tcp_usrreq.c revision 1.190
1 /* $NetBSD: tcp_usrreq.c,v 1.190 2014/07/24 15:12:03 rtr 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 /* 98 * TCP protocol interface to socket abstraction. 99 */ 100 101 #include <sys/cdefs.h> 102 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.190 2014/07/24 15:12:03 rtr Exp $"); 103 104 #include "opt_inet.h" 105 #include "opt_ipsec.h" 106 #include "opt_tcp_debug.h" 107 #include "opt_mbuftrace.h" 108 109 #include <sys/param.h> 110 #include <sys/systm.h> 111 #include <sys/kernel.h> 112 #include <sys/malloc.h> 113 #include <sys/mbuf.h> 114 #include <sys/socket.h> 115 #include <sys/socketvar.h> 116 #include <sys/protosw.h> 117 #include <sys/errno.h> 118 #include <sys/stat.h> 119 #include <sys/proc.h> 120 #include <sys/domain.h> 121 #include <sys/sysctl.h> 122 #include <sys/kauth.h> 123 #include <sys/uidinfo.h> 124 125 #include <net/if.h> 126 #include <net/route.h> 127 128 #include <netinet/in.h> 129 #include <netinet/in_systm.h> 130 #include <netinet/in_var.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/ip_var.h> 134 #include <netinet/in_offload.h> 135 136 #ifdef INET6 137 #ifndef INET 138 #include <netinet/in.h> 139 #endif 140 #include <netinet/ip6.h> 141 #include <netinet6/in6_pcb.h> 142 #include <netinet6/ip6_var.h> 143 #include <netinet6/scope6_var.h> 144 #endif 145 146 #include <netinet/tcp.h> 147 #include <netinet/tcp_fsm.h> 148 #include <netinet/tcp_seq.h> 149 #include <netinet/tcp_timer.h> 150 #include <netinet/tcp_var.h> 151 #include <netinet/tcp_private.h> 152 #include <netinet/tcp_congctl.h> 153 #include <netinet/tcpip.h> 154 #include <netinet/tcp_debug.h> 155 #include <netinet/tcp_vtw.h> 156 157 #include "opt_tcp_space.h" 158 159 static int 160 tcp_debug_capture(struct tcpcb *tp, int req) 161 { 162 #ifdef KPROF 163 tcp_acounts[tp->t_state][req]++; 164 #endif 165 #ifdef TCP_DEBUG 166 return tp->t_state; 167 #endif 168 return 0; 169 } 170 171 static inline void 172 tcp_debug_trace(struct socket *so, struct tcpcb *tp, int ostate, int req) 173 { 174 #ifdef TCP_DEBUG 175 if (tp && (so->so_options & SO_DEBUG)) 176 tcp_trace(TA_USER, ostate, tp, NULL, req); 177 #endif 178 } 179 180 /* 181 * Process a TCP user request for TCP tb. If this is a send request 182 * then m is the mbuf chain of send data. If this is a timer expiration 183 * (called from the software clock routine), then timertype tells which timer. 184 */ 185 static int 186 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 187 struct mbuf *control, struct lwp *l) 188 { 189 struct inpcb *inp; 190 #ifdef INET6 191 struct in6pcb *in6p; 192 #endif 193 struct tcpcb *tp = NULL; 194 int s; 195 int error = 0; 196 int ostate = 0; 197 int family; /* family of the socket */ 198 199 KASSERT(req != PRU_ATTACH); 200 KASSERT(req != PRU_DETACH); 201 KASSERT(req != PRU_ACCEPT); 202 KASSERT(req != PRU_CONTROL); 203 KASSERT(req != PRU_SENSE); 204 KASSERT(req != PRU_PEERADDR); 205 KASSERT(req != PRU_SOCKADDR); 206 KASSERT(req != PRU_RCVOOB); 207 KASSERT(req != PRU_SENDOOB); 208 209 family = so->so_proto->pr_domain->dom_family; 210 211 s = splsoftnet(); 212 213 if (req == PRU_PURGEIF) { 214 mutex_enter(softnet_lock); 215 switch (family) { 216 #ifdef INET 217 case PF_INET: 218 in_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 219 in_purgeif((struct ifnet *)control); 220 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 221 break; 222 #endif 223 #ifdef INET6 224 case PF_INET6: 225 in6_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 226 in6_purgeif((struct ifnet *)control); 227 in6_pcbpurgeif(&tcbtable, (struct ifnet *)control); 228 break; 229 #endif 230 default: 231 mutex_exit(softnet_lock); 232 splx(s); 233 return (EAFNOSUPPORT); 234 } 235 mutex_exit(softnet_lock); 236 splx(s); 237 return (0); 238 } 239 240 KASSERT(solocked(so)); 241 242 switch (family) { 243 #ifdef INET 244 case PF_INET: 245 inp = sotoinpcb(so); 246 #ifdef INET6 247 in6p = NULL; 248 #endif 249 break; 250 #endif 251 #ifdef INET6 252 case PF_INET6: 253 inp = NULL; 254 in6p = sotoin6pcb(so); 255 break; 256 #endif 257 default: 258 splx(s); 259 return EAFNOSUPPORT; 260 } 261 KASSERT(!control || req == PRU_SEND); 262 #ifdef INET6 263 /* XXX: KASSERT((inp != NULL) ^ (in6p != NULL)); */ 264 #endif 265 /* 266 * When a TCP is attached to a socket, then there will be 267 * a (struct inpcb) pointed at by the socket, and this 268 * structure will point at a subsidary (struct tcpcb). 269 */ 270 if (inp == NULL 271 #ifdef INET6 272 && in6p == NULL 273 #endif 274 ) 275 { 276 error = EINVAL; 277 goto release; 278 } 279 #ifdef INET 280 if (inp) { 281 tp = intotcpcb(inp); 282 /* WHAT IF TP IS 0? */ 283 ostate = tcp_debug_capture(tp, req); 284 } 285 #endif 286 #ifdef INET6 287 if (in6p) { 288 tp = in6totcpcb(in6p); 289 /* WHAT IF TP IS 0? */ 290 ostate = tcp_debug_capture(tp, req); 291 } 292 #endif 293 294 switch (req) { 295 296 /* 297 * Prepare to accept connections. 298 */ 299 case PRU_LISTEN: 300 #ifdef INET 301 if (inp && inp->inp_lport == 0) { 302 error = in_pcbbind(inp, NULL); 303 if (error) 304 break; 305 } 306 #endif 307 #ifdef INET6 308 if (in6p && in6p->in6p_lport == 0) { 309 error = in6_pcbbind(in6p, NULL); 310 if (error) 311 break; 312 } 313 #endif 314 tp->t_state = TCPS_LISTEN; 315 break; 316 317 /* 318 * Initiate connection to peer. 319 * Create a template for use in transmissions on this connection. 320 * Enter SYN_SENT state, and mark socket as connecting. 321 * Start keep-alive timer, and seed output sequence space. 322 * Send initial segment on connection. 323 */ 324 case PRU_CONNECT: 325 #ifdef INET 326 if (inp) { 327 if (inp->inp_lport == 0) { 328 error = in_pcbbind(inp, NULL); 329 if (error) 330 break; 331 } 332 error = in_pcbconnect(inp, nam, l); 333 } 334 #endif 335 #ifdef INET6 336 if (in6p) { 337 if (in6p->in6p_lport == 0) { 338 error = in6_pcbbind(in6p, NULL); 339 if (error) 340 break; 341 } 342 error = in6_pcbconnect(in6p, nam, l); 343 if (!error) { 344 /* mapped addr case */ 345 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 346 tp->t_family = AF_INET; 347 else 348 tp->t_family = AF_INET6; 349 } 350 } 351 #endif 352 if (error) 353 break; 354 tp->t_template = tcp_template(tp); 355 if (tp->t_template == 0) { 356 #ifdef INET 357 if (inp) 358 in_pcbdisconnect(inp); 359 #endif 360 #ifdef INET6 361 if (in6p) 362 in6_pcbdisconnect(in6p); 363 #endif 364 error = ENOBUFS; 365 break; 366 } 367 /* 368 * Compute window scaling to request. 369 * XXX: This should be moved to tcp_output(). 370 */ 371 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 372 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 373 tp->request_r_scale++; 374 soisconnecting(so); 375 TCP_STATINC(TCP_STAT_CONNATTEMPT); 376 tp->t_state = TCPS_SYN_SENT; 377 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 378 tp->iss = tcp_new_iss(tp, 0); 379 tcp_sendseqinit(tp); 380 error = tcp_output(tp); 381 break; 382 383 /* 384 * Create a TCP connection between two sockets. 385 */ 386 case PRU_CONNECT2: 387 error = EOPNOTSUPP; 388 break; 389 390 /* 391 * Initiate disconnect from peer. 392 * If connection never passed embryonic stage, just drop; 393 * else if don't need to let data drain, then can just drop anyways, 394 * else have to begin TCP shutdown process: mark socket disconnecting, 395 * drain unread data, state switch to reflect user close, and 396 * send segment (e.g. FIN) to peer. Socket will be really disconnected 397 * when peer sends FIN and acks ours. 398 * 399 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 400 */ 401 case PRU_DISCONNECT: 402 tp = tcp_disconnect(tp); 403 break; 404 405 /* 406 * Mark the connection as being incapable of further output. 407 */ 408 case PRU_SHUTDOWN: 409 socantsendmore(so); 410 tp = tcp_usrclosed(tp); 411 if (tp) 412 error = tcp_output(tp); 413 break; 414 415 /* 416 * After a receive, possibly send window update to peer. 417 */ 418 case PRU_RCVD: 419 /* 420 * soreceive() calls this function when a user receives 421 * ancillary data on a listening socket. We don't call 422 * tcp_output in such a case, since there is no header 423 * template for a listening socket and hence the kernel 424 * will panic. 