr case */ 347 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 348 tp->t_family = AF_INET; 349 else 350 tp->t_family = AF_INET6; 351 } 352 break; 353 #endif 354 } 355 break; 356 357 /* 358 * Prepare to accept connections. 359 */ 360 case PRU_LISTEN: 361 #ifdef INET 362 if (inp) { 363 in_port_t lport; 364 365 inpcb_get_ports(inp, &lport, NULL); 366 if (lport == 0) { 367 error = inpcb_bind(inp, NULL, l); 368 if (error) 369 break; 370 } 371 } 372 #endif 373 #ifdef INET6 374 if (in6p && in6p->in6p_lport == 0) { 375 error = in6_pcbbind(in6p, NULL, l); 376 if (error) 377 break; 378 } 379 #endif 380 tp->t_state = TCPS_LISTEN; 381 break; 382 383 /* 384 * Initiate connection to peer. 385 * Create a template for use in transmissions on this connection. 386 * Enter SYN_SENT state, and mark socket as connecting. 387 * Start keep-alive timer, and seed output sequence space. 388 * Send initial segment on connection. 389 */ 390 case PRU_CONNECT: 391 #ifdef INET 392 if (inp) { 393 in_port_t lport; 394 395 inpcb_get_ports(inp, &lport, NULL); 396 if (lport == 0) { 397 error = inpcb_bind(inp, NULL, l); 398 if (error) 399 break; 400 } 401 error = inpcb_connect(inp, nam, l); 402 } 403 #endif 404 #ifdef INET6 405 if (in6p) { 406 if (in6p->in6p_lport == 0) { 407 error = in6_pcbbind(in6p, NULL, l); 408 if (error) 409 break; 410 } 411 error = in6_pcbconnect(in6p, nam, l); 412 if (!error) { 413 /* mapped addr case */ 414 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 415 tp->t_family = AF_INET; 416 else 417 tp->t_family = AF_INET6; 418 } 419 } 420 #endif 421 if (error) 422 break; 423 tp->t_template = tcp_template(tp); 424 if (tp->t_template == 0) { 425 #ifdef INET 426 if (inp) 427 inpcb_disconnect(inp); 428 #endif 429 #ifdef INET6 430 if (in6p) 431 in6_pcbdisconnect(in6p); 432 #endif 433 error = ENOBUFS; 434 break; 435 } 436 /* 437 * Compute window scaling to request. 438 * XXX: This should be moved to tcp_output(). 439 */ 440 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 441 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 442 tp->request_r_scale++; 443 soisconnecting(so); 444 TCP_STATINC(TCP_STAT_CONNATTEMPT); 445 tp->t_state = TCPS_SYN_SENT; 446 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 447 tp->iss = tcp_new_iss(tp, 0); 448 tcp_sendseqinit(tp); 449 error = tcp_output(tp); 450 break; 451 452 /* 453 * Create a TCP connection between two sockets. 454 */ 455 case PRU_CONNECT2: 456 error = EOPNOTSUPP; 457 break; 458 459 /* 460 * Initiate disconnect from peer. 461 * If connection never passed embryonic stage, just drop; 462 * else if don't need to let data drain, then can just drop anyways, 463 * else have to begin TCP shutdown process: mark socket disconnecting, 464 * drain unread data, state switch to reflect user close, and 465 * send segment (e.g. FIN) to peer. Socket will be really disconnected 466 * when peer sends FIN and acks ours. 467 * 468 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 469 */ 470 case PRU_DISCONNECT: 471 tp = tcp_disconnect(tp); 472 break; 473 474 /* 475 * Accept a connection. Essentially all the work is 476 * done at higher levels; just return the address 477 * of the peer, storing through addr. 478 */ 479 case PRU_ACCEPT: 480 #ifdef INET 481 if (inp) 482 inpcb_fetch_peeraddr(inp, nam); 483 #endif 484 #ifdef INET6 485 if (in6p) 486 in6_setpeeraddr(in6p, nam); 487 #endif 488 break; 489 490 /* 491 * Mark the connection as being incapable of further output. 492 */ 493 case PRU_SHUTDOWN: 494 socantsendmore(so); 495 tp = tcp_usrclosed(tp); 496 if (tp) 497 error = tcp_output(tp); 498 break; 499 500 /* 501 * After a receive, possibly send window update to peer. 502 */ 503 case PRU_RCVD: 504 /* 505 * soreceive() calls this function when a user receives 506 * ancillary data on a listening socket. We don't call 507 * tcp_output in such a case, since there is no header 508 * template for a listening socket and hence the kernel 509 * will panic. 510 */ 511 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 512 (void) tcp_output(tp); 513 break; 514 515 /* 516 * Do a send by putting data in output queue and updating urgent 517 * marker if URG set. Possibly send more data. 518 */ 519 case PRU_SEND: 520 if (control && control->m_len) { 521 m_freem(control); 522 m_freem(m); 523 error = EINVAL; 524 break; 525 } 526 sbappendstream(&so->so_snd, m); 527 error = tcp_output(tp); 528 break; 529 530 /* 531 * Abort the TCP. 532 */ 533 case PRU_ABORT: 534 tp = tcp_drop(tp, ECONNABORTED); 535 break; 536 537 case PRU_SENSE: 538 /* 539 * stat: don't bother with a blocksize. 540 */ 541 splx(s); 542 return (0); 543 544 case PRU_RCVOOB: 545 if (control && control->m_len) { 546 m_freem(control); 547 m_freem(m); 548 error = EINVAL; 549 break; 550 } 551 if ((so->so_oobmark == 0 && 552 (so->so_state & SS_RCVATMARK) == 0) || 553 so->so_options & SO_OOBINLINE || 554 tp->t_oobflags & TCPOOB_HADDATA) { 555 error = EINVAL; 556 break; 557 } 558 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 559 error = EWOULDBLOCK; 560 break; 561 } 562 m->m_len = 1; 563 *mtod(m, char *) = tp->t_iobc; 564 if (((long)nam & MSG_PEEK) == 0) 565 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 566 break; 567 568 case PRU_SENDOOB: 569 if (sbspace(&so->so_snd) < -512) { 570 m_freem(m); 571 error = ENOBUFS; 572 break; 573 } 574 /* 575 * According to RFC961 (Assigned Protocols), 576 * the urgent pointer points to the last octet 577 * of urgent data. We continue, however, 578 * to consider it to indicate the first octet 579 * of data past the urgent section. 580 * Otherwise, snd_up should be one lower. 