tcp_usrreq.c revision 1.133
1 /* $NetBSD: tcp_usrreq.c,v 1.133 2007/06/25 23:35:12 christos 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 * 3. All advertising materials mentioning features or use of this software 53 * must display the following acknowledgement: 54 * This product includes software developed by the NetBSD 55 * Foundation, Inc. and its contributors. 56 * 4. Neither the name of The NetBSD Foundation nor the names of its 57 * contributors may be used to endorse or promote products derived 58 * from this software without specific prior written permission. 59 * 60 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 61 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 62 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 63 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 64 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 65 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 66 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 67 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 68 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 69 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 70 * POSSIBILITY OF SUCH DAMAGE. 71 */ 72 73 /* 74 * Copyright (c) 1982, 1986, 1988, 1993, 1995 75 * The Regents of the University of California. All rights reserved. 76 * 77 * Redistribution and use in source and binary forms, with or without 78 * modification, are permitted provided that the following conditions 79 * are met: 80 * 1. Redistributions of source code must retain the above copyright 81 * notice, this list of conditions and the following disclaimer. 82 * 2. Redistributions in binary form must reproduce the above copyright 83 * notice, this list of conditions and the following disclaimer in the 84 * documentation and/or other materials provided with the distribution. 85 * 3. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 102 */ 103 104 #include <sys/cdefs.h> 105 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.133 2007/06/25 23:35:12 christos Exp $"); 106 107 #include "opt_inet.h" 108 #include "opt_ipsec.h" 109 #include "opt_tcp_debug.h" 110 #include "opt_mbuftrace.h" 111 #include "rnd.h" 112 113 #include <sys/param.h> 114 #include <sys/systm.h> 115 #include <sys/kernel.h> 116 #include <sys/malloc.h> 117 #include <sys/mbuf.h> 118 #include <sys/socket.h> 119 #include <sys/socketvar.h> 120 #include <sys/protosw.h> 121 #include <sys/errno.h> 122 #include <sys/stat.h> 123 #include <sys/proc.h> 124 #include <sys/domain.h> 125 #include <sys/sysctl.h> 126 #include <sys/kauth.h> 127 128 #include <net/if.h> 129 #include <net/route.h> 130 131 #include <netinet/in.h> 132 #include <netinet/in_systm.h> 133 #include <netinet/in_var.h> 134 #include <netinet/ip.h> 135 #include <netinet/in_pcb.h> 136 #include <netinet/ip_var.h> 137 #include <netinet/in_offload.h> 138 139 #ifdef INET6 140 #ifndef INET 141 #include <netinet/in.h> 142 #endif 143 #include <netinet/ip6.h> 144 #include <netinet6/in6_pcb.h> 145 #include <netinet6/ip6_var.h> 146 #endif 147 148 #include <netinet/tcp.h> 149 #include <netinet/tcp_fsm.h> 150 #include <netinet/tcp_seq.h> 151 #include <netinet/tcp_timer.h> 152 #include <netinet/tcp_var.h> 153 #include <netinet/tcp_congctl.h> 154 #include <netinet/tcpip.h> 155 #include <netinet/tcp_debug.h> 156 157 #include "opt_tcp_space.h" 158 159 #ifdef IPSEC 160 #include <netinet6/ipsec.h> 161 #endif /*IPSEC*/ 162 163 /* 164 * TCP protocol interface to socket abstraction. 165 */ 166 167 /* 168 * Process a TCP user request for TCP tb. If this is a send request 169 * then m is the mbuf chain of send data. If this is a timer expiration 170 * (called from the software clock routine), then timertype tells which timer. 171 */ 172 /*ARGSUSED*/ 173 int 174 tcp_usrreq(struct socket *so, int req, 175 struct mbuf *m, struct mbuf *nam, struct mbuf *control, struct lwp *l) 176 { 177 struct inpcb *inp; 178 #ifdef INET6 179 struct in6pcb *in6p; 180 #endif 181 struct tcpcb *tp = NULL; 182 int s; 183 int error = 0; 184 #ifdef TCP_DEBUG 185 int ostate = 0; 186 #endif 187 int family; /* family of the socket */ 188 189 family = so->so_proto->pr_domain->dom_family; 190 191 if (req == PRU_CONTROL) { 192 switch (family) { 193 #ifdef INET 194 case PF_INET: 195 return (in_control(so, (long)m, (void *)nam, 196 (struct ifnet *)control, l)); 197 #endif 198 #ifdef INET6 199 case PF_INET6: 200 return (in6_control(so, (long)m, (void *)nam, 201 (struct ifnet *)control, l)); 202 #endif 203 default: 204 return EAFNOSUPPORT; 205 } 206 } 207 208 s = splsoftnet(); 209 210 if (req == PRU_PURGEIF) { 211 switch (family) { 212 #ifdef INET 213 case PF_INET: 214 in_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 215 in_purgeif((struct ifnet *)control); 216 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 217 break; 218 #endif 219 #ifdef INET6 220 case PF_INET6: 221 in6_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 222 in6_purgeif((struct ifnet *)control); 223 in6_pcbpurgeif(&tcbtable, (struct ifnet *)control); 224 break; 225 #endif 226 default: 227 splx(s); 228 return (EAFNOSUPPORT); 229 } 230 splx(s); 231 return (0); 232 } 233 234 switch (family) { 235 #ifdef INET 236 case PF_INET: 237 inp = sotoinpcb(so); 238 #ifdef INET6 239 in6p = NULL; 240 #endif 241 break; 242 #endif 243 #ifdef INET6 244 case PF_INET6: 245 inp = NULL; 246 in6p = sotoin6pcb(so); 247 break; 248 #endif 249 default: 250 splx(s); 251 return EAFNOSUPPORT; 252 } 253 254 #ifdef DIAGNOSTIC 255 #ifdef INET6 256 if (inp && in6p) 257 panic("tcp_usrreq: both inp and in6p set to non-NULL"); 258 #endif 259 if (req != PRU_SEND && req != PRU_SENDOOB && control) 260 panic("tcp_usrreq: unexpected control mbuf"); 261 #endif 262 /* 263 * When a TCP is attached to a socket, then there will be 264 * a (struct inpcb) pointed at by the socket, and this 265 * structure will point at a subsidary (struct tcpcb). 266 */ 267 #ifndef INET6 268 if (inp == 0 && req != PRU_ATTACH) 269 #else 270 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 271 #endif 272 { 273 error = EINVAL; 274 goto release; 275 } 276 #ifdef INET 277 if (inp) { 278 tp = intotcpcb(inp); 279 /* WHAT IF TP IS 0? */ 280 #ifdef KPROF 281 tcp_acounts[tp->t_state][req]++; 282 #endif 283 #ifdef TCP_DEBUG 284 ostate = tp->t_state; 285 #endif 286 } 287 #endif 288 #ifdef INET6 289 if (in6p) { 290 tp = in6totcpcb(in6p); 291 /* WHAT IF TP IS 0? */ 292 #ifdef KPROF 293 tcp_acounts[tp->t_state][req]++; 294 #endif 295 #ifdef TCP_DEBUG 296 ostate = tp->t_state; 297 #endif 298 } 299 #endif 300 301 switch (req) { 302 303 /* 304 * TCP attaches to socket via PRU_ATTACH, reserving space, 305 * and an internet control block. 