425 */ 426 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 427 (void) tcp_output(tp); 428 break; 429 430 /* 431 * Do a send by putting data in output queue and updating urgent 432 * marker if URG set. Possibly send more data. 433 */ 434 case PRU_SEND: 435 if (control && control->m_len) { 436 m_freem(control); 437 m_freem(m); 438 error = EINVAL; 439 break; 440 } 441 sbappendstream(&so->so_snd, m); 442 error = tcp_output(tp); 443 break; 444 445 /* 446 * Abort the TCP. 447 */ 448 case PRU_ABORT: 449 tp = tcp_drop(tp, ECONNABORTED); 450 break; 451 452 default: 453 panic("tcp_usrreq"); 454 } 455 456 tcp_debug_trace(so, tp, ostate, req); 457 458 release: 459 splx(s); 460 return (error); 461 } 462 463 static void 464 change_keepalive(struct socket *so, struct tcpcb *tp) 465 { 466 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 467 TCP_TIMER_DISARM(tp, TCPT_KEEP); 468 TCP_TIMER_DISARM(tp, TCPT_2MSL); 469 470 if (tp->t_state == TCPS_SYN_RECEIVED || 471 tp->t_state == TCPS_SYN_SENT) { 472 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 473 } else if (so->so_options & SO_KEEPALIVE && 474 tp->t_state <= TCPS_CLOSE_WAIT) { 475 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 476 } else { 477 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 478 } 479 480 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 481 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 482 } 483 484 485 int 486 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 487 { 488 int error = 0, s; 489 struct inpcb *inp; 490 #ifdef INET6 491 struct in6pcb *in6p; 492 #endif 493 struct tcpcb *tp; 494 u_int ui; 495 int family; /* family of the socket */ 496 int level, optname, optval; 497 498 level = sopt->sopt_level; 499 optname = sopt->sopt_name; 500 501 family = so->so_proto->pr_domain->dom_family; 502 503 s = splsoftnet(); 504 switch (family) { 505 #ifdef INET 506 case PF_INET: 507 inp = sotoinpcb(so); 508 #ifdef INET6 509 in6p = NULL; 510 #endif 511 break; 512 #endif 513 #ifdef INET6 514 case PF_INET6: 515 inp = NULL; 516 in6p = sotoin6pcb(so); 517 break; 518 #endif 519 default: 520 splx(s); 521 panic("%s: af %d", __func__, family); 522 } 523 #ifndef INET6 524 if (inp == NULL) 525 #else 526 if (inp == NULL && in6p == NULL) 527 #endif 528 { 529 splx(s); 530 return (ECONNRESET); 531 } 532 if (level != IPPROTO_TCP) { 533 switch (family) { 534 #ifdef INET 535 case PF_INET: 536 error = ip_ctloutput(op, so, sopt); 537 break; 538 #endif 539 #ifdef INET6 540 case PF_INET6: 541 error = ip6_ctloutput(op, so, sopt); 542 break; 543 #endif 544 } 545 splx(s); 546 return (error); 547 } 548 if (inp) 549 tp = intotcpcb(inp); 550 #ifdef INET6 551 else if (in6p) 552 tp = in6totcpcb(in6p); 553 #endif 554 else 555 tp = NULL; 556 557 switch (op) { 558 case PRCO_SETOPT: 559 switch (optname) { 560 #ifdef TCP_SIGNATURE 561 case TCP_MD5SIG: 562 error = sockopt_getint(sopt, &optval); 563 if (error) 564 break; 565 if (optval > 0) 566 tp->t_flags |= TF_SIGNATURE; 567 else 568 tp->t_flags &= ~TF_SIGNATURE; 569 break; 570 #endif /* TCP_SIGNATURE */ 571 572 case TCP_NODELAY: 573 error = sockopt_getint(sopt, &optval); 574 if (error) 575 break; 576 if (optval) 577 tp->t_flags |= TF_NODELAY; 578 else 579 tp->t_flags &= ~TF_NODELAY; 580 break; 581 582 case TCP_MAXSEG: 583 error = sockopt_getint(sopt, &optval); 584 if (error) 585 break; 586 if (optval > 0 && optval <= tp->t_peermss) 587 tp->t_peermss = optval; /* limit on send size */ 588 else 589 error = EINVAL; 590 break; 591 #ifdef notyet 592 case TCP_CONGCTL: 593 /* XXX string overflow XXX */ 594 error = tcp_congctl_select(tp, sopt->sopt_data); 595 break; 596 #endif 597 598 case TCP_KEEPIDLE: 599 error = sockopt_get(sopt, &ui, sizeof(ui)); 600 if (error) 601 break; 602 if (ui > 0) { 603 tp->t_keepidle = ui; 604 change_keepalive(so, tp); 605 } else 606 error = EINVAL; 607 break; 608 609 case TCP_KEEPINTVL: 610 error = sockopt_get(sopt, &ui, sizeof(ui)); 611 if (error) 612 break; 613 if (ui > 0) { 614 tp->t_keepintvl = ui; 615 change_keepalive(so, tp); 616 } else 617 error = EINVAL; 618 break; 619 620 case TCP_KEEPCNT: 621 error = sockopt_get(sopt, &ui, sizeof(ui)); 622 if (error) 623 break; 624 if (ui > 0) { 625 tp->t_keepcnt = ui; 626 change_keepalive(so, tp); 627 } else 628 error = EINVAL; 629 break; 630 631 case TCP_KEEPINIT: 632 error = sockopt_get(sopt, &ui, sizeof(ui)); 633 if (error) 634 break; 635 if (ui > 0) { 636 tp->t_keepinit = ui; 637 change_keepalive(so, tp); 638 } else 639 error = EINVAL; 640 break; 641 642 default: 643 error = ENOPROTOOPT; 644 break; 645 } 646 break; 647 648 case PRCO_GETOPT: 649 switch (optname) { 650 #ifdef TCP_SIGNATURE 651 case TCP_MD5SIG: 652 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 653 error = sockopt_set(sopt, &optval, sizeof(optval)); 654 break; 655 #endif 656 case TCP_NODELAY: 657 optval = tp->t_flags & TF_NODELAY; 658 error = sockopt_set(sopt, &optval, sizeof(optval)); 659 break; 660 case TCP_MAXSEG: 661 optval = tp->t_peermss; 662 error = sockopt_set(sopt, &optval, sizeof(optval)); 663 break; 664 #ifdef notyet 665 case TCP_CONGCTL: 666 break; 667 #endif 668 default: 669 error = ENOPROTOOPT; 670 break; 671 } 672 break; 673 } 674 splx(s); 675 return (error); 676 } 677 678 #ifndef TCP_SENDSPACE 679 #define TCP_SENDSPACE 1024*32 680 #endif 681 int tcp_sendspace = TCP_SENDSPACE; 682 #ifndef TCP_RECVSPACE 683 #define TCP_RECVSPACE 1024*32 684 #endif 685 int tcp_recvspace = TCP_RECVSPACE; 686 687 /* 688 * tcp_attach: attach TCP protocol to socket, allocating internet protocol 689 * control block, TCP control block, buffer space and entering LISTEN state 690 * if to accept connections. 691 */ 692 static int 693 tcp_attach(struct socket *so, int proto) 694 { 695 struct tcpcb *tp; 696 struct inpcb *inp; 697 #ifdef INET6 698 struct in6pcb *in6p; 699 #endif 700 int s, error, family; 701 702 /* Assign the lock (must happen even if we will error out). */ 703 s = splsoftnet(); 704 sosetlock(so); 705 KASSERT(solocked(so)); 706 707 family = so->so_proto->pr_domain->dom_family; 708 switch (family) { 709 #ifdef INET 710 case PF_INET: 711 inp = sotoinpcb(so); 712 #ifdef INET6 713 in6p = NULL; 714 #endif 715 break; 716 #endif 717 #ifdef INET6 718 case PF_INET6: 719 inp = NULL; 720 in6p = sotoin6pcb(so); 721 break; 722 #endif 723 default: 724 error = EAFNOSUPPORT; 725 goto out; 726 } 727 728 KASSERT(inp == NULL); 729 #ifdef INET6 730 KASSERT(in6p == NULL); 731 #endif 732 733 #ifdef MBUFTRACE 734 so->so_mowner = &tcp_sock_mowner; 735 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 736 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 737 #endif 738 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 739 error = soreserve(so, tcp_sendspace, tcp_recvspace); 740 if (error) 741 goto out; 742 } 743 744 so->so_rcv.sb_flags |= SB_AUTOSIZE; 745 so->so_snd.sb_flags |= SB_AUTOSIZE; 746 747 switch (family) { 748 #ifdef INET 749 case PF_INET: 750 error = in_pcballoc(so, &tcbtable); 751 if (error) 752 goto out; 753 inp = sotoinpcb(so); 754 #ifdef INET6 755 in6p = NULL; 756 #endif 757 break; 758 #endif 759 #ifdef INET6 760 case PF_INET6: 761 error = in6_pcballoc(so, &tcbtable); 762 if (error) 763 goto out; 764 inp = NULL; 765 in6p = sotoin6pcb(so); 766 break; 767 #endif 768 default: 769 error = EAFNOSUPPORT; 770 goto out; 771 } 772 if (inp) 773 tp = tcp_newtcpcb(family, (void *)inp); 774 #ifdef INET6 775 else if (in6p) 776 tp = tcp_newtcpcb(family, (void *)in6p); 777 #endif 778 else 779 tp = NULL; 780 781 if (tp == NULL) { 782 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 783 784 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 785 #ifdef INET 786 if (inp) 787 in_pcbdetach(inp); 788 #endif 789 #ifdef INET6 790 if (in6p) 791 in6_pcbdetach(in6p); 792 #endif 793 so->so_state |= nofd; 794 error = ENOBUFS; 795 goto out; 796 } 797 tp->t_state = TCPS_CLOSED; 798 if ((so->so_options & SO_LINGER) && so->so_linger == 0) { 799 so->so_linger = TCP_LINGERTIME; 800 } 801 out: 802 KASSERT(solocked(so)); 803 splx(s); 804 return error; 805 } 806 807 static void 808 tcp_detach(struct socket *so) 809 { 810 struct inpcb *inp; 811 #ifdef INET6 812 struct in6pcb *in6p; 813 #endif 814 struct tcpcb *tp = NULL; 815 int s, family; 816 817 KASSERT(solocked(so)); 818 819 s = splsoftnet(); 820 family = so->so_proto->pr_domain->dom_family; 821 switch (family) { 822 #ifdef INET 823 case PF_INET: 824 inp = sotoinpcb(so); 825 tp = intotcpcb(inp); 826 break; 827 #endif 828 #ifdef INET6 829 case PF_INET6: 830 in6p = sotoin6pcb(so); 831 tp = in6totcpcb(in6p); 832 break; 833 #endif 834 default: 835 splx(s); 836 return; 837 } 838 KASSERT(tp != NULL); 839 (void)tcp_disconnect(tp); 840 splx(s); 841 } 842 843 static int 844 tcp_accept(struct socket *so, struct mbuf *nam) 845 { 846 struct inpcb *inp = NULL; 847 #ifdef INET6 848 struct in6pcb *in6p = NULL; 849 #endif 850 struct tcpcb *tp = NULL; 851 int ostate = 0; 852 853 KASSERT(solocked(so)); 854 855 switch (so->so_proto->pr_domain->dom_family) { 856 #ifdef INET 857 case PF_INET: 858 inp = sotoinpcb(so); 859 break; 860 #endif 861 #ifdef INET6 862 case PF_INET6: 863 in6p = sotoin6pcb(so); 864 break; 865 #endif 866 default: 867 return EAFNOSUPPORT; 868 } 869 870 /* 871 * When a TCP is attached to a socket, then there will be 872 * a (struct inpcb) pointed at by the socket, and this 873 * structure will point at a subsidary (struct tcpcb). 