581 */ 582 sbappendstream(&so->so_snd, m); 583 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 584 tp->t_force = 1; 585 error = tcp_output(tp); 586 tp->t_force = 0; 587 break; 588 589 case PRU_SOCKADDR: 590 #ifdef INET 591 if (inp) 592 inpcb_fetch_sockaddr(inp, nam); 593 #endif 594 #ifdef INET6 595 if (in6p) 596 in6_setsockaddr(in6p, nam); 597 #endif 598 break; 599 600 case PRU_PEERADDR: 601 #ifdef INET 602 if (inp) 603 inpcb_fetch_peeraddr(inp, nam); 604 #endif 605 #ifdef INET6 606 if (in6p) 607 in6_setpeeraddr(in6p, nam); 608 #endif 609 break; 610 611 default: 612 panic("tcp_usrreq"); 613 } 614 #ifdef TCP_DEBUG 615 if (tp && (so->so_options & SO_DEBUG)) 616 tcp_trace(TA_USER, ostate, tp, NULL, req); 617 #endif 618 619 release: 620 splx(s); 621 return (error); 622 } 623 624 static void 625 change_keepalive(struct socket *so, struct tcpcb *tp) 626 { 627 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 628 TCP_TIMER_DISARM(tp, TCPT_KEEP); 629 TCP_TIMER_DISARM(tp, TCPT_2MSL); 630 631 if (tp->t_state == TCPS_SYN_RECEIVED || 632 tp->t_state == TCPS_SYN_SENT) { 633 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 634 } else if (so->so_options & SO_KEEPALIVE && 635 tp->t_state <= TCPS_CLOSE_WAIT) { 636 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 637 } else { 638 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 639 } 640 641 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 642 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 643 } 644 645 646 int 647 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 648 { 649 int error = 0, s; 650 inpcb_t *inp; 651 #ifdef INET6 652 struct in6pcb *in6p; 653 #endif 654 struct tcpcb *tp; 655 u_int ui; 656 int family; /* family of the socket */ 657 int level, optname, optval; 658 659 level = sopt->sopt_level; 660 optname = sopt->sopt_name; 661 662 family = so->so_proto->pr_domain->dom_family; 663 664 s = splsoftnet(); 665 switch (family) { 666 #ifdef INET 667 case PF_INET: 668 inp = sotoinpcb(so); 669 #ifdef INET6 670 in6p = NULL; 671 #endif 672 break; 673 #endif 674 #ifdef INET6 675 case PF_INET6: 676 inp = NULL; 677 in6p = sotoin6pcb(so); 678 break; 679 #endif 680 default: 681 splx(s); 682 panic("%s: af %d", __func__, family); 683 } 684 #ifndef INET6 685 if (inp == NULL) 686 #else 687 if (inp == NULL && in6p == NULL) 688 #endif 689 { 690 splx(s); 691 return (ECONNRESET); 692 } 693 if (level != IPPROTO_TCP) { 694 switch (family) { 695 #ifdef INET 696 case PF_INET: 697 error = ip_ctloutput(op, so, sopt); 698 break; 699 #endif 700 #ifdef INET6 701 case PF_INET6: 702 error = ip6_ctloutput(op, so, sopt); 703 break; 704 #endif 705 } 706 splx(s); 707 return (error); 708 } 709 if (inp) 710 tp = intotcpcb(inp); 711 #ifdef INET6 712 else if (in6p) 713 tp = in6totcpcb(in6p); 714 #endif 715 else 716 tp = NULL; 717 718 switch (op) { 719 case PRCO_SETOPT: 720 switch (optname) { 721 #ifdef TCP_SIGNATURE 722 case TCP_MD5SIG: 723 error = sockopt_getint(sopt, &optval); 724 if (error) 725 break; 726 if (optval > 0) 727 tp->t_flags |= TF_SIGNATURE; 728 else 729 tp->t_flags &= ~TF_SIGNATURE; 730 break; 731 #endif /* TCP_SIGNATURE */ 732 733 case TCP_NODELAY: 734 error = sockopt_getint(sopt, &optval); 735 if (error) 736 break; 737 if (optval) 738 tp->t_flags |= TF_NODELAY; 739 else 740 tp->t_flags &= ~TF_NODELAY; 741 break; 742 743 case TCP_MAXSEG: 744 error = sockopt_getint(sopt, &optval); 745 if (error) 746 break; 747 if (optval > 0 && optval <= tp->t_peermss) 748 tp->t_peermss = optval; /* limit on send size */ 749 else 750 error = EINVAL; 751 break; 752 #ifdef notyet 753 case TCP_CONGCTL: 754 /* XXX string overflow XXX */ 755 error = tcp_congctl_select(tp, sopt->sopt_data); 756 break; 757 #endif 758 759 case TCP_KEEPIDLE: 760 error = sockopt_get(sopt, &ui, sizeof(ui)); 761 if (error) 762 break; 763 if (ui > 0) { 764 tp->t_keepidle = ui; 765 change_keepalive(so, tp); 766 } else 767 error = EINVAL; 768 break; 769 770 case TCP_KEEPINTVL: 771 error = sockopt_get(sopt, &ui, sizeof(ui)); 772 if (error) 773 break; 774 if (ui > 0) { 775 tp->t_keepintvl = ui; 776 change_keepalive(so, tp); 777 } else 778 error = EINVAL; 779 break; 780 781 case TCP_KEEPCNT: 782 error = sockopt_get(sopt, &ui, sizeof(ui)); 783 if (error) 784 break; 785 if (ui > 0) { 786 tp->t_keepcnt = ui; 787 change_keepalive(so, tp); 788 } else 789 error = EINVAL; 790 break; 791 792 case TCP_KEEPINIT: 793 error = sockopt_get(sopt, &ui, sizeof(ui)); 794 if (error) 795 break; 796 if (ui > 0) { 797 tp->t_keepinit = ui; 798 change_keepalive(so, tp); 799 } else 800 error = EINVAL; 801 break; 802 803 default: 804 error = ENOPROTOOPT; 805 break; 806 } 807 break; 808 809 case PRCO_GETOPT: 810 switch (optname) { 811 #ifdef TCP_SIGNATURE 812 case TCP_MD5SIG: 813 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 814 error = sockopt_set(sopt, &optval, sizeof(optval)); 815 break; 816 #endif 817 case TCP_NODELAY: 818 optval = tp->t_flags & TF_NODELAY; 819 error = sockopt_set(sopt, &optval, sizeof(optval)); 820 break; 821 case TCP_MAXSEG: 822 optval = tp->t_peermss; 823 error = sockopt_set(sopt, &optval, sizeof(optval)); 824 break; 825 #ifdef notyet 826 case TCP_CONGCTL: 827 break; 828 #endif 829 default: 830 error = ENOPROTOOPT; 831 break; 832 } 833 break; 834 } 835 splx(s); 836 return (error); 837 } 838 839 #ifndef TCP_SENDSPACE 840 #define TCP_SENDSPACE 1024*32 841 #endif 842 int tcp_sendspace = TCP_SENDSPACE; 843 #ifndef TCP_RECVSPACE 844 #define TCP_RECVSPACE 1024*32 845 #endif 846 int tcp_recvspace = TCP_RECVSPACE; 847 848 /* 849 * Attach TCP protocol to socket, allocating 850 * internet protocol control block, tcp control block, 851 * bufer space, and entering LISTEN state if to accept connections. 