tcp_usrreq.c revision 1.136.2.1
1 /* $NetBSD: tcp_usrreq.c,v 1.136.2.1 2007/10/02 18:29:21 joerg 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 1.136.2.1 2007/10/02 18:29:21 joerg 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 #include <netinet6/scope6_var.h> 147 #endif 148 149 #include <netinet/tcp.h> 150 #include <netinet/tcp_fsm.h> 151 #include <netinet/tcp_seq.h> 152 #include <netinet/tcp_timer.h> 153 #include <netinet/tcp_var.h> 154 #include <netinet/tcp_congctl.h> 155 #include <netinet/tcpip.h> 156 #include <netinet/tcp_debug.h> 157 158 #include "opt_tcp_space.h" 159 160 #ifdef IPSEC 161 #include <netinet6/ipsec.h> 162 #endif /*IPSEC*/ 163 164 /* 165 * TCP protocol interface to socket abstraction. 166 */ 167 168 /* 169 * Process a TCP user request for TCP tb. If this is a send request 170 * then m is the mbuf chain of send data. If this is a timer expiration 171 * (called from the software clock routine), then timertype tells which timer. 172 */ 173 /*ARGSUSED*/ 174 int 175 tcp_usrreq(struct socket *so, int req, 176 struct mbuf *m, struct mbuf *nam, struct mbuf *control, struct lwp *l) 177 { 178 struct inpcb *inp; 179 #ifdef INET6 180 struct in6pcb *in6p; 181 #endif 182 struct tcpcb *tp = NULL; 183 int s; 184 int error = 0; 185 #ifdef TCP_DEBUG 186 int ostate = 0; 187 #endif 188 int family; /* family of the socket */ 189 190 family = so->so_proto->pr_domain->dom_family; 191 192 if (req == PRU_CONTROL) { 193 switch (family) { 194 #ifdef INET 195 case PF_INET: 196 return (in_control(so, (long)m, (void *)nam, 197 (struct ifnet *)control, l)); 198 #endif 199 #ifdef INET6 200 case PF_INET6: 201 return (in6_control(so, (long)m, (void *)nam, 202 (struct ifnet *)control, l)); 203 #endif 204 default: 205 return EAFNOSUPPORT; 206 } 207 } 208 209 s = splsoftnet(); 210 211 if (req == PRU_PURGEIF) { 212 switch (family) { 213 #ifdef INET 214 case PF_INET: 215 in_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 216 in_purgeif((struct ifnet *)control); 217 in_pcbpurgeif(&tcbtable, (struct ifnet *)control); 218 break; 219 #endif 220 #ifdef INET6 221 case PF_INET6: 222 in6_pcbpurgeif0(&tcbtable, (struct ifnet *)control); 223 in6_purgeif((struct ifnet *)control); 224 in6_pcbpurgeif(&tcbtable, (struct ifnet *)control); 225 break; 226 #endif 227 default: 228 splx(s); 229 return (EAFNOSUPPORT); 230 } 231 splx(s); 232 return (0); 233 } 234 235 switch (family) { 236 #ifdef INET 237 case PF_INET: 238 inp = sotoinpcb(so); 239 #ifdef INET6 240 in6p = NULL; 241 #endif 242 break; 243 #endif 244 #ifdef INET6 245 case PF_INET6: 246 inp = NULL; 247 in6p = sotoin6pcb(so); 248 break; 249 #endif 250 default: 251 splx(s); 252 return EAFNOSUPPORT; 253 } 254 255 #ifdef DIAGNOSTIC 256 #ifdef INET6 257 if (inp && in6p) 258 panic("tcp_usrreq: both inp and in6p set to non-NULL"); 259 #endif 260 if (req != PRU_SEND && req != PRU_SENDOOB && control) 261 panic("tcp_usrreq: unexpected control mbuf"); 262 #endif 263 /* 264 * When a TCP is attached to a socket, then there will be 265 * a (struct inpcb) pointed at by the socket, and this 266 * structure will point at a subsidary (struct tcpcb). 267 */ 268 #ifndef INET6 269 if (inp == 0 && req != PRU_ATTACH) 270 #else 271 if ((inp == 0 && in6p == 0) && req != PRU_ATTACH) 272 #endif 273 { 274 error = EINVAL; 275 goto release; 276 } 277 #ifdef INET 278 if (inp) { 279 tp = intotcpcb(inp); 280 /* WHAT IF TP IS 0? */ 281 #ifdef KPROF 282 tcp_acounts[tp->t_state][req]++; 283 #endif 284 #ifdef TCP_DEBUG 285 ostate = tp->t_state; 286 #endif 287 } 288 #endif 289 #ifdef INET6 290 if (in6p) { 291 tp = in6totcpcb(in6p); 292 /* WHAT IF TP IS 0? */ 293 #ifdef KPROF 294 tcp_acounts[tp->t_state][req]++; 295 #endif 296 #ifdef TCP_DEBUG 297 ostate = tp->t_state; 298 #endif 299 } 300 #endif 301 302 switch (req) { 303 304 /* 305 * TCP attaches to socket via PRU_ATTACH, reserving space, 306 * and an internet control block. 307 */ 308 case PRU_ATTACH: 309 #ifndef INET6 310 if (inp != 0) 311 #else 312 if (inp != 0 || in6p != 0) 313 #endif 314 { 315 error = EISCONN; 316 break; 317 } 318 error = tcp_attach(so); 319 if (error) 320 break; 321 if ((so->so_options & SO_LINGER) && so->so_linger == 0) 322 so->so_linger = TCP_LINGERTIME; 323 tp = sototcpcb(so); 324 break; 325 326 /* 327 * PRU_DETACH detaches the TCP protocol from the socket. 328 */ 329 case PRU_DETACH: 330 tp = tcp_disconnect(tp); 331 break; 332 333 /* 334 * Give the socket an address. 335 */ 336 case PRU_BIND: 337 switch (family) { 338 #ifdef INET 339 case PF_INET: 340 error = in_pcbbind(inp, nam, l); 341 break; 342 #endif 343 #ifdef INET6 344 case PF_INET6: 345 error = in6_pcbbind(in6p, nam, l); 346 if (!error) { 347 /* mapped addr case */ 348 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 349 tp->t_family = AF_INET; 350 else 351 tp->t_family = AF_INET6; 352 } 353 break; 354 #endif 355 } 356 break; 357 358 /* 359 * Prepare to accept connections. 360 */ 361 case PRU_LISTEN: 362 #ifdef INET 363 if (inp && inp->inp_lport == 0) { 364 error = in_pcbbind(inp, (struct mbuf *)0, 365 (struct lwp *)0); 366 if (error) 367 break; 368 } 369 #endif 370 #ifdef INET6 371 if (in6p && in6p->in6p_lport == 0) { 372 error = in6_pcbbind(in6p, (struct mbuf *)0, 373 (struct lwp *)0); 374 if (error) 375 break; 376 } 377 #endif 378 tp->t_state = TCPS_LISTEN; 379 break; 380 381 /* 382 * Initiate connection to peer. 383 * Create a template for use in transmissions on this connection. 384 * Enter SYN_SENT state, and mark socket as connecting. 385 * Start keep-alive timer, and seed output sequence space. 386 * Send initial segment on connection. 387 */ 388 case PRU_CONNECT: 389 #ifdef INET 390 if (inp) { 391 if (inp->inp_lport == 0) { 392 error = in_pcbbind(inp, (struct mbuf *)0, 393 (struct lwp *)0); 394 if (error) 395 break; 396 } 397 error = in_pcbconnect(inp, nam, l); 398 } 399 #endif 400 #ifdef INET6 401 if (in6p) { 402 if (in6p->in6p_lport == 0) { 403 error = in6_pcbbind(in6p, (struct mbuf *)0, 404 (struct lwp *)0); 405 if (error) 406 break; 407 } 408 error = in6_pcbconnect(in6p, nam, l); 409 if (!error) { 410 /* mapped addr case */ 411 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 412 tp->t_family = AF_INET; 413 else 414 tp->t_family = AF_INET6; 415 } 416 } 417 #endif 418 if (error) 419 break; 420 tp->t_template = tcp_template(tp); 421 if (tp->t_template == 0) { 422 #ifdef INET 423 if (inp) 424 in_pcbdisconnect(inp); 425 #endif 426 #ifdef INET6 427 if (in6p) 428 in6_pcbdisconnect(in6p); 429 #endif 430 error = ENOBUFS; 431 break; 432 } 433 /* 434 * Compute window scaling to request: 435 * XXX: This should be moved to tcp_output(), 436 * and should be based on the actual MSS. 437 */ 438 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 439 (0x1 << tp->request_r_scale) < tcp_minmss) 440 tp->request_r_scale++; 441 soisconnecting(so); 442 tcpstat.tcps_connattempt++; 443 tp->t_state = TCPS_SYN_SENT; 444 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 445 tp->iss = tcp_new_iss(tp, 0); 446 tcp_sendseqinit(tp); 447 error = tcp_output(tp); 448 break; 449 450 /* 451 * Create a TCP connection between two sockets. 