tcp_usrreq.c revision 1.166.4.5
1 /* $NetBSD: tcp_usrreq.c,v 1.166.4.5 2014/05/18 17:46:13 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 1.166.4.5 2014/05/18 17:46:13 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 static 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 KASSERT(req != PRU_ATTACH); 182 KASSERT(req != PRU_DETACH); 183 184 family = so->so_proto->pr_domain->dom_family; 185 186 if (req == PRU_CONTROL) { 187 switch (family) { 188 #ifdef INET 189 case PF_INET: 190 return (in_control(so, (long)m, (void *)nam, 191 (struct ifnet *)control, l)); 192 #endif 193 #ifdef INET6 194 case PF_INET6: 195 return (in6_control(so, (long)m, (void *)nam, 196 (struct ifnet *)control, l)); 197 #endif 198 default: 199 return EAFNOSUPPORT; 200 } 201 } 202 203 s = splsoftnet(); 204 205 if (req == PRU_PURGEIF) { 206 mutex_enter(softnet_lock); 207 switch (family) { 208 #ifdef INET 209 case PF_INET: 210 inpcb_purgeif0(tcbtable, (struct ifnet *)control); 211 in_purgeif((struct ifnet *)control); 212 inpcb_purgeif(tcbtable, (struct ifnet *)control); 213 break; 214 #endif 215 #ifdef INET6 216 case PF_INET6: 217 in6_pcbpurgeif0(tcbtable, (struct ifnet *)control); 218 in6_purgeif((struct ifnet *)control); 219 in6_pcbpurgeif(tcbtable, (struct ifnet *)control); 220 break; 221 #endif 222 default: 223 mutex_exit(softnet_lock); 224 splx(s); 225 return (EAFNOSUPPORT); 226 } 227 mutex_exit(softnet_lock); 228 splx(s); 229 return (0); 230 } 231 232 KASSERT(solocked(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 INET6 255 KASSERT((inp != NULL) ^ (in6p != NULL)); 256 #endif 257 KASSERT(!control || (req == PRU_SEND || req == PRU_SENDOOB)); 258 259 /* 260 * When a TCP is attached to a socket, then there will be 261 * a (struct inpcb) pointed at by the socket, and this 262 * structure will point at a subsidary (struct tcpcb). 263 */ 264 if ((inp == NULL 265 #ifdef INET6 266 && in6p == NULL 267 #endif 268 ) && req != PRU_SENSE) 269 { 270 error = EINVAL; 271 goto release; 272 } 273 #ifdef INET 274 if (inp) { 275 tp = intotcpcb(inp); 276 /* WHAT IF TP IS 0? */ 277 #ifdef KPROF 278 tcp_acounts[tp->t_state][req]++; 279 #endif 280 #ifdef TCP_DEBUG 281 ostate = tp->t_state; 282 #endif 283 } 284 #endif 285 #ifdef INET6 286 if (in6p) { 287 tp = in6totcpcb(in6p); 288 /* WHAT IF TP IS 0? */ 289 #ifdef KPROF 290 tcp_acounts[tp->t_state][req]++; 291 #endif 292 #ifdef TCP_DEBUG 293 ostate = tp->t_state; 294 #endif 295 } 296 #endif 297 298 switch (req) { 299 /* 300 * Give the socket an address. 301 */ 302 case PRU_BIND: 303 switch (family) { 304 #ifdef INET 305 case PF_INET: 306 error = inpcb_bind(inp, nam, l); 307 break; 308 #endif 309 #ifdef INET6 310 case PF_INET6: 311 error = in6_pcbbind(in6p, nam, l); 312 if (!error) { 313 /* mapped addr case */ 314 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 315 tp->t_family = AF_INET; 316 else 317 tp->t_family = AF_INET6; 318 } 319 break; 320 #endif 321 } 322 break; 323 324 /* 325 * Prepare to accept connections. 326 */ 327 case PRU_LISTEN: 328 #ifdef INET 329 if (inp) { 330 in_port_t lport; 331 332 inpcb_get_ports(inp, &lport, NULL); 333 if (lport == 0) { 334 error = inpcb_bind(inp, NULL, l); 335 if (error) 336 break; 337 } 338 } 339 #endif 340 #ifdef INET6 341 if (in6p && in6p->in6p_lport == 0) { 342 error = in6_pcbbind(in6p, NULL, l); 343 if (error) 344 break; 345 } 346 #endif 347 tp->t_state = TCPS_LISTEN; 348 break; 349 350 /* 351 * Initiate connection to peer. 352 * Create a template for use in transmissions on this connection. 353 * Enter SYN_SENT state, and mark socket as connecting. 354 * Start keep-alive timer, and seed output sequence space. 355 * Send initial segment on connection. 356 */ 357 case PRU_CONNECT: 358 #ifdef INET 359 if (inp) { 360 in_port_t lport; 361 362 inpcb_get_ports(inp, &lport, NULL); 363 if (lport == 0) { 364 error = inpcb_bind(inp, NULL, l); 365 if (error) 366 break; 367 } 368 error = inpcb_connect(inp, nam, l); 369 } 370 #endif 371 #ifdef INET6 372 if (in6p) { 373 if (in6p->in6p_lport == 0) { 374 error = in6_pcbbind(in6p, NULL, l); 375 if (error) 376 break; 377 } 378 error = in6_pcbconnect(in6p, nam, l); 379 if (!error) { 380 /* mapped addr case */ 381 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 382 tp->t_family = AF_INET; 383 else 384 tp->t_family = AF_INET6; 385 } 386 } 387 #endif 388 if (error) 389 break; 390 tp->t_template = tcp_template(tp); 391 if (tp->t_template == 0) { 392 #ifdef INET 393 if (inp) 394 inpcb_disconnect(inp); 395 #endif 396 #ifdef INET6 397 if (in6p) 398 in6_pcbdisconnect(in6p); 399 #endif 400 error = ENOBUFS; 401 break; 402 } 403 /* 404 * Compute window scaling to request. 405 * XXX: This should be moved to tcp_output(). 406 */ 407 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 408 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 409 tp->request_r_scale++; 410 soisconnecting(so); 411 TCP_STATINC(TCP_STAT_CONNATTEMPT); 412 tp->t_state = TCPS_SYN_SENT; 413 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 414 tp->iss = tcp_new_iss(tp, 0); 415 tcp_sendseqinit(tp); 416 error = tcp_output(tp); 417 break; 418 419 /* 420 * Create a TCP connection between two sockets. 421 */ 422 case PRU_CONNECT2: 423 error = EOPNOTSUPP; 424 break; 425 426 /* 427 * Initiate disconnect from peer. 428 * If connection never passed embryonic stage, just drop; 429 * else if don't need to let data drain, then can just drop anyways, 430 * else have to begin TCP shutdown process: mark socket disconnecting, 431 * drain unread data, state switch to reflect user close, and 432 * send segment (e.g. FIN) to peer. Socket will be really disconnected 433 * when peer sends FIN and acks ours. 434 * 435 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 436 */ 437 case PRU_DISCONNECT: 438 tp = tcp_disconnect(tp); 439 break; 440 441 /* 442 * Accept a connection. Essentially all the work is 443 * done at higher levels; just return the address 444 * of the peer, storing through addr. 445 */ 446 case PRU_ACCEPT: 447 #ifdef INET 448 if (inp) 449 inpcb_fetch_peeraddr(inp, nam); 450 #endif 451 #ifdef INET6 452 if (in6p) 453 in6_setpeeraddr(in6p, nam); 454 #endif 455 break; 456 457 /* 458 * Mark the connection as being incapable of further output. 