852 */ 853 int 854 tcp_attach(struct socket *so) 855 { 856 struct tcpcb *tp; 857 inpcb_t *inp; 858 #ifdef INET6 859 struct in6pcb *in6p; 860 #endif 861 int error; 862 int family; /* family of the socket */ 863 864 family = so->so_proto->pr_domain->dom_family; 865 866 #ifdef MBUFTRACE 867 so->so_mowner = &tcp_sock_mowner; 868 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 869 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 870 #endif 871 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 872 error = soreserve(so, tcp_sendspace, tcp_recvspace); 873 if (error) 874 return (error); 875 } 876 877 so->so_rcv.sb_flags |= SB_AUTOSIZE; 878 so->so_snd.sb_flags |= SB_AUTOSIZE; 879 880 switch (family) { 881 #ifdef INET 882 case PF_INET: 883 error = inpcb_create(so, tcbtable); 884 if (error) 885 return (error); 886 inp = sotoinpcb(so); 887 #ifdef INET6 888 in6p = NULL; 889 #endif 890 break; 891 #endif 892 #ifdef INET6 893 case PF_INET6: 894 error = in6_pcballoc(so, tcbtable); 895 if (error) 896 return (error); 897 inp = NULL; 898 in6p = sotoin6pcb(so); 899 break; 900 #endif 901 default: 902 return EAFNOSUPPORT; 903 } 904 if (inp) 905 tp = tcp_newtcpcb(family, (void *)inp); 906 #ifdef INET6 907 else if (in6p) 908 tp = tcp_newtcpcb(family, (void *)in6p); 909 #endif 910 else 911 tp = NULL; 912 913 if (tp == NULL) { 914 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 915 916 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 917 #ifdef INET 918 if (inp) 919 inpcb_destroy(inp); 920 #endif 921 #ifdef INET6 922 if (in6p) 923 in6_pcbdetach(in6p); 924 #endif 925 so->so_state |= nofd; 926 return (ENOBUFS); 927 } 928 tp->t_state = TCPS_CLOSED; 929 return (0); 930 } 931 932 /* 933 * Initiate (or continue) disconnect. 934 * If embryonic state, just send reset (once). 935 * If in ``let data drain'' option and linger null, just drop. 936 * Otherwise (hard), mark socket disconnecting and drop 937 * current input data; switch states based on user close, and 938 * send segment to peer (with FIN). 939 */ 940 struct tcpcb * 941 tcp_disconnect(struct tcpcb *tp) 942 { 943 struct socket *so; 944 945 if (tp->t_inpcb) 946 so = inpcb_get_socket(tp->t_inpcb); 947 #ifdef INET6 948 else if (tp->t_in6pcb) 949 so = tp->t_in6pcb->in6p_socket; 950 #endif 951 else 952 so = NULL; 953 954 KASSERT(so != NULL); 955 956 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 957 tp = tcp_close(tp); 958 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 959 tp = tcp_drop(tp, 0); 960 else { 961 soisdisconnecting(so); 962 sbflush(&so->so_rcv); 963 tp = tcp_usrclosed(tp); 964 if (tp) 965 (void) tcp_output(tp); 966 } 967 return (tp); 968 } 969 970 /* 971 * User issued close, and wish to trail through shutdown states: 972 * if never received SYN, just forget it. If got a SYN from peer, 973 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 974 * If already got a FIN from peer, then almost done; go to LAST_ACK 975 * state. In all other cases, have already sent FIN to peer (e.g. 976 * after PRU_SHUTDOWN), and just have to play tedious game waiting 977 * for peer to send FIN or not respond to keep-alives, etc. 978 * We can let the user exit from the close as soon as the FIN is acked. 979 */ 980 struct tcpcb * 981 tcp_usrclosed(struct tcpcb *tp) 982 { 983 984 switch (tp->t_state) { 985 986 case TCPS_CLOSED: 987 case TCPS_LISTEN: 988 case TCPS_SYN_SENT: 989 tp->t_state = TCPS_CLOSED; 990 tp = tcp_close(tp); 991 break; 992 993 case TCPS_SYN_RECEIVED: 994 case TCPS_ESTABLISHED: 995 tp->t_state = TCPS_FIN_WAIT_1; 996 break; 997 998 case TCPS_CLOSE_WAIT: 999 tp->t_state = TCPS_LAST_ACK; 1000 break; 1001 } 1002 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1003 struct socket *so; 1004 if (tp->t_inpcb) 1005 so = inpcb_get_socket(tp->t_inpcb); 1006 #ifdef INET6 1007 else if (tp->t_in6pcb) 1008 so = tp->t_in6pcb->in6p_socket; 1009 #endif 1010 else 1011 so = NULL; 1012 if (so) 1013 soisdisconnected(so); 1014 /* 1015 * If we are in FIN_WAIT_2, we arrived here because the 1016 * application did a shutdown of the send side. Like the 1017 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1018 * a full close, we start a timer to make sure sockets are 1019 * not left in FIN_WAIT_2 forever. 1020 */ 1021 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1022 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1023 else if (tp->t_state == TCPS_TIME_WAIT 1024 && ((tp->t_inpcb 1025 && (tcp4_vtw_enable & 1) 1026 && vtw_add(AF_INET, tp)) 1027 || 1028 (tp->t_in6pcb 1029 && (tcp6_vtw_enable & 1) 1030 && vtw_add(AF_INET6, tp)))) { 1031 tp = 0; 1032 } 1033 } 1034 return (tp); 1035 } 1036 1037 /* 1038 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1039 * than 32. 1040 */ 1041 static int 1042 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1043 { 1044 int error, mssdflt; 1045 struct sysctlnode node; 1046 1047 mssdflt = tcp_mssdflt; 1048 node = *rnode; 1049 node.sysctl_data = &mssdflt; 1050 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1051 if (error || newp == NULL) 1052 return (error); 1053 1054 if (mssdflt < 32) 1055 return (EINVAL); 1056 tcp_mssdflt = mssdflt; 1057 1058 return (0); 1059 } 1060 1061 /* 1062 * sysctl helper routine for setting port related values under 1063 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1064 * additional checks for each type. this code has placed in 1065 * tcp_input.c since INET and INET6 both use the same tcp code. 1066 * 1067 * this helper is not static so that both inet and inet6 can use it. 1068 */ 1069 int 1070 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1071 { 1072 int error, tmp; 1073 int apmin, apmax; 1074 #ifndef IPNOPRIVPORTS 1075 int lpmin, lpmax; 1076 #endif /* IPNOPRIVPORTS */ 1077 struct sysctlnode node; 1078 1079 if (namelen != 0) 1080 return (EINVAL); 1081 1082 switch (name[-3]) { 1083 #ifdef INET 1084 case PF_INET: 1085 apmin = anonportmin; 1086 apmax = anonportmax; 1087 #ifndef IPNOPRIVPORTS 1088 lpmin = lowportmin; 1089 lpmax = lowportmax; 1090 #endif /* IPNOPRIVPORTS */ 1091 break; 1092 #endif /* INET */ 1093 #ifdef INET6 1094 case PF_INET6: 1095 apmin = ip6_anonportmin; 1096 apmax = ip6_anonportmax; 1097 #ifndef IPNOPRIVPORTS 1098 lpmin = ip6_lowportmin; 1099 lpmax = ip6_lowportmax; 1100 #endif /* IPNOPRIVPORTS */ 1101 break; 1102 #endif /* INET6 */ 1103 default: 