tcp_usrreq.c revision 1.131.2.1
1 /* $NetBSD: tcp_usrreq.c,v 1.131.2.1 2007/07/15 13:27:59 ad 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 1.131.2.1 2007/07/15 13:27:59 ad 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 /* Compute window scaling to request. */ 434 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 435 (TCP_MAXWIN << tp->request_r_scale) < so->so_rcv.sb_hiwat) 436 tp->request_r_scale++; 437 soisconnecting(so); 438 tcpstat.tcps_connattempt++; 439 tp->t_state = TCPS_SYN_SENT; 440 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 441 tp->iss = tcp_new_iss(tp, 0); 442 tcp_sendseqinit(tp); 443 error = tcp_output(tp); 444 break; 445 446 /* 447 * Create a TCP connection between two sockets. 448 */ 449 case PRU_CONNECT2: 450 error = EOPNOTSUPP; 451 break; 452 453 /* 454 * Initiate disconnect from peer. 455 * If connection never passed embryonic stage, just drop; 456 * else if don't need to let data drain, then can just drop anyways, 457 * else have to begin TCP shutdown process: mark socket disconnecting, 458 * drain unread data, state switch to reflect user close, and 459 * send segment (e.g. FIN) to peer. Socket will be really disconnected 460 * when peer sends FIN and acks ours. 461 * 462 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 463 */ 464 case PRU_DISCONNECT: 465 tp = tcp_disconnect(tp); 466 break; 467 468 /* 469 * Accept a connection. Essentially all the work is 470 * done at higher levels; just return the address 471 * of the peer, storing through addr. 472 */ 473 case PRU_ACCEPT: 474 #ifdef INET 475 if (inp) 476 in_setpeeraddr(inp, nam); 477 #endif 478 #ifdef INET6 479 if (in6p) 480 in6_setpeeraddr(in6p, nam); 481 #endif 482 break; 483 484 /* 485 * Mark the connection as being incapable of further output. 486 */ 487 case PRU_SHUTDOWN: 488 socantsendmore(so); 489 tp = tcp_usrclosed(tp); 490 if (tp) 491 error = tcp_output(tp); 492 break; 493 494 /* 495 * After a receive, possibly send window update to peer. 496 */ 497 case PRU_RCVD: 498 /* 499 * soreceive() calls this function when a user receives 500 * ancillary data on a listening socket. We don't call 501 * tcp_output in such a case, since there is no header 502 * template for a listening socket and hence the kernel 503 * will panic. 504 */ 505 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 506 (void) tcp_output(tp); 507 break; 508 509 /* 510 * Do a send by putting data in output queue and updating urgent 511 * marker if URG set. Possibly send more data. 512 */ 513 case PRU_SEND: 514 if (control && control->m_len) { 515 m_freem(control); 516 m_freem(m); 517 error = EINVAL; 518 break; 519 } 520 sbappendstream(&so->so_snd, m); 521 error = tcp_output(tp); 522 break; 523 524 /* 525 * Abort the TCP. 526 */ 527 case PRU_ABORT: 528 tp = tcp_drop(tp, ECONNABORTED); 529 break; 530 531 case PRU_SENSE: 532 /* 533 * stat: don't bother with a blocksize. 534 */ 535 splx(s); 536 return (0); 537 538 case PRU_RCVOOB: 539 if (control && control->m_len) { 540 m_freem(control); 541 m_freem(m); 542 error = EINVAL; 543 break; 544 } 545 if ((so->so_oobmark == 0 && 546 (so->so_state & SS_RCVATMARK) == 0) || 547 so->so_options & SO_OOBINLINE || 548 tp->t_oobflags & TCPOOB_HADDATA) { 549 error = EINVAL; 550 break; 551 } 552 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 553 error = EWOULDBLOCK; 554 break; 555 } 556 m->m_len = 1; 557 *mtod(m, char *) = tp->t_iobc; 558 if (((long)nam & MSG_PEEK) == 0) 559 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 560 break; 561 562 case PRU_SENDOOB: 563 if (sbspace(&so->so_snd) < -512) { 564 m_freem(m); 565 error = ENOBUFS; 566 break; 567 } 568 /* 569 * According to RFC961 (Assigned Protocols), 570 * the urgent pointer points to the last octet 571 * of urgent data. We continue, however, 572 * to consider it to indicate the first octet 573 * of data past the urgent section. 574 * Otherwise, snd_up should be one lower. 575 */ 576 sbappendstream(&so->so_snd, m); 577 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 578 tp->t_force = 1; 579 error = tcp_output(tp); 580 tp->t_force = 0; 581 break; 582 583 case PRU_SOCKADDR: 584 #ifdef INET 585 if (inp) 586 in_setsockaddr(inp, nam); 587 #endif 588 #ifdef INET6 589 if (in6p) 590 in6_setsockaddr(in6p, nam); 591 #endif 592 break; 593 594 case PRU_PEERADDR: 595 #ifdef INET 596 if (inp) 597 in_setpeeraddr(inp, nam); 598 #endif 599 #ifdef INET6 600 if (in6p) 601 in6_setpeeraddr(in6p, nam); 602 #endif 603 break; 604 605 default: 606 panic("tcp_usrreq"); 607 } 608 #ifdef TCP_DEBUG 609 if (tp && (so->so_options & SO_DEBUG)) 610 tcp_trace(TA_USER, ostate, tp, NULL, req); 611 #endif 612 613 release: 614 splx(s); 615 return (error); 616 } 617 618 static void 619 change_keepalive(struct socket *so, struct tcpcb *tp) 620 { 621 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 622 TCP_TIMER_DISARM(tp, TCPT_KEEP); 623 TCP_TIMER_DISARM(tp, TCPT_2MSL); 624 625 if (tp->t_state == TCPS_SYN_RECEIVED || 626 tp->t_state == TCPS_SYN_SENT) { 627 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 628 } else if (so->so_options & SO_KEEPALIVE && 629 tp->t_state <= TCPS_CLOSE_WAIT) { 630 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 631 } else { 632 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 633 } 634 635 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 636 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 637 } 638 639 640 int 641 tcp_ctloutput(int op, struct socket *so, int level, int optname, 642 struct mbuf **mp) 643 { 644 int error = 0, s; 645 struct inpcb *inp; 646 #ifdef INET6 647 