306 */ 307 case PRU_ATTACH: 308 #ifndef INET6 309 if (inp != 0) 310 #else 311 if (inp != 0 || in6p != 0) 312 #endif 313 { 314 error = EISCONN; 315 break; 316 } 317 error = tcp_attach(so); 318 if (error) 319 break; 320 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 321 so->so_linger = TCP_LINGERTIME; 322 tp = sototcpcb(so); 323 break; 324 325 /* 326 * PRU_DETACH detaches the TCP protocol from the socket. 327 */ 328 case PRU_DETACH: 329 tp = tcp_disconnect(tp); 330 break; 331 332 /* 333 * Give the socket an address. 334 */ 335 case PRU_BIND: 336 switch (family) { 337 #ifdef INET 338 case PF_INET: 339 error = in_pcbbind(inp, nam, l); 340 break; 341 #endif 342 #ifdef INET6 343 case PF_INET6: 344 error = in6_pcbbind(in6p, nam, l); 345 if (!error) { 346 /* mapped addr case */ 347 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 348 tp->t_family = AF_INET; 349 else 350 tp->t_family = AF_INET6; 351 } 352 break; 353 #endif 354 } 355 break; 356 357 /* 358 * Prepare to accept connections. 359 */ 360 case PRU_LISTEN: 361 #ifdef INET 362 if (inp && inp->inp_lport == 0) { 363 error = in_pcbbind(inp, (struct mbuf *)0, 364 (struct lwp *)0); 365 if (error) 366 break; 367 } 368 #endif 369 #ifdef INET6 370 if (in6p && in6p->in6p_lport == 0) { 371 error = in6_pcbbind(in6p, (struct mbuf *)0, 372 (struct lwp *)0); 373 if (error) 374 break; 375 } 376 #endif 377 tp->t_state = TCPS_LISTEN; 378 break; 379 380 /* 381 * Initiate connection to peer. 382 * Create a template for use in transmissions on this connection. 383 * Enter SYN_SENT state, and mark socket as connecting. 384 * Start keep-alive timer, and seed output sequence space. 385 * Send initial segment on connection. 386 */ 387 case PRU_CONNECT: 388 #ifdef INET 389 if (inp) { 390 if (inp->inp_lport == 0) { 391 error = in_pcbbind(inp, (struct mbuf *)0, 392 (struct lwp *)0); 393 if (error) 394 break; 395 } 396 error = in_pcbconnect(inp, nam, l); 397 } 398 #endif 399 #ifdef INET6 400 if (in6p) { 401 if (in6p->in6p_lport == 0) { 402 error = in6_pcbbind(in6p, (struct mbuf *)0, 403 (struct lwp *)0); 404 if (error) 405 break; 406 } 407 error = in6_pcbconnect(in6p, nam, l); 408 if (!error) { 409 /* mapped addr case */ 410 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 411 tp->t_family = AF_INET; 412 else 413 tp->t_family = AF_INET6; 414 } 415 } 416 #endif 417 if (error) 418 break; 419 tp->t_template = tcp_template(tp); 420 if (tp->t_template == 0) { 421 #ifdef INET 422 if (inp) 423 in_pcbdisconnect(inp); 424 #endif 425 #ifdef INET6 426 if (in6p) 427 in6_pcbdisconnect(in6p); 428 #endif 429 error = ENOBUFS; 430 break; 431 } 432 /* Compute window scaling to request. */ 433 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 434 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 435 tp->request_r_scale++; 436 soisconnecting(so); 437 tcpstat.tcps_connattempt++; 438 tp->t_state = TCPS_SYN_SENT; 439 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 440 tp->iss = tcp_new_iss(tp, 0); 441 tcp_sendseqinit(tp); 442 error = tcp_output(tp); 443 break; 444 445 /* 446 * Create a TCP connection between two sockets. 447 */ 448 case PRU_CONNECT2: 449 error = EOPNOTSUPP; 450 break; 451 452 /* 453 * Initiate disconnect from peer. 454 * If connection never passed embryonic stage, just drop; 455 * else if don't need to let data drain, then can just drop anyways, 456 * else have to begin TCP shutdown process: mark socket disconnecting, 457 * drain unread data, state switch to reflect user close, and 458 * send segment (e.g. FIN) to peer. Socket will be really disconnected 459 * when peer sends FIN and acks ours. 460 * 461 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 462 */ 463 case PRU_DISCONNECT: 464 tp = tcp_disconnect(tp); 465 break; 466 467 /* 468 * Accept a connection. Essentially all the work is 469 * done at higher levels; just return the address 470 * of the peer, storing through addr. 471 */ 472 case PRU_ACCEPT: 473 #ifdef INET 474 if (inp) 475 in_setpeeraddr(inp, nam); 476 #endif 477 #ifdef INET6 478 if (in6p) 479 in6_setpeeraddr(in6p, nam); 480 #endif 481 break; 482 483 /* 484 * Mark the connection as being incapable of further output. 485 */ 486 case PRU_SHUTDOWN: 487 socantsendmore(so); 488 tp = tcp_usrclosed(tp); 489 if (tp) 490 error = tcp_output(tp); 491 break; 492 493 /* 494 * After a receive, possibly send window update to peer. 495 */ 496 case PRU_RCVD: 497 /* 498 * soreceive() calls this function when a user receives 499 * ancillary data on a listening socket. We don't call 500 * tcp_output in such a case, since there is no header 501 * template for a listening socket and hence the kernel 502 * will panic. 503 */ 504 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 505 (void) tcp_output(tp); 506 break; 507 508 /* 509 * Do a send by putting data in output queue and updating urgent 510 * marker if URG set. Possibly send more data. 511 */ 512 case PRU_SEND: 513 if (control && control->m_len) { 514 m_freem(control); 515 m_freem(m); 516 error = EINVAL; 517 break; 518 } 519 sbappendstream(&so->so_snd, m); 520 error = tcp_output(tp); 521 break; 522 523 /* 524 * Abort the TCP. 525 */ 526 case PRU_ABORT: 527 tp = tcp_drop(tp, ECONNABORTED); 528 break; 529 530 case PRU_SENSE: 531 /* 532 * stat: don't bother with a blocksize. 533 */ 534 splx(s); 535 return (0); 536 537 case PRU_RCVOOB: 538 if (control && control->m_len) { 539 m_freem(control); 540 m_freem(m); 541 error = EINVAL; 542 break; 543 } 544 if ((so->so_oobmark == 0 && 545 (so->so_state & SS_RCVATMARK) == 0) || 546 so->so_options & SO_OOBINLINE || 547 tp->t_oobflags & TCPOOB_HADDATA) { 548 error = EINVAL; 549 break; 550 } 551 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 552 error = EWOULDBLOCK; 553 break; 554 } 555 m->m_len = 1; 556 *mtod(m, char *) = tp->t_iobc; 557 if (((long)nam & MSG_PEEK) == 0) 558 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 559 break; 560 561 case PRU_SENDOOB: 562 if (sbspace(&so->so_snd) < -512) { 563 m_freem(m); 564 error = ENOBUFS; 565 break; 566 } 567 /* 568 * According to RFC961 (Assigned Protocols), 569 * the urgent pointer points to the last octet 570 * of urgent data. We continue, however, 571 * to consider it to indicate the first octet 572 * of data past the urgent section. 