452 */ 453 case PRU_CONNECT2: 454 error = EOPNOTSUPP; 455 break; 456 457 /* 458 * Initiate disconnect from peer. 459 * If connection never passed embryonic stage, just drop; 460 * else if don't need to let data drain, then can just drop anyways, 461 * else have to begin TCP shutdown process: mark socket disconnecting, 462 * drain unread data, state switch to reflect user close, and 463 * send segment (e.g. FIN) to peer. Socket will be really disconnected 464 * when peer sends FIN and acks ours. 465 * 466 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 467 */ 468 case PRU_DISCONNECT: 469 tp = tcp_disconnect(tp); 470 break; 471 472 /* 473 * Accept a connection. Essentially all the work is 474 * done at higher levels; just return the address 475 * of the peer, storing through addr. 476 */ 477 case PRU_ACCEPT: 478 #ifdef INET 479 if (inp) 480 in_setpeeraddr(inp, nam); 481 #endif 482 #ifdef INET6 483 if (in6p) 484 in6_setpeeraddr(in6p, nam); 485 #endif 486 break; 487 488 /* 489 * Mark the connection as being incapable of further output. 490 */ 491 case PRU_SHUTDOWN: 492 socantsendmore(so); 493 tp = tcp_usrclosed(tp); 494 if (tp) 495 error = tcp_output(tp); 496 break; 497 498 /* 499 * After a receive, possibly send window update to peer. 500 */ 501 case PRU_RCVD: 502 /* 503 * soreceive() calls this function when a user receives 504 * ancillary data on a listening socket. We don't call 505 * tcp_output in such a case, since there is no header 506 * template for a listening socket and hence the kernel 507 * will panic. 508 */ 509 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 510 (void) tcp_output(tp); 511 break; 512 513 /* 514 * Do a send by putting data in output queue and updating urgent 515 * marker if URG set. Possibly send more data. 516 */ 517 case PRU_SEND: 518 if (control && control->m_len) { 519 m_freem(control); 520 m_freem(m); 521 error = EINVAL; 522 break; 523 } 524 sbappendstream(&so->so_snd, m); 525 error = tcp_output(tp); 526 break; 527 528 /* 529 * Abort the TCP. 530 */ 531 case PRU_ABORT: 532 tp = tcp_drop(tp, ECONNABORTED); 533 break; 534 535 case PRU_SENSE: 536 /* 537 * stat: don't bother with a blocksize. 538 */ 539 splx(s); 540 return (0); 541 542 case PRU_RCVOOB: 543 if (control && control->m_len) { 544 m_freem(control); 545 m_freem(m); 546 error = EINVAL; 547 break; 548 } 549 if ((so->so_oobmark == 0 && 550 (so->so_state & SS_RCVATMARK) == 0) || 551 so->so_options & SO_OOBINLINE || 552 tp->t_oobflags & TCPOOB_HADDATA) { 553 error = EINVAL; 554 break; 555 } 556 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 557 error = EWOULDBLOCK; 558 break; 559 } 560 m->m_len = 1; 561 *mtod(m, char *) = tp->t_iobc; 562 if (((long)nam & MSG_PEEK) == 0) 563 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 564 break; 565 566 case PRU_SENDOOB: 567 if (sbspace(&so->so_snd) < -512) { 568 m_freem(m); 569 error = ENOBUFS; 570 break; 571 } 572 /* 573 * According to RFC961 (Assigned Protocols), 574 * the urgent pointer points to the last octet 575 * of urgent data. We continue, however, 576 * to consider it to indicate the first octet 577 * of data past the urgent section. 578 * Otherwise, snd_up should be one lower. 579 */ 580 sbappendstream(&so->so_snd, m); 581 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 582 tp->t_force = 1; 583 error = tcp_output(tp); 584 tp->t_force = 0; 585 break; 586 587 case PRU_SOCKADDR: 588 #ifdef INET 589 if (inp) 590 in_setsockaddr(inp, nam); 591 #endif 592 #ifdef INET6 593 if (in6p) 594 in6_setsockaddr(in6p, nam); 595 #endif 596 break; 597 598 case PRU_PEERADDR: 599 #ifdef INET 600 if (inp) 601 in_setpeeraddr(inp, nam); 602 #endif 603 #ifdef INET6 604 if (in6p) 605 in6_setpeeraddr(in6p, nam); 606 #endif 607 break; 608 609 default: 610 panic("tcp_usrreq"); 611 } 612 #ifdef TCP_DEBUG 613 if (tp && (so->so_options & SO_DEBUG)) 614 tcp_trace(TA_USER, ostate, tp, NULL, req); 615 #endif 616 617 release: 618 splx(s); 619 return (error); 620 } 621 622 static void 623 change_keepalive(struct socket *so, struct tcpcb *tp) 624 { 625 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 626 TCP_TIMER_DISARM(tp, TCPT_KEEP); 627 TCP_TIMER_DISARM(tp, TCPT_2MSL); 628 629 if (tp->t_state == TCPS_SYN_RECEIVED || 630 tp->t_state == TCPS_SYN_SENT) { 631 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 632 } else if (so->so_options & SO_KEEPALIVE && 633 tp->t_state <= TCPS_CLOSE_WAIT) { 634 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 635 } else { 636 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 637 } 638 639 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 640 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 641 } 642 643 644 int 645 tcp_ctloutput(int op, struct socket *so, int level, int optname, 646 struct mbuf **mp) 647 { 648 int error = 0, s; 649 struct inpcb *inp; 650 #ifdef INET6 651 struct in6pcb *in6p; 652 #endif 653 struct tcpcb *tp; 654 struct mbuf *m; 655 int i; 656 u_int ui; 657 int family; /* family of the socket */ 658 659 family = so->so_proto->pr_domain->dom_family; 660 661 s = splsoftnet(); 662 switch (family) { 663 #ifdef INET 664 case PF_INET: 665 inp = sotoinpcb(so); 666 #ifdef INET6 667 in6p = NULL; 668 #endif 669 break; 670 #endif 671 #ifdef INET6 672 case PF_INET6: 673 inp = NULL; 674 in6p = sotoin6pcb(so); 675 break; 676 #endif 677 default: 678 splx(s); 679 panic("%s: af %d", __func__, family); 680 } 681 #ifndef INET6 682 if (inp == NULL) 683 #else 684 if (inp == NULL && in6p == NULL) 685 #endif 686 { 687 splx(s); 688 if (op == PRCO_SETOPT && *mp) 689 (void) m_free(*mp); 690 return (ECONNRESET); 691 } 692 if (level != IPPROTO_TCP) { 693 switch (family) { 694 #ifdef INET 695 case PF_INET: 696 error = ip_ctloutput(op, so, level, optname, mp); 697 break; 698 #endif 699 #ifdef INET6 700 case PF_INET6: 701 error = ip6_ctloutput(op, so, level, optname, mp); 702 break; 703 #endif 704 } 705 splx(s); 706 return (error); 707 } 708 if (inp) 709 tp = intotcpcb(inp); 710 #ifdef INET6 711 else if (in6p) 712 tp = in6totcpcb(in6p); 713 #endif 714 else 715 tp = NULL; 716 717 switch (op) { 718 719 case PRCO_SETOPT: 720 m = *mp; 721 switch (optname) { 722 723 #ifdef TCP_SIGNATURE 724 case TCP_MD5SIG: 