struct in6pcb *in6p; 648 #endif 649 struct tcpcb *tp; 650 struct mbuf *m; 651 int i; 652 u_int ui; 653 int family; /* family of the socket */ 654 655 family = so->so_proto->pr_domain->dom_family; 656 657 s = splsoftnet(); 658 switch (family) { 659 #ifdef INET 660 case PF_INET: 661 inp = sotoinpcb(so); 662 #ifdef INET6 663 in6p = NULL; 664 #endif 665 break; 666 #endif 667 #ifdef INET6 668 case PF_INET6: 669 inp = NULL; 670 in6p = sotoin6pcb(so); 671 break; 672 #endif 673 default: 674 splx(s); 675 panic("%s: af %d", __func__, family); 676 } 677 #ifndef INET6 678 if (inp == NULL) 679 #else 680 if (inp == NULL && in6p == NULL) 681 #endif 682 { 683 splx(s); 684 if (op == PRCO_SETOPT && *mp) 685 (void) m_free(*mp); 686 return (ECONNRESET); 687 } 688 if (level != IPPROTO_TCP) { 689 switch (family) { 690 #ifdef INET 691 case PF_INET: 692 error = ip_ctloutput(op, so, level, optname, mp); 693 break; 694 #endif 695 #ifdef INET6 696 case PF_INET6: 697 error = ip6_ctloutput(op, so, level, optname, mp); 698 break; 699 #endif 700 } 701 splx(s); 702 return (error); 703 } 704 if (inp) 705 tp = intotcpcb(inp); 706 #ifdef INET6 707 else if (in6p) 708 tp = in6totcpcb(in6p); 709 #endif 710 else 711 tp = NULL; 712 713 switch (op) { 714 715 case PRCO_SETOPT: 716 m = *mp; 717 switch (optname) { 718 719 #ifdef TCP_SIGNATURE 720 case TCP_MD5SIG: 721 if (m == NULL || m->m_len < sizeof (int)) 722 error = EINVAL; 723 if (error) 724 break; 725 if (*mtod(m, int *) > 0) 726 tp->t_flags |= TF_SIGNATURE; 727 else 728 tp->t_flags &= ~TF_SIGNATURE; 729 break; 730 #endif /* TCP_SIGNATURE */ 731 732 case TCP_NODELAY: 733 if (m == NULL || m->m_len < sizeof (int)) 734 error = EINVAL; 735 else if (*mtod(m, int *)) 736 tp->t_flags |= TF_NODELAY; 737 else 738 tp->t_flags &= ~TF_NODELAY; 739 break; 740 741 case TCP_MAXSEG: 742 if (m && m->m_len >= sizeof(int) && 743 (i = *mtod(m, int *)) > 0 && 744 i <= tp->t_peermss) 745 tp->t_peermss = i; /* limit on send size */ 746 else 747 error = EINVAL; 748 break; 749 #ifdef notyet 750 case TCP_CONGCTL: 751 if (m == NULL) 752 error = EINVAL; 753 error = tcp_congctl_select(tp, mtod(m, char *)); 754 #endif 755 break; 756 757 case TCP_KEEPIDLE: 758 if (m && m->m_len >= sizeof(u_int) && 759 (ui = *mtod(m, u_int *)) > 0) { 760 tp->t_keepidle = ui; 761 change_keepalive(so, tp); 762 } else 763 error = EINVAL; 764 break; 765 766 case TCP_KEEPINTVL: 767 if (m && m->m_len >= sizeof(u_int) && 768 (ui = *mtod(m, u_int *)) > 0) { 769 tp->t_keepintvl = ui; 770 change_keepalive(so, tp); 771 } else 772 error = EINVAL; 773 break; 774 775 case TCP_KEEPCNT: 776 if (m && m->m_len >= sizeof(u_int) && 777 (ui = *mtod(m, u_int *)) > 0) { 778 tp->t_keepcnt = ui; 779 change_keepalive(so, tp); 780 } else 781 error = EINVAL; 782 break; 783 784 case TCP_KEEPINIT: 785 if (m && m->m_len >= sizeof(u_int) && 786 (ui = *mtod(m, u_int *)) > 0) { 787 tp->t_keepinit = ui; 788 change_keepalive(so, tp); 789 } else 790 error = EINVAL; 791 break; 792 793 default: 794 error = ENOPROTOOPT; 795 break; 796 } 797 if (m) 798 (void) m_free(m); 799 break; 800 801 case PRCO_GETOPT: 802 *mp = m = m_get(M_WAIT, MT_SOOPTS); 803 m->m_len = sizeof(int); 804 MCLAIM(m, so->so_mowner); 805 806 switch (optname) { 807 #ifdef TCP_SIGNATURE 808 case TCP_MD5SIG: 809 *mtod(m, int *) = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 810 break; 811 #endif 812 case TCP_NODELAY: 813 *mtod(m, int *) = tp->t_flags & TF_NODELAY; 814 break; 815 case TCP_MAXSEG: 816 *mtod(m, int *) = tp->t_peermss; 817 break; 818 #ifdef notyet 819 case TCP_CONGCTL: 820 break; 821 #endif 822 default: 823 error = ENOPROTOOPT; 824 break; 825 } 826 break; 827 } 828 splx(s); 829 return (error); 830 } 831 832 #ifndef TCP_SENDSPACE 833 #define TCP_SENDSPACE 1024*32 834 #endif 835 int tcp_sendspace = TCP_SENDSPACE; 836 #ifndef TCP_RECVSPACE 837 #define TCP_RECVSPACE 1024*32 838 #endif 839 int tcp_recvspace = TCP_RECVSPACE; 840 841 /* 842 * Attach TCP protocol to socket, allocating 843 * internet protocol control block, tcp control block, 844 * bufer space, and entering LISTEN state if to accept connections. 845 */ 846 int 847 tcp_attach(struct socket *so) 848 { 849 struct tcpcb *tp; 850 struct inpcb *inp; 851 #ifdef INET6 852 struct in6pcb *in6p; 853 #endif 854 int error; 855 int family; /* family of the socket */ 856 857 family = so->so_proto->pr_domain->dom_family; 858 859 #ifdef MBUFTRACE 860 so->so_mowner = &tcp_sock_mowner; 861 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 862 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 863 #endif 864 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 865 error = soreserve(so, tcp_sendspace, tcp_recvspace); 866 if (error) 867 return (error); 868 } 869 switch (family) { 870 #ifdef INET 871 case PF_INET: 872 error = in_pcballoc(so, &tcbtable); 873 if (error) 874 return (error); 875 inp = sotoinpcb(so); 876 #ifdef INET6 877 in6p = NULL; 878 #endif 879 break; 880 #endif 881 #ifdef INET6 882 case PF_INET6: 883 error = in6_pcballoc(so, &tcbtable); 884 if (error) 885 return (error); 886 inp = NULL; 887 in6p = sotoin6pcb(so); 888 break; 889 #endif 890 default: 891 return EAFNOSUPPORT; 892 } 893 if (inp) 894 tp = tcp_newtcpcb(family, (void *)inp); 895 #ifdef INET6 896 else if (in6p) 897 tp = tcp_newtcpcb(family, (void *)in6p); 898 #endif 899 else 900 tp = NULL; 901 902 if (tp == 0) { 903 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 904 905 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 906 #ifdef INET 907 if (inp) 908 in_pcbdetach(inp); 909 #endif 910 #ifdef INET6 911 if (in6p) 912 in6_pcbdetach(in6p); 913 #endif 914 so->so_state |= nofd; 915 return (ENOBUFS); 916 } 917 tp->t_state = TCPS_CLOSED; 918 return (0); 919 } 920 921 /* 922 * Initiate (or continue) disconnect. 