573 * Otherwise, snd_up should be one lower. 574 */ 575 sbappendstream(&so->so_snd, m); 576 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 577 tp->t_force = 1; 578 error = tcp_output(tp); 579 tp->t_force = 0; 580 break; 581 582 case PRU_SOCKADDR: 583 #ifdef INET 584 if (inp) 585 in_setsockaddr(inp, nam); 586 #endif 587 #ifdef INET6 588 if (in6p) 589 in6_setsockaddr(in6p, nam); 590 #endif 591 break; 592 593 case PRU_PEERADDR: 594 #ifdef INET 595 if (inp) 596 in_setpeeraddr(inp, nam); 597 #endif 598 #ifdef INET6 599 if (in6p) 600 in6_setpeeraddr(in6p, nam); 601 #endif 602 break; 603 604 default: 605 panic("tcp_usrreq"); 606 } 607 #ifdef TCP_DEBUG 608 if (tp && (so->so_options & SO_DEBUG)) 609 tcp_trace(TA_USER, ostate, tp, NULL, req); 610 #endif 611 612 release: 613 splx(s); 614 return (error); 615 } 616 617 static void 618 change_keepalive(struct socket *so, struct tcpcb *tp) 619 { 620 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 621 TCP_TIMER_DISARM(tp, TCPT_KEEP); 622 TCP_TIMER_DISARM(tp, TCPT_2MSL); 623 624 if (tp->t_state == TCPS_SYN_RECEIVED || 625 tp->t_state == TCPS_SYN_SENT) { 626 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 627 } else if (so->so_options & SO_KEEPALIVE && 628 tp->t_state <= TCPS_CLOSE_WAIT) { 629 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 630 } else { 631 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 632 } 633 634 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 635 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 636 } 637 638 639 int 640 tcp_ctloutput(int op, struct socket *so, int level, int optname, 641 struct mbuf **mp) 642 { 643 int error = 0, s; 644 struct inpcb *inp; 645 #ifdef INET6 646 struct in6pcb *in6p; 647 #endif 648 struct tcpcb *tp; 649 struct mbuf *m; 650 int i; 651 u_int ui; 652 int family; /* family of the socket */ 653 654 family = so->so_proto->pr_domain->dom_family; 655 656 s = splsoftnet(); 657 switch (family) { 658 #ifdef INET 659 case PF_INET: 660 inp = sotoinpcb(so); 661 #ifdef INET6 662 in6p = NULL; 663 #endif 664 break; 665 #endif 666 #ifdef INET6 667 case PF_INET6: 668 inp = NULL; 669 in6p = sotoin6pcb(so); 670 break; 671 #endif 672 default: 673 splx(s); 674 panic("%s: af %d", __func__, family); 675 } 676 #ifndef INET6 677 if (inp == NULL) 678 #else 679 if (inp == NULL && in6p == NULL) 680 #endif 681 { 682 splx(s); 683 if (op == PRCO_SETOPT && *mp) 684 (void) m_free(*mp); 685 return (ECONNRESET); 686 } 687 if (level != IPPROTO_TCP) { 688 switch (family) { 689 #ifdef INET 690 case PF_INET: 691 error = ip_ctloutput(op, so, level, optname, mp); 692 break; 693 #endif 694 #ifdef INET6 695 case PF_INET6: 696 error = ip6_ctloutput(op, so, level, optname, mp); 697 break; 698 #endif 699 } 700 splx(s); 701 return (error); 702 } 703 if (inp) 704 tp = intotcpcb(inp); 705 #ifdef INET6 706 else if (in6p) 707 tp = in6totcpcb(in6p); 708 #endif 709 else 710 tp = NULL; 711 712 switch (op) { 713 714 case PRCO_SETOPT: 715 m = *mp; 716 switch (optname) { 717 718 #ifdef TCP_SIGNATURE 719 case TCP_MD5SIG: 720 if (m == NULL || m->m_len < sizeof (int)) 721 error = EINVAL; 722 if (error) 723 break; 724 if (*mtod(m, int *) > 0) 725 tp->t_flags |= TF_SIGNATURE; 726 else 727 tp->t_flags &= ~TF_SIGNATURE; 728 break; 729 #endif /* TCP_SIGNATURE */ 730 731 case TCP_NODELAY: 732 if (m == NULL || m->m_len < sizeof (int)) 733 error = EINVAL; 734 else if (*mtod(m, int *)) 735 tp->t_flags |= TF_NODELAY; 736 else 737 tp->t_flags &= ~TF_NODELAY; 738 break; 739 740 case TCP_MAXSEG: 741 if (m && m->m_len >= sizeof(int) && 742 (i = *mtod(m, int *)) > 0 && 743 i <= tp->t_peermss) 744 tp->t_peermss = i; /* limit on send size */ 745 else 746 error = EINVAL; 747 break; 748 #ifdef notyet 749 case TCP_CONGCTL: 750 if (m == NULL) 751 error = EINVAL; 752 error = tcp_congctl_select(tp, mtod(m, char *)); 753 #endif 754 break; 755 756 case TCP_KEEPIDLE: 757 if (m && m->m_len >= sizeof(u_int) && 758 (ui = *mtod(m, u_int *)) > 0) { 759 tp->t_keepidle = ui; 760 change_keepalive(so, tp); 761 } else 762 error = EINVAL; 763 break; 764 765 case TCP_KEEPINTVL: 766 if (m && m->m_len >= sizeof(u_int) && 767 (ui = *mtod(m, u_int *)) > 0) { 768 tp->t_keepintvl = ui; 769 change_keepalive(so, tp); 770 } else 771 error = EINVAL; 772 break; 773 774 case TCP_KEEPCNT: 775 if (m && m->m_len >= sizeof(u_int) && 776 (ui = *mtod(m, u_int *)) > 0) { 777 tp->t_keepcnt = ui; 778 change_keepalive(so, tp); 779 } else 780 error = EINVAL; 781 break; 782 783 case TCP_KEEPINIT: 784 if (m && m->m_len >= sizeof(u_int) && 785 (ui = *mtod(m, u_int *)) > 0) { 786 tp->t_keepinit = ui; 787 change_keepalive(so, tp); 788 } else 789 error = EINVAL; 790 break; 791 792 default: 793 error = ENOPROTOOPT; 794 break; 795 } 796 if (m) 797 (void) m_free(m); 798 break; 799 800 case PRCO_GETOPT: 801 *mp = m = m_get(M_WAIT, MT_SOOPTS); 802 m->m_len = sizeof(int); 803 MCLAIM(m, so->so_mowner); 804 805 switch (optname) { 806 #ifdef TCP_SIGNATURE 807 case TCP_MD5SIG: 808 *mtod(m, int *) = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 809 break; 810 #endif 811 case TCP_NODELAY: 812 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 813 break; 814 case TCP_MAXSEG: 815 *mtod(m, int *) = tp->t_peermss; 816 break; 817 #ifdef notyet 818 case TCP_CONGCTL: 819 break; 820 #endif 821 default: 822 error = ENOPROTOOPT; 823 break; 824 } 825 break; 826 } 827 splx(s); 828 return (error); 829 } 830 831 #ifndef TCP_SENDSPACE 832 #define TCP_SENDSPACE 1024*32 833 #endif 834 int tcp_sendspace = TCP_SENDSPACE; 835 #ifndef TCP_RECVSPACE 836 #define TCP_RECVSPACE 1024*32 837 #endif 838 int tcp_recvspace = TCP_RECVSPACE; 839 840 /* 841 * Attach TCP protocol to socket, allocating 842 * internet protocol control block, tcp control block, 843 * bufer space, and entering LISTEN state if to accept connections. 