459 */ 460 case PRU_SHUTDOWN: 461 socantsendmore(so); 462 tp = tcp_usrclosed(tp); 463 if (tp) 464 error = tcp_output(tp); 465 break; 466 467 /* 468 * After a receive, possibly send window update to peer. 469 */ 470 case PRU_RCVD: 471 /* 472 * soreceive() calls this function when a user receives 473 * ancillary data on a listening socket. We don't call 474 * tcp_output in such a case, since there is no header 475 * template for a listening socket and hence the kernel 476 * will panic. 477 */ 478 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 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 sbappendstream(&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, char *) = 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 sbappendstream(&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 inpcb_fetch_sockaddr(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 inpcb_fetch_peeraddr(inp, nam); 571 #endif 572 #ifdef INET6 573 if (in6p) 574 in6_setpeeraddr(in6p, nam); 575 #endif 576 break; 577 578 default: 579 panic("tcp_usrreq"); 580 } 581 #ifdef TCP_DEBUG 582 if (tp && (so->so_options & SO_DEBUG)) 583 tcp_trace(TA_USER, ostate, tp, NULL, req); 584 #endif 585 586 release: 587 splx(s); 588 return (error); 589 } 590 591 static void 592 change_keepalive(struct socket *so, struct tcpcb *tp) 593 { 594 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 595 TCP_TIMER_DISARM(tp, TCPT_KEEP); 596 TCP_TIMER_DISARM(tp, TCPT_2MSL); 597 598 if (tp->t_state == TCPS_SYN_RECEIVED || 599 tp->t_state == TCPS_SYN_SENT) { 600 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 601 } else if (so->so_options & SO_KEEPALIVE && 602 tp->t_state <= TCPS_CLOSE_WAIT) { 603 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 604 } else { 605 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 606 } 607 608 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 609 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 610 } 611 612 613 int 614 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 615 { 616 int error = 0, s; 617 inpcb_t *inp; 618 #ifdef INET6 619 struct in6pcb *in6p; 620 #endif 621 struct tcpcb *tp; 622 u_int ui; 623 int family; /* family of the socket */ 624 int level, optname, optval; 625 626 level = sopt->sopt_level; 627 optname = sopt->sopt_name; 628 629 family = so->so_proto->pr_domain->dom_family; 630 631 s = splsoftnet(); 632 switch (family) { 633 #ifdef INET 634 case PF_INET: 635 inp = sotoinpcb(so); 636 #ifdef INET6 637 in6p = NULL; 638 #endif 639 break; 640 #endif 641 #ifdef INET6 642 case PF_INET6: 643 inp = NULL; 644 in6p = sotoin6pcb(so); 645 break; 646 #endif 647 default: 648 splx(s); 649 panic("%s: af %d", __func__, family); 650 } 651 #ifndef INET6 652 if (inp == NULL) 653 #else 654 if (inp == NULL && in6p == NULL) 655 #endif 656 { 657 splx(s); 658 return (ECONNRESET); 659 } 660 if (level != IPPROTO_TCP) { 661 switch (family) { 662 #ifdef INET 663 case PF_INET: 664 error = ip_ctloutput(op, so, sopt); 665 break; 666 #endif 667 #ifdef INET6 668 case PF_INET6: 669 error = ip6_ctloutput(op, so, sopt); 670 break; 671 #endif 672 } 673 splx(s); 674 return (error); 675 } 676 if (inp) 677 tp = intotcpcb(inp); 678 #ifdef INET6 679 else if (in6p) 680 tp = in6totcpcb(in6p); 681 #endif 682 else 683 tp = NULL; 684 685 switch (op) { 686 case PRCO_SETOPT: 687 switch (optname) { 688 #ifdef TCP_SIGNATURE 689 case TCP_MD5SIG: 690 error = sockopt_getint(sopt, &optval); 691 if (error) 692 break; 693 if (optval > 0) 694 tp->t_flags |= TF_SIGNATURE; 695 else 696 tp->t_flags &= ~TF_SIGNATURE; 697 break; 698 #endif /* TCP_SIGNATURE */ 699 700 case TCP_NODELAY: 701 error = sockopt_getint(sopt, &optval); 702 if (error) 703 break; 704 if (optval) 705 tp->t_flags |= TF_NODELAY; 706 else 707 tp->t_flags &= ~TF_NODELAY; 708 break; 709 710 case TCP_MAXSEG: 711 error = sockopt_getint(sopt, &optval); 712 if (error) 713 break; 714 if (optval > 0 && optval <= tp->t_peermss) 715 tp->t_peermss = optval; /* limit on send size */ 716 else 717 error = EINVAL; 718 break; 719 #ifdef notyet 720 case TCP_CONGCTL: 721 /* XXX string overflow XXX */ 722 error = tcp_congctl_select(tp, sopt->sopt_data); 723 break; 724 #endif 725 726 case TCP_KEEPIDLE: 727 error = sockopt_get(sopt, &ui, sizeof(ui)); 728 if (error) 729 break; 730 if (ui > 0) { 731 tp->t_keepidle = ui; 732 change_keepalive(so, tp); 733 } else 734 error = EINVAL; 735 break; 736 737 case TCP_KEEPINTVL: 738 error = sockopt_get(sopt, &ui, sizeof(ui)); 739 if (error) 740 break; 741 if (ui > 0) { 742 tp->t_keepintvl = ui; 743 change_keepalive(so, tp); 744 } else 745 error = EINVAL; 746 break; 747 748 case TCP_KEEPCNT: 749 error = sockopt_get(sopt, &ui, sizeof(ui)); 750 if (error) 751 break; 752 if (ui > 0) { 753 tp->t_keepcnt = ui; 754 change_keepalive(so, tp); 755 } else 756 error = EINVAL; 757 break; 758 759 case TCP_KEEPINIT: 760 error = sockopt_get(sopt, &ui, sizeof(ui)); 761 if (error) 762 break; 763 if (ui > 0) { 764 tp->t_keepinit = ui; 765 change_keepalive(so, tp); 766 } else 767 error = EINVAL; 768 break; 769 770 default: 771 error = ENOPROTOOPT; 772 break; 773 } 774 break; 775 776 case PRCO_GETOPT: 777 switch (optname) { 778 #ifdef TCP_SIGNATURE 779 case TCP_MD5SIG: 780 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 781 error = sockopt_set(sopt, &optval, sizeof(optval)); 782 break; 783 #endif 784 case TCP_NODELAY: 785 optval = tp->t_flags & TF_NODELAY; 786 error = sockopt_set(sopt, &optval, sizeof(optval)); 787 break; 788 case TCP_MAXSEG: 789 optval = tp->t_peermss; 790 error = sockopt_set(sopt, &optval, sizeof(optval)); 791 break; 792 #ifdef notyet 793 case TCP_CONGCTL: 794 break; 795 #endif 796 default: 797 error = ENOPROTOOPT; 798 break; 799 } 800 break; 801 } 802 splx(s); 803 return (error); 804 } 805 806 #ifndef TCP_SENDSPACE 807 #define TCP_SENDSPACE 1024*32 808 #endif 809 int tcp_sendspace = TCP_SENDSPACE; 810 #ifndef TCP_RECVSPACE 811 #define TCP_RECVSPACE 1024*32 812 #endif 813 int tcp_recvspace = TCP_RECVSPACE; 814 815 /* 816 * tcp_attach: attach TCP protocol to socket, allocating internet protocol 817 * control block, TCP control block, buffer space and entering LISTEN state 818 * if to accept connections. 