1104 return (EINVAL); 1105 } 1106 1107 /* 1108 * insert temporary copy into node, perform lookup on 1109 * temporary, then restore pointer 1110 */ 1111 node = *rnode; 1112 tmp = *(int*)rnode->sysctl_data; 1113 node.sysctl_data = &tmp; 1114 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1115 if (error || newp == NULL) 1116 return (error); 1117 1118 /* 1119 * simple port range check 1120 */ 1121 if (tmp < 0 || tmp > 65535) 1122 return (EINVAL); 1123 1124 /* 1125 * per-node range checks 1126 */ 1127 switch (rnode->sysctl_num) { 1128 case IPCTL_ANONPORTMIN: 1129 case IPV6CTL_ANONPORTMIN: 1130 if (tmp >= apmax) 1131 return (EINVAL); 1132 #ifndef IPNOPRIVPORTS 1133 if (tmp < IPPORT_RESERVED) 1134 return (EINVAL); 1135 #endif /* IPNOPRIVPORTS */ 1136 break; 1137 1138 case IPCTL_ANONPORTMAX: 1139 case IPV6CTL_ANONPORTMAX: 1140 if (apmin >= tmp) 1141 return (EINVAL); 1142 #ifndef IPNOPRIVPORTS 1143 if (tmp < IPPORT_RESERVED) 1144 return (EINVAL); 1145 #endif /* IPNOPRIVPORTS */ 1146 break; 1147 1148 #ifndef IPNOPRIVPORTS 1149 case IPCTL_LOWPORTMIN: 1150 case IPV6CTL_LOWPORTMIN: 1151 if (tmp >= lpmax || 1152 tmp > IPPORT_RESERVEDMAX || 1153 tmp < IPPORT_RESERVEDMIN) 1154 return (EINVAL); 1155 break; 1156 1157 case IPCTL_LOWPORTMAX: 1158 case IPV6CTL_LOWPORTMAX: 1159 if (lpmin >= tmp || 1160 tmp > IPPORT_RESERVEDMAX || 1161 tmp < IPPORT_RESERVEDMIN) 1162 return (EINVAL); 1163 break; 1164 #endif /* IPNOPRIVPORTS */ 1165 1166 default: 1167 return (EINVAL); 1168 } 1169 1170 *(int*)rnode->sysctl_data = tmp; 1171 1172 return (0); 1173 } 1174 1175 static inline int 1176 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1177 { 1178 size_t sz; 1179 int error; 1180 uid_t uid; 1181 1182 uid = kauth_cred_geteuid(sockp->so_cred); 1183 if (oldp) { 1184 sz = MIN(sizeof(uid), *oldlenp); 1185 error = copyout(&uid, oldp, sz); 1186 if (error) 1187 return error; 1188 } 1189 *oldlenp = sizeof(uid); 1190 return 0; 1191 } 1192 1193 static inline int 1194 inet4_ident_core(struct in_addr raddr, u_int rport, 1195 struct in_addr laddr, u_int lport, 1196 void *oldp, size_t *oldlenp, 1197 struct lwp *l, int dodrop) 1198 { 1199 inpcb_t *inp; 1200 struct socket *so; 1201 1202 inp = inpcb_lookup_connect(tcbtable, raddr, rport, laddr, lport, 0); 1203 if (inp == NULL || (so = inpcb_get_socket(inp)) == NULL) 1204 return ESRCH; 1205 1206 if (dodrop) { 1207 struct tcpcb *tp; 1208 int error; 1209 1210 if ((tp = intotcpcb(inp)) == NULL || 1211 (so->so_options & SO_ACCEPTCONN) != 0) 1212 return ESRCH; 1213 1214 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1215 KAUTH_REQ_NETWORK_SOCKET_DROP, so, tp, NULL); 1216 if (error) { 1217 return error; 1218 } 1219 1220 (void)tcp_drop(tp, ECONNABORTED); 1221 return 0; 1222 } 1223 1224 return copyout_uid(so, oldp, oldlenp); 1225 } 1226 1227 #ifdef INET6 1228 static inline int 1229 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1230 struct in6_addr *laddr, u_int lport, 1231 void *oldp, size_t *oldlenp, 1232 struct lwp *l, int dodrop) 1233 { 1234 struct in6pcb *in6p; 1235 struct socket *sockp; 1236 1237 in6p = in6_pcblookup_connect(tcbtable, raddr, rport, laddr, lport, 0, 0); 1238 1239 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1240 return ESRCH; 1241 1242 if (dodrop) { 1243 struct tcpcb *tp; 1244 int error; 1245 1246 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1247 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1248 return ESRCH; 1249 1250 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1251 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1252 if (error) 1253 return (error); 1254 1255 (void)tcp_drop(tp, ECONNABORTED); 1256 return 0; 1257 } 1258 else 1259 return copyout_uid(sockp, oldp, oldlenp); 1260 } 1261 #endif 1262 1263 /* 1264 * sysctl helper routine for the net.inet.tcp.drop and 1265 * net.inet6.tcp6.drop nodes. 1266 */ 1267 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1268 1269 /* 1270 * sysctl helper routine for the net.inet.tcp.ident and 1271 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1272 * old way of looking up the ident information for ipv4 which involves 1273 * stuffing the port/addr pairs into the mib lookup. 1274 */ 1275 static int 1276 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1277 { 1278 #ifdef INET 1279 struct sockaddr_in *si4[2]; 1280 #endif /* INET */ 1281 #ifdef INET6 1282 struct sockaddr_in6 *si6[2]; 1283 #endif /* INET6 */ 1284 struct sockaddr_storage sa[2]; 1285 int error, pf, dodrop; 1286 1287 dodrop = name[-1] == TCPCTL_DROP; 1288 if (dodrop) { 1289 if (oldp != NULL || *oldlenp != 0) 1290 return EINVAL; 1291 if (newp == NULL) 1292 return EPERM; 1293 if (newlen < sizeof(sa)) 1294 return ENOMEM; 1295 } 1296 if (namelen != 4 && namelen != 0) 1297 return EINVAL; 1298 if (name[-2] != IPPROTO_TCP) 1299 return EINVAL; 1300 pf = name[-3]; 1301 1302 /* old style lookup, ipv4 only */ 1303 if (namelen == 4) { 1304 #ifdef INET 1305 struct in_addr laddr, raddr; 1306 u_int lport, rport; 1307 1308 if (pf != PF_INET) 1309 return EPROTONOSUPPORT; 1310 raddr.s_addr = (uint32_t)name[0]; 1311 rport = (u_int)name[1]; 1312 laddr.s_addr = (uint32_t)name[2]; 1313 lport = (u_int)name[3]; 1314 1315 mutex_enter(softnet_lock); 1316 error = inet4_ident_core(raddr, rport, laddr, lport, 1317 oldp, oldlenp, l, dodrop); 1318 mutex_exit(softnet_lock); 1319 return error; 1320 #else /* INET */ 1321 return EINVAL; 1322 #endif /* INET */ 1323 } 1324 1325 if (newp == NULL || newlen != sizeof(sa)) 1326 return EINVAL; 1327 error = copyin(newp, &sa, newlen); 1328 if (error) 1329 return error; 1330 1331 /* 1332 * requested families must match 1333 */ 1334 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1335 return EINVAL; 1336 1337 switch (pf) { 1338 #ifdef INET6 1339 case PF_INET6: 1340 si6[0] = (struct sockaddr_in6*)&sa[0]; 1341 si6[1] = (struct sockaddr_in6*)&sa[1]; 1342 