tcp_usrreq.c revision 1.200.2.2.2.1
1 /* $NetBSD: tcp_usrreq.c,v 1.200.2.2.2.1 2019/01/29 08:09:00 msaitoh Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1997, 1998, 2005, 2006 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 38 * Facility, NASA Ames Research Center. 39 * This code is derived from software contributed to The NetBSD Foundation 40 * by Charles M. Hannum. 41 * This code is derived from software contributed to The NetBSD Foundation 42 * by Rui Paulo. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 54 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 55 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 56 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 57 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 58 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 59 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 60 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 61 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 62 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 63 * POSSIBILITY OF SUCH DAMAGE. 64 */ 65 66 /* 67 * Copyright (c) 1982, 1986, 1988, 1993, 1995 68 * The Regents of the University of California. All rights reserved. 69 * 70 * Redistribution and use in source and binary forms, with or without 71 * modification, are permitted provided that the following conditions 72 * are met: 73 * 1. Redistributions of source code must retain the above copyright 74 * notice, this list of conditions and the following disclaimer. 75 * 2. Redistributions in binary form must reproduce the above copyright 76 * notice, this list of conditions and the following disclaimer in the 77 * documentation and/or other materials provided with the distribution. 78 * 3. Neither the name of the University nor the names of its contributors 79 * may be used to endorse or promote products derived from this software 80 * without specific prior written permission. 81 * 82 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 83 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 84 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 85 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 86 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 87 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 88 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 89 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 90 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 91 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 92 * SUCH DAMAGE. 93 * 94 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 95 */ 96 97 /* 98 * TCP protocol interface to socket abstraction. 99 */ 100 101 #include <sys/cdefs.h> 102 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.200.2.2.2.1 2019/01/29 08:09:00 msaitoh Exp $"); 103 104 #include "opt_inet.h" 105 #include "opt_ipsec.h" 106 #include "opt_tcp_debug.h" 107 #include "opt_mbuftrace.h" 108 109 #include <sys/param.h> 110 #include <sys/systm.h> 111 #include <sys/kernel.h> 112 #include <sys/mbuf.h> 113 #include <sys/socket.h> 114 #include <sys/socketvar.h> 115 #include <sys/protosw.h> 116 #include <sys/errno.h> 117 #include <sys/stat.h> 118 #include <sys/proc.h> 119 #include <sys/domain.h> 120 #include <sys/sysctl.h> 121 #include <sys/kauth.h> 122 #include <sys/kernel.h> 123 #include <sys/uidinfo.h> 124 125 #include <net/if.h> 126 #include <net/route.h> 127 128 #include <netinet/in.h> 129 #include <netinet/in_systm.h> 130 #include <netinet/in_var.