819 */ 820 static int 821 tcp_attach(struct socket *so, int proto) 822 { 823 inpcb_t *inp; 824 struct tcpcb *tp; 825 #ifdef INET6 826 struct in6pcb *in6p; 827 #endif 828 int s, error, family; 829 830 s = splsoftnet(); 831 sosetlock(so); 832 solock(so); 833 834 family = so->so_proto->pr_domain->dom_family; 835 switch (family) { 836 #ifdef INET 837 case PF_INET: 838 inp = sotoinpcb(so); 839 #ifdef INET6 840 in6p = NULL; 841 #endif 842 break; 843 #endif 844 #ifdef INET6 845 case PF_INET6: 846 inp = NULL; 847 in6p = sotoin6pcb(so); 848 break; 849 #endif 850 default: 851 error = EAFNOSUPPORT; 852 goto out; 853 } 854 855 KASSERT(inp == NULL); 856 #ifdef INET6 857 KASSERT(in6p == NULL); 858 #endif 859 860 #ifdef MBUFTRACE 861 so->so_mowner = &tcp_sock_mowner; 862 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 863 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 864 #endif 865 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 866 error = soreserve(so, tcp_sendspace, tcp_recvspace); 867 if (error) { 868 goto out; 869 } 870 } 871 872 so->so_rcv.sb_flags |= SB_AUTOSIZE; 873 so->so_snd.sb_flags |= SB_AUTOSIZE; 874 875 tp = NULL; 876 switch (family) { 877 #ifdef INET 878 case PF_INET: 879 error = inpcb_create(so, tcbtable); 880 if ((inp = sotoinpcb(so)) != NULL) { 881 tp = tcp_newtcpcb(family, (void *)inp); 882 } 883 #ifdef INET6 884 in6p = NULL; 885 #endif 886 break; 887 #endif 888 #ifdef INET6 889 case PF_INET6: 890 error = in6_pcballoc(so, tcbtable); 891 if ((in6p = sotoin6pcb(so)) != NULL) { 892 tp = tcp_newtcpcb(family, (void *)in6p); 893 } 894 inp = NULL; 895 break; 896 #endif 897 default: 898 error = EAFNOSUPPORT; 899 } 900 901 if (error) { 902 goto out; 903 } 904 905 if (tp == NULL) { 906 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 907 908 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 909 #ifdef INET 910 if (inp) 911 inpcb_destroy(inp); 912 #endif 913 #ifdef INET6 914 if (in6p) 915 in6_pcbdetach(in6p); 916 #endif 917 so->so_state |= nofd; 918 error = ENOBUFS; 919 goto out; 920 } 921 tp->t_state = TCPS_CLOSED; 922 923 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 924 so->so_linger = TCP_LINGERTIME; 925 tp = sototcpcb(so); 926 out: 927 sounlock(so); 928 splx(s); 929 return error; 930 } 931 932 static void 933 tcp_detach(struct socket *so) 934 { 935 inpcb_t *inp; 936 #ifdef INET6 937 struct in6pcb *in6p; 938 #endif 939 struct tcpcb *tp = NULL; 940 int s, family; 941 942 KASSERT(solocked(so)); 943 944 s = splsoftnet(); 945 family = so->so_proto->pr_domain->dom_family; 946 switch (family) { 947 #ifdef INET 948 case PF_INET: 949 inp = sotoinpcb(so); 950 tp = intotcpcb(inp); 951 break; 952 #endif 953 #ifdef INET6 954 case PF_INET6: 955 in6p = sotoin6pcb(so); 956 tp = in6totcpcb(in6p); 957 break; 958 #endif 959 default: 960 splx(s); 961 return; 962 } 963 KASSERT(tp != NULL); 964 (void)tcp_disconnect(tp); 965 splx(s); 966 } 967 968 /* 969 * Initiate (or continue) disconnect. 970 * If embryonic state, just send reset (once). 971 * If in ``let data drain'' option and linger null, just drop. 972 * Otherwise (hard), mark socket disconnecting and drop 973 * current input data; switch states based on user close, and 974 * send segment to peer (with FIN). 975 */ 976 struct tcpcb * 977 tcp_disconnect(struct tcpcb *tp) 978 { 979 struct socket *so; 980 981 if (tp->t_inpcb) 982 so = inpcb_get_socket(tp->t_inpcb); 983 #ifdef INET6 984 else if (tp->t_in6pcb) 985 so = tp->t_in6pcb->in6p_socket; 986 #endif 987 else 988 so = NULL; 989 990 KASSERT(so != NULL); 991 992 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 993 tp = tcp_close(tp); 994 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 995 tp = tcp_drop(tp, 0); 996 else { 997 soisdisconnecting(so); 998 sbflush(&so->so_rcv); 999 tp = tcp_usrclosed(tp); 1000 if (tp) 1001 (void) tcp_output(tp); 1002 } 1003 return (tp); 1004 } 1005 1006 /* 1007 * User issued close, and wish to trail through shutdown states: 1008 * if never received SYN, just forget it. If got a SYN from peer, 1009 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 1010 * If already got a FIN from peer, then almost done; go to LAST_ACK 1011 * state. In all other cases, have already sent FIN to peer (e.g. 1012 * after PRU_SHUTDOWN), and just have to play tedious game waiting 1013 * for peer to send FIN or not respond to keep-alives, etc. 1014 * We can let the user exit from the close as soon as the FIN is acked. 1015 */ 1016 struct tcpcb * 1017 tcp_usrclosed(struct tcpcb *tp) 1018 { 1019 1020 switch (tp->t_state) { 1021 1022 case TCPS_CLOSED: 1023 case TCPS_LISTEN: 1024 case TCPS_SYN_SENT: 1025 tp->t_state = TCPS_CLOSED; 1026 tp = tcp_close(tp); 1027 break; 1028 1029 case TCPS_SYN_RECEIVED: 1030 case TCPS_ESTABLISHED: 1031 tp->t_state = TCPS_FIN_WAIT_1; 1032 break; 1033 1034 case TCPS_CLOSE_WAIT: 1035 tp->t_state = TCPS_LAST_ACK; 1036 break; 1037 } 1038 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1039 struct socket *so; 1040 if (tp->t_inpcb) 1041 so = inpcb_get_socket(tp->t_inpcb); 1042 #ifdef INET6 1043 else if (tp->t_in6pcb) 1044 so = tp->t_in6pcb->in6p_socket; 1045 #endif 1046 else 1047 so = NULL; 1048 if (so) 1049 soisdisconnected(so); 1050 /* 1051 * If we are in FIN_WAIT_2, we arrived here because the 1052 * application did a shutdown of the send side. Like the 1053 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1054 * a full close, we start a timer to make sure sockets are 1055 * not left in FIN_WAIT_2 forever. 1056 */ 1057 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1058 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1059 else if (tp->t_state == TCPS_TIME_WAIT 1060 && ((tp->t_inpcb 1061 && (tcp4_vtw_enable & 1) 1062 && vtw_add(AF_INET, tp)) 1063 || 1064 (tp->t_in6pcb 1065 && (tcp6_vtw_enable & 1) 1066 && vtw_add(AF_INET6, tp)))) { 1067 tp = 0; 1068 } 1069 } 1070 return (tp); 1071 } 1072 1073 /* 1074 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1075 * than 32. 