923 * If embryonic state, just send reset (once). 924 * If in ``let data drain'' option and linger null, just drop. 925 * Otherwise (hard), mark socket disconnecting and drop 926 * current input data; switch states based on user close, and 927 * send segment to peer (with FIN). 928 */ 929 struct tcpcb * 930 tcp_disconnect(struct tcpcb *tp) 931 { 932 struct socket *so; 933 934 if (tp->t_inpcb) 935 so = tp->t_inpcb->inp_socket; 936 #ifdef INET6 937 else if (tp->t_in6pcb) 938 so = tp->t_in6pcb->in6p_socket; 939 #endif 940 else 941 so = NULL; 942 943 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 944 tp = tcp_close(tp); 945 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 946 tp = tcp_drop(tp, 0); 947 else { 948 soisdisconnecting(so); 949 sbflush(&so->so_rcv); 950 tp = tcp_usrclosed(tp); 951 if (tp) 952 (void) tcp_output(tp); 953 } 954 return (tp); 955 } 956 957 /* 958 * User issued close, and wish to trail through shutdown states: 959 * if never received SYN, just forget it. If got a SYN from peer, 960 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 961 * If already got a FIN from peer, then almost done; go to LAST_ACK 962 * state. In all other cases, have already sent FIN to peer (e.g. 963 * after PRU_SHUTDOWN), and just have to play tedious game waiting 964 * for peer to send FIN or not respond to keep-alives, etc. 965 * We can let the user exit from the close as soon as the FIN is acked. 966 */ 967 struct tcpcb * 968 tcp_usrclosed(struct tcpcb *tp) 969 { 970 971 switch (tp->t_state) { 972 973 case TCPS_CLOSED: 974 case TCPS_LISTEN: 975 case TCPS_SYN_SENT: 976 tp->t_state = TCPS_CLOSED; 977 tp = tcp_close(tp); 978 break; 979 980 case TCPS_SYN_RECEIVED: 981 case TCPS_ESTABLISHED: 982 tp->t_state = TCPS_FIN_WAIT_1; 983 break; 984 985 case TCPS_CLOSE_WAIT: 986 tp->t_state = TCPS_LAST_ACK; 987 break; 988 } 989 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 990 struct socket *so; 991 if (tp->t_inpcb) 992 so = tp->t_inpcb->inp_socket; 993 #ifdef INET6 994 else if (tp->t_in6pcb) 995 so = tp->t_in6pcb->in6p_socket; 996 #endif 997 else 998 so = NULL; 999 if (so) 1000 soisdisconnected(so); 1001 /* 1002 * If we are in FIN_WAIT_2, we arrived here because the 1003 * application did a shutdown of the send side. Like the 1004 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1005 * a full close, we start a timer to make sure sockets are 1006 * not left in FIN_WAIT_2 forever. 1007 */ 1008 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1009 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1010 } 1011 return (tp); 1012 } 1013 1014 /* 1015 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1016 * than 32. 1017 */ 1018 static int 1019 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1020 { 1021 int error, mssdflt; 1022 struct sysctlnode node; 1023 1024 mssdflt = tcp_mssdflt; 1025 node = *rnode; 1026 node.sysctl_data = &mssdflt; 1027 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1028 if (error || newp == NULL) 1029 return (error); 1030 1031 if (mssdflt < 32) 1032 return (EINVAL); 1033 tcp_mssdflt = mssdflt; 1034 1035 return (0); 1036 } 1037 1038 /* 1039 * sysctl helper routine for setting port related values under 1040 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1041 * additional checks for each type. this code has placed in 1042 * tcp_input.c since INET and INET6 both use the same tcp code. 1043 * 1044 * this helper is not static so that both inet and inet6 can use it. 1045 */ 1046 int 1047 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1048 { 1049 int error, tmp; 1050 int apmin, apmax; 1051 #ifndef IPNOPRIVPORTS 1052 int lpmin, lpmax; 1053 #endif /* IPNOPRIVPORTS */ 1054 struct sysctlnode node; 1055 1056 if (namelen != 0) 1057 return (EINVAL); 1058 1059 switch (name[-3]) { 1060 #ifdef INET 1061 case PF_INET: 1062 apmin = anonportmin; 1063 apmax = anonportmax; 1064 #ifndef IPNOPRIVPORTS 1065 lpmin = lowportmin; 1066 lpmax = lowportmax; 1067 #endif /* IPNOPRIVPORTS */ 1068 break; 1069 #endif /* INET */ 1070 #ifdef INET6 1071 case PF_INET6: 1072 apmin = ip6_anonportmin; 1073 apmax = ip6_anonportmax; 1074 #ifndef IPNOPRIVPORTS 1075 lpmin = ip6_lowportmin; 1076 lpmax = ip6_lowportmax; 1077 #endif /* IPNOPRIVPORTS */ 1078 break; 1079 #endif /* INET6 */ 1080 default: 1081 return (EINVAL); 1082 } 1083 1084 /* 1085 * insert temporary copy into node, perform lookup on 1086 * temporary, then restore pointer 1087 */ 1088 node = *rnode; 1089 tmp = *(int*)rnode->sysctl_data; 1090 node.sysctl_data = &tmp; 1091 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1092 if (error || newp == NULL) 1093 return (error); 1094 1095 /* 1096 * simple port range check 1097 */ 1098 if (tmp < 0 || tmp > 65535) 1099 return (EINVAL); 1100 1101 /* 1102 * per-node range checks 1103 */ 1104 switch (rnode->sysctl_num) { 1105 case IPCTL_ANONPORTMIN: 1106 if (tmp >= apmax) 1107 return (EINVAL); 1108 #ifndef IPNOPRIVPORTS 1109 if (tmp < IPPORT_RESERVED) 1110 return (EINVAL); 1111 #endif /* IPNOPRIVPORTS */ 1112 break; 1113 1114 case IPCTL_ANONPORTMAX: 1115 if (apmin >= tmp) 1116 return (EINVAL); 1117 #ifndef IPNOPRIVPORTS 1118 if (tmp < IPPORT_RESERVED) 1119 return (EINVAL); 1120 #endif /* IPNOPRIVPORTS */ 1121 break; 1122 1123 #ifndef IPNOPRIVPORTS 1124 case IPCTL_LOWPORTMIN: 1125 if (tmp >= lpmax || 1126 tmp > IPPORT_RESERVEDMAX || 1127 tmp < IPPORT_RESERVEDMIN) 1128 return (EINVAL); 1129 break; 1130 1131 case IPCTL_LOWPORTMAX: 1132 if (lpmin >= tmp || 1133 tmp > IPPORT_RESERVEDMAX || 1134 tmp < IPPORT_RESERVEDMIN) 1135 return (EINVAL); 1136 break; 1137 #endif /* IPNOPRIVPORTS */ 1138 1139 default: 1140 return (EINVAL); 1141 } 1142 1143 *(int*)rnode->sysctl_data = tmp; 1144 1145 return (0); 1146 } 1147 1148 /* 1149 * The superuser can drop any connection. Normal users can only drop 1150 * their own connections. 