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/ip_var.h> 134 #include <netinet/in_offload.h> 135 136 #ifdef INET6 137 #ifndef INET 138 #include <netinet/in.h> 139 #endif 140 #include <netinet/ip6.h> 141 #include <netinet6/in6_pcb.h> 142 #include <netinet6/ip6_var.h> 143 #include <netinet6/scope6_var.h> 144 #endif 145 146 #include <netinet/tcp.h> 147 #include <netinet/tcp_fsm.h> 148 #include <netinet/tcp_seq.h> 149 #include <netinet/tcp_timer.h> 150 #include <netinet/tcp_var.h> 151 #include <netinet/tcp_private.h> 152 #include <netinet/tcp_congctl.h> 153 #include <netinet/tcpip.h> 154 #include <netinet/tcp_debug.h> 155 #include <netinet/tcp_vtw.h> 156 157 #include "opt_tcp_space.h" 158 159 static int 160 tcp_debug_capture(struct tcpcb *tp, int req) 161 { 162 #ifdef KPROF 163 tcp_acounts[tp->t_state][req]++; 164 #endif 165 #ifdef TCP_DEBUG 166 return tp->t_state; 167 #endif 168 return 0; 169 } 170 171 static inline void 172 tcp_debug_trace(struct socket *so, struct tcpcb *tp, int ostate, int req) 173 { 174 #ifdef TCP_DEBUG 175 if (tp && (so->so_options & SO_DEBUG)) 176 tcp_trace(TA_USER, ostate, tp, NULL, req); 177 #endif 178 } 179 180 static int 181 tcp_getpcb(struct socket *so, struct inpcb **inp, 182 struct in6pcb **in6p, struct tcpcb **tp) 183 { 184 185 KASSERT(solocked(so)); 186 187 /* 188 * When a TCP is attached to a socket, then there will be 189 * a (struct inpcb) pointed at by the socket, and this 190 * structure will point at a subsidary (struct tcpcb). 191 */ 192 switch (so->so_proto->pr_domain->dom_family) { 193 #ifdef INET 194 case PF_INET: 195 *inp = sotoinpcb(so); 196 if (*inp == NULL) 197 return EINVAL; 198 *tp = intotcpcb(*inp); 199 break; 200 #endif 201 #ifdef INET6 202 case PF_INET6: 203 *in6p = sotoin6pcb(so); 204 if (*in6p == NULL) 205 return EINVAL; 206 *tp = in6totcpcb(*in6p); 207 break; 208 #endif 209 default: 210 return EAFNOSUPPORT; 211 } 212 213 KASSERT(tp != NULL); 214 215 return 0; 216 } 217 218 /* 219 * Process a TCP user request for TCP tb. If this is a send request 220 * then m is the mbuf chain of send data. If this is a timer expiration 221 * (called from the software clock routine), then timertype tells which timer. 222 */ 223 static int 224 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 225 struct mbuf *control, struct lwp *l) 226 { 227 KASSERT(req != PRU_ATTACH); 228 KASSERT(req != PRU_DETACH); 229 KASSERT(req != PRU_ACCEPT); 230 KASSERT(req != PRU_BIND); 231 KASSERT(req != PRU_LISTEN); 232 KASSERT(req != PRU_CONNECT); 233 KASSERT(req != PRU_CONNECT2); 234 KASSERT(req != PRU_DISCONNECT); 235 KASSERT(req != PRU_SHUTDOWN); 236 KASSERT(req != PRU_ABORT); 237 KASSERT(req != PRU_CONTROL); 238 KASSERT(req != PRU_SENSE); 239 KASSERT(req != PRU_PEERADDR); 240 KASSERT(req != PRU_SOCKADDR); 241 KASSERT(req != PRU_RCVD); 242 KASSERT(req != PRU_RCVOOB); 243 KASSERT(req != PRU_SEND); 244 KASSERT(req != PRU_SENDOOB); 245 KASSERT(req != PRU_PURGEIF); 246 247 KASSERT(solocked(so)); 248 249 panic("tcp_usrreq"); 250 251 return 0; 252 } 253 254 static void 255 change_keepalive(struct socket *so, struct tcpcb *tp) 256 { 257 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 258 TCP_TIMER_DISARM(tp, TCPT_KEEP); 259 TCP_TIMER_DISARM(tp, TCPT_2MSL); 260 261 if (tp->t_state == TCPS_SYN_RECEIVED || 262 tp->t_state == TCPS_SYN_SENT) { 263 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 264 } else if (so->so_options & SO_KEEPALIVE && 265 tp->t_state <= TCPS_CLOSE_WAIT) { 266 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 267 } else { 268 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 269 } 270 271 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 272 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 273 } 274 275 /* 276 * Export TCP internal state information via a struct tcp_info, based on the 277 * Linux 2.6 API. Not ABI compatible as our constants are mapped differently 278 * (TCP state machine, etc). We export all information using FreeBSD-native 279 * constants -- for example, the numeric values for tcpi_state will differ 280 * from Linux. 