1076 */ 1077 static int 1078 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1079 { 1080 int error, mssdflt; 1081 struct sysctlnode node; 1082 1083 mssdflt = tcp_mssdflt; 1084 node = *rnode; 1085 node.sysctl_data = &mssdflt; 1086 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1087 if (error || newp == NULL) 1088 return (error); 1089 1090 if (mssdflt < 32) 1091 return (EINVAL); 1092 tcp_mssdflt = mssdflt; 1093 1094 mutex_enter(softnet_lock); 1095 tcp_tcpcb_template(); 1096 mutex_exit(softnet_lock); 1097 1098 return (0); 1099 } 1100 1101 /* 1102 * sysctl helper for TCP CB template update 1103 */ 1104 static int 1105 sysctl_update_tcpcb_template(SYSCTLFN_ARGS) 1106 { 1107 int t, error; 1108 struct sysctlnode node; 1109 1110 /* follow procedures in sysctl(9) manpage */ 1111 t = *(int *)rnode->sysctl_data; 1112 node = *rnode; 1113 node.sysctl_data = &t; 1114 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1115 if (error || newp == NULL) 1116 return error; 1117 1118 if (t < 0) 1119 return EINVAL; 1120 1121 *(int *)rnode->sysctl_data = t; 1122 1123 mutex_enter(softnet_lock); 1124 tcp_tcpcb_template(); 1125 mutex_exit(softnet_lock); 1126 1127 return 0; 1128 } 1129 1130 /* 1131 * sysctl helper routine for setting port related values under 1132 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1133 * additional checks for each type. this code has placed in 1134 * tcp_input.c since INET and INET6 both use the same tcp code. 1135 * 1136 * this helper is not static so that both inet and inet6 can use it. 1137 */ 1138 int 1139 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1140 { 1141 int error, tmp; 1142 int apmin, apmax; 1143 #ifndef IPNOPRIVPORTS 1144 int lpmin, lpmax; 1145 #endif /* IPNOPRIVPORTS */ 1146 struct sysctlnode node; 1147 1148 if (namelen != 0) 1149 return (EINVAL); 1150 1151 switch (name[-3]) { 1152 #ifdef INET 1153 case PF_INET: 1154 apmin = anonportmin; 1155 apmax = anonportmax; 1156 #ifndef IPNOPRIVPORTS 1157 lpmin = lowportmin; 1158 lpmax = lowportmax; 1159 #endif /* IPNOPRIVPORTS */ 1160 break; 1161 #endif /* INET */ 1162 #ifdef INET6 1163 case PF_INET6: 1164 apmin = ip6_anonportmin; 1165 apmax = ip6_anonportmax; 1166 #ifndef IPNOPRIVPORTS 1167 lpmin = ip6_lowportmin; 1168 lpmax = ip6_lowportmax; 1169 #endif /* IPNOPRIVPORTS */ 1170 break; 1171 #endif /* INET6 */ 1172 default: 1173 return (EINVAL); 1174 } 1175 1176 /* 1177 * insert temporary copy into node, perform lookup on 1178 * temporary, then restore pointer 1179 */ 1180 node = *rnode; 1181 tmp = *(int*)rnode->sysctl_data; 1182 node.sysctl_data = &tmp; 1183 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1184 if (error || newp == NULL) 1185 return (error); 1186 1187 /* 1188 * simple port range check 1189 */ 1190 if (tmp < 0 || tmp > 65535) 1191 return (EINVAL); 1192 1193 /* 1194 * per-node range checks 1195 */ 1196 switch (rnode->sysctl_num) { 1197 case IPCTL_ANONPORTMIN: 1198 case IPV6CTL_ANONPORTMIN: 1199 if (tmp >= apmax) 1200 return (EINVAL); 1201 #ifndef IPNOPRIVPORTS 1202 if (tmp < IPPORT_RESERVED) 1203 return (EINVAL); 1204 #endif /* IPNOPRIVPORTS */ 1205 break; 1206 1207 case IPCTL_ANONPORTMAX: 1208 case IPV6CTL_ANONPORTMAX: 1209 if (apmin >= tmp) 1210 return (EINVAL); 1211 #ifndef IPNOPRIVPORTS 1212 if (tmp < IPPORT_RESERVED) 1213 return (EINVAL); 1214 #endif /* IPNOPRIVPORTS */ 1215 break; 1216 1217 #ifndef IPNOPRIVPORTS 1218 case IPCTL_LOWPORTMIN: 1219 case IPV6CTL_LOWPORTMIN: 1220 if (tmp >= lpmax || 1221 tmp > IPPORT_RESERVEDMAX || 1222 tmp < IPPORT_RESERVEDMIN) 1223 return (EINVAL); 1224 break; 1225 1226 case IPCTL_LOWPORTMAX: 1227 case IPV6CTL_LOWPORTMAX: 1228 if (lpmin >= tmp || 1229 tmp > IPPORT_RESERVEDMAX || 1230 tmp < IPPORT_RESERVEDMIN) 1231 return (EINVAL); 1232 break; 1233 #endif /* IPNOPRIVPORTS */ 1234 1235 default: 1236 return (EINVAL); 1237 } 1238 1239 *(int*)rnode->sysctl_data = tmp; 1240 1241 return (0); 1242 } 1243 1244 static inline int 1245 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1246 { 1247 if (oldp) { 1248 size_t sz; 1249 uid_t uid; 1250 int error; 1251 1252 if (sockp->so_cred == NULL) 1253 return EPERM; 1254 1255 uid = kauth_cred_geteuid(sockp->so_cred); 1256 sz = MIN(sizeof(uid), *oldlenp); 1257 if ((error = copyout(&uid, oldp, sz)) != 0) 1258 return error; 1259 } 1260 *oldlenp = sizeof(uid_t); 1261 return 0; 1262 } 1263 1264 static inline int 1265 inet4_ident_core(struct in_addr raddr, u_int rport, 1266 struct in_addr laddr, u_int lport, 1267 void *oldp, size_t *oldlenp, 1268 struct lwp *l, int dodrop) 1269 { 1270 inpcb_t *inp; 1271 struct socket *so; 1272 1273 inp = inpcb_lookup(tcbtable, raddr, rport, laddr, lport, NULL); 1274 if (inp == NULL || (so = inpcb_get_socket(inp)) == NULL) 1275 return ESRCH; 1276 1277 if (dodrop) { 1278 struct tcpcb *tp; 1279 int error; 1280 1281 if ((tp = intotcpcb(inp)) == NULL || 1282 (so->so_options & SO_ACCEPTCONN) != 0) 1283 return ESRCH; 1284 1285 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1286 KAUTH_REQ_NETWORK_SOCKET_DROP, so, tp, NULL); 1287 if (error) { 1288 return error; 1289 } 1290 1291 (void)tcp_drop(tp, ECONNABORTED); 1292 return 0; 1293 } 1294 1295 return copyout_uid(so, oldp, oldlenp); 1296 } 1297 1298 #ifdef INET6 1299 static inline int 1300 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1301 struct in6_addr *laddr, u_int lport, 1302 void *oldp, size_t *oldlenp, 1303 struct lwp *l, int dodrop) 1304 { 1305 struct in6pcb *in6p; 1306 struct socket *sockp; 1307 1308 in6p = in6_pcblookup_connect(tcbtable, raddr, rport, laddr, lport, 0, 0); 1309 1310 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1311 return ESRCH; 1312 1313 if (dodrop) { 1314 struct tcpcb *tp; 1315 int error; 1316 1317 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1318 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1319 return ESRCH; 1320 1321 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1322 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1323 if (error) 1324 return (error); 1325 1326 (void)tcp_drop(tp, ECONNABORTED); 1327 return 0; 1328 } 1329 else 1330 return copyout_uid(sockp, oldp, oldlenp); 1331 } 1332 #endif 1333 1334 /* 1335 * sysctl helper routine for the net.inet.tcp.drop and 1336 * net.inet6.tcp6.drop nodes. 1337 */ 1338 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1339 1340 /* 1341 * sysctl helper routine for the net.inet.tcp.ident and 1342 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1343 * old way of looking up the ident information for ipv4 which involves 1344 * stuffing the port/addr pairs into the mib lookup. 