725 if (m == NULL || m->m_len < sizeof (int)) 726 error = EINVAL; 727 if (error) 728 break; 729 if (*mtod(m, int *) > 0) 730 tp->t_flags |= TF_SIGNATURE; 731 else 732 tp->t_flags &= ~TF_SIGNATURE; 733 break; 734 #endif /* TCP_SIGNATURE */ 735 736 case TCP_NODELAY: 737 if (m == NULL || m->m_len < sizeof (int)) 738 error = EINVAL; 739 else if (*mtod(m, int *)) 740 tp->t_flags |= TF_NODELAY; 741 else 742 tp->t_flags &= ~TF_NODELAY; 743 break; 744 745 case TCP_MAXSEG: 746 if (m && m->m_len >= sizeof(int) && 747 (i = *mtod(m, int *)) > 0 && 748 i <= tp->t_peermss) 749 tp->t_peermss = i; /* limit on send size */ 750 else 751 error = EINVAL; 752 break; 753 #ifdef notyet 754 case TCP_CONGCTL: 755 if (m == NULL) 756 error = EINVAL; 757 error = tcp_congctl_select(tp, mtod(m, char *)); 758 #endif 759 break; 760 761 case TCP_KEEPIDLE: 762 if (m && m->m_len >= sizeof(u_int) && 763 (ui = *mtod(m, u_int *)) > 0) { 764 tp->t_keepidle = ui; 765 change_keepalive(so, tp); 766 } else 767 error = EINVAL; 768 break; 769 770 case TCP_KEEPINTVL: 771 if (m && m->m_len >= sizeof(u_int) && 772 (ui = *mtod(m, u_int *)) > 0) { 773 tp->t_keepintvl = ui; 774 change_keepalive(so, tp); 775 } else 776 error = EINVAL; 777 break; 778 779 case TCP_KEEPCNT: 780 if (m && m->m_len >= sizeof(u_int) && 781 (ui = *mtod(m, u_int *)) > 0) { 782 tp->t_keepcnt = ui; 783 change_keepalive(so, tp); 784 } else 785 error = EINVAL; 786 break; 787 788 case TCP_KEEPINIT: 789 if (m && m->m_len >= sizeof(u_int) && 790 (ui = *mtod(m, u_int *)) > 0) { 791 tp->t_keepinit = ui; 792 change_keepalive(so, tp); 793 } else 794 error = EINVAL; 795 break; 796 797 default: 798 error = ENOPROTOOPT; 799 break; 800 } 801 if (m) 802 (void) m_free(m); 803 break; 804 805 case PRCO_GETOPT: 806 *mp = m = m_intopt(so, 0); 807 808 switch (optname) { 809 #ifdef TCP_SIGNATURE 810 case TCP_MD5SIG: 811 *mtod(m, int *) = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 812 break; 813 #endif 814 case TCP_NODELAY: 815 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 816 break; 817 case TCP_MAXSEG: 818 *mtod(m, int *) = tp->t_peermss; 819 break; 820 #ifdef notyet 821 case TCP_CONGCTL: 822 break; 823 #endif 824 default: 825 error = ENOPROTOOPT; 826 break; 827 } 828 break; 829 } 830 splx(s); 831 return (error); 832 } 833 834 #ifndef TCP_SENDSPACE 835 #define TCP_SENDSPACE 1024*32 836 #endif 837 int tcp_sendspace = TCP_SENDSPACE; 838 #ifndef TCP_RECVSPACE 839 #define TCP_RECVSPACE 1024*32 840 #endif 841 int tcp_recvspace = TCP_RECVSPACE; 842 843 /* 844 * Attach TCP protocol to socket, allocating 845 * internet protocol control block, tcp control block, 846 * bufer space, and entering LISTEN state if to accept connections. 847 */ 848 int 849 tcp_attach(struct socket *so) 850 { 851 struct tcpcb *tp; 852 struct inpcb *inp; 853 #ifdef INET6 854 struct in6pcb *in6p; 855 #endif 856 int error; 857 int family; /* family of the socket */ 858 859 family = so->so_proto->pr_domain->dom_family; 860 861 #ifdef MBUFTRACE 862 so->so_mowner = &tcp_sock_mowner; 863 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 864 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 865 #endif 866 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 867 error = soreserve(so, tcp_sendspace, tcp_recvspace); 868 if (error) 869 return (error); 870 } 871 872 so->so_rcv.sb_flags |= SB_AUTOSIZE; 873 so->so_snd.sb_flags |= SB_AUTOSIZE; 874 875 switch (family) { 876 #ifdef INET 877 case PF_INET: 878 error = in_pcballoc(so, &tcbtable); 879 if (error) 880 return (error); 881 inp = sotoinpcb(so); 882 #ifdef INET6 883 in6p = NULL; 884 #endif 885 break; 886 #endif 887 #ifdef INET6 888 case PF_INET6: 889 error = in6_pcballoc(so, &tcbtable); 890 if (error) 891 return (error); 892 inp = NULL; 893 in6p = sotoin6pcb(so); 894 break; 895 #endif 896 default: 897 return EAFNOSUPPORT; 898 } 899 if (inp) 900 tp = tcp_newtcpcb(family, (void *)inp); 901 #ifdef INET6 902 else if (in6p) 903 tp = tcp_newtcpcb(family, (void *)in6p); 904 #endif 905 else 906 tp = NULL; 907 908 if (tp == 0) { 909 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 910 911 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 912 #ifdef INET 913 if (inp) 914 in_pcbdetach(inp); 915 #endif 916 #ifdef INET6 917 if (in6p) 918 in6_pcbdetach(in6p); 919 #endif 920 so->so_state |= nofd; 921 return (ENOBUFS); 922 } 923 tp->t_state = TCPS_CLOSED; 924 return (0); 925 } 926 927 /* 928 * Initiate (or continue) disconnect. 929 * If embryonic state, just send reset (once). 930 * If in ``let data drain'' option and linger null, just drop. 931 * Otherwise (hard), mark socket disconnecting and drop 932 * current input data; switch states based on user close, and 933 * send segment to peer (with FIN). 934 */ 935 struct tcpcb * 936 tcp_disconnect(struct tcpcb *tp) 937 { 938 struct socket *so; 939 940 if (tp->t_inpcb) 941 so = tp->t_inpcb->inp_socket; 942 #ifdef INET6 943 else if (tp->t_in6pcb) 944 so = tp->t_in6pcb->in6p_socket; 945 #endif 946 else 947 so = NULL; 948 949 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 950 tp = tcp_close(tp); 951 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 952 tp = tcp_drop(tp, 0); 953 else { 954 soisdisconnecting(so); 955 sbflush(&so->so_rcv); 956 tp = tcp_usrclosed(tp); 957 if (tp) 958 (void) tcp_output(tp); 959 } 960 return (tp); 961 } 962 963 /* 964 * User issued close, and wish to trail through shutdown states: 965 * if never received SYN, just forget it. If got a SYN from peer, 966 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 967 * If already got a FIN from peer, then almost done; go to LAST_ACK 968 * state. In all other cases, have already sent FIN to peer (e.g. 969 * after PRU_SHUTDOWN), and just have to play tedious game waiting 970 * for peer to send FIN or not respond to keep-alives, etc. 971 * We can let the user exit from the close as soon as the FIN is acked. 972 */ 973 struct tcpcb * 974 tcp_usrclosed(struct tcpcb *tp) 975 { 976 977 switch (tp->t_state) { 978 979 case TCPS_CLOSED: 980 case TCPS_LISTEN: 981 case TCPS_SYN_SENT: 982 tp->t_state = TCPS_CLOSED; 983 tp = tcp_close(tp); 984 break; 985 986 case TCPS_SYN_RECEIVED: 987 case TCPS_ESTABLISHED: 988 tp->t_state = TCPS_FIN_WAIT_1; 989 break; 990 991 case TCPS_CLOSE_WAIT: 992 tp->t_state = TCPS_LAST_ACK; 993 break; 994 } 995 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 996 struct socket *so; 997 if (tp->t_inpcb) 998 so = tp->t_inpcb->inp_socket; 999 #ifdef INET6 1000 else if (tp->t_in6pcb) 1001 so = tp->t_in6pcb->in6p_socket; 1002 #endif 1003 else 1004 so = NULL; 1005 if (so) 1006 soisdisconnected(so); 1007 /* 1008 * If we are in FIN_WAIT_2, we arrived here because the 1009 * application did a shutdown of the send side. Like the 1010 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1011 * a full close, we start a timer to make sure sockets are 1012 * not left in FIN_WAIT_2 forever. 