281 */ 282 static void 283 tcp_fill_info(struct tcpcb *tp, struct tcp_info *ti) 284 { 285 286 bzero(ti, sizeof(*ti)); 287 288 ti->tcpi_state = tp->t_state; 289 if ((tp->t_flags & TF_REQ_TSTMP) && (tp->t_flags & TF_RCVD_TSTMP)) 290 ti->tcpi_options |= TCPI_OPT_TIMESTAMPS; 291 if (tp->t_flags & TF_SACK_PERMIT) 292 ti->tcpi_options |= TCPI_OPT_SACK; 293 if ((tp->t_flags & TF_REQ_SCALE) && (tp->t_flags & TF_RCVD_SCALE)) { 294 ti->tcpi_options |= TCPI_OPT_WSCALE; 295 ti->tcpi_snd_wscale = tp->snd_scale; 296 ti->tcpi_rcv_wscale = tp->rcv_scale; 297 } 298 if (tp->t_flags & TF_ECN_PERMIT) { 299 ti->tcpi_options |= TCPI_OPT_ECN; 300 } 301 302 ti->tcpi_rto = tp->t_rxtcur * tick; 303 ti->tcpi_last_data_recv = (long)(hardclock_ticks - 304 (int)tp->t_rcvtime) * tick; 305 ti->tcpi_rtt = ((u_int64_t)tp->t_srtt * tick) >> TCP_RTT_SHIFT; 306 ti->tcpi_rttvar = ((u_int64_t)tp->t_rttvar * tick) >> TCP_RTTVAR_SHIFT; 307 308 ti->tcpi_snd_ssthresh = tp->snd_ssthresh; 309 /* Linux API wants these in # of segments, apparently */ 310 ti->tcpi_snd_cwnd = tp->snd_cwnd / tp->t_segsz; 311 ti->tcpi_snd_wnd = tp->snd_wnd / tp->t_segsz; 312 313 /* 314 * FreeBSD-specific extension fields for tcp_info. 315 */ 316 ti->tcpi_rcv_space = tp->rcv_wnd; 317 ti->tcpi_rcv_nxt = tp->rcv_nxt; 318 ti->tcpi_snd_bwnd = 0; /* Unused, kept for compat. */ 319 ti->tcpi_snd_nxt = tp->snd_nxt; 320 ti->tcpi_snd_mss = tp->t_segsz; 321 ti->tcpi_rcv_mss = tp->t_segsz; 322 #ifdef TF_TOE 323 if (tp->t_flags & TF_TOE) 324 ti->tcpi_options |= TCPI_OPT_TOE; 325 #endif 326 /* From the redundant department of redundancies... */ 327 ti->__tcpi_retransmits = ti->__tcpi_retrans = 328 ti->tcpi_snd_rexmitpack = tp->t_sndrexmitpack; 329 330 ti->tcpi_rcv_ooopack = tp->t_rcvoopack; 331 ti->tcpi_snd_zerowin = tp->t_sndzerowin; 332 } 333 334 int 335 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 336 { 337 int error = 0, s; 338 struct inpcb *inp; 339 #ifdef INET6 340 struct in6pcb *in6p; 341 #endif 342 struct tcpcb *tp; 343 struct tcp_info ti; 344 u_int ui; 345 int family; /* family of the socket */ 346 int level, optname, optval; 347 348 level = sopt->sopt_level; 349 optname = sopt->sopt_name; 350 351 family = so->so_proto->pr_domain->dom_family; 352 353 s = splsoftnet(); 354 switch (family) { 355 #ifdef INET 356 case PF_INET: 357 inp = sotoinpcb(so); 358 #ifdef INET6 359 in6p = NULL; 360 #endif 361 break; 362 #endif 363 #ifdef INET6 364 case PF_INET6: 365 inp = NULL; 366 in6p = sotoin6pcb(so); 367 break; 368 #endif 369 default: 370 splx(s); 371 panic("%s: af %d", __func__, family); 372 } 373 #ifndef INET6 374 if (inp == NULL) 375 #else 376 if (inp == NULL && in6p == NULL) 377 #endif 378 { 379 splx(s); 380 return (ECONNRESET); 381 } 382 if (level != IPPROTO_TCP) { 383 switch (family) { 384 #ifdef INET 385 case PF_INET: 386 error = ip_ctloutput(op, so, sopt); 387 break; 388 #endif 389 #ifdef INET6 390 case PF_INET6: 391 error = ip6_ctloutput(op, so, sopt); 392 break; 393 #endif 394 } 395 splx(s); 396 return (error); 397 } 398 if (inp) 399 tp = intotcpcb(inp); 400 #ifdef INET6 401 else if (in6p) 402 tp = in6totcpcb(in6p); 403 #endif 404 else 405 tp = NULL; 406 407 switch (op) { 408 case PRCO_SETOPT: 409 switch (optname) { 410 #ifdef TCP_SIGNATURE 411 case TCP_MD5SIG: 412 error = sockopt_getint(sopt, &optval); 413 if (error) 414 break; 415 if (optval > 0) 416 tp->t_flags |= TF_SIGNATURE; 417 else 418 tp->t_flags &= ~TF_SIGNATURE; 419 break; 420 #endif /* TCP_SIGNATURE */ 421 422 case TCP_NODELAY: 423 error = sockopt_getint(sopt, &optval); 424 if (error) 425 break; 426 if (optval) 427 tp->t_flags |= TF_NODELAY; 428 else 429 tp->t_flags &= ~TF_NODELAY; 430 break; 431 432 case TCP_MAXSEG: 433 error = sockopt_getint(sopt, &optval); 