1345 */ 1346 static int 1347 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1348 { 1349 #ifdef INET 1350 struct sockaddr_in *si4[2]; 1351 #endif /* INET */ 1352 #ifdef INET6 1353 struct sockaddr_in6 *si6[2]; 1354 #endif /* INET6 */ 1355 struct sockaddr_storage sa[2]; 1356 int error, pf, dodrop; 1357 1358 dodrop = name[-1] == TCPCTL_DROP; 1359 if (dodrop) { 1360 if (oldp != NULL || *oldlenp != 0) 1361 return EINVAL; 1362 if (newp == NULL) 1363 return EPERM; 1364 if (newlen < sizeof(sa)) 1365 return ENOMEM; 1366 } 1367 if (namelen != 4 && namelen != 0) 1368 return EINVAL; 1369 if (name[-2] != IPPROTO_TCP) 1370 return EINVAL; 1371 pf = name[-3]; 1372 1373 /* old style lookup, ipv4 only */ 1374 if (namelen == 4) { 1375 #ifdef INET 1376 struct in_addr laddr, raddr; 1377 u_int lport, rport; 1378 1379 if (pf != PF_INET) 1380 return EPROTONOSUPPORT; 1381 raddr.s_addr = (uint32_t)name[0]; 1382 rport = (u_int)name[1]; 1383 laddr.s_addr = (uint32_t)name[2]; 1384 lport = (u_int)name[3]; 1385 1386 mutex_enter(softnet_lock); 1387 error = inet4_ident_core(raddr, rport, laddr, lport, 1388 oldp, oldlenp, l, dodrop); 1389 mutex_exit(softnet_lock); 1390 return error; 1391 #else /* INET */ 1392 return EINVAL; 1393 #endif /* INET */ 1394 } 1395 1396 if (newp == NULL || newlen != sizeof(sa)) 1397 return EINVAL; 1398 error = copyin(newp, &sa, newlen); 1399 if (error) 1400 return error; 1401 1402 /* 1403 * requested families must match 1404 */ 1405 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1406 return EINVAL; 1407 1408 switch (pf) { 1409 #ifdef INET6 1410 case PF_INET6: 1411 si6[0] = (struct sockaddr_in6*)&sa[0]; 1412 si6[1] = (struct sockaddr_in6*)&sa[1]; 1413 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1414 si6[1]->sin6_len != sizeof(*si6[1])) 1415 return EINVAL; 1416 1417 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1418 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1419 error = sa6_embedscope(si6[0], ip6_use_defzone); 1420 if (error) 1421 return error; 1422 error = sa6_embedscope(si6[1], ip6_use_defzone); 1423 if (error) 1424 return error; 1425 1426 mutex_enter(softnet_lock); 1427 error = inet6_ident_core(&si6[0]->sin6_addr, 1428 si6[0]->sin6_port, &si6[1]->sin6_addr, 1429 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1430 mutex_exit(softnet_lock); 1431 return error; 1432 } 1433 1434 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1435 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1436 return EINVAL; 1437 1438 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1439 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1440 /*FALLTHROUGH*/ 1441 #endif /* INET6 */ 1442 #ifdef INET 1443 case PF_INET: 1444 si4[0] = (struct sockaddr_in*)&sa[0]; 1445 si4[1] = (struct sockaddr_in*)&sa[1]; 1446 if (si4[0]->sin_len != sizeof(*si4[0]) || 1447 si4[0]->sin_len != sizeof(*si4[1])) 1448 return EINVAL; 1449 1450 mutex_enter(softnet_lock); 1451 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1452 si4[1]->sin_addr, si4[1]->sin_port, 1453 oldp, oldlenp, l, dodrop); 1454 mutex_exit(softnet_lock); 1455 return error; 1456 #endif /* INET */ 1457 default: 1458 return EPROTONOSUPPORT; 1459 } 1460 } 1461 1462 static int 1463 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1464 { 1465 struct sysctlnode node; 1466 int error; 1467 char newname[TCPCC_MAXLEN]; 1468 1469 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1470 1471 node = *rnode; 1472 node.sysctl_data = newname; 1473 node.sysctl_size = sizeof(newname); 1474 1475 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1476 1477 if (error || 1478 newp == NULL || 1479 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1480 return error; 1481 1482 mutex_enter(softnet_lock); 1483 error = tcp_congctl_select(NULL, newname); 1484 mutex_exit(softnet_lock); 1485 1486 return error; 1487 } 1488 1489 static int 1490 sysctl_tcp_init_win(SYSCTLFN_ARGS) 1491 { 1492 int error; 1493 u_int iw; 1494 struct sysctlnode node; 1495 1496 iw = *(u_int *)rnode->sysctl_data; 1497 node = *rnode; 1498 node.sysctl_data = &iw; 1499 node.sysctl_size = sizeof(iw); 1500 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1501 if (error || newp == NULL) 1502 return error; 1503 1504 if (iw >= __arraycount(tcp_init_win_max)) 1505 return EINVAL; 1506 *(u_int *)rnode->sysctl_data = iw; 1507 return 0; 1508 } 1509 1510 static int 1511 sysctl_tcp_keep(SYSCTLFN_ARGS) 1512 { 1513 int error; 1514 u_int tmp; 1515 struct sysctlnode node; 1516 1517 node = *rnode; 1518 tmp = *(u_int *)rnode->sysctl_data; 1519 node.sysctl_data = &tmp; 1520 1521 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1522 if (error || newp == NULL) 1523 return error; 1524 1525 mutex_enter(softnet_lock); 1526 1527 *(u_int *)rnode->sysctl_data = tmp; 1528 tcp_tcpcb_template(); /* update the template */ 1529 1530 mutex_exit(softnet_lock); 1531 return 0; 1532 } 1533 1534 static int 1535 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 1536 { 1537 1538 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 1539 } 1540 1541 /* 1542 * this (second stage) setup routine is a replacement for tcp_sysctl() 1543 * (which is currently used for ipv4 and ipv6) 1544 */ 1545 static void 1546 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1547 const char *tcpname) 1548 { 1549 const struct sysctlnode *sack_node; 1550 const struct sysctlnode *abc_node; 1551 const struct sysctlnode *ecn_node; 1552 const struct sysctlnode *congctl_node; 1553 const struct sysctlnode *mslt_node; 1554 const struct sysctlnode *vtw_node; 1555 #ifdef TCP_DEBUG 1556 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1557 extern int tcp_debx; 1558 #endif 1559 1560 sysctl_createv(clog, 0, NULL, NULL, 1561 CTLFLAG_PERMANENT, 1562 CTLTYPE_NODE, pfname, NULL, 1563 NULL, 0, NULL, 0, 1564 CTL_NET, pf, CTL_EOL); 1565 sysctl_createv(clog, 0, NULL, NULL, 1566 CTLFLAG_PERMANENT, 1567 CTLTYPE_NODE, tcpname, 1568 SYSCTL_DESCR("TCP related settings"), 1569 NULL, 0, NULL, 0, 1570 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1571 1572 sysctl_createv(clog, 0, NULL, NULL, 1573 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1574 CTLTYPE_INT, "rfc1323", 1575 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1576 sysctl_update_tcpcb_template, 0, &tcp_do_rfc1323, 0, 1577 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1578 sysctl_createv(clog, 0, NULL, NULL, 1579 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1580 CTLTYPE_INT, "sendspace", 1581 SYSCTL_DESCR("Default TCP send buffer size"), 1582 NULL, 