844 */ 845 int 846 tcp_attach(struct socket *so) 847 { 848 struct tcpcb *tp; 849 struct inpcb *inp; 850 #ifdef INET6 851 struct in6pcb *in6p; 852 #endif 853 int error; 854 int family; /* family of the socket */ 855 856 family = so->so_proto->pr_domain->dom_family; 857 858 #ifdef MBUFTRACE 859 so->so_mowner = &tcp_sock_mowner; 860 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 861 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 862 #endif 863 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 864 error = soreserve(so, tcp_sendspace, tcp_recvspace); 865 if (error) 866 return (error); 867 } 868 switch (family) { 869 #ifdef INET 870 case PF_INET: 871 error = in_pcballoc(so, &tcbtable); 872 if (error) 873 return (error); 874 inp = sotoinpcb(so); 875 #ifdef INET6 876 in6p = NULL; 877 #endif 878 break; 879 #endif 880 #ifdef INET6 881 case PF_INET6: 882 error = in6_pcballoc(so, &tcbtable); 883 if (error) 884 return (error); 885 inp = NULL; 886 in6p = sotoin6pcb(so); 887 break; 888 #endif 889 default: 890 return EAFNOSUPPORT; 891 } 892 if (inp) 893 tp = tcp_newtcpcb(family, (void *)inp); 894 #ifdef INET6 895 else if (in6p) 896 tp = tcp_newtcpcb(family, (void *)in6p); 897 #endif 898 else 899 tp = NULL; 900 901 if (tp == 0) { 902 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 903 904 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 905 #ifdef INET 906 if (inp) 907 in_pcbdetach(inp); 908 #endif 909 #ifdef INET6 910 if (in6p) 911 in6_pcbdetach(in6p); 912 #endif 913 so->so_state |= nofd; 914 return (ENOBUFS); 915 } 916 tp->t_state = TCPS_CLOSED; 917 return (0); 918 } 919 920 /* 921 * Initiate (or continue) disconnect. 922 * If embryonic state, just send reset (once). 923 * If in ``let data drain'' option and linger null, just drop. 924 * Otherwise (hard), mark socket disconnecting and drop 925 * current input data; switch states based on user close, and 926 * send segment to peer (with FIN). 927 */ 928 struct tcpcb * 929 tcp_disconnect(struct tcpcb *tp) 930 { 931 struct socket *so; 932 933 if (tp->t_inpcb) 934 so = tp->t_inpcb->inp_socket; 935 #ifdef INET6 936 else if (tp->t_in6pcb) 937 so = tp->t_in6pcb->in6p_socket; 938 #endif 939 else 940 so = NULL; 941 942 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 943 tp = tcp_close(tp); 944 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 945 tp = tcp_drop(tp, 0); 946 else { 947 soisdisconnecting(so); 948 sbflush(&so->so_rcv); 949 tp = tcp_usrclosed(tp); 950 if (tp) 951 (void) tcp_output(tp); 952 } 953 return (tp); 954 } 955 956 /* 957 * User issued close, and wish to trail through shutdown states: 958 * if never received SYN, just forget it. If got a SYN from peer, 959 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 960 * If already got a FIN from peer, then almost done; go to LAST_ACK 961 * state. In all other cases, have already sent FIN to peer (e.g. 962 * after PRU_SHUTDOWN), and just have to play tedious game waiting 963 * for peer to send FIN or not respond to keep-alives, etc. 964 * We can let the user exit from the close as soon as the FIN is acked. 965 */ 966 struct tcpcb * 967 tcp_usrclosed(struct tcpcb *tp) 968 { 969 970 switch (tp->t_state) { 971 972 case TCPS_CLOSED: 973 case TCPS_LISTEN: 974 case TCPS_SYN_SENT: 975 tp->t_state = TCPS_CLOSED; 976 tp = tcp_close(tp); 977 break; 978 979 case TCPS_SYN_RECEIVED: 980 case TCPS_ESTABLISHED: 981 tp->t_state = TCPS_FIN_WAIT_1; 982 break; 983 984 case TCPS_CLOSE_WAIT: 985 tp->t_state = TCPS_LAST_ACK; 986 break; 987 } 988 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 989 struct socket *so; 990 if (tp->t_inpcb) 991 so = tp->t_inpcb->inp_socket; 992 #ifdef INET6 993 else if (tp->t_in6pcb) 994 so = tp->t_in6pcb->in6p_socket; 995 #endif 996 else 997 so = NULL; 998 if (so) 999 soisdisconnected(so); 1000 /* 1001 * If we are in FIN_WAIT_2, we arrived here because the 1002 * application did a shutdown of the send side. Like the 1003 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1004 * a full close, we start a timer to make sure sockets are 1005 * not left in FIN_WAIT_2 forever. 1006 */ 1007 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1008 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1009 } 1010 return (tp); 1011 } 1012 1013 /* 1014 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1015 * than 32. 1016 */ 1017 static int 1018 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1019 { 1020 int error, mssdflt; 1021 struct sysctlnode node; 1022 1023 mssdflt = tcp_mssdflt; 1024 node = *rnode; 1025 node.sysctl_data = &mssdflt; 1026 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1027 if (error || newp == NULL) 1028 return (error); 1029 1030 if (mssdflt < 32) 1031 return (EINVAL); 1032 tcp_mssdflt = mssdflt; 1033 1034 return (0); 1035 } 1036 1037 /* 1038 * sysctl helper routine for setting port related values under 1039 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1040 * additional checks for each type. this code has placed in 1041 * tcp_input.c since INET and INET6 both use the same tcp code. 1042 * 1043 * this helper is not static so that both inet and inet6 can use it. 1044 */ 1045 int 1046 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1047 { 1048 int error, tmp; 1049 int apmin, apmax; 1050 #ifndef IPNOPRIVPORTS 1051 int lpmin, lpmax; 1052 #endif /* IPNOPRIVPORTS */ 1053 struct sysctlnode node; 1054 1055 if (namelen != 0) 1056 return (EINVAL); 1057 1058 switch (name[-3]) { 1059 #ifdef INET 1060 case PF_INET: 1061 apmin = anonportmin; 1062 apmax = anonportmax; 1063 #ifndef IPNOPRIVPORTS 1064 lpmin = lowportmin; 1065 lpmax = lowportmax; 1066 #endif /* IPNOPRIVPORTS */ 1067 break; 1068 #endif /* INET */ 1069 #ifdef INET6 1070 case PF_INET6: 1071 apmin = ip6_anonportmin; 1072 apmax = ip6_anonportmax; 1073 #ifndef IPNOPRIVPORTS 1074 lpmin = ip6_lowportmin; 1075 lpmax = ip6_lowportmax; 1076 #endif /* IPNOPRIVPORTS */ 1077 break; 1078 #endif /* INET6 */ 1079 default: 1080 return (EINVAL); 1081 } 1082 1083 /* 1084 * insert temporary copy into node, perform lookup on 1085 * temporary, then restore pointer 1086 */ 1087 node = *rnode; 1088 tmp = *(int*)rnode->sysctl_data; 1089 node.sysctl_data = &tmp; 1090 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1091 if (error || newp == NULL) 1092 return (error); 1093 1094 /* 1095 * simple port range check 1096 */ 1097 if (tmp < 0 || tmp > 65535) 1098 return (EINVAL); 1099 1100 /* 1101 * per-node range checks 1102 */ 1103 switch (rnode->sysctl_num) { 1104 case IPCTL_ANONPORTMIN: 1105 if (tmp >= apmax) 1106 return (EINVAL); 1107 #ifndef IPNOPRIVPORTS 1108 if (tmp < IPPORT_RESERVED) 1109 return (EINVAL); 1110 #endif /* IPNOPRIVPORTS */ 1111 break; 1112 1113 case IPCTL_ANONPORTMAX: 1114 if (apmin >= tmp) 1115 return (EINVAL); 1116 #ifndef IPNOPRIVPORTS 1117 if (tmp < IPPORT_RESERVED) 1118 return (EINVAL); 1119 #endif /* IPNOPRIVPORTS */ 1120 break; 1121 1122 #ifndef IPNOPRIVPORTS 1123 case IPCTL_LOWPORTMIN: 1124 if (tmp >= lpmax || 1125 tmp > IPPORT_RESERVEDMAX || 1126 tmp < IPPORT_RESERVEDMIN) 1127 return (EINVAL); 1128 break; 1129 1130 case IPCTL_LOWPORTMAX: 1131 if (lpmin >= tmp || 1132 tmp > IPPORT_RESERVEDMAX || 1133 tmp < IPPORT_RESERVEDMIN) 1134 return (EINVAL); 1135 break; 1136 #endif /* IPNOPRIVPORTS */ 1137 1138 default: 1139 return (EINVAL); 1140 } 1141 1142 *(int*)rnode->sysctl_data = tmp; 1143 1144 return (0); 1145 } 1146 1147 /* 1148 * The superuser can drop any connection. Normal users can only drop 1149 * their own connections. 