1151 */ 1152 static inline int 1153 check_sockuid(struct socket *sockp, kauth_cred_t cred) 1154 { 1155 uid_t sockuid; 1156 1157 sockuid = sockp->so_uidinfo->ui_uid; 1158 if (kauth_authorize_generic(cred, KAUTH_GENERIC_ISSUSER, NULL) == 0 || 1159 sockuid == kauth_cred_getuid(cred) || 1160 sockuid == kauth_cred_geteuid(cred)) 1161 return 0; 1162 return EACCES; 1163 } 1164 1165 static inline int 1166 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1167 { 1168 size_t sz; 1169 int error; 1170 uid_t uid; 1171 1172 uid = sockp->so_uidinfo->ui_uid; 1173 if (oldp) { 1174 sz = MIN(sizeof(uid), *oldlenp); 1175 error = copyout(&uid, oldp, sz); 1176 if (error) 1177 return error; 1178 } 1179 *oldlenp = sizeof(uid); 1180 return 0; 1181 } 1182 1183 static inline int 1184 inet4_ident_core(struct in_addr raddr, u_int rport, 1185 struct in_addr laddr, u_int lport, 1186 void *oldp, size_t *oldlenp, 1187 struct lwp *l, int dodrop) 1188 { 1189 struct inpcb *inp; 1190 struct socket *sockp; 1191 1192 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport); 1193 1194 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1195 return ESRCH; 1196 1197 if (dodrop) { 1198 struct tcpcb *tp; 1199 1200 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1201 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1202 return ESRCH; 1203 1204 if (check_sockuid(inp->inp_socket, l->l_cred) != 0) 1205 return EACCES; 1206 1207 (void)tcp_drop(tp, ECONNABORTED); 1208 return 0; 1209 } 1210 else 1211 return copyout_uid(sockp, oldp, oldlenp); 1212 } 1213 1214 #ifdef INET6 1215 static inline int 1216 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1217 struct in6_addr *laddr, u_int lport, 1218 void *oldp, size_t *oldlenp, 1219 struct lwp *l, int dodrop) 1220 { 1221 struct in6pcb *in6p; 1222 struct socket *sockp; 1223 1224 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1225 1226 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1227 return ESRCH; 1228 1229 if (dodrop) { 1230 struct tcpcb *tp; 1231 1232 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1233 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1234 return ESRCH; 1235 1236 if (check_sockuid(in6p->in6p_socket, l->l_cred) != 0) 1237 return EACCES; 1238 1239 (void)tcp_drop(tp, ECONNABORTED); 1240 return 0; 1241 } 1242 else 1243 return copyout_uid(sockp, oldp, oldlenp); 1244 } 1245 #endif 1246 1247 /* 1248 * sysctl helper routine for the net.inet.tcp.drop and 1249 * net.inet6.tcp6.drop nodes. 1250 */ 1251 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1252 1253 /* 1254 * sysctl helper routine for the net.inet.tcp.ident and 1255 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1256 * old way of looking up the ident information for ipv4 which involves 1257 * stuffing the port/addr pairs into the mib lookup. 1258 */ 1259 static int 1260 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1261 { 1262 #ifdef INET 1263 struct sockaddr_in *si4[2]; 1264 #endif /* INET */ 1265 #ifdef INET6 1266 struct sockaddr_in6 *si6[2]; 1267 #endif /* INET6 */ 1268 struct sockaddr_storage sa[2]; 1269 int error, pf, dodrop, s; 1270 1271 dodrop = name[-1] == TCPCTL_DROP; 1272 if (dodrop) { 1273 if (oldp != NULL || *oldlenp != 0) 1274 return EINVAL; 1275 if (newp == NULL) 1276 return EPERM; 1277 if (newlen < sizeof(sa)) 1278 return ENOMEM; 1279 } 1280 if (namelen != 4 && namelen != 0) 1281 return EINVAL; 1282 if (name[-2] != IPPROTO_TCP) 1283 return EINVAL; 1284 pf = name[-3]; 1285 1286 /* old style lookup, ipv4 only */ 1287 if (namelen == 4) { 1288 #ifdef INET 1289 struct in_addr laddr, raddr; 1290 u_int lport, rport; 1291 1292 if (pf != PF_INET) 1293 return EPROTONOSUPPORT; 1294 raddr.s_addr = (uint32_t)name[0]; 1295 rport = (u_int)name[1]; 1296 laddr.s_addr = (uint32_t)name[2]; 1297 lport = (u_int)name[3]; 1298 1299 s = splsoftnet(); 1300 error = inet4_ident_core(raddr, rport, laddr, lport, 1301 oldp, oldlenp, l, dodrop); 1302 splx(s); 1303 return error; 1304 #else /* INET */ 1305 return EINVAL; 1306 #endif /* INET */ 1307 } 1308 1309 if (newp == NULL || newlen != sizeof(sa)) 1310 return EINVAL; 1311 error = copyin(newp, &sa, newlen); 1312 if (error) 1313 return error; 1314 1315 /* 1316 * requested families must match 1317 */ 1318 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1319 return EINVAL; 1320 1321 switch (pf) { 1322 #ifdef INET6 1323 case PF_INET6: 1324 si6[0] = (struct sockaddr_in6*)&sa[0]; 1325 si6[1] = (struct sockaddr_in6*)&sa[1]; 1326 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1327 si6[1]->sin6_len != sizeof(*si6[1])) 1328 return EINVAL; 1329 1330 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1331 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1332 error = sa6_embedscope(si6[0], ip6_use_defzone); 1333 if (error) 1334 return error; 1335 error = sa6_embedscope(si6[1], ip6_use_defzone); 1336 if (error) 1337 return error; 1338 1339 s = splsoftnet(); 1340 error = inet6_ident_core(&si6[0]->sin6_addr, 1341 si6[0]->sin6_port, &si6[1]->sin6_addr, 1342 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1343 splx(s); 1344 return error; 1345 } 1346 1347 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1348 