0, &tcp_sendspace, 0, 1583 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1584 sysctl_createv(clog, 0, NULL, NULL, 1585 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1586 CTLTYPE_INT, "recvspace", 1587 SYSCTL_DESCR("Default TCP receive buffer size"), 1588 NULL, 0, &tcp_recvspace, 0, 1589 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1590 sysctl_createv(clog, 0, NULL, NULL, 1591 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1592 CTLTYPE_INT, "mssdflt", 1593 SYSCTL_DESCR("Default maximum segment size"), 1594 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1595 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1596 sysctl_createv(clog, 0, NULL, NULL, 1597 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1598 CTLTYPE_INT, "minmss", 1599 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 1600 NULL, 0, &tcp_minmss, 0, 1601 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1602 sysctl_createv(clog, 0, NULL, NULL, 1603 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1604 CTLTYPE_INT, "msl", 1605 SYSCTL_DESCR("Maximum Segment Life"), 1606 NULL, 0, &tcp_msl, 0, 1607 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 1608 sysctl_createv(clog, 0, NULL, NULL, 1609 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1610 CTLTYPE_INT, "syn_cache_limit", 1611 SYSCTL_DESCR("Maximum number of entries in the TCP " 1612 "compressed state engine"), 1613 NULL, 0, &tcp_syn_cache_limit, 0, 1614 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1615 CTL_EOL); 1616 sysctl_createv(clog, 0, NULL, NULL, 1617 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1618 CTLTYPE_INT, "syn_bucket_limit", 1619 SYSCTL_DESCR("Maximum number of entries per hash " 1620 "bucket in the TCP compressed state " 1621 "engine"), 1622 NULL, 0, &tcp_syn_bucket_limit, 0, 1623 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1624 CTL_EOL); 1625 #if 0 /* obsoleted */ 1626 sysctl_createv(clog, 0, NULL, NULL, 1627 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1628 CTLTYPE_INT, "syn_cache_interval", 1629 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1630 NULL, 0, &tcp_syn_cache_interval, 0, 1631 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1632 CTL_EOL); 1633 #endif 1634 sysctl_createv(clog, 0, NULL, NULL, 1635 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1636 CTLTYPE_INT, "init_win", 1637 SYSCTL_DESCR("Initial TCP congestion window"), 1638 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 1639 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1640 sysctl_createv(clog, 0, NULL, NULL, 1641 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1642 CTLTYPE_INT, "mss_ifmtu", 1643 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1644 NULL, 0, &tcp_mss_ifmtu, 0, 1645 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1646 sysctl_createv(clog, 0, NULL, &sack_node, 1647 CTLFLAG_PERMANENT, 1648 CTLTYPE_NODE, "sack", 1649 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1650 NULL, 0, NULL, 0, 1651 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1652 1653 /* Congctl subtree */ 1654 sysctl_createv(clog, 0, NULL, &congctl_node, 1655 CTLFLAG_PERMANENT, 1656 CTLTYPE_NODE, "congctl", 1657 SYSCTL_DESCR("TCP Congestion Control"), 1658 NULL, 0, NULL, 0, 1659 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1660 sysctl_createv(clog, 0, &congctl_node, NULL, 1661 CTLFLAG_PERMANENT, 1662 CTLTYPE_STRING, "available", 1663 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1664 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1665 sysctl_createv(clog, 0, &congctl_node, NULL, 1666 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1667 CTLTYPE_STRING, "selected", 1668 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1669 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1670 CTL_CREATE, CTL_EOL); 1671 1672 sysctl_createv(clog, 0, NULL, NULL, 1673 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1674 CTLTYPE_INT, "win_scale", 1675 SYSCTL_DESCR("Use RFC1323 window scale options"), 1676 sysctl_update_tcpcb_template, 0, &tcp_do_win_scale, 0, 1677 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1678 sysctl_createv(clog, 0, NULL, NULL, 1679 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1680 CTLTYPE_INT, "timestamps", 1681 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1682 sysctl_update_tcpcb_template, 0, &tcp_do_timestamps, 0, 1683 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1684 sysctl_createv(clog, 0, NULL, NULL, 1685 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1686 CTLTYPE_INT, "compat_42", 1687 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1688 NULL, 0, &tcp_compat_42, 0, 1689 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1690 sysctl_createv(clog, 0, NULL, NULL, 1691 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1692 CTLTYPE_INT, "cwm", 1693 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1694 "Monitoring"), 1695 NULL, 0, &tcp_cwm, 0, 1696 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1697 sysctl_createv(clog, 0, NULL, NULL, 1698 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1699 CTLTYPE_INT, "cwm_burstsize", 1700 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1701 "burst count in packets"), 1702 NULL, 0, &tcp_cwm_burstsize, 0, 1703 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1704 CTL_EOL); 1705 sysctl_createv(clog, 0, NULL, NULL, 1706 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1707 CTLTYPE_INT, "ack_on_push", 1708 SYSCTL_DESCR("Immediately return ACK when PSH is " 1709 "received"), 1710 NULL, 0, &tcp_ack_on_push, 0, 1711 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1712 sysctl_createv(clog, 0, NULL, NULL, 1713 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1714 CTLTYPE_INT, "keepidle", 1715 SYSCTL_DESCR("Allowed connection idle ticks before a " 1716 "keepalive probe is sent"), 1717 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1718 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1719 sysctl_createv(clog, 0, NULL, NULL, 1720 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1721 CTLTYPE_INT, "keepintvl", 1722 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1723 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1724 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1725 sysctl_createv(clog, 