1150 */ 1151 static inline int 1152 check_sockuid(struct socket *sockp, kauth_cred_t cred) 1153 { 1154 uid_t sockuid; 1155 1156 sockuid = sockp->so_uidinfo->ui_uid; 1157 if (kauth_authorize_generic(cred, KAUTH_GENERIC_ISSUSER, NULL) == 0 || 1158 sockuid == kauth_cred_getuid(cred) || 1159 sockuid == kauth_cred_geteuid(cred)) 1160 return 0; 1161 return EACCES; 1162 } 1163 1164 static inline int 1165 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1166 { 1167 size_t sz; 1168 int error; 1169 uid_t uid; 1170 1171 uid = sockp->so_uidinfo->ui_uid; 1172 if (oldp) { 1173 sz = MIN(sizeof(uid), *oldlenp); 1174 error = copyout(&uid, oldp, sz); 1175 if (error) 1176 return error; 1177 } 1178 *oldlenp = sizeof(uid); 1179 return 0; 1180 } 1181 1182 static inline int 1183 inet4_ident_core(struct in_addr raddr, u_int rport, 1184 struct in_addr laddr, u_int lport, 1185 void *oldp, size_t *oldlenp, 1186 struct lwp *l, int dodrop) 1187 { 1188 struct inpcb *inp; 1189 struct socket *sockp; 1190 1191 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport); 1192 1193 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1194 return ESRCH; 1195 1196 if (dodrop) { 1197 struct tcpcb *tp; 1198 1199 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1200 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1201 return ESRCH; 1202 1203 if (check_sockuid(inp->inp_socket, l->l_cred) != 0) 1204 return EACCES; 1205 1206 (void)tcp_drop(tp, ECONNABORTED); 1207 return 0; 1208 } 1209 else 1210 return copyout_uid(sockp, oldp, oldlenp); 1211 } 1212 1213 static inline int 1214 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1215 struct in6_addr *laddr, u_int lport, 1216 void *oldp, size_t *oldlenp, 1217 struct lwp *l, int dodrop) 1218 { 1219 struct in6pcb *in6p; 1220 struct socket *sockp; 1221 1222 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1223 1224 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1225 return ESRCH; 1226 1227 if (dodrop) { 1228 struct tcpcb *tp; 1229 1230 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1231 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1232 return ESRCH; 1233 1234 if (check_sockuid(in6p->in6p_socket, l->l_cred) != 0) 1235 return EACCES; 1236 1237 (void)tcp_drop(tp, ECONNABORTED); 1238 return 0; 1239 } 1240 else 1241 return copyout_uid(sockp, oldp, oldlenp); 1242 } 1243 1244 /* 1245 * sysctl helper routine for the net.inet.tcp.drop and 1246 * net.inet6.tcp6.drop nodes. 1247 */ 1248 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1249 1250 /* 1251 * sysctl helper routine for the net.inet.tcp.ident and 1252 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1253 * old way of looking up the ident information for ipv4 which involves 1254 * stuffing the port/addr pairs into the mib lookup. 1255 */ 1256 static int 1257 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1258 { 1259 #ifdef INET 1260 struct sockaddr_in *si4[2]; 1261 #endif /* INET */ 1262 #ifdef INET6 1263 struct sockaddr_in6 *si6[2]; 1264 #endif /* INET6 */ 1265 struct sockaddr_storage sa[2]; 1266 int error, pf, dodrop, s; 1267 1268 dodrop = name[-1] == TCPCTL_DROP; 1269 if (dodrop) { 1270 if (oldp != NULL || *oldlenp != 0) 1271 return EINVAL; 1272 if (newp == NULL) 1273 return EPERM; 1274 if (newlen < sizeof(sa)) 1275 return ENOMEM; 1276 } 1277 if (namelen != 4 && namelen != 0) 1278 return EINVAL; 1279 if (name[-2] != IPPROTO_TCP) 1280 return EINVAL; 1281 pf = name[-3]; 1282 1283 /* old style lookup, ipv4 only */ 1284 if (namelen == 4) { 1285 #ifdef INET 1286 struct in_addr laddr, raddr; 1287 u_int lport, rport; 1288 1289 if (pf != PF_INET) 1290 return EPROTONOSUPPORT; 1291 raddr.s_addr = (uint32_t)name[0]; 1292 rport = (u_int)name[1]; 1293 laddr.s_addr = (uint32_t)name[2]; 1294 lport = (u_int)name[3]; 1295 1296 s = splsoftnet(); 1297 error = inet4_ident_core(raddr, rport, laddr, lport, 1298 oldp, oldlenp, l, dodrop); 1299 splx(s); 1300 return error; 1301 #else /* INET */ 1302 return EINVAL; 1303 #endif /* INET */ 1304 } 1305 1306 if (newp == NULL || newlen != sizeof(sa)) 1307 return EINVAL; 1308 error = copyin(newp, &sa, newlen); 1309 if (error) 1310 return error; 1311 1312 /* 1313 * requested families must match 1314 */ 1315 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1316 return EINVAL; 1317 1318 switch (pf) { 1319 #ifdef INET 1320 case PF_INET: 1321 si4[0] = (struct sockaddr_in*)&sa[0]; 1322 si4[1] = (struct sockaddr_in*)&sa[1]; 1323 if (si4[0]->sin_len != sizeof(*si4[0]) || 1324 si4[0]->sin_len != si4[1]->sin_len) 1325 return EINVAL; 1326 1327 s = splsoftnet(); 1328 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1329 si4[1]->sin_addr, si4[1]->sin_port, 1330 oldp, oldlenp, l, dodrop); 1331 splx(s); 1332 return error; 1333 #endif /* INET */ 1334 #ifdef INET6 1335 case PF_INET6: 1336 si6[0] = (struct sockaddr_in6*)&sa[0]; 1337 si6[1] = (struct sockaddr_in6*)&sa[1]; 1338 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1339 si6[0]->sin6_len != si6[1]->sin6_len) 1340 return EINVAL; 1341 1342 s = splsoftnet(); 1343 error = inet6_ident_core(&si6[0]->sin6_addr, si6[0]->sin6_port, 1344 &si6[1]->sin6_addr, si6[1]->sin6_port, 1345 oldp, oldlenp, l, dodrop); 1346 splx(s); 1347 return error; 1348 #endif /* INET6 */ 1349 default: 1350 return EPROTONOSUPPORT; 1351 } 1352 /* NOTREACHED */ 1353 } 1354 1355 /* 1356 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1357 * inet/inet6, as well as raw pcbs for each. specifically not 1358 * declared static so that raw sockets and udp/udp6 can use it as 1359 * well. 1360 */ 1361 int 1362 sysctl_inpcblist(SYSCTLFN_ARGS) 1363 { 1364 #ifdef