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1349 return EINVAL; 1350 1351 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1352 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1353 /*FALLTHROUGH*/ 1354 #endif /* INET6 */ 1355 #ifdef INET 1356 case PF_INET: 1357 si4[0] = (struct sockaddr_in*)&sa[0]; 1358 si4[1] = (struct sockaddr_in*)&sa[1]; 1359 if (si4[0]->sin_len != sizeof(*si4[0]) || 1360 si4[0]->sin_len != sizeof(*si4[1])) 1361 return EINVAL; 1362 1363 s = splsoftnet(); 1364 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1365 si4[1]->sin_addr, si4[1]->sin_port, 1366 oldp, oldlenp, l, dodrop); 1367 splx(s); 1368 return error; 1369 #endif /* INET */ 1370 default: 1371 return EPROTONOSUPPORT; 1372 } 1373 } 1374 1375 /* 1376 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1377 * inet/inet6, as well as raw pcbs for each. specifically not 1378 * declared static so that raw sockets and udp/udp6 can use it as 1379 * well. 1380 */ 1381 int 1382 sysctl_inpcblist(SYSCTLFN_ARGS) 1383 { 1384 #ifdef INET 1385 struct sockaddr_in *in; 1386 const struct inpcb *inp; 1387 #endif 1388 #ifdef INET6 1389 struct sockaddr_in6 *in6; 1390 const struct in6pcb *in6p; 1391 #endif 1392 /* 1393 * sysctl_data is const, but CIRCLEQ_FOREACH can't use a const 1394 * struct inpcbtable pointer, so we have to discard const. :-/ 1395 */ 1396 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1397 const struct inpcb_hdr *inph; 1398 struct tcpcb *tp; 1399 struct kinfo_pcb pcb; 1400 char *dp; 1401 u_int op, arg; 1402 size_t len, needed, elem_size, out_size; 1403 int error, elem_count, pf, proto, pf2; 1404 1405 if (namelen != 4) 1406 return (EINVAL); 1407 1408 if (oldp != NULL) { 1409 len = *oldlenp; 1410 elem_size = name[2]; 1411 elem_count = name[3]; 1412 if (elem_size != sizeof(pcb)) 1413 return EINVAL; 1414 } else { 1415 len = 0; 1416 elem_count = INT_MAX; 1417 elem_size = sizeof(pcb); 1418 } 1419 error = 0; 1420 dp = oldp; 1421 op = name[0]; 1422 arg = name[1]; 1423 out_size = elem_size; 1424 needed = 0; 1425 1426 if (namelen == 1 && name[0] == CTL_QUERY) 1427 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1428 1429 if (name - oname != 4) 1430 return (EINVAL); 1431 1432 pf = oname[1]; 1433 proto = oname[2]; 1434 pf2 = (oldp != NULL) ? pf : 0; 1435 1436 CIRCLEQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1437 #ifdef INET 1438 inp = (const struct inpcb *)inph; 1439 #endif 1440 #ifdef INET6 1441 in6p = (const struct in6pcb *)inph; 1442 #endif 1443 1444 if (inph->inph_af != pf) 1445 continue; 1446 1447 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1448 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1449 NULL) != 0) 1450 continue; 1451 1452 memset(&pcb, 0, sizeof(pcb)); 1453 1454 pcb.ki_family = pf; 1455 pcb.ki_type = proto; 1456 1457 switch (pf2) { 1458 case 0: 1459 /* just probing for size */ 1460 break; 1461 #ifdef INET 1462 case PF_INET: 1463 pcb.ki_family = inp->inp_socket->so_proto-> 1464 pr_domain->dom_family; 1465 pcb.ki_type = inp->inp_socket->so_proto-> 1466 pr_type; 1467 pcb.ki_protocol = inp->inp_socket->so_proto-> 1468 pr_protocol; 1469 pcb.ki_pflags = inp->inp_flags; 1470 1471 pcb.ki_sostate = inp->inp_socket->so_state; 1472 pcb.ki_prstate = inp->inp_state; 1473 if (proto == IPPROTO_TCP) { 1474 tp = intotcpcb(inp); 1475 pcb.ki_tstate = tp->t_state; 1476 pcb.ki_tflags = tp->t_flags; 1477 } 1478 1479 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1480 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1481 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1482 1483 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1484 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1485 1486 in = satosin(&pcb.ki_src); 1487 in->sin_len = sizeof(*in); 1488 in->sin_family = pf; 1489 in->sin_port = inp->inp_lport; 1490 in->sin_addr = inp->inp_laddr; 1491 if (pcb.ki_prstate >= INP_CONNECTED) { 1492 in = satosin(&pcb.ki_dst); 1493 in->sin_len = sizeof(*in); 1494 in->sin_family = pf; 1495 in->sin_port = inp->inp_fport; 1496 in->sin_addr = inp->inp_faddr; 1497 } 1498 break; 1499 #endif 1500 #ifdef INET6 1501 case PF_INET6: 1502 pcb.ki_family = in6p->in6p_socket->so_proto-> 1503 pr_domain->dom_family; 1504 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1505 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1506 pr_protocol; 1507 pcb.ki_pflags = in6p->in6p_flags; 1508 1509 pcb.ki_sostate = in6p->in6p_socket->so_state; 1510 pcb.ki_prstate = in6p->in6p_state; 1511 if (proto == IPPROTO_TCP) { 1512 tp = in6totcpcb(in6p); 1513 pcb.ki_tstate = tp->t_state; 1514 pcb.ki_tflags = tp->t_flags; 1515 } 1516 1517 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1518 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1519 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1520 1521 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1522 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1523 1524 in6 = satosin6(&pcb.ki_src); 1525 in6->sin6_len = sizeof(*in6); 1526 in6->sin6_family = pf; 1527 in6->sin6_port = in6p->in6p_lport; 1528 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1529 in6->sin6_addr = in6p->in6p_laddr; 1530 in6->sin6_scope_id = 0; /* XXX? */ 1531 1532 if (pcb.ki_prstate >= IN6P_CONNECTED) { 1533 in6 = satosin6(&pcb.ki_dst); 1534 in6->sin6_len = sizeof(*in6); 1535 in6->sin6_family = pf; 1536 in6->sin6_port = in6p->in6p_fport; 1537 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1538 in6->sin6_addr = in6p->in6p_faddr; 1539 in6->sin6_scope_id = 0; /* XXX? */ 1540 } 1541 break; 1542 #endif 1543 } 1544 1545 if (len >= elem_size && elem_count > 0) { 1546 error = copyout(&pcb, dp, out_size); 1547 if (error) 1548 return (error); 1549 dp += elem_size; 1550 len -= elem_size; 1551 } 1552 if (elem_count > 0) { 1553 needed += elem_size; 1554 if (elem_count != INT_MAX) 1555 elem_count--; 1556 } 1557 } 1558 1559 *oldlenp = needed; 1560 if (oldp == NULL) 1561 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 1562 1563 return (error); 1564 } 1565 1566 static int 1567 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1568 { 1569 struct sysctlnode node; 1570 int error, r; 1571 char newname[TCPCC_MAXLEN]; 1572 1573 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1574 1575 node = *rnode; 1576 node.sysctl_data = newname; 1577 node.sysctl_size = sizeof(newname); 1578 1579 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1580 1581 if (error || 1582 newp == NULL || 1583 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1584 return error; 1585 1586 if ((r = tcp_congctl_select(NULL, newname))) 1587 return r; 1588 1589 return error; 1590 } 1591 1592 static int 1593 sysctl_tcp_keep(SYSCTLFN_ARGS) 1594 { 1595 int error; 1596 u_int tmp; 1597 struct sysctlnode node; 1598 1599 node = *rnode; 1600 tmp = *(u_int *)rnode->sysctl_data; 1601 node.sysctl_data = &tmp; 1602 1603 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1604 if (error || newp == NULL) 1605 return error; 1606 1607 *(u_int *)rnode->sysctl_data = tmp; 1608 tcp_tcpcb_template(); /* update the template */ 1609 return 0; 1610 } 1611 1612 1613 /* 1614 * this (second stage) setup routine is a replacement for tcp_sysctl() 1615 * (which is currently used for ipv4 and ipv6) 1616 */ 1617 static void 1618 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 1619 const char *tcpname) 1620 { 1621 const struct sysctlnode *sack_node; 1622 const struct sysctlnode *abc_node; 1623 const struct sysctlnode *ecn_node; 1624 const struct sysctlnode *congctl_node; 1625 #ifdef TCP_DEBUG 1626 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 1627 extern int tcp_debx; 1628 #endif 1629 1630 sysctl_createv(clog, 0, NULL, NULL, 1631 CTLFLAG_PERMANENT, 1632 CTLTYPE_NODE, "net", NULL, 1633 NULL, 0, NULL, 0, 1634 CTL_NET, CTL_EOL); 1635 sysctl_createv(clog, 0, NULL, NULL, 1636 CTLFLAG_PERMANENT, 1637 CTLTYPE_NODE, pfname, NULL, 1638 NULL, 0, NULL, 0, 1639 CTL_NET, pf, CTL_EOL); 1640 sysctl_createv(clog, 0, NULL, NULL, 1641 CTLFLAG_PERMANENT, 1642 CTLTYPE_NODE, tcpname, 1643 SYSCTL_DESCR("TCP related settings"), 1644 NULL, 0, NULL, 0, 1645 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 1646 1647 sysctl_createv(clog, 0, NULL, NULL, 1648 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1649 CTLTYPE_INT, "rfc1323", 1650 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 1651 NULL, 0, &tcp_do_rfc1323, 0, 1652 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 1653 sysctl_createv(clog, 0, NULL, NULL, 1654 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1655 CTLTYPE_INT, "sendspace", 1656 SYSCTL_DESCR("Default TCP send buffer size"), 1657 NULL, 0, &tcp_sendspace, 0, 1658 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 1659 sysctl_createv(clog, 0, NULL, NULL, 1660 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1661 CTLTYPE_INT, "recvspace", 1662 SYSCTL_DESCR("Default TCP receive buffer size"), 1663 NULL, 0, &tcp_recvspace, 0, 1664 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 1665 sysctl_createv(clog, 0, NULL, NULL, 1666 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1667 CTLTYPE_INT, "mssdflt", 1668 SYSCTL_DESCR("Default maximum segment size"), 1669 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 1670 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 1671 sysctl_createv(clog, 0, NULL, NULL, 1672 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1673 CTLTYPE_INT, "syn_cache_limit", 1674 SYSCTL_DESCR("Maximum number of entries in the TCP " 1675 "compressed state engine"), 1676 NULL, 0, &tcp_syn_cache_limit, 0, 1677 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 1678 CTL_EOL); 1679 sysctl_createv(clog, 0, NULL, NULL, 1680 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1681 CTLTYPE_INT, "syn_bucket_limit", 1682 SYSCTL_DESCR("Maximum number of entries per hash " 1683 "bucket in the TCP compressed state " 1684 "engine"), 1685 NULL, 0, &tcp_syn_bucket_limit, 0, 1686 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 1687 CTL_EOL); 1688 #if 0 /* obsoleted */ 1689 sysctl_createv(clog, 0, NULL, NULL, 1690 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1691 CTLTYPE_INT, "syn_cache_interval", 1692 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 1693 NULL, 0, &tcp_syn_cache_interval, 0, 1694 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 1695 CTL_EOL); 1696 #endif 1697 sysctl_createv(clog, 0, NULL, NULL, 1698 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1699 CTLTYPE_INT, "init_win", 1700 SYSCTL_DESCR("Initial TCP congestion window"), 1701 NULL, 0, &tcp_init_win, 0, 1702 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 1703 sysctl_createv(clog, 0, NULL, NULL, 1704 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1705 CTLTYPE_INT, "mss_ifmtu", 1706 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 1707 NULL, 0, &tcp_mss_ifmtu, 0, 1708 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 1709 sysctl_createv(clog, 0, NULL, &sack_node, 1710 CTLFLAG_PERMANENT, 1711 CTLTYPE_NODE, "sack", 1712 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 1713 NULL, 0, NULL, 0, 1714 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 1715 1716 /* Congctl subtree */ 1717 