0, NULL, NULL, 1726 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1727 CTLTYPE_INT, "keepcnt", 1728 SYSCTL_DESCR("Number of keepalive probes to send"), 1729 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1730 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1731 sysctl_createv(clog, 0, NULL, NULL, 1732 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1733 CTLTYPE_INT, "slowhz", 1734 SYSCTL_DESCR("Keepalive ticks per second"), 1735 NULL, PR_SLOWHZ, NULL, 0, 1736 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1737 sysctl_createv(clog, 0, NULL, NULL, 1738 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1739 CTLTYPE_INT, "log_refused", 1740 SYSCTL_DESCR("Log refused TCP connections"), 1741 NULL, 0, &tcp_log_refused, 0, 1742 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1743 #if 0 /* obsoleted */ 1744 sysctl_createv(clog, 0, NULL, NULL, 1745 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1746 CTLTYPE_INT, "rstratelimit", NULL, 1747 NULL, 0, &tcp_rst_ratelim, 0, 1748 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1749 #endif 1750 sysctl_createv(clog, 0, NULL, NULL, 1751 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1752 CTLTYPE_INT, "rstppslimit", 1753 SYSCTL_DESCR("Maximum number of RST packets to send " 1754 "per second"), 1755 NULL, 0, &tcp_rst_ppslim, 0, 1756 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1757 sysctl_createv(clog, 0, NULL, NULL, 1758 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1759 CTLTYPE_INT, "delack_ticks", 1760 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1761 NULL, 0, &tcp_delack_ticks, 0, 1762 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1763 sysctl_createv(clog, 0, NULL, NULL, 1764 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1765 CTLTYPE_INT, "init_win_local", 1766 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1767 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 1768 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1769 CTL_EOL); 1770 sysctl_createv(clog, 0, NULL, NULL, 1771 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1772 CTLTYPE_STRUCT, "ident", 1773 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1774 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1775 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1776 sysctl_createv(clog, 0, NULL, NULL, 1777 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1778 CTLTYPE_INT, "do_loopback_cksum", 1779 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1780 NULL, 0, &tcp_do_loopback_cksum, 0, 1781 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1782 CTL_EOL); 1783 sysctl_createv(clog, 0, NULL, NULL, 1784 CTLFLAG_PERMANENT, 1785 CTLTYPE_STRUCT, "pcblist", 1786 SYSCTL_DESCR("TCP protocol control block list"), 1787 sysctl_inpcblist, 0, tcbtable, 0, 1788 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1789 CTL_EOL); 1790 sysctl_createv(clog, 0, NULL, NULL, 1791 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1792 CTLTYPE_INT, "keepinit", 1793 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1794 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 1795 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1796 1797 /* TCP socket buffers auto-sizing nodes */ 1798 sysctl_createv(clog, 0, NULL, NULL, 1799 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1800 CTLTYPE_INT, "recvbuf_auto", 1801 SYSCTL_DESCR("Enable automatic receive " 1802 "buffer sizing (experimental)"), 1803 NULL, 0, &tcp_do_autorcvbuf, 0, 1804 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1805 sysctl_createv(clog, 0, NULL, NULL, 1806 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1807 CTLTYPE_INT, "recvbuf_inc", 1808 SYSCTL_DESCR("Incrementor step size of " 1809 "automatic receive buffer"), 1810 NULL, 0, &tcp_autorcvbuf_inc, 0, 1811 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1812 sysctl_createv(clog, 0, NULL, NULL, 1813 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1814 CTLTYPE_INT, "recvbuf_max", 1815 SYSCTL_DESCR("Max size of automatic receive buffer"), 1816 NULL, 0, &tcp_autorcvbuf_max, 0, 1817 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1818 1819 sysctl_createv(clog, 0, NULL, NULL, 1820 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1821 CTLTYPE_INT, "sendbuf_auto", 1822 SYSCTL_DESCR("Enable automatic send " 1823 "buffer sizing (experimental)"), 1824 NULL, 0, &tcp_do_autosndbuf, 0, 1825 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1826 sysctl_createv(clog, 0, NULL, NULL, 1827 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1828 CTLTYPE_INT, "sendbuf_inc", 1829 SYSCTL_DESCR("Incrementor step size of " 1830 "automatic send buffer"), 1831 NULL, 0, &tcp_autosndbuf_inc, 0, 1832 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1833 sysctl_createv(clog, 0, NULL, NULL, 1834 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1835 CTLTYPE_INT, "sendbuf_max", 1836 SYSCTL_DESCR("Max size of automatic send buffer"), 1837 NULL, 0, &tcp_autosndbuf_max, 0, 1838 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1839 1840 /* ECN subtree */ 1841 sysctl_createv(clog, 0, NULL, &ecn_node, 1842 CTLFLAG_PERMANENT, 1843 CTLTYPE_NODE, "ecn", 1844 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 1845 NULL, 0, NULL, 0, 1846 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1847 sysctl_createv(clog, 0, &ecn_node, NULL, 1848 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1849 CTLTYPE_INT, "enable", 1850 SYSCTL_DESCR("Enable TCP Explicit Congestion " 1851 "Notification"), 1852 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 1853 sysctl_createv(clog, 0, &ecn_node, NULL, 1854 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1855 CTLTYPE_INT, "maxretries", 1856 SYSCTL_DESCR("Number of times to retry ECN setup " 1857 "before disabling ECN on the connection"), 1858 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 1859 1860 /* SACK gets it's own little subtree. */ 1861 sysctl_createv(clog, 0, NULL, &sack_node, 1862 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1863 CTLTYPE_INT, "enable", 1864 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 1865 NULL, 0, &tcp_do_sack, 0, 1866 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1867 sysctl_createv(clog, 0, NULL, &sack_node, 1868 