INET 1365 struct sockaddr_in *in; 1366 const struct inpcb *inp; 1367 #endif 1368 #ifdef INET6 1369 struct sockaddr_in6 *in6; 1370 const struct in6pcb *in6p; 1371 #endif 1372 /* 1373 * sysctl_data is const, but CIRCLEQ_FOREACH can't use a const 1374 * struct inpcbtable pointer, so we have to discard const. :-/ 1375 */ 1376 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1377 const struct inpcb_hdr *inph; 1378 struct tcpcb *tp; 1379 struct kinfo_pcb pcb; 1380 char *dp; 1381 u_int op, arg; 1382 size_t len, needed, elem_size, out_size; 1383 int error, elem_count, pf, proto, pf2; 1384 1385 if (namelen != 4) 1386 return (EINVAL); 1387 1388 if (oldp != NULL) { 1389 len = *oldlenp; 1390 elem_size = name[2]; 1391 elem_count = name[3]; 1392 if (elem_size != sizeof(pcb)) 1393 return EINVAL; 1394 } else { 1395 len = 0; 1396 elem_count = INT_MAX; 1397 elem_size = sizeof(pcb); 1398 } 1399 error = 0; 1400 dp = oldp; 1401 op = name[0]; 1402 arg = name[1]; 1403 out_size = elem_size; 1404 needed = 0; 1405 1406 if (namelen == 1 && name[0] == CTL_QUERY) 1407 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1408 1409 if (name - oname != 4) 1410 return (EINVAL); 1411 1412 pf = oname[1]; 1413 proto = oname[2]; 1414 pf2 = (oldp != NULL) ? pf : 0; 1415 1416 CIRCLEQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1417 #ifdef INET 1418 inp = (const struct inpcb *)inph; 1419 #endif 1420 #ifdef INET6 1421 in6p = (const struct in6pcb *)inph; 1422 #endif 1423 1424 if (inph->inph_af != pf) 1425 continue; 1426 1427 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1428 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1429 NULL) != 0) 1430 continue; 1431 1432 memset(&pcb, 0, sizeof(pcb)); 1433 1434 pcb.ki_family = pf; 1435 pcb.ki_type = proto; 1436 1437 switch (pf2) { 1438 case 0: 1439 /* just probing for size */ 1440 break; 1441 #ifdef INET 1442 case PF_INET: 1443 pcb.ki_family = inp->inp_socket->so_proto-> 1444 pr_domain->dom_family; 1445 pcb.ki_type = inp->inp_socket->so_proto-> 1446 pr_type; 1447 pcb.ki_protocol = inp->inp_socket->so_proto-> 1448 pr_protocol; 1449 pcb.ki_pflags = inp->inp_flags; 1450 1451 pcb.ki_sostate = inp->inp_socket->so_state; 1452 pcb.ki_prstate = inp->inp_state; 1453 if (proto == IPPROTO_TCP) { 1454 tp = intotcpcb(inp); 1455 pcb.ki_tstate = tp->t_state; 1456 pcb.ki_tflags = tp->t_flags; 1457 } 1458 1459 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1460 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1461 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1462 1463 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1464 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1465 1466 in = satosin(&pcb.ki_src); 1467 in->sin_len = sizeof(*in); 1468 in->sin_family = pf; 1469 in->sin_port = inp->inp_lport; 1470 in->sin_addr = inp->inp_laddr; 1471 if (pcb.ki_prstate >= INP_CONNECTED) { 1472 in = satosin(&pcb.ki_dst); 1473 in->sin_len = sizeof(*in); 1474 in->sin_family = pf; 1475 in->sin_port = inp->inp_fport; 1476 in->sin_addr = inp->inp_faddr; 1477 } 1478 break; 1479 #endif 1480 #ifdef INET6 1481 case PF_INET6: 1482 pcb.ki_family = in6p->in6p_socket->so_proto-> 1483 pr_domain->dom_family; 1484 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1485 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1486 pr_protocol; 1487 pcb.ki_pflags = in6p->in6p_flags; 1488 1489 pcb.ki_sostate = in6p->in6p_socket->so_state; 1490 pcb.ki_prstate = in6p->in6p_state; 1491 if (proto == IPPROTO_TCP) { 1492 tp = in6totcpcb(in6p); 1493 pcb.ki_tstate = tp->t_state; 1494 pcb.ki_tflags = tp->t_flags; 1495 } 1496 1497 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1498 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1499 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1500 1501 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1502 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1503 1504 in6 = satosin6(&pcb.ki_src); 1505 in6->sin6_len = sizeof(*in6); 1506 in6->sin6_family = pf; 1507 in6->sin6_port = in6p->in6p_lport; 1508 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1509 in6->sin6_addr = in6p->in6p_laddr; 1510 in6->sin6_scope_id = 0; /* XXX? */ 1511 1512 if (pcb.ki_prstate >= IN6P_CONNECTED) { 1513 in6 = satosin6(&pcb.ki_dst); 1514 in6->sin6_len = sizeof(*in6); 1515 in6->sin6_family = pf; 1516 in6->sin6_port = in6p->in6p_fport; 1517 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1518 in6->sin6_addr = in6p->in6p_faddr; 1519 in6->sin6_scope_id = 0; /* XXX? */ 1520 } 1521 break; 1522 #endif 1523 } 1524 1525 if (len >= elem_size && elem_count > 0) { 1526 error = copyout(&pcb, dp, out_size); 1527 if (error) 1528 return (error); 1529 dp += elem_size; 1530 len -= elem_size; 1531 } 1532 if (elem_count > 0) { 1533 needed += elem_size; 1534 if (elem_count != INT_MAX) 1535 elem_count--; 1536 } 1537 } 1538 1539 *oldlenp = needed; 1540 if (oldp == NULL) 1541 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 1542 1543 return (error); 1544 } 1545 1546 static int 1547 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1548 { 1549 struct sysctlnode node; 1550 int error, r; 1551 char newname[TCPCC_MAXLEN]; 1552 1553 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1554 1555 node = *rnode; 1556 node.sysctl_data = newname; 1557 node.sysctl_size = sizeof(newname); 1558 1559 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1560 1561 if (error || 1562 newp == NULL || 1563 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1564 return error; 1565 1566 if ((r = tcp_congctl_select(NULL, newname))) 1567 return r; 1568 1569 return error; 1570 } 1571 1572 static int 1573 sysctl_tcp_keep(SYSCTLFN_ARGS) 1574 { 1575 int error; 1576 u_int tmp; 1577 struct sysctlnode node; 1578 1579 node = *rnode; 1580 tmp = *(u_int *)rnode->sysctl_data; 1581 node.sysctl_data = &tmp; 1582 1583 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1584 if (error || newp == NULL) 1585 return error; 1586 1587 *(u_int *)rnode->sysctl_data = tmp; 1588 tcp_tcpcb_template(); /* update the template */ 1589 return 0; 1590 } 1591 1592 1593 /* 1594 * this (second stage) setup routine is a replacement for tcp_sysctl() 1595 * (which is currently used for ipv4 and ipv6) 1596 */ 1597 static void 1598 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1599 const char *tcpname) 1600 { 1601 const struct sysctlnode *sack_node; 1602 const struct sysctlnode *abc_node; 1603 const struct sysctlnode *ecn_node; 1604 const struct sysctlnode *congctl_node; 1605 #ifdef TCP_DEBUG 1606 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1607 extern int tcp_debx; 1608 #endif 1609 1610 sysctl_createv(clog, 0, NULL, NULL, 1611 CTLFLAG_PERMANENT, 1612 CTLTYPE_NODE, "net", NULL, 1613 NULL, 0, NULL, 0, 1614 CTL_NET, CTL_EOL); 1615 sysctl_createv(clog, 0, NULL, NULL, 1616 CTLFLAG_PERMANENT, 1617 CTLTYPE_NODE, pfname, NULL, 1618 NULL, 0, NULL, 0, 1619 CTL_NET, pf, CTL_EOL); 1620 sysctl_createv(clog, 0, NULL, NULL, 1621 CTLFLAG_PERMANENT, 1622 CTLTYPE_NODE, tcpname, 1623 SYSCTL_DESCR("TCP related settings"), 1624 NULL, 0, NULL, 0, 1625 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1626 1627 sysctl_createv(clog, 0, NULL, NULL, 1628 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1629 CTLTYPE_INT, "rfc1323", 1630 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1631 NULL, 0, &tcp_do_rfc1323, 0, 1632 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1633 sysctl_createv(clog, 0, NULL, NULL, 1634 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1635 CTLTYPE_INT, "sendspace", 1636 SYSCTL_DESCR("Default TCP send buffer size"), 1637 NULL, 0, &tcp_sendspace, 0, 1638 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1639 sysctl_createv(clog, 0, NULL, NULL, 1640 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1641 CTLTYPE_INT, "recvspace", 1642 SYSCTL_DESCR("Default TCP receive buffer size"), 1643 NULL, 0, &tcp_recvspace, 0, 1644 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1645 sysctl_createv(clog, 0, NULL, NULL, 1646 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1647 CTLTYPE_INT, "mssdflt", 1648 SYSCTL_DESCR("Default maximum segment size"), 1649 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1650 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1651 sysctl_createv(clog, 0, NULL, NULL, 1652 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1653 CTLTYPE_INT, "syn_cache_limit", 1654 SYSCTL_DESCR("Maximum number of entries in the TCP " 1655 "compressed state engine"), 1656 NULL, 0, &tcp_syn_cache_limit, 0, 1657 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1658 CTL_EOL); 1659 sysctl_createv(clog, 0, NULL, NULL, 1660 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1661 CTLTYPE_INT, "syn_bucket_limit", 1662 SYSCTL_DESCR("Maximum number of entries per hash " 1663 "bucket in the TCP compressed state " 1664 "engine"), 1665 NULL, 0, &tcp_syn_bucket_limit, 0, 1666 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1667 CTL_EOL); 1668 #if 0 /* obsoleted */ 1669 sysctl_createv(clog, 0, NULL, NULL, 1670 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1671 CTLTYPE_INT, "syn_cache_interval", 1672 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1673 NULL, 0, &tcp_syn_cache_interval, 0, 1674 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1675 CTL_EOL); 1676 #endif 1677 sysctl_createv(clog, 0, NULL, NULL, 1678 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1679 CTLTYPE_INT, "init_win", 1680 SYSCTL_DESCR("Initial TCP congestion window"), 1681 NULL, 0, &tcp_init_win, 0, 1682 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1683 sysctl_createv(clog, 0, NULL, NULL, 1684 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1685 CTLTYPE_INT, "mss_ifmtu", 1686 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1687 NULL, 0, &tcp_mss_ifmtu, 0, 1688 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1689 sysctl_createv(clog, 0, NULL, &sack_node, 1690 CTLFLAG_PERMANENT, 1691 CTLTYPE_NODE, "sack", 1692 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1693 NULL, 0, NULL, 0, 1694 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1695 1696 /* Congctl subtree */ 1697 sysctl_createv(clog, 0, NULL, &congctl_node, 1698 CTLFLAG_PERMANENT, 1699 CTLTYPE_NODE, "congctl", 1700 SYSCTL_DESCR("TCP Congestion Control"), 1701 NULL, 0, NULL, 0, 1702 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1703 sysctl_createv(clog, 0, &congctl_node, NULL, 1704 CTLFLAG_PERMANENT, 1705 CTLTYPE_STRING, "available", 1706 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1707 NULL, 0, &tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1708 sysctl_createv(clog, 0, &congctl_node, NULL, 1709 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1710 CTLTYPE_STRING, "selected", 1711 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1712 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1713 CTL_CREATE, CTL_EOL); 1714 1715 sysctl_createv(clog, 0, NULL, NULL, 1716 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1717 CTLTYPE_INT, "win_scale", 1718 SYSCTL_DESCR("Use RFC1323 window scale options"), 1719 NULL, 0, &tcp_do_win_scale, 0, 1720 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1721 sysctl_createv(clog, 0, NULL, NULL, 1722 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1723 CTLTYPE_INT, "timestamps", 1724 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1725 NULL, 0, &tcp_do_timestamps, 0, 1726 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1727 sysctl_createv(clog, 0, NULL, NULL, 1728 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1729 CTLTYPE_INT, "compat_42", 1730 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1731 NULL, 0, &tcp_compat_42, 0, 1732 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1733 sysctl_createv(clog, 0, NULL, NULL, 1734 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1735 CTLTYPE_INT, "cwm", 1736 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1737 "Monitoring"), 1738 NULL, 0, &tcp_cwm, 0, 1739 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1740 sysctl_createv(clog, 0, NULL, NULL, 1741 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1742 CTLTYPE_INT, "cwm_burstsize", 1743 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1744 "burst count in packets"), 1745 NULL, 0, &tcp_cwm_burstsize, 0, 1746 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1747 CTL_EOL); 1748 sysctl_createv(clog, 0, NULL, NULL, 1749 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1750 CTLTYPE_INT, "ack_on_push", 1751 SYSCTL_DESCR("Immediately return ACK when PSH is " 1752 "received"), 1753 NULL, 0, &tcp_ack_on_push, 0, 1754 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1755 sysctl_createv(clog, 0, NULL, NULL, 1756 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1757 CTLTYPE_INT, "keepidle", 