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1343 si6[1]->sin6_len != sizeof(*si6[1])) 1344 return EINVAL; 1345 1346 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1347 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1348 error = sa6_embedscope(si6[0], ip6_use_defzone); 1349 if (error) 1350 return error; 1351 error = sa6_embedscope(si6[1], ip6_use_defzone); 1352 if (error) 1353 return error; 1354 1355 mutex_enter(softnet_lock); 1356 error = inet6_ident_core(&si6[0]->sin6_addr, 1357 si6[0]->sin6_port, &si6[1]->sin6_addr, 1358 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1359 mutex_exit(softnet_lock); 1360 return error; 1361 } 1362 1363 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1364 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1365 return EINVAL; 1366 1367 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1368 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1369 /*FALLTHROUGH*/ 1370 #endif /* INET6 */ 1371 #ifdef INET 1372 case PF_INET: 1373 si4[0] = (struct sockaddr_in*)&sa[0]; 1374 si4[1] = (struct sockaddr_in*)&sa[1]; 1375 if (si4[0]->sin_len != sizeof(*si4[0]) || 1376 si4[0]->sin_len != sizeof(*si4[1])) 1377 return EINVAL; 1378 1379 mutex_enter(softnet_lock); 1380 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1381 si4[1]->sin_addr, si4[1]->sin_port, 1382 oldp, oldlenp, l, dodrop); 1383 mutex_exit(softnet_lock); 1384 return error; 1385 #endif /* INET */ 1386 default: 1387 return EPROTONOSUPPORT; 1388 } 1389 } 1390 1391 static int 1392 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1393 { 1394 struct sysctlnode node; 1395 int error; 1396 char newname[TCPCC_MAXLEN]; 1397 1398 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1399 1400 node = *rnode; 1401 node.sysctl_data = newname; 1402 node.sysctl_size = sizeof(newname); 1403 1404 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1405 1406 if (error || 1407 newp == NULL || 1408 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1409 return error; 1410 1411 mutex_enter(softnet_lock); 1412 error = tcp_congctl_select(NULL, newname); 1413 mutex_exit(softnet_lock); 1414 1415 return error; 1416 } 1417 1418 static int 1419 sysctl_tcp_init_win(SYSCTLFN_ARGS) 1420 { 1421 int error; 1422 u_int iw; 1423 struct sysctlnode node; 1424 1425 iw = *(u_int *)rnode->sysctl_data; 1426 node = *rnode; 1427 node.sysctl_data = &iw; 1428 node.sysctl_size = sizeof(iw); 1429 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1430 if (error || newp == NULL) 1431 return error; 1432 1433 if (iw >= __arraycount(tcp_init_win_max)) 1434 return EINVAL; 1435 *(u_int *)rnode->sysctl_data = iw; 1436 return 0; 1437 } 1438 1439 static int 1440 sysctl_tcp_keep(SYSCTLFN_ARGS) 1441 { 1442 int error; 1443 u_int tmp; 1444 struct sysctlnode node; 1445 1446 node = *rnode; 1447 tmp = *(u_int *)rnode->sysctl_data; 1448 node.sysctl_data = &tmp; 1449 1450 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1451 if (error || newp == NULL) 1452 return error; 1453 1454 mutex_enter(softnet_lock); 1455 1456 *(u_int *)rnode->sysctl_data = tmp; 1457 tcp_tcpcb_template(); /* update the template */ 1458 1459 mutex_exit(softnet_lock); 1460 return 0; 1461 } 1462 1463 static int 1464 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 1465 { 1466 1467 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 1468 } 1469 1470 /* 1471 * this (second stage) setup routine is a replacement for tcp_sysctl() 1472 * (which is currently used for ipv4 and ipv6) 1473 */ 1474 static void 1475 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1476 const char *tcpname) 1477 { 1478 const struct sysctlnode *sack_node; 1479 const struct sysctlnode *abc_node; 1480 const struct sysctlnode *ecn_node; 1481 const struct sysctlnode *congctl_node; 1482 const struct sysctlnode *mslt_node; 1483 const struct sysctlnode *vtw_node; 1484 #ifdef TCP_DEBUG 1485 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1486 extern int tcp_debx; 1487 #endif 1488 1489 sysctl_createv(clog, 0, NULL, NULL, 1490 CTLFLAG_PERMANENT, 1491 CTLTYPE_NODE, "net", NULL, 1492 NULL, 0, NULL, 0, 1493 CTL_NET, CTL_EOL); 1494 sysctl_createv(clog, 0, NULL, NULL, 1495 CTLFLAG_PERMANENT, 1496 CTLTYPE_NODE, pfname, NULL, 1497 NULL, 0, NULL, 0, 1498 CTL_NET, pf, CTL_EOL); 1499 sysctl_createv(clog, 0, NULL, NULL, 1500 CTLFLAG_PERMANENT, 1501 CTLTYPE_NODE, tcpname, 1502 SYSCTL_DESCR("TCP related settings"), 1503 NULL, 0, NULL, 0, 1504 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1505 1506 sysctl_createv(clog, 0, NULL, NULL, 1507 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1508 CTLTYPE_INT, "rfc1323", 1509 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1510 NULL, 0, &tcp_do_rfc1323, 0, 1511 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1512 sysctl_createv(clog, 0, NULL, NULL, 1513 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1514 CTLTYPE_INT, "sendspace", 1515 SYSCTL_DESCR("Default TCP send buffer size"), 1516 NULL, 0, &tcp_sendspace, 0, 1517 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1518 sysctl_createv(clog, 0, NULL, NULL, 1519 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1520 CTLTYPE_INT, "recvspace", 1521 SYSCTL_DESCR("Default TCP receive buffer size"), 1522 NULL, 0, &tcp_recvspace, 0, 1523 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1524 sysctl_createv(clog, 0, NULL, NULL, 1525 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1526 CTLTYPE_INT, "mssdflt", 1527 SYSCTL_DESCR("Default maximum segment size"), 1528 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1529 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1530 sysctl_createv(clog, 0, NULL, NULL, 1531 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1532 CTLTYPE_INT, "minmss", 1533 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 1534 NULL, 0, &tcp_minmss, 0, 1535 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1536 sysctl_createv(clog, 