1013 */ 1014 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1015 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1016 } 1017 return (tp); 1018 } 1019 1020 /* 1021 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1022 * than 32. 1023 */ 1024 static int 1025 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1026 { 1027 int error, mssdflt; 1028 struct sysctlnode node; 1029 1030 mssdflt = tcp_mssdflt; 1031 node = *rnode; 1032 node.sysctl_data = &mssdflt; 1033 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1034 if (error || newp == NULL) 1035 return (error); 1036 1037 if (mssdflt < 32) 1038 return (EINVAL); 1039 tcp_mssdflt = mssdflt; 1040 1041 return (0); 1042 } 1043 1044 /* 1045 * sysctl helper routine for setting port related values under 1046 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1047 * additional checks for each type. this code has placed in 1048 * tcp_input.c since INET and INET6 both use the same tcp code. 1049 * 1050 * this helper is not static so that both inet and inet6 can use it. 1051 */ 1052 int 1053 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1054 { 1055 int error, tmp; 1056 int apmin, apmax; 1057 #ifndef IPNOPRIVPORTS 1058 int lpmin, lpmax; 1059 #endif /* IPNOPRIVPORTS */ 1060 struct sysctlnode node; 1061 1062 if (namelen != 0) 1063 return (EINVAL); 1064 1065 switch (name[-3]) { 1066 #ifdef INET 1067 case PF_INET: 1068 apmin = anonportmin; 1069 apmax = anonportmax; 1070 #ifndef IPNOPRIVPORTS 1071 lpmin = lowportmin; 1072 lpmax = lowportmax; 1073 #endif /* IPNOPRIVPORTS */ 1074 break; 1075 #endif /* INET */ 1076 #ifdef INET6 1077 case PF_INET6: 1078 apmin = ip6_anonportmin; 1079 apmax = ip6_anonportmax; 1080 #ifndef IPNOPRIVPORTS 1081 lpmin = ip6_lowportmin; 1082 lpmax = ip6_lowportmax; 1083 #endif /* IPNOPRIVPORTS */ 1084 break; 1085 #endif /* INET6 */ 1086 default: 1087 return (EINVAL); 1088 } 1089 1090 /* 1091 * insert temporary copy into node, perform lookup on 1092 * temporary, then restore pointer 1093 */ 1094 node = *rnode; 1095 tmp = *(int*)rnode->sysctl_data; 1096 node.sysctl_data = &tmp; 1097 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1098 if (error || newp == NULL) 1099 return (error); 1100 1101 /* 1102 * simple port range check 1103 */ 1104 if (tmp < 0 || tmp > 65535) 1105 return (EINVAL); 1106 1107 /* 1108 * per-node range checks 1109 */ 1110 switch (rnode->sysctl_num) { 1111 case IPCTL_ANONPORTMIN: 1112 if (tmp >= apmax) 1113 return (EINVAL); 1114 #ifndef IPNOPRIVPORTS 1115 if (tmp < IPPORT_RESERVED) 1116 return (EINVAL); 1117 #endif /* IPNOPRIVPORTS */ 1118 break; 1119 1120 case IPCTL_ANONPORTMAX: 1121 if (apmin >= tmp) 1122 return (EINVAL); 1123 #ifndef IPNOPRIVPORTS 1124 if (tmp < IPPORT_RESERVED) 1125 return (EINVAL); 1126 #endif /* IPNOPRIVPORTS */ 1127 break; 1128 1129 #ifndef IPNOPRIVPORTS 1130 case IPCTL_LOWPORTMIN: 1131 if (tmp >= lpmax || 1132 tmp > IPPORT_RESERVEDMAX || 1133 tmp < IPPORT_RESERVEDMIN) 1134 return (EINVAL); 1135 break; 1136 1137 case IPCTL_LOWPORTMAX: 1138 if (lpmin >= tmp || 1139 tmp > IPPORT_RESERVEDMAX || 1140 tmp < IPPORT_RESERVEDMIN) 1141 return (EINVAL); 1142 break; 1143 #endif /* IPNOPRIVPORTS */ 1144 1145 default: 1146 return (EINVAL); 1147 } 1148 1149 *(int*)rnode->sysctl_data = tmp; 1150 1151 return (0); 1152 } 1153 1154 /* 1155 * The superuser can drop any connection. Normal users can only drop 1156 * their own connections. 1157 */ 1158 static inline int 1159 check_sockuid(struct socket *sockp, kauth_cred_t cred) 1160 { 1161 uid_t sockuid; 1162 1163 sockuid = sockp->so_uidinfo->ui_uid; 1164 if (kauth_authorize_generic(cred, KAUTH_GENERIC_ISSUSER, NULL) == 0 || 1165 sockuid == kauth_cred_getuid(cred) || 1166 sockuid == kauth_cred_geteuid(cred)) 1167 return 0; 1168 return EACCES; 1169 } 1170 1171 static inline int 1172 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1173 { 1174 size_t sz; 1175 int error; 1176 uid_t uid; 1177 1178 uid = sockp->so_uidinfo->ui_uid; 1179 if (oldp) { 1180 sz = MIN(sizeof(uid), *oldlenp); 1181 error = copyout(&uid, oldp, sz); 1182 if (error) 1183 return error; 1184 } 1185 *oldlenp = sizeof(uid); 1186 return 0; 1187 } 1188 1189 static inline int 1190 inet4_ident_core(struct in_addr raddr, u_int rport, 1191 struct in_addr laddr, u_int lport, 1192 void *oldp, size_t *oldlenp, 1193 struct lwp *l, int dodrop) 1194 { 1195 struct inpcb *inp; 1196 struct socket *sockp; 1197 1198 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport); 1199 1200 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1201 return ESRCH; 1202 1203 if (dodrop) { 1204 struct tcpcb *tp; 1205 1206 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1207 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1208 return ESRCH; 1209 1210 if (check_sockuid(inp->inp_socket, l->l_cred) != 0) 1211 return EACCES; 1212 1213 (void)tcp_drop(tp, ECONNABORTED); 1214 return 0; 1215 } 1216 else 1217 return copyout_uid(sockp, oldp, oldlenp); 1218 } 1219 1220 #ifdef INET6 1221 static inline int 1222 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1223 struct in6_addr *laddr, u_int lport, 1224 void *oldp, size_t *oldlenp, 1225 struct lwp *l, int dodrop) 1226 { 1227 struct in6pcb *in6p; 1228 struct socket *sockp; 1229 1230 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1231 1232 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1233 return ESRCH; 1234 1235 if (dodrop) { 1236 struct tcpcb *tp; 1237 1238 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1239 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1240 return ESRCH; 1241 1242 if (check_sockuid(in6p->in6p_socket, l->l_cred) != 0) 1243 return EACCES; 1244 1245 (void)tcp_drop(tp, ECONNABORTED); 1246 return 0; 1247 } 1248 else 1249 return copyout_uid(sockp, oldp, oldlenp); 1250 } 1251 #endif 1252 1253 /* 1254 * sysctl helper routine for the net.inet.tcp.drop and 1255 * net.inet6.tcp6.drop nodes. 1256 */ 1257 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1258 1259 /* 1260 * sysctl helper routine for the net.inet.tcp.ident and 1261 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1262 * old way of looking up the ident information for ipv4 which involves 1263 * stuffing the port/addr pairs into the mib lookup. 