1758 SYSCTL_DESCR("Allowed connection idle ticks before a " 1759 "keepalive probe is sent"), 1760 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1761 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1762 sysctl_createv(clog, 0, NULL, NULL, 1763 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1764 CTLTYPE_INT, "keepintvl", 1765 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1766 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1767 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1768 sysctl_createv(clog, 0, NULL, NULL, 1769 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1770 CTLTYPE_INT, "keepcnt", 1771 SYSCTL_DESCR("Number of keepalive probes to send"), 1772 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1773 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1774 sysctl_createv(clog, 0, NULL, NULL, 1775 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1776 CTLTYPE_INT, "slowhz", 1777 SYSCTL_DESCR("Keepalive ticks per second"), 1778 NULL, PR_SLOWHZ, NULL, 0, 1779 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1780 sysctl_createv(clog, 0, NULL, NULL, 1781 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1782 CTLTYPE_INT, "log_refused", 1783 SYSCTL_DESCR("Log refused TCP connections"), 1784 NULL, 0, &tcp_log_refused, 0, 1785 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1786 #if 0 /* obsoleted */ 1787 sysctl_createv(clog, 0, NULL, NULL, 1788 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1789 CTLTYPE_INT, "rstratelimit", NULL, 1790 NULL, 0, &tcp_rst_ratelim, 0, 1791 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1792 #endif 1793 sysctl_createv(clog, 0, NULL, NULL, 1794 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1795 CTLTYPE_INT, "rstppslimit", 1796 SYSCTL_DESCR("Maximum number of RST packets to send " 1797 "per second"), 1798 NULL, 0, &tcp_rst_ppslim, 0, 1799 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1800 sysctl_createv(clog, 0, NULL, NULL, 1801 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1802 CTLTYPE_INT, "delack_ticks", 1803 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1804 NULL, 0, &tcp_delack_ticks, 0, 1805 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1806 sysctl_createv(clog, 0, NULL, NULL, 1807 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1808 CTLTYPE_INT, "init_win_local", 1809 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1810 NULL, 0, &tcp_init_win_local, 0, 1811 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1812 CTL_EOL); 1813 sysctl_createv(clog, 0, NULL, NULL, 1814 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1815 CTLTYPE_STRUCT, "ident", 1816 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1817 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1818 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1819 sysctl_createv(clog, 0, NULL, NULL, 1820 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1821 CTLTYPE_INT, "do_loopback_cksum", 1822 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1823 NULL, 0, &tcp_do_loopback_cksum, 0, 1824 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1825 CTL_EOL); 1826 sysctl_createv(clog, 0, NULL, NULL, 1827 CTLFLAG_PERMANENT, 1828 CTLTYPE_STRUCT, "pcblist", 1829 SYSCTL_DESCR("TCP protocol control block list"), 1830 sysctl_inpcblist, 0, &tcbtable, 0, 1831 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1832 CTL_EOL); 1833 sysctl_createv(clog, 0, NULL, NULL, 1834 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1835 CTLTYPE_INT, "keepinit", 1836 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1837 sysctl_tcp_keep, 0, &tcp_keepinit, 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 NULL, 0, &tcpstat, sizeof(tcpstat), 1891 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1892 CTL_EOL); 1893 #ifdef TCP_DEBUG 1894 sysctl_createv(clog, 0, NULL, NULL, 1895 CTLFLAG_PERMANENT, 1896 CTLTYPE_STRUCT, "debug", 1897 SYSCTL_DESCR("TCP sockets debug information"), 1898 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1899 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1900 CTL_EOL); 1901 sysctl_createv(clog, 0, NULL, NULL, 1902 CTLFLAG_PERMANENT, 1903 CTLTYPE_INT, "debx", 1904 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1905 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1906 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1907 CTL_EOL); 1908 #endif 1909 sysctl_createv(clog, 0, NULL, NULL, 1910 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1911 CTLTYPE_STRUCT, "drop", 1912 SYSCTL_DESCR("TCP drop connection"), 1913 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1914 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1915 #if NRND > 0 1916 sysctl_createv(clog, 0, NULL, NULL, 1917 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1918 CTLTYPE_INT, "iss_hash", 1919 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1920 "hash computation"), 1921 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1922 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1923 CTL_EOL); 1924 #endif 1925 1926 /* ABC subtree */ 1927 1928 sysctl_createv(clog, 0, NULL, &abc_node, 1929 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 1930 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 1931 NULL, 0, NULL, 0, 1932 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1933 sysctl_createv(clog, 0, &abc_node, NULL, 1934 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1935 CTLTYPE_INT, "enable", 1936 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 1937 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 1938 sysctl_createv(clog, 0, &abc_node, NULL, 1939 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1940 CTLTYPE_INT, "aggressive", 1941 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 1942 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 1943 } 1944 1945 /* 1946 * Sysctl for tcp variables. 1947 */ 1948 #ifdef INET 1949 SYSCTL_SETUP(sysctl_net_inet_tcp_setup, "sysctl net.inet.tcp subtree setup") 1950 { 1951 1952 sysctl_net_inet_tcp_setup2(clog, PF_INET, "inet", "tcp"); 1953 } 1954 #endif /* INET */ 1955 1956 #ifdef INET6 1957 SYSCTL_SETUP(sysctl_net_inet6_tcp6_setup, "sysctl net.inet6.tcp6 subtree setup") 1958 { 1959 1960 sysctl_net_inet_tcp_setup2(clog, PF_INET6, "inet6", "tcp6"); 1961 } 1962 #endif /* INET6 */ 1963
Indexes created Sun Sep 21 18:09:52 GMT 2025