434 if (error) 435 break; 436 if (optval > 0 && optval <= tp->t_peermss) 437 tp->t_peermss = optval; /* limit on send size */ 438 else 439 error = EINVAL; 440 break; 441 #ifdef notyet 442 case TCP_CONGCTL: 443 /* XXX string overflow XXX */ 444 error = tcp_congctl_select(tp, sopt->sopt_data); 445 break; 446 #endif 447 448 case TCP_KEEPIDLE: 449 error = sockopt_get(sopt, &ui, sizeof(ui)); 450 if (error) 451 break; 452 if (ui > 0) { 453 tp->t_keepidle = ui; 454 change_keepalive(so, tp); 455 } else 456 error = EINVAL; 457 break; 458 459 case TCP_KEEPINTVL: 460 error = sockopt_get(sopt, &ui, sizeof(ui)); 461 if (error) 462 break; 463 if (ui > 0) { 464 tp->t_keepintvl = ui; 465 change_keepalive(so, tp); 466 } else 467 error = EINVAL; 468 break; 469 470 case TCP_KEEPCNT: 471 error = sockopt_get(sopt, &ui, sizeof(ui)); 472 if (error) 473 break; 474 if (ui > 0) { 475 tp->t_keepcnt = ui; 476 change_keepalive(so, tp); 477 } else 478 error = EINVAL; 479 break; 480 481 case TCP_KEEPINIT: 482 error = sockopt_get(sopt, &ui, sizeof(ui)); 483 if (error) 484 break; 485 if (ui > 0) { 486 tp->t_keepinit = ui; 487 change_keepalive(so, tp); 488 } else 489 error = EINVAL; 490 break; 491 492 default: 493 error = ENOPROTOOPT; 494 break; 495 } 496 break; 497 498 case PRCO_GETOPT: 499 switch (optname) { 500 #ifdef TCP_SIGNATURE 501 case TCP_MD5SIG: 502 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 503 error = sockopt_set(sopt, &optval, sizeof(optval)); 504 break; 505 #endif 506 case TCP_NODELAY: 507 optval = tp->t_flags & TF_NODELAY; 508 error = sockopt_set(sopt, &optval, sizeof(optval)); 509 break; 510 case TCP_MAXSEG: 511 optval = tp->t_peermss; 512 error = sockopt_set(sopt, &optval, sizeof(optval)); 513 break; 514 case TCP_INFO: 515 tcp_fill_info(tp, &ti); 516 error = sockopt_set(sopt, &ti, sizeof ti); 517 break; 518 #ifdef notyet 519 case TCP_CONGCTL: 520 break; 521 #endif 522 default: 523 error = ENOPROTOOPT; 524 break; 525 } 526 break; 527 } 528 splx(s); 529 return (error); 530 } 531 532 #ifndef TCP_SENDSPACE 533 #define TCP_SENDSPACE 1024*32 534 #endif 535 int tcp_sendspace = TCP_SENDSPACE; 536 #ifndef TCP_RECVSPACE 537 #define TCP_RECVSPACE 1024*32 538 #endif 539 int tcp_recvspace = TCP_RECVSPACE; 540 541 /* 542 * tcp_attach: attach TCP protocol to socket, allocating internet protocol 543 * control block, TCP control block, buffer space and entering LISTEN state 544 * if to accept connections. 545 */ 546 static int 547 tcp_attach(struct socket *so, int proto) 548 { 549 struct tcpcb *tp; 550 struct inpcb *inp; 551 #ifdef INET6 552 struct in6pcb *in6p; 553 #endif 554 int s, error, family; 555 556 /* Assign the lock (must happen even if we will error out). */ 557 s = splsoftnet(); 558 sosetlock(so); 559 KASSERT(solocked(so)); 560 561 family = so->so_proto->pr_domain->dom_family; 562 switch (family) { 563 #ifdef INET 564 case PF_INET: 565 inp = sotoinpcb(so); 566 #ifdef INET6 567 in6p = NULL; 568 #endif 569 break; 570 #endif 571 #ifdef INET6 572 case PF_INET6: 573 inp = NULL; 574 in6p = sotoin6pcb(so); 575 break; 576 #endif 577 default: 578 error = EAFNOSUPPORT; 579 goto out; 580 } 581 582 KASSERT(inp == NULL); 583 #ifdef INET6 584 KASSERT(in6p == NULL); 585 #endif 586 587 #ifdef MBUFTRACE 588 so->so_mowner = &tcp_sock_mowner; 589 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 590 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 591 #endif 592 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 593 error = soreserve(so, tcp_sendspace, tcp_recvspace); 594 if (error) 595 goto out; 596 } 597 598 so->so_rcv.sb_flags |= SB_AUTOSIZE; 599 so->so_snd.sb_flags |= SB_AUTOSIZE; 600 601 switch (family) { 602 #ifdef INET 603 case PF_INET: 604 error = in_pcballoc(so, &tcbtable); 605 if (error) 