1264 */ 1265 static int 1266 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1267 { 1268 #ifdef INET 1269 struct sockaddr_in *si4[2]; 1270 #endif /* INET */ 1271 #ifdef INET6 1272 struct sockaddr_in6 *si6[2]; 1273 #endif /* INET6 */ 1274 struct sockaddr_storage sa[2]; 1275 int error, pf, dodrop, s; 1276 1277 dodrop = name[-1] == TCPCTL_DROP; 1278 if (dodrop) { 1279 if (oldp != NULL || *oldlenp != 0) 1280 return EINVAL; 1281 if (newp == NULL) 1282 return EPERM; 1283 if (newlen < sizeof(sa)) 1284 return ENOMEM; 1285 } 1286 if (namelen != 4 && namelen != 0) 1287 return EINVAL; 1288 if (name[-2] != IPPROTO_TCP) 1289 return EINVAL; 1290 pf = name[-3]; 1291 1292 /* old style lookup, ipv4 only */ 1293 if (namelen == 4) { 1294 #ifdef INET 1295 struct in_addr laddr, raddr; 1296 u_int lport, rport; 1297 1298 if (pf != PF_INET) 1299 return EPROTONOSUPPORT; 1300 raddr.s_addr = (uint32_t)name[0]; 1301 rport = (u_int)name[1]; 1302 laddr.s_addr = (uint32_t)name[2]; 1303 lport = (u_int)name[3]; 1304 1305 s = splsoftnet(); 1306 error = inet4_ident_core(raddr, rport, laddr, lport, 1307 oldp, oldlenp, l, dodrop); 1308 splx(s); 1309 return error; 1310 #else /* INET */ 1311 return EINVAL; 1312 #endif /* INET */ 1313 } 1314 1315 if (newp == NULL || newlen != sizeof(sa)) 1316 return EINVAL; 1317 error = copyin(newp, &sa, newlen); 1318 if (error) 1319 return error; 1320 1321 /* 1322 * requested families must match 1323 */ 1324 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1325 return EINVAL; 1326 1327 switch (pf) { 1328 #ifdef INET6 1329 case PF_INET6: 1330 si6[0] = (struct sockaddr_in6*)&sa[0]; 1331 si6[1] = (struct sockaddr_in6*)&sa[1]; 1332 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1333 si6[1]->sin6_len != sizeof(*si6[1])) 1334 return EINVAL; 1335 1336 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1337 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1338 error = sa6_embedscope(si6[0], ip6_use_defzone); 1339 if (error) 1340 return error; 1341 error = sa6_embedscope(si6[1], ip6_use_defzone); 1342 if (error) 1343 return error; 1344 1345 s = splsoftnet(); 1346 error = inet6_ident_core(&si6[0]->sin6_addr, 1347 si6[0]->sin6_port, &si6[1]->sin6_addr, 1348 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1349 splx(s); 1350 return error; 1351 } 1352 1353 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1354 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1355 return EINVAL; 1356 1357 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1358 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1359 /*FALLTHROUGH*/ 1360 #endif /* INET6 */ 1361 #ifdef INET 1362 case PF_INET: 1363 si4[0] = (struct sockaddr_in*)&sa[0]; 1364 si4[1] = (struct sockaddr_in*)&sa[1]; 1365 if (si4[0]->sin_len != sizeof(*si4[0]) || 1366 si4[0]->sin_len != sizeof(*si4[1])) 1367 return EINVAL; 1368 1369 s = splsoftnet(); 1370 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1371 si4[1]->sin_addr, si4[1]->sin_port, 1372 oldp, oldlenp, l, dodrop); 1373 splx(s); 1374 return error; 1375 #endif /* INET */ 1376 default: 1377 return EPROTONOSUPPORT; 1378 } 1379 } 1380 1381 /* 1382 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1383 * inet/inet6, as well as raw pcbs for each. specifically not 1384 * declared static so that raw sockets and udp/udp6 can use it as 1385 * well. 1386 */ 1387 int 1388 sysctl_inpcblist(SYSCTLFN_ARGS) 1389 { 1390 #ifdef INET 1391 struct sockaddr_in *in; 1392 const struct inpcb *inp; 1393 #endif 1394 #ifdef INET6 1395 struct sockaddr_in6 *in6; 1396 const struct in6pcb *in6p; 1397 #endif 1398 /* 1399 * sysctl_data is const, but CIRCLEQ_FOREACH can't use a const 1400 * struct inpcbtable pointer, so we have to discard const. :-/ 1401 */ 1402 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1403 const struct inpcb_hdr *inph; 1404 struct tcpcb *tp; 1405 struct kinfo_pcb pcb; 1406 char *dp; 1407 u_int op, arg; 1408 size_t len, needed, elem_size, out_size; 1409 int error, elem_count, pf, proto, pf2; 1410 1411 if (namelen != 4) 1412 return (EINVAL); 1413 1414 if (oldp != NULL) { 1415 len = *oldlenp; 1416 elem_size = name[2]; 1417 elem_count = name[3]; 1418 if (elem_size != sizeof(pcb)) 1419 return EINVAL; 1420 } else { 1421 len = 0; 1422 elem_count = INT_MAX; 1423 elem_size = sizeof(pcb); 1424 } 1425 error = 0; 1426 dp = oldp; 1427 op = name[0]; 1428 arg = name[1]; 1429 out_size = elem_size; 1430 needed = 0; 1431 1432 if (namelen == 1 && name[0] == CTL_QUERY) 1433 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1434 1435 if (name - oname != 4) 1436 return (EINVAL); 1437 1438 pf = oname[1]; 1439 proto = oname[2]; 1440 pf2 = (oldp != NULL) ? pf : 0; 1441 1442 CIRCLEQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1443 #ifdef INET 1444 inp = (const struct inpcb *)inph; 1445 #endif 1446 #ifdef INET6 1447 in6p = (const struct in6pcb *)inph; 1448 #endif 1449 1450 if (inph->inph_af != pf) 1451 continue; 1452 1453 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1454 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1455 NULL) != 0) 1456 continue; 1457 1458 memset(&pcb, 0, sizeof(pcb)); 1459 1460 pcb.ki_family = pf; 1461 pcb.ki_type = proto; 1462 1463 switch (pf2) { 1464 case 0: 1465 /* just probing for size */ 1466 break; 1467 #ifdef INET 1468 case PF_INET: 1469 pcb.ki_family = inp->inp_socket->so_proto-> 1470 pr_domain->dom_family; 1471 pcb.ki_type = inp->inp_socket->so_proto-> 1472 pr_type; 1473 pcb.ki_protocol = inp->inp_socket->so_proto-> 1474 pr_protocol; 1475 pcb.ki_pflags = inp->inp_flags; 1476 1477 pcb.ki_sostate = inp->inp_socket->so_state; 1478 pcb.ki_prstate = inp->inp_state; 1479 if (proto == IPPROTO_TCP) { 1480 tp = intotcpcb(inp); 1481 pcb.ki_tstate = tp->t_state; 1482 pcb.ki_tflags = tp->t_flags; 1483 } 1484 1485 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1486 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1487 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1488 1489 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1490 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1491 1492 in = satosin(&pcb.ki_src); 1493 in->sin_len = sizeof(*in); 1494 in->sin_family = pf; 1495 in->sin_port = inp->inp_lport; 1496 in->sin_addr = inp->inp_laddr; 1497 if (pcb.ki_prstate >= INP_CONNECTED) { 1498 in = satosin(&pcb.ki_dst); 1499 in->sin_len = sizeof(*in); 1500 in->sin_family = pf; 1501 in->sin_port = inp->inp_fport; 1502 in->sin_addr = inp->inp_faddr; 1503 } 1504 break; 1505 #endif 1506 #ifdef INET6 1507 case PF_INET6: 1508 pcb.ki_family = in6p->in6p_socket->so_proto-> 1509 pr_domain->dom_family; 1510 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1511 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1512 pr_protocol; 1513 pcb.ki_pflags = in6p->in6p_flags; 1514 1515 pcb.ki_sostate = in6p->in6p_socket->so_state; 1516 pcb.ki_prstate = in6p->in6p_state; 1517 if (proto == IPPROTO_TCP) { 1518 tp = in6totcpcb(in6p); 1519 pcb.ki_tstate = tp->t_state; 1520 pcb.ki_tflags = tp->t_flags; 1521 } 1522 1523 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1524 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1525 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1526 1527 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1528 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1529 1530 in6 = satosin6(&pcb.ki_src); 1531 in6->sin6_len = sizeof(*in6); 1532 in6->sin6_family = pf; 1533 in6->sin6_port = in6p->in6p_lport; 1534 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1535 in6->sin6_addr = in6p->in6p_laddr; 1536 in6->sin6_scope_id = 0; /* XXX? */ 1537 1538 if (pcb.ki_prstate >= IN6P_CONNECTED) { 1539 in6 = satosin6(&pcb.ki_dst); 1540 in6->sin6_len = sizeof(*in6); 1541 in6->sin6_family = pf; 1542 in6->sin6_port = in6p->in6p_fport; 1543 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1544 in6->sin6_addr = in6p->in6p_faddr; 1545 in6->sin6_scope_id = 0; /* XXX? */ 1546 } 1547 break; 1548 #endif 1549 } 1550 1551 if (len >= elem_size && elem_count > 0) { 1552 error = copyout(&pcb, dp, out_size); 1553 if (error) 1554 return (error); 1555 dp += elem_size; 1556 len -= elem_size; 1557 } 1558 if (elem_count > 0) { 1559 needed += elem_size; 1560 if (elem_count != INT_MAX) 1561 elem_count--; 1562 } 1563 } 1564 1565 *oldlenp = needed; 1566 if (oldp == NULL) 1567 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 1568 1569 return (error); 1570 } 1571 1572 static int 1573 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1574 { 1575 struct sysctlnode node; 1576 int error, r; 1577 char newname[TCPCC_MAXLEN]; 1578 1579 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1580 1581 node = *rnode; 1582 node.sysctl_data = newname; 1583 node.sysctl_size = sizeof(newname); 1584 1585 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1586 1587 if (error || 1588 newp == NULL || 1589 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1590 return error; 1591 1592 if ((r = tcp_congctl_select(NULL, newname))) 1593 return r; 1594 1595 return error; 1596 } 1597 1598 static int 1599 sysctl_tcp_keep(SYSCTLFN_ARGS) 1600 { 1601 int error; 1602 u_int tmp; 1603 struct sysctlnode node; 1604 1605 node = *rnode; 1606 tmp = *(u_int *)rnode->sysctl_data; 1607 node.sysctl_data = &tmp; 1608 1609 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1610 if (error || newp == NULL) 1611 return error; 1612 1613 *(u_int *)rnode->sysctl_data = tmp; 1614 tcp_tcpcb_template(); /* update the template */ 1615 return 0; 1616 } 1617 1618 1619 /* 1620 * this (second stage) setup routine is a replacement for tcp_sysctl() 1621 * (which is currently used for ipv4 and ipv6) 1622 */ 1623 static void 1624 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1625 const char *tcpname) 1626 { 1627 const struct sysctlnode *sack_node; 1628 const struct sysctlnode *abc_node; 1629 const struct sysctlnode *ecn_node; 1630 const struct sysctlnode *congctl_node; 1631 #ifdef TCP_DEBUG 1632 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1633 extern int tcp_debx; 1634 #endif 1635 1636 sysctl_createv(clog, 0, NULL, NULL, 1637 CTLFLAG_PERMANENT, 1638 CTLTYPE_NODE, "net", NULL, 1639 NULL, 0, NULL, 0, 1640 CTL_NET, CTL_EOL); 1641 sysctl_createv(clog, 0, NULL, NULL, 1642 CTLFLAG_PERMANENT, 1643 CTLTYPE_NODE, pfname, NULL, 1644 NULL, 0, NULL, 0, 1645 CTL_NET, pf, CTL_EOL); 1646 sysctl_createv(clog, 0, NULL, NULL, 1647 CTLFLAG_PERMANENT, 1648 CTLTYPE_NODE, tcpname, 1649 SYSCTL_DESCR("TCP related settings"), 1650 NULL, 0, NULL, 0, 1651 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1652 1653 sysctl_createv(clog, 0, NULL, NULL, 1654 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1655 CTLTYPE_INT, "rfc1323", 1656 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1657 NULL, 0, &tcp_do_rfc1323, 0, 1658 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1659 sysctl_createv(clog, 0, NULL, NULL, 1660 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1661 CTLTYPE_INT, "sendspace", 1662 SYSCTL_DESCR("Default TCP send buffer size"), 1663 NULL, 0, &tcp_sendspace, 0, 1664 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1665 sysctl_createv(clog, 0, NULL, NULL, 1666 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1667 CTLTYPE_INT, "recvspace", 1668 SYSCTL_DESCR("Default TCP receive buffer size"), 1669 NULL, 0, &tcp_recvspace, 0, 1670 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1671 sysctl_createv(clog, 0, NULL, NULL, 1672 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1673 CTLTYPE_INT, "mssdflt", 1674 SYSCTL_DESCR("Default maximum segment size"), 1675 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1676 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1677 sysctl_createv(clog, 0, NULL, NULL, 1678 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1679 CTLTYPE_INT, "minmss", 1680 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 1681 NULL, 0, &tcp_minmss, 0, 1682 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1683 sysctl_createv(clog, 0, NULL, NULL, 1684 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1685 CTLTYPE_INT, "syn_cache_limit", 1686 SYSCTL_DESCR("Maximum number of entries in the TCP " 1687 "compressed state engine"), 1688 NULL, 0, &tcp_syn_cache_limit, 0, 1689 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1690 CTL_EOL); 1691 sysctl_createv(clog, 0, NULL, NULL, 1692 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1693 CTLTYPE_INT, "syn_bucket_limit", 1694 SYSCTL_DESCR("Maximum number of entries per hash " 1695 "bucket in the TCP compressed state " 1696 "engine"), 1697 NULL, 0, &tcp_syn_bucket_limit, 0, 1698 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1699 CTL_EOL); 1700 #if 0 /* obsoleted */ 1701 sysctl_createv(clog, 0, NULL, NULL, 1702 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1703 CTLTYPE_INT, "syn_cache_interval", 1704 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1705 NULL, 0, &tcp_syn_cache_interval, 0, 1706 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1707 CTL_EOL); 1708 #endif 1709 sysctl_createv(clog, 0, NULL, NULL, 1710 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1711 CTLTYPE_INT, "init_win", 1712 SYSCTL_DESCR("Initial TCP congestion window"), 1713 NULL, 0, &tcp_init_win, 0, 1714 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1715 sysctl_createv(clog, 0, NULL, NULL, 1716 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1717 CTLTYPE_INT, "mss_ifmtu", 1718 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1719 NULL, 0, &tcp_mss_ifmtu, 0, 1720 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1721 