sysctl_createv(clog, 0, NULL, &congctl_node, 1718 CTLFLAG_PERMANENT, 1719 CTLTYPE_NODE, "congctl", 1720 SYSCTL_DESCR("TCP Congestion Control"), 1721 NULL, 0, NULL, 0, 1722 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1723 sysctl_createv(clog, 0, &congctl_node, NULL, 1724 CTLFLAG_PERMANENT, 1725 CTLTYPE_STRING, "available", 1726 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 1727 NULL, 0, &tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 1728 sysctl_createv(clog, 0, &congctl_node, NULL, 1729 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1730 CTLTYPE_STRING, "selected", 1731 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 1732 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 1733 CTL_CREATE, CTL_EOL); 1734 1735 sysctl_createv(clog, 0, NULL, NULL, 1736 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1737 CTLTYPE_INT, "win_scale", 1738 SYSCTL_DESCR("Use RFC1323 window scale options"), 1739 NULL, 0, &tcp_do_win_scale, 0, 1740 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 1741 sysctl_createv(clog, 0, NULL, NULL, 1742 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1743 CTLTYPE_INT, "timestamps", 1744 SYSCTL_DESCR("Use RFC1323 time stamp options"), 1745 NULL, 0, &tcp_do_timestamps, 0, 1746 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 1747 sysctl_createv(clog, 0, NULL, NULL, 1748 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1749 CTLTYPE_INT, "compat_42", 1750 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 1751 NULL, 0, &tcp_compat_42, 0, 1752 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 1753 sysctl_createv(clog, 0, NULL, NULL, 1754 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1755 CTLTYPE_INT, "cwm", 1756 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 1757 "Monitoring"), 1758 NULL, 0, &tcp_cwm, 0, 1759 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 1760 sysctl_createv(clog, 0, NULL, NULL, 1761 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1762 CTLTYPE_INT, "cwm_burstsize", 1763 SYSCTL_DESCR("Congestion Window Monitoring allowed " 1764 "burst count in packets"), 1765 NULL, 0, &tcp_cwm_burstsize, 0, 1766 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 1767 CTL_EOL); 1768 sysctl_createv(clog, 0, NULL, NULL, 1769 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1770 CTLTYPE_INT, "ack_on_push", 1771 SYSCTL_DESCR("Immediately return ACK when PSH is " 1772 "received"), 1773 NULL, 0, &tcp_ack_on_push, 0, 1774 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 1775 sysctl_createv(clog, 0, NULL, NULL, 1776 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1777 CTLTYPE_INT, "keepidle", 1778 SYSCTL_DESCR("Allowed connection idle ticks before a " 1779 "keepalive probe is sent"), 1780 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 1781 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 1782 sysctl_createv(clog, 0, NULL, NULL, 1783 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1784 CTLTYPE_INT, "keepintvl", 1785 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 1786 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 1787 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 1788 sysctl_createv(clog, 0, NULL, NULL, 1789 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1790 CTLTYPE_INT, "keepcnt", 1791 SYSCTL_DESCR("Number of keepalive probes to send"), 1792 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 1793 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 1794 sysctl_createv(clog, 0, NULL, NULL, 1795 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 1796 CTLTYPE_INT, "slowhz", 1797 SYSCTL_DESCR("Keepalive ticks per second"), 1798 NULL, PR_SLOWHZ, NULL, 0, 1799 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 1800 sysctl_createv(clog, 0, NULL, NULL, 1801 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1802 CTLTYPE_INT, "log_refused", 1803 SYSCTL_DESCR("Log refused TCP connections"), 1804 NULL, 0, &tcp_log_refused, 0, 1805 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 1806 #if 0 /* obsoleted */ 1807 sysctl_createv(clog, 0, NULL, NULL, 1808 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1809 CTLTYPE_INT, "rstratelimit", NULL, 1810 NULL, 0, &tcp_rst_ratelim, 0, 1811 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 1812 #endif 1813 sysctl_createv(clog, 0, NULL, NULL, 1814 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1815 CTLTYPE_INT, "rstppslimit", 1816 SYSCTL_DESCR("Maximum number of RST packets to send " 1817 "per second"), 1818 NULL, 0, &tcp_rst_ppslim, 0, 1819 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 1820 sysctl_createv(clog, 0, NULL, NULL, 1821 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1822 CTLTYPE_INT, "delack_ticks", 1823 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 1824 NULL, 0, &tcp_delack_ticks, 0, 1825 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 1826 sysctl_createv(clog, 0, NULL, NULL, 1827 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1828 CTLTYPE_INT, "init_win_local", 1829 SYSCTL_DESCR("Initial TCP window size (in segments)"), 1830 NULL, 0, &tcp_init_win_local, 0, 1831 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 1832 CTL_EOL); 1833 sysctl_createv(clog, 0, NULL, NULL, 1834 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1835 CTLTYPE_STRUCT, "ident", 1836 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 1837 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 1838 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 1839 sysctl_createv(clog, 0, NULL, NULL, 1840 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1841 