0, NULL, NULL, 1537 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1538 CTLTYPE_INT, "msl", 1539 SYSCTL_DESCR("Maximum Segment Life"), 1540 NULL, 0, &tcp_msl, 0, 1541 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 1542 sysctl_createv(clog, 0, NULL, NULL, 1543 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1544 CTLTYPE_INT, "syn_cache_limit", 1545 SYSCTL_DESCR("Maximum number of entries in the TCP " 1546 "compressed state engine"), 1547 NULL, 0, &tcp_syn_cache_limit, 0, 1548 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1549 CTL_EOL); 1550 sysctl_createv(clog, 0, NULL, NULL, 1551 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1552 CTLTYPE_INT, "syn_bucket_limit", 1553 SYSCTL_DESCR("Maximum number of entries per hash " 1554 "bucket in the TCP compressed state " 1555 "engine"), 1556 NULL, 0, &tcp_syn_bucket_limit, 0, 1557 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1558 CTL_EOL); 1559 #if 0 /* obsoleted */ 1560 sysctl_createv(clog, 0, NULL, NULL, 1561 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1562 CTLTYPE_INT, "syn_cache_interval", 1563 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1564 NULL, 0, &tcp_syn_cache_interval, 0, 1565 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1566 CTL_EOL); 1567 #endif 1568 sysctl_createv(clog, 0, NULL, NULL, 1569 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1570 CTLTYPE_INT, "init_win", 1571 SYSCTL_DESCR("Initial TCP congestion window"), 1572 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 1573 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1574 sysctl_createv(clog, 0, NULL, NULL, 1575 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1576 CTLTYPE_INT, "mss_ifmtu", 1577 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1578 NULL, 0, &tcp_mss_ifmtu, 0, 1579 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1580 sysctl_createv(clog, 0, NULL, &sack_node, 1581 CTLFLAG_PERMANENT, 1582 CTLTYPE_NODE, "sack", 1583 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1584 NULL, 0, NULL, 0, 1585 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1586 1587 /* Congctl subtree */ 1588 sysctl_createv(clog, 0, NULL, &congctl_node, 1589 CTLFLAG_PERMANENT, 1590 CTLTYPE_NODE, "congctl", 1591 SYSCTL_DESCR("TCP Congestion Control"), 1592 NULL, 0, NULL, 0, 1593 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1594 sysctl_createv(clog, 0, &congctl_node, NULL, 1595 CTLFLAG_PERMANENT, 1596 CTLTYPE_STRING, "available", 1597 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1598 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1599 sysctl_createv(clog, 0, &congctl_node, NULL, 1600 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1601 CTLTYPE_STRING, "selected", 1602 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1603 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1604 CTL_CREATE, CTL_EOL); 1605 1606 sysctl_createv(clog, 0, NULL, NULL, 1607 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1608 CTLTYPE_INT, "win_scale", 1609 SYSCTL_DESCR("Use RFC1323 window scale options"), 1610 NULL, 0, &tcp_do_win_scale, 0, 1611 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1612 sysctl_createv(clog, 0, NULL, NULL, 1613 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1614 CTLTYPE_INT, "timestamps", 1615 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1616 NULL, 0, &tcp_do_timestamps, 0, 1617 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1618 sysctl_createv(clog, 0, NULL, NULL, 1619 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1620 CTLTYPE_INT, "compat_42", 1621 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1622 NULL, 0, &tcp_compat_42, 0, 1623 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1624 sysctl_createv(clog, 0, NULL, NULL, 1625 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1626 CTLTYPE_INT, "cwm", 1627 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1628 "Monitoring"), 1629 NULL, 0, &tcp_cwm, 0, 1630 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1631 sysctl_createv(clog, 0, NULL, NULL, 1632 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1633 CTLTYPE_INT, "cwm_burstsize", 1634 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1635 "burst count in packets"), 1636 NULL, 0, &tcp_cwm_burstsize, 0, 1637 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1638 CTL_EOL); 1639 sysctl_createv(clog, 0, NULL, NULL, 1640 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1641 CTLTYPE_INT, "ack_on_push", 1642 SYSCTL_DESCR("Immediately return ACK when PSH is " 1643 "received"), 1644 NULL, 0, &tcp_ack_on_push, 0, 1645 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1646 sysctl_createv(clog, 0, NULL, NULL, 1647 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1648 CTLTYPE_INT, "keepidle", 1649 SYSCTL_DESCR("Allowed connection idle ticks before a " 1650 "keepalive probe is sent"), 1651 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1652 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1653 sysctl_createv(clog, 0, NULL, NULL, 1654 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1655 CTLTYPE_INT, "keepintvl", 1656 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1657 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1658 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1659 sysctl_createv(clog, 0, NULL, NULL, 1660 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1661 CTLTYPE_INT, "keepcnt", 1662 SYSCTL_DESCR("Number of keepalive probes to send"), 1663 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1664 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1665 sysctl_createv(clog, 0, NULL, NULL, 1666 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1667 CTLTYPE_INT, "slowhz", 1668 SYSCTL_DESCR("Keepalive ticks per second"), 1669 NULL, PR_SLOWHZ, NULL, 0, 1670 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1671 sysctl_createv(clog, 0, NULL, NULL, 