sysctl_createv(clog, 0, NULL, &sack_node, 1722 CTLFLAG_PERMANENT, 1723 CTLTYPE_NODE, "sack", 1724 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1725 NULL, 0, NULL, 0, 1726 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1727 1728 /* Congctl subtree */ 1729 sysctl_createv(clog, 0, NULL, &congctl_node, 1730 CTLFLAG_PERMANENT, 1731 CTLTYPE_NODE, "congctl", 1732 SYSCTL_DESCR("TCP Congestion Control"), 1733 NULL, 0, NULL, 0, 1734 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1735 sysctl_createv(clog, 0, &congctl_node, NULL, 1736 CTLFLAG_PERMANENT, 1737 CTLTYPE_STRING, "available", 1738 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1739 NULL, 0, &tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1740 sysctl_createv(clog, 0, &congctl_node, NULL, 1741 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1742 CTLTYPE_STRING, "selected", 1743 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1744 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1745 CTL_CREATE, CTL_EOL); 1746 1747 sysctl_createv(clog, 0, NULL, NULL, 1748 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1749 CTLTYPE_INT, "win_scale", 1750 SYSCTL_DESCR("Use RFC1323 window scale options"), 1751 NULL, 0, &tcp_do_win_scale, 0, 1752 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1753 sysctl_createv(clog, 0, NULL, NULL, 1754 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1755 CTLTYPE_INT, "timestamps", 1756 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1757 NULL, 0, &tcp_do_timestamps, 0, 1758 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1759 sysctl_createv(clog, 0, NULL, NULL, 1760 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1761 CTLTYPE_INT, "compat_42", 1762 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1763 NULL, 0, &tcp_compat_42, 0, 1764 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1765 sysctl_createv(clog, 0, NULL, NULL, 1766 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1767 CTLTYPE_INT, "cwm", 1768 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1769 "Monitoring"), 1770 NULL, 0, &tcp_cwm, 0, 1771 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1772 sysctl_createv(clog, 0, NULL, NULL, 1773 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1774 CTLTYPE_INT, "cwm_burstsize", 1775 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1776 "burst count in packets"), 1777 NULL, 0, &tcp_cwm_burstsize, 0, 1778 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1779 CTL_EOL); 1780 sysctl_createv(clog, 0, NULL, NULL, 1781 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1782 CTLTYPE_INT, "ack_on_push", 1783 SYSCTL_DESCR("Immediately return ACK when PSH is " 1784 "received"), 1785 NULL, 0, &tcp_ack_on_push, 0, 1786 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1787 sysctl_createv(clog, 0, NULL, NULL, 1788 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1789 CTLTYPE_INT, "keepidle", 1790 SYSCTL_DESCR("Allowed connection idle ticks before a " 1791 "keepalive probe is sent"), 1792 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1793 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1794 sysctl_createv(clog, 0, NULL, NULL, 1795 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1796 CTLTYPE_INT, "keepintvl", 1797 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1798 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1799 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1800 sysctl_createv(clog, 0, NULL, NULL, 1801 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1802 CTLTYPE_INT, "keepcnt", 1803 SYSCTL_DESCR("Number of keepalive probes to send"), 1804 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1805 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1806 sysctl_createv(clog, 0, NULL, NULL, 1807 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1808 CTLTYPE_INT, "slowhz", 1809 SYSCTL_DESCR("Keepalive ticks per second"), 1810 NULL, PR_SLOWHZ, NULL, 0, 1811 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1812 sysctl_createv(clog, 0, NULL, NULL, 1813 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1814 CTLTYPE_INT, "log_refused", 1815 SYSCTL_DESCR("Log refused TCP connections"), 1816 NULL, 0, &tcp_log_refused, 0, 1817 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1818 #if 0 /* obsoleted */ 1819 sysctl_createv(clog, 0, NULL, NULL, 1820 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1821 CTLTYPE_INT, "rstratelimit", NULL, 1822 NULL, 0, &tcp_rst_ratelim, 0, 1823 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1824 #endif 1825 sysctl_createv(clog, 0, NULL, NULL, 1826 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1827 CTLTYPE_INT, "rstppslimit", 1828 SYSCTL_DESCR("Maximum number of RST packets to send " 1829 "per second"), 1830 NULL, 0, &tcp_rst_ppslim, 0, 1831 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1832 sysctl_createv(clog, 0, NULL, NULL, 1833 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1834 CTLTYPE_INT, "delack_ticks", 1835 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1836 NULL, 0, &tcp_delack_ticks, 0, 1837 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1838 sysctl_createv(clog, 0, NULL, NULL, 1839 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1840 CTLTYPE_INT, "init_win_local", 1841 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1842 NULL, 0, &tcp_init_win_local, 0, 1843 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1844 CTL_EOL); 1845 sysctl_createv(clog, 0, NULL, NULL, 1846 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1847 CTLTYPE_STRUCT, "ident", 1848 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1849 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1850 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1851 sysctl_createv(clog, 0, NULL, NULL, 1852 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1853 CTLTYPE_INT, "do_loopback_cksum", 1854 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1855 NULL, 0, &tcp_do_loopback_cksum, 0, 1856 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1857 CTL_EOL); 1858 sysctl_createv(clog, 0, NULL, NULL, 1859 CTLFLAG_PERMANENT, 1860 CTLTYPE_STRUCT, "pcblist", 1861 SYSCTL_DESCR("TCP protocol control block list"), 1862 sysctl_inpcblist, 0, &tcbtable, 0, 1863 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1864 CTL_EOL); 1865 sysctl_createv(clog, 0, NULL, NULL, 1866 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1867 CTLTYPE_INT, "keepinit", 1868 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1869 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 1870 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1871 1872 /* TCP socket buffers auto-sizing