874 */ 875 if (inp == NULL 876 #ifdef INET6 877 && in6p == NULL 878 #endif 879 ) 880 { 881 return EINVAL; 882 } 883 884 /* 885 * Accept a connection. Essentially all the work is 886 * done at higher levels; just return the address 887 * of the peer, storing through addr. 888 */ 889 #ifdef INET 890 if (inp) { 891 tp = intotcpcb(inp); 892 KASSERT(tp != NULL); 893 ostate = tcp_debug_capture(tp, PRU_ACCEPT); 894 in_setpeeraddr(inp, nam); 895 } 896 #endif 897 #ifdef INET6 898 if (in6p) { 899 tp = in6totcpcb(in6p); 900 KASSERT(tp != NULL); 901 ostate = tcp_debug_capture(tp, PRU_ACCEPT); 902 in6_setpeeraddr(in6p, nam); 903 } 904 #endif 905 tcp_debug_trace(so, tp, ostate, PRU_ACCEPT); 906 return 0; 907 } 908 909 static int 910 tcp_bind(struct socket *so, struct mbuf *nam) 911 { 912 struct inpcb *inp; 913 #ifdef INET6 914 struct in6pcb *in6p; 915 #endif 916 struct tcpcb *tp = NULL; 917 int s; 918 int error = 0; 919 int ostate = 0; 920 int family; /* family of the socket */ 921 922 KASSERT(solocked(so)); 923 924 925 s = splsoftnet(); 926 927 family = so->so_proto->pr_domain->dom_family; 928 switch (family) { 929 #ifdef INET 930 case PF_INET: 931 inp = sotoinpcb(so); 932 #ifdef INET6 933 in6p = NULL; 934 #endif 935 break; 936 #endif 937 #ifdef INET6 938 case PF_INET6: 939 inp = NULL; 940 in6p = sotoin6pcb(so); 941 break; 942 #endif 943 default: 944 splx(s); 945 return EAFNOSUPPORT; 946 } 947 948 /* 949 * When a TCP is attached to a socket, then there will be 950 * a (struct inpcb) pointed at by the socket, and this 951 * structure will point at a subsidary (struct tcpcb). 952 */ 953 if (inp == NULL 954 #ifdef INET6 955 && in6p == NULL 956 #endif 957 ) 958 { 959 error = EINVAL; 960 goto release; 961 } 962 #ifdef INET 963 if (inp) { 964 tp = intotcpcb(inp); 965 ostate = tcp_debug_capture(tp, PRU_BIND); 966 } 967 #endif 968 #ifdef INET6 969 if (in6p) { 970 tp = in6totcpcb(in6p); 971 ostate = tcp_debug_capture(tp, PRU_BIND); 972 } 973 #endif 974 975 /* 976 * Give the socket an address. 977 */ 978 switch (family) { 979 #ifdef INET 980 case PF_INET: 981 error = in_pcbbind(inp, nam); 982 break; 983 #endif 984 #ifdef INET6 985 case PF_INET6: 986 error = in6_pcbbind(in6p, nam); 987 if (!error) { 988 /* mapped addr case */ 989 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 990 tp->t_family = AF_INET; 991 else 992 tp->t_family = AF_INET6; 993 } 994 break; 995 #endif 996 } 997 998 tcp_debug_trace(so, tp, ostate, PRU_BIND); 999 1000 release: 1001 splx(s); 1002 return (error); 1003 } 1004 1005 static int 1006 tcp_listen(struct socket *so) 1007 { 1008 struct inpcb *inp; 1009 #ifdef INET6 1010 struct in6pcb *in6p; 1011 #endif 1012 struct tcpcb *tp = NULL; 1013 int s; 1014 int error = 0; 1015 int ostate = 0; 1016 int family; /* family of the socket */ 1017 1018 KASSERT(solocked(so)); 1019 1020 s = splsoftnet(); 1021 1022 family = so->so_proto->pr_domain->dom_family; 1023 switch (family) { 1024 #ifdef INET 1025 case PF_INET: 1026 inp = sotoinpcb(so); 1027 #ifdef INET6 1028 in6p = NULL; 1029 #endif 1030 break; 1031 #endif 1032 #ifdef INET6 1033 case PF_INET6: 1034 inp = NULL; 1035 in6p = sotoin6pcb(so); 1036 break; 1037 #endif 1038 default: 1039 splx(s); 1040 return EAFNOSUPPORT; 1041 } 1042 1043 /* 1044 * When a TCP is attached to a socket, then there will be 1045 * a (struct inpcb) pointed at by the socket, and this 1046 * structure will point at a subsidary (struct tcpcb). 1047 */ 1048 if (inp == NULL 1049 #ifdef INET6 1050 && in6p == NULL 1051 #endif 1052 ) 1053 { 1054 error = EINVAL; 1055 goto release; 1056 } 1057 #ifdef INET 1058 if (inp) { 1059 tp = intotcpcb(inp); 1060 ostate = tcp_debug_capture(tp, PRU_LISTEN); 1061 } 1062 #endif 1063 #ifdef INET6 1064 if (in6p) { 1065 tp = in6totcpcb(in6p); 1066 ostate = tcp_debug_capture(tp, PRU_LISTEN); 1067 } 1068 #endif 1069 1070 /* 1071 * Prepare to accept connections. 1072 */ 1073 #ifdef INET 1074 if (inp && inp->inp_lport == 0) { 1075 error = in_pcbbind(inp, NULL); 1076 if (error) 1077 goto release; 1078 } 1079 #endif 1080 #ifdef INET6 1081 if (in6p && in6p->in6p_lport == 0) { 1082 error = in6_pcbbind(in6p, NULL); 1083 if (error) 1084 goto release; 1085 } 1086 #endif 1087 tp->t_state = TCPS_LISTEN; 1088 1089 tcp_debug_trace(so, tp, ostate, PRU_LISTEN); 1090 1091 release: 1092 splx(s); 1093 return (error); 1094 } 1095 1096 static int 1097 tcp_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp) 1098 { 1099 switch (so->so_proto->pr_domain->dom_family) { 1100 #ifdef INET 1101 case PF_INET: 1102 return in_control(so, cmd, nam, ifp); 1103 #endif 1104 #ifdef INET6 1105 case PF_INET6: 1106 return in6_control(so, cmd, nam, ifp); 1107 #endif 1108 default: 1109 return EAFNOSUPPORT; 1110 } 1111 } 1112 1113 static int 1114 tcp_stat(struct socket *so, struct stat *ub) 1115 { 1116 KASSERT(solocked(so)); 1117 1118 /* stat: don't bother with a blocksize. */ 1119 return 0; 1120 } 1121 1122 static int 1123 tcp_peeraddr(struct socket *so, struct mbuf *nam) 1124 { 1125 struct inpcb *inp = NULL; 1126 #ifdef INET6 1127 struct in6pcb *in6p = NULL; 1128 #endif 1129 struct tcpcb *tp = NULL; 1130 int ostate = 0; 1131 1132 switch (so->so_proto->pr_domain->dom_family) { 1133 #ifdef INET 1134 case PF_INET: 1135 inp = sotoinpcb(so); 1136 break; 1137 #endif 1138 #ifdef INET6 1139 case PF_INET6: 1140 in6p = sotoin6pcb(so); 1141 break; 1142 #endif 1143 default: 1144 return EAFNOSUPPORT; 1145 } 1146 1147 if (inp == NULL 1148 #ifdef INET6 1149 && in6p == NULL 1150 #endif 1151 ) 1152 return EINVAL; 1153 1154 #ifdef INET 1155 if (inp) { 1156 tp = intotcpcb(inp); 1157 ostate = tcp_debug_capture(tp, PRU_PEERADDR); 1158 in_setpeeraddr(inp, nam); 1159 } 1160 #endif 1161 #ifdef INET6 1162 if (in6p) { 1163 tp = in6totcpcb(in6p); 1164 ostate = tcp_debug_capture(tp, PRU_PEERADDR); 1165 in6_setpeeraddr(in6p, nam); 1166 } 1167 #endif 1168 1169 tcp_debug_trace(so, tp, ostate, PRU_PEERADDR); 1170 1171 return 0; 1172 } 1173 1174 static int 1175 tcp_sockaddr(struct socket *so, struct mbuf *nam) 1176 { 1177 struct inpcb *inp = NULL; 1178 #ifdef INET6 1179 struct in6pcb *in6p = NULL; 1180 #endif 1181 struct tcpcb *tp = NULL; 1182 int ostate = 0; 1183 1184 switch (so->so_proto->pr_domain->dom_family) { 1185 #ifdef INET 1186 case PF_INET: 1187 inp = sotoinpcb(so); 1188 break; 1189 #endif 1190 #ifdef INET6 1191 case PF_INET6: 1192 in6p = sotoin6pcb(so); 1193 break; 1194 #endif 1195 default: 1196 return EAFNOSUPPORT; 1197 } 1198 1199 if (inp == NULL 1200 #ifdef INET6 1201 && in6p == NULL 1202 #endif 1203 ) 1204 return EINVAL; 1205 1206 #ifdef INET 1207 if (inp) { 1208 tp = intotcpcb(inp); 1209 ostate = tcp_debug_capture(tp, PRU_SOCKADDR); 1210 in_setsockaddr(inp, nam); 1211 } 1212 #endif 1213 #ifdef INET6 1214 if (in6p) { 1215 tp = in6totcpcb(in6p); 1216 ostate = tcp_debug_capture(tp, PRU_SOCKADDR); 1217 in6_setsockaddr(in6p, nam); 1218 } 1219 #endif 1220 1221 tcp_debug_trace(so, tp, ostate, PRU_SOCKADDR); 1222 1223 return 0; 1224 } 1225 1226 static int 1227 tcp_recvoob(struct socket *so, struct mbuf *m, int flags) 1228 { 1229 struct inpcb *inp = NULL; 1230 #ifdef INET6 1231 struct in6pcb *in6p = NULL; 1232 #endif 1233 struct tcpcb *tp = NULL; 1234 int ostate = 0; 1235 1236 switch (so->so_proto->pr_domain->dom_family) { 1237 #ifdef INET 1238 case PF_INET: 1239 inp = sotoinpcb(so); 1240 break; 1241 #endif 1242 #ifdef INET6 1243 case PF_INET6: 1244 in6p = sotoin6pcb(so); 1245 break; 1246 #endif 1247 default: 1248 return EAFNOSUPPORT; 1249 } 1250 1251 if (inp == NULL 1252 #ifdef INET6 1253 && in6p == NULL 1254 #endif 1255 ) 1256 return EINVAL; 1257 1258 #ifdef INET 1259 if (inp) { 1260 tp = intotcpcb(inp); 1261 ostate = tcp_debug_capture(tp, PRU_RCVOOB); 1262 } 1263 #endif 1264 #ifdef INET6 1265 if (in6p) { 1266 tp = in6totcpcb(in6p); 1267 ostate = tcp_debug_capture(tp, PRU_RCVOOB); 1268 } 1269 #endif 1270 1271 if ((so->so_oobmark == 0 && 1272 (so->so_state & SS_RCVATMARK) == 0) || 1273 so->so_options & SO_OOBINLINE || 1274 tp->t_oobflags & TCPOOB_HADDATA) 1275 return EINVAL; 1276 1277 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) 1278 