1672 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1673 CTLTYPE_INT, "log_refused", 1674 SYSCTL_DESCR("Log refused TCP connections"), 1675 NULL, 0, &tcp_log_refused, 0, 1676 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1677 #if 0 /* obsoleted */ 1678 sysctl_createv(clog, 0, NULL, NULL, 1679 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1680 CTLTYPE_INT, "rstratelimit", NULL, 1681 NULL, 0, &tcp_rst_ratelim, 0, 1682 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1683 #endif 1684 sysctl_createv(clog, 0, NULL, NULL, 1685 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1686 CTLTYPE_INT, "rstppslimit", 1687 SYSCTL_DESCR("Maximum number of RST packets to send " 1688 "per second"), 1689 NULL, 0, &tcp_rst_ppslim, 0, 1690 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1691 sysctl_createv(clog, 0, NULL, NULL, 1692 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1693 CTLTYPE_INT, "delack_ticks", 1694 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1695 NULL, 0, &tcp_delack_ticks, 0, 1696 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1697 sysctl_createv(clog, 0, NULL, NULL, 1698 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1699 CTLTYPE_INT, "init_win_local", 1700 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1701 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 1702 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1703 CTL_EOL); 1704 sysctl_createv(clog, 0, NULL, NULL, 1705 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1706 CTLTYPE_STRUCT, "ident", 1707 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1708 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1709 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1710 sysctl_createv(clog, 0, NULL, NULL, 1711 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1712 CTLTYPE_INT, "do_loopback_cksum", 1713 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1714 NULL, 0, &tcp_do_loopback_cksum, 0, 1715 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1716 CTL_EOL); 1717 sysctl_createv(clog, 0, NULL, NULL, 1718 CTLFLAG_PERMANENT, 1719 CTLTYPE_STRUCT, "pcblist", 1720 SYSCTL_DESCR("TCP protocol control block list"), 1721 sysctl_inpcblist, 0, tcbtable, 0, 1722 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1723 CTL_EOL); 1724 sysctl_createv(clog, 0, NULL, NULL, 1725 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1726 CTLTYPE_INT, "keepinit", 1727 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1728 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 1729 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1730 1731 /* TCP socket buffers auto-sizing nodes */ 1732 sysctl_createv(clog, 0, NULL, NULL, 1733 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1734 CTLTYPE_INT, "recvbuf_auto", 1735 SYSCTL_DESCR("Enable automatic receive " 1736 "buffer sizing (experimental)"), 1737 NULL, 0, &tcp_do_autorcvbuf, 0, 1738 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1739 sysctl_createv(clog, 0, NULL, NULL, 1740 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1741 CTLTYPE_INT, "recvbuf_inc", 1742 SYSCTL_DESCR("Incrementor step size of " 1743 "automatic receive buffer"), 1744 NULL, 0, &tcp_autorcvbuf_inc, 0, 1745 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1746 sysctl_createv(clog, 0, NULL, NULL, 1747 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1748 CTLTYPE_INT, "recvbuf_max", 1749 SYSCTL_DESCR("Max size of automatic receive buffer"), 1750 NULL, 0, &tcp_autorcvbuf_max, 0, 1751 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1752 1753 sysctl_createv(clog, 0, NULL, NULL, 1754 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1755 CTLTYPE_INT, "sendbuf_auto", 1756 SYSCTL_DESCR("Enable automatic send " 1757 "buffer sizing (experimental)"), 1758 NULL, 0, &tcp_do_autosndbuf, 0, 1759 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1760 sysctl_createv(clog, 0, NULL, NULL, 1761 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1762 CTLTYPE_INT, "sendbuf_inc", 1763 SYSCTL_DESCR("Incrementor step size of " 1764 "automatic send buffer"), 1765 NULL, 0, &tcp_autosndbuf_inc, 0, 1766 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1767 sysctl_createv(clog, 0, NULL, NULL, 1768 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1769 CTLTYPE_INT, "sendbuf_max", 1770 SYSCTL_DESCR("Max size of automatic send buffer"), 1771 NULL, 0, &tcp_autosndbuf_max, 0, 1772 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1773 1774 /* ECN subtree */ 1775 sysctl_createv(clog, 0, NULL, &ecn_node, 1776 CTLFLAG_PERMANENT, 1777 CTLTYPE_NODE, "ecn", 1778 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 1779 NULL, 0, NULL, 0, 1780 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1781 sysctl_createv(clog, 0, &ecn_node, NULL, 1782 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1783 CTLTYPE_INT, "enable", 1784 SYSCTL_DESCR("Enable TCP Explicit Congestion " 1785 "Notification"), 1786 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 1787 sysctl_createv(clog, 0, &ecn_node, NULL, 1788 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1789 CTLTYPE_INT, "maxretries", 1790 SYSCTL_DESCR("Number of times to retry ECN setup " 1791 "before disabling ECN on the connection"), 1792 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 1793 1794 /* SACK gets it's own little subtree. */ 1795 sysctl_createv(clog, 0, NULL, &sack_node, 1796 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1797 CTLTYPE_INT, "enable", 1798 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 1799 NULL, 0, &tcp_do_sack, 0, 1800 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1801 sysctl_createv(clog, 0, NULL, &sack_node, 1802 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1803 CTLTYPE_INT, "maxholes", 1804 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 1805 NULL, 0, &tcp_sack_tp_maxholes, 0, 1806 