nodes */ 1873 sysctl_createv(clog, 0, NULL, NULL, 1874 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1875 CTLTYPE_INT, "recvbuf_auto", 1876 SYSCTL_DESCR("Enable automatic receive " 1877 "buffer sizing (experimental)"), 1878 NULL, 0, &tcp_do_autorcvbuf, 0, 1879 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1880 sysctl_createv(clog, 0, NULL, NULL, 1881 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1882 CTLTYPE_INT, "recvbuf_inc", 1883 SYSCTL_DESCR("Incrementor step size of " 1884 "automatic receive buffer"), 1885 NULL, 0, &tcp_autorcvbuf_inc, 0, 1886 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1887 sysctl_createv(clog, 0, NULL, NULL, 1888 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1889 CTLTYPE_INT, "recvbuf_max", 1890 SYSCTL_DESCR("Max size of automatic receive buffer"), 1891 NULL, 0, &tcp_autorcvbuf_max, 0, 1892 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1893 1894 sysctl_createv(clog, 0, NULL, NULL, 1895 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1896 CTLTYPE_INT, "sendbuf_auto", 1897 SYSCTL_DESCR("Enable automatic send " 1898 "buffer sizing (experimental)"), 1899 NULL, 0, &tcp_do_autosndbuf, 0, 1900 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1901 sysctl_createv(clog, 0, NULL, NULL, 1902 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1903 CTLTYPE_INT, "sendbuf_inc", 1904 SYSCTL_DESCR("Incrementor step size of " 1905 "automatic send buffer"), 1906 NULL, 0, &tcp_autosndbuf_inc, 0, 1907 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1908 sysctl_createv(clog, 0, NULL, NULL, 1909 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1910 CTLTYPE_INT, "sendbuf_max", 1911 SYSCTL_DESCR("Max size of automatic send buffer"), 1912 NULL, 0, &tcp_autosndbuf_max, 0, 1913 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1914 1915 /* ECN subtree */ 1916 sysctl_createv(clog, 0, NULL, &ecn_node, 1917 CTLFLAG_PERMANENT, 1918 CTLTYPE_NODE, "ecn", 1919 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 1920 NULL, 0, NULL, 0, 1921 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1922 sysctl_createv(clog, 0, &ecn_node, NULL, 1923 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1924 CTLTYPE_INT, "enable", 1925 SYSCTL_DESCR("Enable TCP Explicit Congestion " 1926 "Notification"), 1927 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 1928 sysctl_createv(clog, 0, &ecn_node, NULL, 1929 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1930 CTLTYPE_INT, "maxretries", 1931 SYSCTL_DESCR("Number of times to retry ECN setup " 1932 "before disabling ECN on the connection"), 1933 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 1934 1935 /* SACK gets it's own little subtree. */ 1936 sysctl_createv(clog, 0, NULL, &sack_node, 1937 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1938 CTLTYPE_INT, "enable", 1939 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 1940 NULL, 0, &tcp_do_sack, 0, 1941 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1942 sysctl_createv(clog, 0, NULL, &sack_node, 1943 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1944 CTLTYPE_INT, "maxholes", 1945 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 1946 NULL, 0, &tcp_sack_tp_maxholes, 0, 1947 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1948 sysctl_createv(clog, 0, NULL, &sack_node, 1949 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1950 CTLTYPE_INT, "globalmaxholes", 1951 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 1952 NULL, 0, &tcp_sack_globalmaxholes, 0, 1953 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1954 sysctl_createv(clog, 0, NULL, &sack_node, 1955 CTLFLAG_PERMANENT, 1956 CTLTYPE_INT, "globalholes", 1957 SYSCTL_DESCR("Global number of TCP SACK holes"), 1958 NULL, 0, &tcp_sack_globalholes, 0, 1959 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1960 1961 sysctl_createv(clog, 0, NULL, NULL, 1962 CTLFLAG_PERMANENT, 1963 CTLTYPE_STRUCT, "stats", 1964 SYSCTL_DESCR("TCP statistics"), 1965 NULL, 0, &tcpstat, sizeof(tcpstat), 1966 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1967 CTL_EOL); 1968 #ifdef TCP_DEBUG 1969 sysctl_createv(clog, 0, NULL, NULL, 1970 CTLFLAG_PERMANENT, 1971 CTLTYPE_STRUCT, "debug", 1972 SYSCTL_DESCR("TCP sockets debug information"), 1973 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1974 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1975 CTL_EOL); 1976 sysctl_createv(clog, 0, NULL, NULL, 1977 CTLFLAG_PERMANENT, 1978 CTLTYPE_INT, "debx", 1979 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1980 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1981 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1982 CTL_EOL); 1983 #endif 1984 sysctl_createv(clog, 0, NULL, NULL, 1985 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1986 CTLTYPE_STRUCT, "drop", 1987 SYSCTL_DESCR("TCP drop connection"), 1988 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1989 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1990 #if NRND > 0 1991 sysctl_createv(clog, 0, NULL, NULL, 1992 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1993 CTLTYPE_INT, "iss_hash", 1994 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1995 "hash computation"), 1996 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1997 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1998 CTL_EOL); 1999 #endif 2000 2001 /* ABC subtree */ 2002 2003 sysctl_createv(clog, 0, NULL, &abc_node, 2004 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 2005 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 2006 NULL, 0, NULL, 0, 2007 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2008 sysctl_createv(clog, 0, &abc_node, NULL, 2009 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2010 CTLTYPE_INT, "enable", 2011 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 2012 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 2013 sysctl_createv(clog, 0, &abc_node, NULL, 2014 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2015 CTLTYPE_INT, "aggressive", 2016 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 2017 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 2018 } 2019 2020 /* 2021 * Sysctl for tcp variables. 2022 */ 2023 #ifdef INET 2024 SYSCTL_SETUP(sysctl_net_inet_tcp_setup, "sysctl net.inet.tcp subtree setup") 2025 { 2026 2027 sysctl_net_inet_tcp_setup2(clog, PF_INET, "inet", "tcp"); 2028 } 2029 #endif /* INET */ 2030 2031 #ifdef INET6 2032 SYSCTL_SETUP(sysctl_net_inet6_tcp6_setup, "sysctl net.inet6.tcp6 subtree setup") 2033 { 2034 2035 sysctl_net_inet_tcp_setup2(clog, PF_INET6, "inet6", "tcp6"); 2036 } 2037 #endif /* INET6 */ 2038
Indexes created Sun Sep 21 16:09:51 GMT 2025