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1869 CTLTYPE_INT, "maxholes", 1870 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 1871 NULL, 0, &tcp_sack_tp_maxholes, 0, 1872 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1873 sysctl_createv(clog, 0, NULL, &sack_node, 1874 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1875 CTLTYPE_INT, "globalmaxholes", 1876 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 1877 NULL, 0, &tcp_sack_globalmaxholes, 0, 1878 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1879 sysctl_createv(clog, 0, NULL, &sack_node, 1880 CTLFLAG_PERMANENT, 1881 CTLTYPE_INT, "globalholes", 1882 SYSCTL_DESCR("Global number of TCP SACK holes"), 1883 NULL, 0, &tcp_sack_globalholes, 0, 1884 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1885 1886 sysctl_createv(clog, 0, NULL, NULL, 1887 CTLFLAG_PERMANENT, 1888 CTLTYPE_STRUCT, "stats", 1889 SYSCTL_DESCR("TCP statistics"), 1890 sysctl_net_inet_tcp_stats, 0, NULL, 0, 1891 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1892 CTL_EOL); 1893 sysctl_createv(clog, 0, NULL, NULL, 1894 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1895 CTLTYPE_INT, "local_by_rtt", 1896 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 1897 "are local"), 1898 NULL, 0, &tcp_rttlocal, 0, 1899 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1900 #ifdef TCP_DEBUG 1901 sysctl_createv(clog, 0, NULL, NULL, 1902 CTLFLAG_PERMANENT, 1903 CTLTYPE_STRUCT, "debug", 1904 SYSCTL_DESCR("TCP sockets debug information"), 1905 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1906 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1907 CTL_EOL); 1908 sysctl_createv(clog, 0, NULL, NULL, 1909 CTLFLAG_PERMANENT, 1910 CTLTYPE_INT, "debx", 1911 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1912 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1913 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1914 CTL_EOL); 1915 #endif 1916 sysctl_createv(clog, 0, NULL, NULL, 1917 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1918 CTLTYPE_STRUCT, "drop", 1919 SYSCTL_DESCR("TCP drop connection"), 1920 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1921 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1922 sysctl_createv(clog, 0, NULL, NULL, 1923 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1924 CTLTYPE_INT, "iss_hash", 1925 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1926 "hash computation"), 1927 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1928 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1929 CTL_EOL); 1930 1931 /* ABC subtree */ 1932 1933 sysctl_createv(clog, 0, NULL, &abc_node, 1934 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 1935 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 1936 NULL, 0, NULL, 0, 1937 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1938 sysctl_createv(clog, 0, &abc_node, NULL, 1939 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1940 CTLTYPE_INT, "enable", 1941 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 1942 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 1943 sysctl_createv(clog, 0, &abc_node, NULL, 1944 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1945 CTLTYPE_INT, "aggressive", 1946 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 1947 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 1948 1949 /* MSL tuning subtree */ 1950 1951 sysctl_createv(clog, 0, NULL, &mslt_node, 1952 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 1953 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 1954 NULL, 0, NULL, 0, 1955 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1956 sysctl_createv(clog, 0, &mslt_node, NULL, 1957 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1958 CTLTYPE_INT, "enable", 1959 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 1960 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 1961 sysctl_createv(clog, 0, &mslt_node, NULL, 1962 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1963 CTLTYPE_INT, "loopback", 1964 SYSCTL_DESCR("MSL value to use for loopback connections"), 1965 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 1966 sysctl_createv(clog, 0, &mslt_node, NULL, 1967 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1968 CTLTYPE_INT, "local", 1969 SYSCTL_DESCR("MSL value to use for local connections"), 1970 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 1971 sysctl_createv(clog, 0, &mslt_node, NULL, 1972 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1973 CTLTYPE_INT, "remote", 1974 SYSCTL_DESCR("MSL value to use for remote connections"), 1975 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 1976 sysctl_createv(clog, 0, &mslt_node, NULL, 1977 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1978 CTLTYPE_INT, "remote_threshold", 1979 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 1980 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 1981 1982 /* vestigial TIME_WAIT tuning subtree */ 1983 1984 sysctl_createv(clog, 0, NULL, &vtw_node, 1985 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 1986 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 1987 NULL, 0, NULL, 0, 1988 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1989 sysctl_createv(clog, 0, &vtw_node, NULL, 1990 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1991 CTLTYPE_INT, "enable", 1992 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 1993 sysctl_tcp_vtw_enable, 0, 1994 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 1995 0, CTL_CREATE, CTL_EOL); 1996 sysctl_createv(clog, 0, &vtw_node, NULL, 1997 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 1998 CTLTYPE_INT, "entries", 1999 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 2000 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 2001 } 2002 2003 void 2004 tcp_usrreq_init(void) 2005 { 2006 2007 #ifdef INET 2008 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 2009 #endif 2010 #ifdef INET6 2011 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 2012 #endif 2013 } 2014 2015 PR_WRAP_USRREQ(tcp_usrreq) 2016 2017 #define tcp_usrreq tcp_usrreq_wrapper 2018 2019 const struct pr_usrreqs tcp_usrreqs = { 2020 .pr_attach = tcp_attach, 2021 .pr_detach = tcp_detach, 2022 .pr_generic = tcp_usrreq, 2023 }; 2024
Indexes created Sat Sep 20 22:09:52 GMT 2025