CTLTYPE_INT, "do_loopback_cksum", 1842 SYSCTL_DESCR("Perform TCP checksum on loopback"), 1843 NULL, 0, &tcp_do_loopback_cksum, 0, 1844 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 1845 CTL_EOL); 1846 sysctl_createv(clog, 0, NULL, NULL, 1847 CTLFLAG_PERMANENT, 1848 CTLTYPE_STRUCT, "pcblist", 1849 SYSCTL_DESCR("TCP protocol control block list"), 1850 sysctl_inpcblist, 0, &tcbtable, 0, 1851 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1852 CTL_EOL); 1853 sysctl_createv(clog, 0, NULL, NULL, 1854 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1855 CTLTYPE_INT, "keepinit", 1856 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 1857 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 1858 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1859 1860 /* ECN subtree */ 1861 sysctl_createv(clog, 0, NULL, &ecn_node, 1862 CTLFLAG_PERMANENT, 1863 CTLTYPE_NODE, "ecn", 1864 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 1865 NULL, 0, NULL, 0, 1866 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1867 sysctl_createv(clog, 0, &ecn_node, NULL, 1868 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1869 CTLTYPE_INT, "enable", 1870 SYSCTL_DESCR("Enable TCP Explicit Congestion " 1871 "Notification"), 1872 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 1873 sysctl_createv(clog, 0, &ecn_node, NULL, 1874 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1875 CTLTYPE_INT, "maxretries", 1876 SYSCTL_DESCR("Number of times to retry ECN setup " 1877 "before disabling ECN on the connection"), 1878 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 1879 1880 /* SACK gets it's own little subtree. */ 1881 sysctl_createv(clog, 0, NULL, &sack_node, 1882 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1883 CTLTYPE_INT, "enable", 1884 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 1885 NULL, 0, &tcp_do_sack, 0, 1886 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1887 sysctl_createv(clog, 0, NULL, &sack_node, 1888 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1889 CTLTYPE_INT, "maxholes", 1890 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 1891 NULL, 0, &tcp_sack_tp_maxholes, 0, 1892 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1893 sysctl_createv(clog, 0, NULL, &sack_node, 1894 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1895 CTLTYPE_INT, "globalmaxholes", 1896 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 1897 NULL, 0, &tcp_sack_globalmaxholes, 0, 1898 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1899 sysctl_createv(clog, 0, NULL, &sack_node, 1900 CTLFLAG_PERMANENT, 1901 CTLTYPE_INT, "globalholes", 1902 SYSCTL_DESCR("Global number of TCP SACK holes"), 1903 NULL, 0, &tcp_sack_globalholes, 0, 1904 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 1905 1906 sysctl_createv(clog, 0, NULL, NULL, 1907 CTLFLAG_PERMANENT, 1908 CTLTYPE_STRUCT, "stats", 1909 SYSCTL_DESCR("TCP statistics"), 1910 NULL, 0, &tcpstat, sizeof(tcpstat), 1911 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 1912 CTL_EOL); 1913 #ifdef TCP_DEBUG 1914 sysctl_createv(clog, 0, NULL, NULL, 1915 CTLFLAG_PERMANENT, 1916 CTLTYPE_STRUCT, "debug", 1917 SYSCTL_DESCR("TCP sockets debug information"), 1918 NULL, 0, &tcp_debug, sizeof(tcp_debug), 1919 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 1920 CTL_EOL); 1921 sysctl_createv(clog, 0, NULL, NULL, 1922 CTLFLAG_PERMANENT, 1923 CTLTYPE_INT, "debx", 1924 SYSCTL_DESCR("Number of TCP debug sockets messages"), 1925 NULL, 0, &tcp_debx, sizeof(tcp_debx), 1926 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 1927 CTL_EOL); 1928 #endif 1929 sysctl_createv(clog, 0, NULL, NULL, 1930 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1931 CTLTYPE_STRUCT, "drop", 1932 SYSCTL_DESCR("TCP drop connection"), 1933 sysctl_net_inet_tcp_drop, 0, NULL, 0, 1934 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 1935 #if NRND > 0 1936 sysctl_createv(clog, 0, NULL, NULL, 1937 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1938 CTLTYPE_INT, "iss_hash", 1939 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 1940 "hash computation"), 1941 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 1942 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 1943 CTL_EOL); 1944 #endif 1945 1946 /* ABC subtree */ 1947 1948 sysctl_createv(clog, 0, NULL, &abc_node, 1949 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 1950 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 1951 NULL, 0, NULL, 0, 1952 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 1953 sysctl_createv(clog, 0, &abc_node, NULL, 1954 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1955 CTLTYPE_INT, "enable", 1956 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 1957 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 1958 sysctl_createv(clog, 0, &abc_node, NULL, 1959 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 1960 CTLTYPE_INT, "aggressive", 1961 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 1962 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 1963 } 1964 1965 /* 1966 * Sysctl for tcp variables. 1967 */ 1968 #ifdef INET 1969 SYSCTL_SETUP(sysctl_net_inet_tcp_setup, "sysctl net.inet.tcp subtree setup") 1970 { 1971 1972 sysctl_net_inet_tcp_setup2(clog, PF_INET, "inet", "tcp"); 1973 } 1974 #endif /* INET */ 1975 1976 #ifdef INET6 1977 SYSCTL_SETUP(sysctl_net_inet6_tcp6_setup, "sysctl net.inet6.tcp6 subtree setup") 1978 { 1979 1980 sysctl_net_inet_tcp_setup2(clog, PF_INET6, "inet6", "tcp6"); 1981 } 1982 #endif /* INET6 */ 1983
Indexes created Sun Sep 21 20:09:37 GMT 2025