return EWOULDBLOCK; 1279 1280 m->m_len = 1; 1281 *mtod(m, char *) = tp->t_iobc; 1282 if ((flags & MSG_PEEK) == 0) 1283 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 1284 1285 tcp_debug_trace(so, tp, ostate, PRU_RCVOOB); 1286 1287 return 0; 1288 } 1289 1290 static int 1291 tcp_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control) 1292 { 1293 struct inpcb *inp = NULL; 1294 #ifdef INET6 1295 struct in6pcb *in6p = NULL; 1296 #endif 1297 struct tcpcb *tp = NULL; 1298 int ostate = 0; 1299 int error = 0; 1300 1301 switch (so->so_proto->pr_domain->dom_family) { 1302 #ifdef INET 1303 case PF_INET: 1304 inp = sotoinpcb(so); 1305 break; 1306 #endif 1307 #ifdef INET6 1308 case PF_INET6: 1309 in6p = sotoin6pcb(so); 1310 break; 1311 #endif 1312 default: 1313 return EAFNOSUPPORT; 1314 } 1315 1316 if (inp == NULL 1317 #ifdef INET6 1318 && in6p == NULL 1319 #endif 1320 ) 1321 return EINVAL; 1322 1323 #ifdef INET 1324 if (inp) { 1325 tp = intotcpcb(inp); 1326 ostate = tcp_debug_capture(tp, PRU_SENDOOB); 1327 } 1328 #endif 1329 #ifdef INET6 1330 if (in6p) { 1331 tp = in6totcpcb(in6p); 1332 ostate = tcp_debug_capture(tp, PRU_SENDOOB); 1333 } 1334 #endif 1335 1336 if (sbspace(&so->so_snd) < -512) { 1337 m_freem(m); 1338 return ENOBUFS; 1339 } 1340 /* 1341 * According to RFC961 (Assigned Protocols), 1342 * the urgent pointer points to the last octet 1343 * of urgent data. We continue, however, 1344 * to consider it to indicate the first octet 1345 * of data past the urgent section. 1346 * Otherwise, snd_up should be one lower. 1347 */ 1348 sbappendstream(&so->so_snd, m); 1349 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 1350 tp->t_force = 1; 1351 error = tcp_output(tp); 1352 tp->t_force = 0; 1353 1354 tcp_debug_trace(so, tp, ostate, PRU_SENDOOB); 1355 1356 return error; 1357 } 1358 1359 /* 1360 * Initiate (or continue) disconnect. 1361 * If embryonic state, just send reset (once). 1362 * If in ``let data drain'' option and linger null, just drop. 1363 * Otherwise (hard), mark socket disconnecting and drop 1364 * current input data; switch states based on user close, and 1365 * send segment to peer (with FIN). 1366 */ 1367 struct tcpcb * 1368 tcp_disconnect(struct tcpcb *tp) 1369 { 1370 struct socket *so; 1371 1372 if (tp->t_inpcb) 1373 so = tp->t_inpcb->inp_socket; 1374 #ifdef INET6 1375 else if (tp->t_in6pcb) 1376 so = tp->t_in6pcb->in6p_socket; 1377 #endif 1378 else 1379 so = NULL; 1380 1381 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 1382 tp = tcp_close(tp); 1383 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 1384 tp = tcp_drop(tp, 0); 1385 else { 1386 soisdisconnecting(so); 1387 sbflush(&so->so_rcv); 1388 tp = tcp_usrclosed(tp); 1389 if (tp) 1390 (void) tcp_output(tp); 1391 } 1392 return (tp); 1393 } 1394 1395 /* 1396 * User issued close, and wish to trail through shutdown states: 1397 * if never received SYN, just forget it. If got a SYN from peer, 1398 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 1399 * If already got a FIN from peer, then almost done; go to LAST_ACK 1400 * state. In all other cases, have already sent FIN to peer (e.g. 1401 * after PRU_SHUTDOWN), and just have to play tedious game waiting 1402 * for peer to send FIN or not respond to keep-alives, etc. 1403 * We can let the user exit from the close as soon as the FIN is acked. 1404 */ 1405 struct tcpcb * 1406 tcp_usrclosed(struct tcpcb *tp) 1407 { 1408 1409 switch (tp->t_state) { 1410 1411 case TCPS_CLOSED: 1412 case TCPS_LISTEN: 1413 case TCPS_SYN_SENT: 1414 tp->t_state = TCPS_CLOSED; 1415 tp = tcp_close(tp); 1416 break; 1417 1418 case TCPS_SYN_RECEIVED: 1419 case TCPS_ESTABLISHED: 1420 tp->t_state = TCPS_FIN_WAIT_1; 1421 break; 1422 1423 case TCPS_CLOSE_WAIT: 1424 tp->t_state = TCPS_LAST_ACK; 1425 break; 1426 } 1427 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1428 struct socket *so; 1429 if (tp->t_inpcb) 1430 so = tp->t_inpcb->inp_socket; 1431 #ifdef INET6 1432 else if (tp->t_in6pcb) 1433 so = tp->t_in6pcb->in6p_socket; 1434 #endif 1435 else 1436 so = NULL; 1437 if (so) 1438 soisdisconnected(so); 1439 /* 1440 * If we are in FIN_WAIT_2, we arrived here because the 1441 * application did a shutdown of the send side. Like the 1442 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1443 * a full close, we start a timer to make sure sockets are 1444 * not left in FIN_WAIT_2 forever. 1445 */ 1446 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1447 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1448 else if (tp->t_state == TCPS_TIME_WAIT 1449 && ((tp->t_inpcb 1450 && (tcp4_vtw_enable & 1) 1451 && vtw_add(AF_INET, tp)) 1452 || 1453 (tp->t_in6pcb 1454 && (tcp6_vtw_enable & 1) 1455 && vtw_add(AF_INET6, tp)))) { 1456 tp = 0; 1457 } 1458 } 1459 return (tp); 1460 } 1461 1462 /* 1463 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1464 * than 32. 1465 */ 1466 static int 1467 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1468 { 1469 int error, mssdflt; 1470 struct sysctlnode node; 1471 1472 mssdflt = tcp_mssdflt; 1473 node = *rnode; 1474 node.sysctl_data = &mssdflt; 1475 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1476 if (error || newp == NULL) 1477 return (error); 1478 1479 if (mssdflt < 32) 1480 return (EINVAL); 1481 tcp_mssdflt = mssdflt; 1482 1483 mutex_enter(softnet_lock); 1484 tcp_tcpcb_template(); 1485 mutex_exit(softnet_lock); 1486 1487 return (0); 1488 } 1489 1490 /* 1491 * sysctl helper for TCP CB template update 1492 */ 1493 static int 1494 sysctl_update_tcpcb_template(SYSCTLFN_ARGS) 1495 { 1496 int t, error; 1497 struct sysctlnode node; 1498 1499 /* follow procedures in sysctl(9) manpage */ 1500 t = *(int *)rnode->sysctl_data; 1501 node = *rnode; 1502 node.sysctl_data = &t; 1503 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1504 if (error || newp == NULL) 1505 return error; 1506 1507 if (t < 0) 1508 return EINVAL; 1509 1510 *(int *)rnode->sysctl_data = t; 1511 1512 mutex_enter(softnet_lock); 1513 tcp_tcpcb_template(); 1514 mutex_exit(softnet_lock); 1515 1516 return 0; 1517 } 1518 1519 /* 1520 * sysctl helper routine for setting port related values under 1521 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1522 * additional checks for each type. this code has placed in 1523 * tcp_input.c since INET and INET6 both use the same tcp code. 1524 * 1525 * this helper is not static so that both inet and inet6 can use it. 1526 */ 1527 int 1528 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1529 { 1530 int error, tmp; 1531 int apmin, apmax; 1532 #ifndef IPNOPRIVPORTS 1533 int lpmin, lpmax; 1534 #endif /* IPNOPRIVPORTS */ 1535 struct sysctlnode node; 1536 1537 if (namelen != 0) 1538 return (EINVAL); 1539 1540 switch (name[-3]) { 1541 #ifdef INET 1542 case PF_INET: 1543 apmin = anonportmin; 1544 apmax = anonportmax; 1545 #ifndef IPNOPRIVPORTS 1546 lpmin = lowportmin; 1547 lpmax = lowportmax; 1548 #endif /* IPNOPRIVPORTS */ 1549 break; 1550 #endif /* INET */ 1551 #ifdef INET6 1552 case PF_INET6: 1553 apmin = ip6_anonportmin; 1554 apmax = ip6_anonportmax; 1555 #ifndef IPNOPRIVPORTS 1556 lpmin = ip6_lowportmin; 1557 lpmax = ip6_lowportmax; 1558 #endif /* IPNOPRIVPORTS */ 1559 break; 1560 #endif /* INET6 */ 1561 default: 1562 return (EINVAL); 1563 } 1564 1565 /* 1566 * insert temporary copy into node, perform lookup on 1567 * temporary, then restore pointer 1568 */ 1569 node = *rnode; 1570 tmp = *(int*)rnode->sysctl_data; 1571 node.sysctl_data = &tmp; 1572 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1573 if (error || newp == NULL) 1574 return (error); 1575 1576 /* 1577 * simple port range check 1578 */ 1579 if (tmp < 0 || tmp > 65535) 1580 return (EINVAL); 1581 1582 /* 1583 * per-node range checks 1584 */ 1585 switch (rnode->sysctl_num) { 1586 case IPCTL_ANONPORTMIN: 1587 case IPV6CTL_ANONPORTMIN: 1588 if (tmp >= apmax) 1589 return (EINVAL); 1590 #ifndef IPNOPRIVPORTS 1591 if (tmp < IPPORT_RESERVED) 1592 return (EINVAL); 1593 #endif /* IPNOPRIVPORTS */ 1594 break; 1595 1596 case IPCTL_ANONPORTMAX: 1597 case IPV6CTL_ANONPORTMAX: 1598 if (apmin >= tmp) 1599 return (EINVAL); 1600 #ifndef IPNOPRIVPORTS 1601 if (tmp < IPPORT_RESERVED) 1602 return (EINVAL); 1603 #endif /* IPNOPRIVPORTS */ 1604 break; 1605 1606 #ifndef IPNOPRIVPORTS 1607 case IPCTL_LOWPORTMIN: 1608 case IPV6CTL_LOWPORTMIN: 1609 if (tmp >= lpmax || 1610 tmp > IPPORT_RESERVEDMAX || 1611 tmp < IPPORT_RESERVEDMIN) 1612 return (EINVAL); 1613 break; 1614 1615 case IPCTL_LOWPORTMAX: 1616 case IPV6CTL_LOWPORTMAX: 1617 if (lpmin >= tmp || 1618 tmp > IPPORT_RESERVEDMAX || 1619 tmp < IPPORT_RESERVEDMIN) 1620 return (EINVAL); 1621 break; 1622 #endif /* IPNOPRIVPORTS */ 1623 1624 default: 1625 return (EINVAL); 1626 } 1627 1628 *(int*)rnode->sysctl_data = tmp; 1629 1630 return (0); 1631 } 1632 1633 static inline int 1634 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1635 { 1636 if (oldp) { 1637 size_t sz; 1638 uid_t uid; 1639 int error; 1640 1641 if (sockp->so_cred == NULL) 1642 return EPERM; 1643 1644 uid = kauth_cred_geteuid(sockp->so_cred); 1645 sz = MIN(sizeof(uid), *oldlenp); 1646 if ((error = copyout(&uid, oldp, sz)) != 0) 1647 return error; 1648 } 1649 *oldlenp = sizeof(uid_t); 1650 return 0; 1651 } 1652 1653 static inline int 1654 inet4_ident_core(struct in_addr raddr, u_int rport, 1655 struct in_addr laddr, u_int lport, 1656 void *oldp, size_t *oldlenp, 1657 struct lwp *l, int dodrop) 1658 { 1659 struct inpcb *inp; 1660 struct socket *sockp; 1661 1662 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1663 1664 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1665 return ESRCH; 1666 1667 if (dodrop) { 1668 struct tcpcb *tp; 1669 int error; 1670 1671 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1672 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1673 return ESRCH; 1674 1675 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1676 KAUTH_REQ_NETWORK_SOCKET_DROP, inp->inp_socket, tp, NULL); 1677 if (error) 1678 return (error); 1679 1680 (void)tcp_drop(tp, ECONNABORTED); 1681 return 0; 1682 } 1683 else 1684 return copyout_uid(sockp, oldp, oldlenp); 1685 } 1686 1687 #ifdef INET6 1688 static inline int 1689 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1690 struct in6_addr *laddr, u_int lport, 1691 void *oldp, size_t *oldlenp, 1692 struct lwp *l, int dodrop) 1693 { 1694 struct in6pcb *in6p; 1695 struct socket *sockp; 1696 1697 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0, 0); 1698 1699 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1700 return ESRCH; 1701 1702 if (dodrop) { 1703 struct tcpcb *tp; 1704 int error; 1705 1706 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1707 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1708 return ESRCH; 1709 1710 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1711 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1712 if (error) 1713 return (error); 1714 1715 (void)tcp_drop(tp, ECONNABORTED); 1716 return 0; 1717 } 1718 else 1719 return copyout_uid(sockp, oldp, oldlenp); 1720 } 1721 #endif 1722 1723 /* 1724 * sysctl helper routine for the net.inet.tcp.drop and 1725 * net.inet6.tcp6.drop nodes. 1726 */ 1727 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1728 1729 /* 1730 * sysctl helper routine for the net.inet.tcp.ident and 1731 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1732 * old way of looking up the ident information for ipv4 which involves 1733 * stuffing the port/addr pairs into the mib lookup. 1734 */ 1735 static int 1736 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1737 { 1738 #ifdef INET 1739 struct sockaddr_in *si4[2]; 1740 #endif /* INET */ 1741 #ifdef INET6 1742 struct sockaddr_in6 *si6[2]; 1743 #endif /* INET6 */ 1744 struct sockaddr_storage sa[2]; 1745 int error, pf, dodrop; 1746 1747 dodrop = name[-1] == TCPCTL_DROP; 1748 if (dodrop) { 1749 if (oldp != NULL || *oldlenp != 0) 1750 return EINVAL; 1751 if (newp == NULL) 1752 return EPERM; 1753 if (newlen < sizeof(sa)) 1754 return ENOMEM; 1755 } 1756 if (namelen != 4 && namelen != 0) 1757 return EINVAL; 1758 if (name[-2] != IPPROTO_TCP) 1759 return EINVAL; 1760 pf = name[-3]; 1761 1762 /* old style lookup, ipv4 only */ 1763 if (namelen == 4) { 1764 #ifdef INET 1765 struct in_addr laddr, raddr; 1766 u_int lport, rport; 1767 1768 if (pf != PF_INET) 1769 return EPROTONOSUPPORT; 1770 raddr.s_addr = (uint32_t)name[0]; 1771 rport = (u_int)name[1]; 1772 laddr.s_addr = (uint32_t)name[2]; 1773 lport = (u_int)name[3]; 1774 1775 mutex_enter(softnet_lock); 1776 error = inet4_ident_core(raddr, rport, laddr, lport, 1777 oldp, oldlenp, l, dodrop); 1778 mutex_exit(softnet_lock); 1779 return error; 1780 #else /* INET */ 1781 return EINVAL; 1782 #endif /* INET */ 1783 } 1784 1785 if (newp == NULL || newlen != sizeof(sa)) 1786 return EINVAL; 1787 error = copyin(newp, &sa, newlen); 1788 if (error) 1789 return error; 1790 1791 /* 1792 * requested families must match 1793 */ 1794 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1795 return EINVAL; 1796 1797 switch (pf) { 1798 #ifdef INET6 1799 case PF_INET6: 1800 si6[0] = (struct sockaddr_in6*)&sa[0]; 1801 si6[1] = (struct sockaddr_in6*)&sa[1]; 1802 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1803 si6[1]->sin6_len != sizeof(*si6[1])) 1804 return EINVAL; 1805 1806 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1807 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1808 error = sa6_embedscope(si6[0], ip6_use_defzone); 1809 if (error) 1810 return error; 1811 error = sa6_embedscope(si6[1], ip6_use_defzone); 1812 if (error) 1813 return error; 1814 1815 mutex_enter(softnet_lock); 1816 error = inet6_ident_core(&si6[0]->sin6_addr, 1817 si6[0]->sin6_port, &si6[1]->sin6_addr, 1818 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1819 mutex_exit(softnet_lock); 1820 return error; 1821 } 1822 1823 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1824 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1825 return EINVAL; 1826 1827 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1828 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1829 /*FALLTHROUGH*/ 1830 #endif /* INET6 */ 1831 #ifdef INET 1832 case PF_INET: 1833 si4[0] = (struct sockaddr_in*)&sa[0]; 1834 si4[1] = (struct sockaddr_in*)&sa[1]; 1835 if (si4[0]->sin_len != sizeof(*si4[0]) || 1836 si4[0]->sin_len != sizeof(*si4[1])) 1837 return EINVAL; 1838 1839 mutex_enter(softnet_lock); 1840 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1841 si4[1]->sin_addr, si4[1]->sin_port, 1842 oldp, oldlenp, l, dodrop); 1843 mutex_exit(softnet_lock); 1844 return error; 1845 #endif /* INET */ 1846 default: 1847 return EPROTONOSUPPORT; 1848 } 1849 } 1850 1851 /* 1852 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1853 * inet/inet6, as well as raw pcbs for each. specifically not 1854 * declared static so that raw sockets and udp/udp6 can use it as 1855 * well. 1856 */ 1857 int 1858 sysctl_inpcblist(SYSCTLFN_ARGS) 1859 { 1860 #ifdef INET 1861 struct sockaddr_in *in; 1862 const struct inpcb *inp; 1863 #endif 1864 #ifdef INET6 1865 struct sockaddr_in6 *in6; 1866 const struct in6pcb *in6p; 1867 #endif 1868 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1869 const struct inpcb_hdr *inph; 1870 struct tcpcb *tp; 1871 struct kinfo_pcb pcb; 1872 char *dp; 1873 size_t len, needed, elem_size, out_size; 1874 int error, elem_count, pf, proto, pf2; 1875 1876 if (namelen != 4) 1877 return (EINVAL); 1878 1879 if (oldp != NULL) { 1880 len = *oldlenp; 1881 elem_size = name[2]; 1882 elem_count = name[3]; 1883 if (elem_size != sizeof(pcb)) 1884 return EINVAL; 1885 } else { 1886 len = 0; 1887 elem_count = INT_MAX; 1888 elem_size = sizeof(pcb); 1889 } 1890 error = 0; 1891 dp = oldp; 1892 out_size = elem_size; 1893 needed = 0; 1894 1895 if (namelen == 1 && name[0] == CTL_QUERY) 1896 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1897 1898 if (name - oname != 4) 1899 return (EINVAL); 1900 1901 pf = oname[1]; 1902 proto = oname[2]; 1903 pf2 = (oldp != NULL) ? pf : 0; 1904 1905 mutex_enter(softnet_lock); 1906 1907 TAILQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1908 #ifdef INET 1909 inp = (const struct inpcb *)inph; 1910 #endif 1911 #ifdef INET6 1912 in6p = (const struct in6pcb *)inph; 1913 #endif 1914 1915 if (inph->inph_af != pf) 1916 continue; 1917 1918 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1919 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1920 NULL) != 0) 1921 continue; 1922 1923 memset(&pcb, 0, sizeof(pcb)); 1924 1925 pcb.ki_family = pf; 1926 pcb.ki_type = proto; 1927 1928 switch (pf2) { 1929 case 0: 1930 /* just probing for size */ 1931 break; 1932 #ifdef INET 1933 case PF_INET: 1934 pcb.ki_family = inp->inp_socket->so_proto-> 1935 pr_domain->dom_family; 1936 pcb.ki_type = inp->inp_socket->so_proto-> 1937 pr_type; 1938 pcb.ki_protocol = inp->inp_socket->so_proto-> 1939 pr_protocol; 1940 pcb.ki_pflags = inp->inp_flags; 1941 1942 pcb.ki_sostate = inp->inp_socket->so_state; 1943 pcb.ki_prstate = inp->inp_state; 1944 if (proto == IPPROTO_TCP) { 1945 tp = intotcpcb(inp); 1946 pcb.ki_tstate = tp->t_state; 1947 pcb.ki_tflags = tp->t_flags; 1948 } 1949 1950 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1951 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1952 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1953 1954 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1955 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1956 1957 in = satosin(&pcb.ki_src); 1958 in->sin_len = sizeof(*in); 1959 in->sin_family = pf; 1960 in->sin_port = inp->inp_lport; 1961 in->sin_addr = inp->inp_laddr; 1962 if (pcb.ki_prstate >= INP_CONNECTED) { 1963 in = satosin(&pcb.ki_dst); 1964 in->sin_len = sizeof(*in); 1965 in->sin_family = pf; 1966 in->sin_port = inp->inp_fport; 1967 in->sin_addr = inp->inp_faddr; 1968 } 1969 break; 1970 #endif 1971 #ifdef INET6 1972 case PF_INET6: 1973 pcb.ki_family = in6p->in6p_socket->so_proto-> 1974 pr_domain->dom_family; 1975 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1976 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1977 pr_protocol; 1978 pcb.ki_pflags = in6p->in6p_flags; 1979 1980 pcb.ki_sostate = in6p->in6p_socket->so_state; 1981 pcb.ki_prstate = in6p->in6p_state; 1982 if (proto == IPPROTO_TCP) { 1983 tp = in6totcpcb(in6p); 1984 pcb.ki_tstate = tp->t_state; 1985 pcb.ki_tflags = tp->t_flags; 1986 } 1987 1988 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1989 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1990 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1991 1992 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1993 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1994 1995 in6 = satosin6(&pcb.ki_src); 1996 in6->sin6_len = sizeof(*in6); 1997 in6->sin6_family = pf; 1998 in6->sin6_port = in6p->in6p_lport; 1999 in6->sin6_flowinfo = in6p->in6p_flowinfo; 2000 in6->sin6_addr = in6p->in6p_laddr; 2001 in6->sin6_scope_id = 0; /* XXX? */ 2002 2003 if (pcb.ki_prstate >= IN6P_CONNECTED) { 2004 in6 = satosin6(&pcb.ki_dst); 2005 in6->sin6_len = sizeof(*in6); 2006 in6->sin6_family = pf; 2007 in6->sin6_port = in6p->in6p_fport; 2008 in6->sin6_flowinfo = in6p->in6p_flowinfo; 2009 in6->sin6_addr = in6p->in6p_faddr; 2010 in6->sin6_scope_id = 0; /* XXX? */ 2011 } 2012 break; 2013 #endif 2014 } 2015 2016 if (len >= elem_size && elem_count > 0) { 2017 error = copyout(&pcb, dp, out_size); 2018 if (error) { 2019 mutex_exit(softnet_lock); 2020 return (error); 2021 } 2022 dp += elem_size; 2023 len -= elem_size; 2024 } 2025 needed += elem_size; 2026 if (elem_count > 0 && elem_count != INT_MAX) 2027 elem_count--; 2028 } 2029 2030 *oldlenp = needed; 2031 if (oldp == NULL) 2032 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 2033 2034 mutex_exit(softnet_lock); 2035 2036 return (error); 2037 } 2038 2039 static int 2040 sysctl_tcp_congctl(SYSCTLFN_ARGS) 2041 { 2042 struct sysctlnode node; 2043 int error; 2044 char newname[TCPCC_MAXLEN]; 2045 2046 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 2047 2048 node = *rnode; 2049 node.sysctl_data = newname; 2050 node.sysctl_size = sizeof(newname); 2051 2052 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2053 2054 if (error || 2055 newp == NULL || 2056 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 2057 return error; 2058 2059 mutex_enter(softnet_lock); 2060 error = tcp_congctl_select(NULL, newname); 2061 mutex_exit(softnet_lock); 2062 2063 return error; 2064 } 2065 2066 static int 2067 sysctl_tcp_init_win(SYSCTLFN_ARGS) 2068 { 2069 int error; 2070 u_int iw; 2071 struct sysctlnode node; 2072 2073 iw = *(u_int *)rnode->sysctl_data; 2074 node = *rnode; 2075 node.sysctl_data = &iw; 2076 node.sysctl_size = sizeof(iw); 2077 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2078 if (error || newp == NULL) 2079 return error; 2080 2081 if (iw >= __arraycount(tcp_init_win_max)) 2082 return EINVAL; 2083 *(u_int *)rnode->sysctl_data = iw; 2084 return 0; 2085 } 2086 2087 static int 2088 sysctl_tcp_keep(SYSCTLFN_ARGS) 2089 { 2090 int error; 2091 u_int tmp; 2092 struct sysctlnode node; 2093 2094 node = *rnode; 2095 tmp = *(u_int *)rnode->sysctl_data; 2096 node.sysctl_data = &tmp; 2097 2098 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2099 if (error || newp == NULL) 2100 return error; 2101 2102 mutex_enter(softnet_lock); 2103 2104 *(u_int *)rnode->sysctl_data = tmp; 2105 tcp_tcpcb_template(); /* update the template */ 2106 2107 mutex_exit(softnet_lock); 2108 return 0; 2109 } 2110 2111 static int 2112 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 2113 { 2114 2115 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 2116 } 2117 2118 /* 2119 * this (second stage) setup routine is a replacement for tcp_sysctl() 2120 * (which is currently used for ipv4 and ipv6) 2121 */ 2122 static void 2123 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 2124 const char *tcpname) 2125 { 2126 const struct sysctlnode *sack_node; 2127 const struct sysctlnode *abc_node; 2128 const struct sysctlnode *ecn_node; 2129 const struct sysctlnode *congctl_node; 2130 const struct sysctlnode *mslt_node; 2131 const struct sysctlnode *vtw_node; 2132 #ifdef TCP_DEBUG 2133 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 2134 extern int tcp_debx; 2135 #endif 2136 2137 sysctl_createv(clog, 0, NULL, NULL, 2138 CTLFLAG_PERMANENT, 2139 CTLTYPE_NODE, pfname, NULL, 2140 NULL, 0, NULL, 0, 2141 CTL_NET, pf, CTL_EOL); 2142 sysctl_createv(clog, 0, NULL, NULL, 2143 CTLFLAG_PERMANENT, 2144 CTLTYPE_NODE, tcpname, 2145 SYSCTL_DESCR("TCP related settings"), 2146 NULL, 0, NULL, 0, 2147 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 2148 2149 sysctl_createv(clog, 0, NULL, NULL, 2150 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2151 CTLTYPE_INT, "rfc1323", 2152 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 2153 sysctl_update_tcpcb_template, 0, &tcp_do_rfc1323, 0, 2154 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 2155 sysctl_createv(clog, 0, NULL, NULL, 2156 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2157 CTLTYPE_INT, "sendspace", 2158 SYSCTL_DESCR("Default TCP send buffer size"), 2159 NULL, 0, &tcp_sendspace, 0, 2160 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 2161 sysctl_createv(clog, 0, NULL, NULL, 2162 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2163 CTLTYPE_INT, "recvspace", 2164 SYSCTL_DESCR("Default TCP receive buffer size"), 2165 NULL, 0, &tcp_recvspace, 0, 2166 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 2167 sysctl_createv(clog, 0, NULL, NULL, 2168 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2169 CTLTYPE_INT, "mssdflt", 2170 SYSCTL_DESCR("Default maximum segment size"), 2171 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 2172 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 2173 sysctl_createv(clog, 0, NULL, NULL, 2174 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2175 CTLTYPE_INT, "minmss", 2176 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 2177 NULL, 0, &tcp_minmss, 0, 2178 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2179 sysctl_createv(clog, 0, NULL, NULL, 2180 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2181 CTLTYPE_INT, "msl", 2182 SYSCTL_DESCR("Maximum Segment Life"), 2183 NULL, 0, &tcp_msl, 0, 2184 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 2185 sysctl_createv(clog, 0, NULL, NULL, 2186 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2187 CTLTYPE_INT, "syn_cache_limit", 2188 SYSCTL_DESCR("Maximum number of entries in the TCP " 2189 "compressed state engine"), 2190 NULL, 0, &tcp_syn_cache_limit, 0, 2191 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 2192 CTL_EOL); 2193 sysctl_createv(clog, 0, NULL, NULL, 2194 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2195 CTLTYPE_INT, "syn_bucket_limit", 2196 SYSCTL_DESCR("Maximum number of entries per hash " 2197 "bucket in the TCP compressed state " 2198 "engine"), 2199 NULL, 0, &tcp_syn_bucket_limit, 0, 2200 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 2201 CTL_EOL); 2202 #if 0 /* obsoleted */ 2203 sysctl_createv(clog, 0, NULL, NULL, 2204 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2205 CTLTYPE_INT, "syn_cache_interval", 2206 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 2207 NULL, 0, &tcp_syn_cache_interval, 0, 2208 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 2209 CTL_EOL); 2210 #endif 2211 sysctl_createv(clog, 0, NULL, NULL, 2212 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2213 CTLTYPE_INT, "init_win", 2214 SYSCTL_DESCR("Initial TCP congestion window"), 2215 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 2216 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 