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1807 sysctl_createv(clog, 0, NULL, &sack_node, 1808 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1809 CTLTYPE_INT, "globalmaxholes", 1810 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 1811 NULL, 0, &tcp_sack_globalmaxholes, 0, 1812 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1813 sysctl_createv(clog, 0, NULL, &sack_node, 1814 CTLFLAG_PERMANENT, 1815 CTLTYPE_INT, "globalholes", 1816 SYSCTL_DESCR("Global number of TCP SACK holes"), 1817 NULL, 0, &tcp_sack_globalholes, 0, 1818 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1819 1820 sysctl_createv(clog, 0, NULL, NULL, 1821 CTLFLAG_PERMANENT, 1822 CTLTYPE_STRUCT, "stats", 1823 SYSCTL_DESCR("TCP statistics"), 1824 sysctl_net_inet_tcp_stats, 0, NULL, 0, 1825 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1826 CTL_EOL); 1827 sysctl_createv(clog, 0, NULL, NULL, 1828 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1829 CTLTYPE_INT, "local_by_rtt", 1830 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 1831 "are local"), 1832 NULL, 0, &tcp_rttlocal, 0, 1833 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1834 #ifdef TCP_DEBUG 1835 sysctl_createv(clog, 0, NULL, NULL, 1836 CTLFLAG_PERMANENT, 1837 CTLTYPE_STRUCT, "debug", 1838 SYSCTL_DESCR("TCP sockets debug information"), 1839 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1840 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1841 CTL_EOL); 1842 sysctl_createv(clog, 0, NULL, NULL, 1843 CTLFLAG_PERMANENT, 1844 CTLTYPE_INT, "debx", 1845 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1846 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1847 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1848 CTL_EOL); 1849 #endif 1850 sysctl_createv(clog, 0, NULL, NULL, 1851 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1852 CTLTYPE_STRUCT, "drop", 1853 SYSCTL_DESCR("TCP drop connection"), 1854 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1855 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1856 sysctl_createv(clog, 0, NULL, NULL, 1857 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1858 CTLTYPE_INT, "iss_hash", 1859 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1860 "hash computation"), 1861 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1862 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1863 CTL_EOL); 1864 1865 /* ABC subtree */ 1866 1867 sysctl_createv(clog, 0, NULL, &abc_node, 1868 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 1869 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 1870 NULL, 0, NULL, 0, 1871 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1872 sysctl_createv(clog, 0, &abc_node, NULL, 1873 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1874 CTLTYPE_INT, "enable", 1875 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 1876 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 1877 sysctl_createv(clog, 0, &abc_node, NULL, 1878 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1879 CTLTYPE_INT, "aggressive", 1880 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 1881 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 1882 1883 /* MSL tuning subtree */ 1884 1885 sysctl_createv(clog, 0, NULL, &mslt_node, 1886 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 1887 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 1888 NULL, 0, NULL, 0, 1889 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1890 sysctl_createv(clog, 0, &mslt_node, NULL, 1891 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1892 CTLTYPE_INT, "enable", 1893 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 1894 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 1895 sysctl_createv(clog, 0, &mslt_node, NULL, 1896 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1897 CTLTYPE_INT, "loopback", 1898 SYSCTL_DESCR("MSL value to use for loopback connections"), 1899 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 1900 sysctl_createv(clog, 0, &mslt_node, NULL, 1901 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1902 CTLTYPE_INT, "local", 1903 SYSCTL_DESCR("MSL value to use for local connections"), 1904 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 1905 sysctl_createv(clog, 0, &mslt_node, NULL, 1906 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1907 CTLTYPE_INT, "remote", 1908 SYSCTL_DESCR("MSL value to use for remote connections"), 1909 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 1910 sysctl_createv(clog, 0, &mslt_node, NULL, 1911 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1912 CTLTYPE_INT, "remote_threshold", 1913 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 1914 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 1915 1916 /* vestigial TIME_WAIT tuning subtree */ 1917 1918 sysctl_createv(clog, 0, NULL, &vtw_node, 1919 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 1920 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 1921 NULL, 0, NULL, 0, 1922 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1923 sysctl_createv(clog, 0, &vtw_node, NULL, 1924 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1925 CTLTYPE_INT, "enable", 1926 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 1927 sysctl_tcp_vtw_enable, 0, 1928 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 1929 0, CTL_CREATE, CTL_EOL); 1930 sysctl_createv(clog, 0, &vtw_node, NULL, 1931 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 1932 CTLTYPE_INT, "entries", 1933 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 1934 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 1935 } 1936 1937 void 1938 tcp_usrreq_init(void) 1939 { 1940 1941 #ifdef INET 1942 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 1943 #endif 1944 #ifdef INET6 1945 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 1946 #endif 1947 } 1948
Indexes created Wed Sep 24 05:09:52 GMT 2025