2217 sysctl_createv(clog, 0, NULL, NULL, 2218 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2219 CTLTYPE_INT, "mss_ifmtu", 2220 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 2221 NULL, 0, &tcp_mss_ifmtu, 0, 2222 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 2223 sysctl_createv(clog, 0, NULL, &sack_node, 2224 CTLFLAG_PERMANENT, 2225 CTLTYPE_NODE, "sack", 2226 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 2227 NULL, 0, NULL, 0, 2228 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 2229 2230 /* Congctl subtree */ 2231 sysctl_createv(clog, 0, NULL, &congctl_node, 2232 CTLFLAG_PERMANENT, 2233 CTLTYPE_NODE, "congctl", 2234 SYSCTL_DESCR("TCP Congestion Control"), 2235 NULL, 0, NULL, 0, 2236 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2237 sysctl_createv(clog, 0, &congctl_node, NULL, 2238 CTLFLAG_PERMANENT, 2239 CTLTYPE_STRING, "available", 2240 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 2241 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 2242 sysctl_createv(clog, 0, &congctl_node, NULL, 2243 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2244 CTLTYPE_STRING, "selected", 2245 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 2246 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 2247 CTL_CREATE, CTL_EOL); 2248 2249 sysctl_createv(clog, 0, NULL, NULL, 2250 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2251 CTLTYPE_INT, "win_scale", 2252 SYSCTL_DESCR("Use RFC1323 window scale options"), 2253 sysctl_update_tcpcb_template, 0, &tcp_do_win_scale, 0, 2254 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 2255 sysctl_createv(clog, 0, NULL, NULL, 2256 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2257 CTLTYPE_INT, "timestamps", 2258 SYSCTL_DESCR("Use RFC1323 time stamp options"), 2259 sysctl_update_tcpcb_template, 0, &tcp_do_timestamps, 0, 2260 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 2261 sysctl_createv(clog, 0, NULL, NULL, 2262 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2263 CTLTYPE_INT, "compat_42", 2264 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 2265 NULL, 0, &tcp_compat_42, 0, 2266 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 2267 sysctl_createv(clog, 0, NULL, NULL, 2268 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2269 CTLTYPE_INT, "cwm", 2270 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 2271 "Monitoring"), 2272 NULL, 0, &tcp_cwm, 0, 2273 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 2274 sysctl_createv(clog, 0, NULL, NULL, 2275 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2276 CTLTYPE_INT, "cwm_burstsize", 2277 SYSCTL_DESCR("Congestion Window Monitoring allowed " 2278 "burst count in packets"), 2279 NULL, 0, &tcp_cwm_burstsize, 0, 2280 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 2281 CTL_EOL); 2282 sysctl_createv(clog, 0, NULL, NULL, 2283 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2284 CTLTYPE_INT, "ack_on_push", 2285 SYSCTL_DESCR("Immediately return ACK when PSH is " 2286 "received"), 2287 NULL, 0, &tcp_ack_on_push, 0, 2288 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 2289 sysctl_createv(clog, 0, NULL, NULL, 2290 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2291 CTLTYPE_INT, "keepidle", 2292 SYSCTL_DESCR("Allowed connection idle ticks before a " 2293 "keepalive probe is sent"), 2294 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 2295 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 2296 sysctl_createv(clog, 0, NULL, NULL, 2297 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2298 CTLTYPE_INT, "keepintvl", 2299 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 2300 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 2301 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 2302 sysctl_createv(clog, 0, NULL, NULL, 2303 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2304 CTLTYPE_INT, "keepcnt", 2305 SYSCTL_DESCR("Number of keepalive probes to send"), 2306 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 2307 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 2308 sysctl_createv(clog, 0, NULL, NULL, 2309 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 2310 CTLTYPE_INT, "slowhz", 2311 SYSCTL_DESCR("Keepalive ticks per second"), 2312 NULL, PR_SLOWHZ, NULL, 0, 2313 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 2314 sysctl_createv(clog, 0, NULL, NULL, 2315 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2316 CTLTYPE_INT, "log_refused", 2317 SYSCTL_DESCR("Log refused TCP connections"), 2318 NULL, 0, &tcp_log_refused, 0, 2319 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 2320 #if 0 /* obsoleted */ 2321 sysctl_createv(clog, 0, NULL, NULL, 2322 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2323 CTLTYPE_INT, "rstratelimit", NULL, 2324 NULL, 0, &tcp_rst_ratelim, 0, 2325 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 2326 #endif 2327 sysctl_createv(clog, 0, NULL, NULL, 2328 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2329 CTLTYPE_INT, "rstppslimit", 2330 SYSCTL_DESCR("Maximum number of RST packets to send " 2331 "per second"), 2332 NULL, 0, &tcp_rst_ppslim, 0, 2333 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 2334 sysctl_createv(clog, 0, NULL, NULL, 2335 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2336 CTLTYPE_INT, "delack_ticks", 2337 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 2338 NULL, 0, &tcp_delack_ticks, 0, 2339 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 2340 sysctl_createv(clog, 0, NULL, NULL, 2341 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2342 CTLTYPE_INT, "init_win_local", 2343 SYSCTL_DESCR("Initial TCP window size (in segments)"), 2344 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 2345 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 2346 CTL_EOL); 2347 sysctl_createv(clog, 0, NULL, NULL, 2348 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2349 CTLTYPE_STRUCT, "ident", 2350 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 2351 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 2352 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 2353 sysctl_createv(clog, 0, NULL, NULL, 2354 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2355 CTLTYPE_INT, "do_loopback_cksum", 2356 SYSCTL_DESCR("Perform TCP checksum on loopback"), 2357 NULL, 0, &tcp_do_loopback_cksum, 0, 2358 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 2359 CTL_EOL); 2360 sysctl_createv(clog, 0, NULL, NULL, 2361 CTLFLAG_PERMANENT, 2362 CTLTYPE_STRUCT, "pcblist", 2363 SYSCTL_DESCR("TCP protocol control block list"), 2364 sysctl_inpcblist, 0, &tcbtable, 0, 2365 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2366 CTL_EOL); 2367 sysctl_createv(clog, 0, NULL, NULL, 2368 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2369 CTLTYPE_INT, "keepinit", 2370 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 2371 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 2372 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2373 2374 /* TCP socket buffers auto-sizing nodes */ 2375 sysctl_createv(clog, 0, NULL, NULL, 2376 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2377 CTLTYPE_INT, "recvbuf_auto", 2378 SYSCTL_DESCR("Enable automatic receive " 2379 "buffer sizing (experimental)"), 2380 NULL, 0, &tcp_do_autorcvbuf, 0, 2381 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2382 sysctl_createv(clog, 0, NULL, NULL, 2383 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2384 CTLTYPE_INT, "recvbuf_inc", 2385 SYSCTL_DESCR("Incrementor step size of " 2386 "automatic receive buffer"), 2387 NULL, 0, &tcp_autorcvbuf_inc, 0, 2388 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2389 sysctl_createv(clog, 0, NULL, NULL, 2390 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2391 CTLTYPE_INT, "recvbuf_max", 2392 SYSCTL_DESCR("Max size of automatic receive buffer"), 2393 NULL, 0, &tcp_autorcvbuf_max, 0, 2394 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2395 2396 sysctl_createv(clog, 0, NULL, NULL, 2397 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2398 CTLTYPE_INT, "sendbuf_auto", 2399 SYSCTL_DESCR("Enable automatic send " 2400 "buffer sizing (experimental)"), 2401 NULL, 0, &tcp_do_autosndbuf, 0, 2402 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2403 sysctl_createv(clog, 0, NULL, NULL, 2404 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2405 CTLTYPE_INT, "sendbuf_inc", 2406 SYSCTL_DESCR("Incrementor step size of " 2407 "automatic send buffer"), 2408 NULL, 0, &tcp_autosndbuf_inc, 0, 2409 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2410 sysctl_createv(clog, 0, NULL, NULL, 2411 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2412 CTLTYPE_INT, "sendbuf_max", 2413 SYSCTL_DESCR("Max size of automatic send buffer"), 2414 NULL, 0, &tcp_autosndbuf_max, 0, 2415 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2416 2417 /* ECN subtree */ 2418 sysctl_createv(clog, 0, NULL, &ecn_node, 2419 CTLFLAG_PERMANENT, 2420 CTLTYPE_NODE, "ecn", 2421 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 2422 NULL, 0, NULL, 0, 2423 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2424 sysctl_createv(clog, 0, &ecn_node, NULL, 2425 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2426 CTLTYPE_INT, "enable", 2427 SYSCTL_DESCR("Enable TCP Explicit Congestion " 2428 "Notification"), 2429 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 2430 sysctl_createv(clog, 0, &ecn_node, NULL, 2431 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2432 CTLTYPE_INT, "maxretries", 2433 SYSCTL_DESCR("Number of times to retry ECN setup " 2434 "before disabling ECN on the connection"), 2435 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 2436 2437 /* SACK gets it's own little subtree. */ 2438 sysctl_createv(clog, 0, NULL, &sack_node, 2439 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2440 CTLTYPE_INT, "enable", 2441 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 2442 NULL, 0, &tcp_do_sack, 0, 2443 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2444 sysctl_createv(clog, 0, NULL, &sack_node, 2445 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2446 CTLTYPE_INT, "maxholes", 2447 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 2448 NULL, 0, &tcp_sack_tp_maxholes, 0, 2449 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2450 sysctl_createv(clog, 0, NULL, &sack_node, 2451 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2452 CTLTYPE_INT, "globalmaxholes", 2453 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 2454 NULL, 0, &tcp_sack_globalmaxholes, 0, 2455 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2456 sysctl_createv(clog, 0, NULL, &sack_node, 2457 CTLFLAG_PERMANENT, 2458 CTLTYPE_INT, "globalholes", 2459 SYSCTL_DESCR("Global number of TCP SACK holes"), 2460 NULL, 0, &tcp_sack_globalholes, 0, 2461 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2462 2463 sysctl_createv(clog, 0, NULL, NULL, 2464 CTLFLAG_PERMANENT, 2465 CTLTYPE_STRUCT, "stats", 2466 SYSCTL_DESCR("TCP statistics"), 2467 sysctl_net_inet_tcp_stats, 0, NULL, 0, 2468 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 2469 CTL_EOL); 2470 sysctl_createv(clog, 0, NULL, NULL, 2471 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2472 CTLTYPE_INT, "local_by_rtt", 2473 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 2474 "are local"), 2475 NULL, 0, &tcp_rttlocal, 0, 2476 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2477 #ifdef TCP_DEBUG 2478 sysctl_createv(clog, 0, NULL, NULL, 2479 CTLFLAG_PERMANENT, 2480 CTLTYPE_STRUCT, "debug", 2481 SYSCTL_DESCR("TCP sockets debug information"), 2482 NULL, 0, &tcp_debug, sizeof(tcp_debug), 2483 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 2484 CTL_EOL); 2485 sysctl_createv(clog, 0, NULL, NULL, 2486 CTLFLAG_PERMANENT, 2487 CTLTYPE_INT, "debx", 2488 SYSCTL_DESCR("Number of TCP debug sockets messages"), 2489 NULL, 0, &tcp_debx, sizeof(tcp_debx), 2490 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 2491 CTL_EOL); 2492 #endif 2493 sysctl_createv(clog, 0, NULL, NULL, 2494 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2495 CTLTYPE_STRUCT, "drop", 2496 SYSCTL_DESCR("TCP drop connection"), 2497 sysctl_net_inet_tcp_drop, 0, NULL, 0, 2498 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 2499 sysctl_createv(clog, 0, NULL, NULL, 2500 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2501 CTLTYPE_INT, "iss_hash", 2502 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 2503 "hash computation"), 2504 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 2505 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2506 CTL_EOL); 2507 2508 /* ABC subtree */ 2509 2510 sysctl_createv(clog, 0, NULL, &abc_node, 2511 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 2512 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 2513 NULL, 0, NULL, 0, 2514 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2515 sysctl_createv(clog, 0, &abc_node, NULL, 2516 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2517 CTLTYPE_INT, "enable", 2518 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 2519 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 2520 sysctl_createv(clog, 0, &abc_node, NULL, 2521 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2522 CTLTYPE_INT, "aggressive", 2523 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 2524 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 2525 2526 /* MSL tuning subtree */ 2527 2528 sysctl_createv(clog, 0, NULL, &mslt_node, 2529 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 2530 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 2531 NULL, 0, NULL, 0, 2532 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2533 sysctl_createv(clog, 0, &mslt_node, NULL, 2534 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2535 CTLTYPE_INT, "enable", 2536 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 2537 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 2538 sysctl_createv(clog, 0, &mslt_node, NULL, 2539 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2540 CTLTYPE_INT, "loopback", 2541 SYSCTL_DESCR("MSL value to use for loopback connections"), 2542 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 2543 sysctl_createv(clog, 0, &mslt_node, NULL, 2544 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2545 CTLTYPE_INT, "local", 2546 SYSCTL_DESCR("MSL value to use for local connections"), 2547 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 2548 sysctl_createv(clog, 0, &mslt_node, NULL, 2549 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2550 CTLTYPE_INT, "remote", 2551 SYSCTL_DESCR("MSL value to use for remote connections"), 2552 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 2553 sysctl_createv(clog, 0, &mslt_node, NULL, 2554 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2555 CTLTYPE_INT, "remote_threshold", 2556 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 2557 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 2558 2559 /* vestigial TIME_WAIT tuning subtree */ 2560 2561 sysctl_createv(clog, 0, NULL, &vtw_node, 2562 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 2563 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 2564 NULL, 0, NULL, 0, 2565 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2566 sysctl_createv(clog, 0, &vtw_node, NULL, 2567 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2568 CTLTYPE_INT, "enable", 2569 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 2570 sysctl_tcp_vtw_enable, 0, 2571 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 2572 0, CTL_CREATE, CTL_EOL); 2573 sysctl_createv(clog, 0, &vtw_node, NULL, 2574 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 2575 CTLTYPE_INT, "entries", 2576 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 2577 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 2578 } 2579 2580 void 2581 tcp_usrreq_init(void) 2582 { 2583 2584 #ifdef INET 2585 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 2586 #endif 2587 #ifdef INET6 2588 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 2589 #endif 2590 } 2591 2592 PR_WRAP_USRREQS(tcp) 2593 #define tcp_attach tcp_attach_wrapper 2594 #define tcp_detach tcp_detach_wrapper 2595 #define tcp_accept tcp_accept_wrapper 2596 #define tcp_bind tcp_bind_wrapper 2597 #define tcp_listen tcp_listen_wrapper 2598 #define tcp_ioctl tcp_ioctl_wrapper 2599 #define tcp_stat tcp_stat_wrapper 2600 #define tcp_peeraddr tcp_peeraddr_wrapper 2601 #define tcp_sockaddr tcp_sockaddr_wrapper 2602 #define tcp_recvoob tcp_recvoob_wrapper 2603 #define tcp_sendoob tcp_sendoob_wrapper 2604 #define tcp_usrreq tcp_usrreq_wrapper 2605 2606 const struct pr_usrreqs tcp_usrreqs = { 2607 .pr_attach = tcp_attach, 2608 .pr_detach = tcp_detach, 2609 .pr_accept = tcp_accept, 2610 .pr_bind = tcp_bind, 2611 .pr_listen = tcp_listen, 2612 .pr_ioctl = tcp_ioctl, 2613 .pr_stat = tcp_stat, 2614 .pr_peeraddr = tcp_peeraddr, 2615 .pr_sockaddr = tcp_sockaddr, 2616 .pr_recvoob = tcp_recvoob, 2617 .pr_sendoob = tcp_sendoob, 2618 .pr_generic = tcp_usrreq, 2619 }; 2620
Indexes created Sun Sep 21 18:09:52 GMT 2025