606 goto out; 607 inp = sotoinpcb(so); 608 #ifdef INET6 609 in6p = NULL; 610 #endif 611 break; 612 #endif 613 #ifdef INET6 614 case PF_INET6: 615 error = in6_pcballoc(so, &tcbtable); 616 if (error) 617 goto out; 618 inp = NULL; 619 in6p = sotoin6pcb(so); 620 break; 621 #endif 622 default: 623 error = EAFNOSUPPORT; 624 goto out; 625 } 626 if (inp) 627 tp = tcp_newtcpcb(family, (void *)inp); 628 #ifdef INET6 629 else if (in6p) 630 tp = tcp_newtcpcb(family, (void *)in6p); 631 #endif 632 else 633 tp = NULL; 634 635 if (tp == NULL) { 636 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 637 638 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 639 #ifdef INET 640 if (inp) 641 in_pcbdetach(inp); 642 #endif 643 #ifdef INET6 644 if (in6p) 645 in6_pcbdetach(in6p); 646 #endif 647 so->so_state |= nofd; 648 error = ENOBUFS; 649 goto out; 650 } 651 tp->t_state = TCPS_CLOSED; 652 if ((so->so_options & SO_LINGER) && so->so_linger == 0) { 653 so->so_linger = TCP_LINGERTIME; 654 } 655 out: 656 KASSERT(solocked(so)); 657 splx(s); 658 return error; 659 } 660 661 static void 662 tcp_detach(struct socket *so) 663 { 664 struct inpcb *inp = NULL; 665 struct in6pcb *in6p = NULL; 666 struct tcpcb *tp = NULL; 667 int s; 668 669 if (tcp_getpcb(so, &inp, &in6p, &tp) != 0) 670 return; 671 672 s = splsoftnet(); 673 (void)tcp_disconnect1(tp); 674 splx(s); 675 } 676 677 static int 678 tcp_accept(struct socket *so, struct mbuf *nam) 679 { 680 struct inpcb *inp = NULL; 681 struct in6pcb *in6p = NULL; 682 struct tcpcb *tp = NULL; 683 int ostate = 0; 684 int error = 0; 685 int s; 686 687 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 688 return error; 689 690 ostate = tcp_debug_capture(tp, PRU_ACCEPT); 691 692 /* 693 * Accept a connection. Essentially all the work is 694 * done at higher levels; just return the address 695 * of the peer, storing through addr. 696 */ 697 s = splsoftnet(); 698 #ifdef INET 699 if (inp) { 700 in_setpeeraddr(inp, nam); 701 } 702 #endif 703 #ifdef INET6 704 if (in6p) { 705 in6_setpeeraddr(in6p, nam); 706 } 707 #endif 708 tcp_debug_trace(so, tp, ostate, PRU_ACCEPT); 709 splx(s); 710 711 return 0; 712 } 713 714 static int 715 tcp_bind(struct socket *so, struct mbuf *nam, struct lwp *l) 716 { 717 struct inpcb *inp = NULL; 718 struct in6pcb *in6p = NULL; 719 struct tcpcb *tp = NULL; 720 int s; 721 int error = 0; 722 int ostate = 0; 723 724 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 725 return error; 726 727 ostate = tcp_debug_capture(tp, PRU_BIND); 728 729 /* 730 * Give the socket an address. 731 */ 732 s = splsoftnet(); 733 switch (so->so_proto->pr_domain->dom_family) { 734 #ifdef INET 735 case PF_INET: 736 error = in_pcbbind(inp, nam, l); 737 break; 738 #endif 739 #ifdef INET6 740 case PF_INET6: 741 error = in6_pcbbind(in6p, nam, l); 742 if (!error) { 743 /* mapped addr case */ 744 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 745 tp->t_family = AF_INET; 746 else 747 tp->t_family = AF_INET6; 748 } 749 break; 750 #endif 751 } 752 tcp_debug_trace(so, tp, ostate, PRU_BIND); 753 splx(s); 754 755 return error; 756 } 757 758 static int 759 tcp_listen(struct socket *so, struct lwp *l) 760 { 761 struct inpcb *inp = NULL; 762 struct in6pcb *in6p = NULL; 763 struct tcpcb *tp = NULL; 764 int error = 0; 765 int ostate = 0; 766 int s; 767 768 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 769 return error; 770 771 ostate = tcp_debug_capture(tp, PRU_LISTEN); 772 773 /* 774 * Prepare to accept connections. 775 */ 776 s = splsoftnet(); 777 #ifdef INET 778 if (inp && inp->inp_lport == 0) { 779 error = in_pcbbind(inp, NULL, l); 780 if (error) 781 goto release; 782 } 783 #endif 784 #ifdef INET6 785 if (in6p && in6p->in6p_lport == 0) { 786 error = in6_pcbbind(in6p, NULL, l); 787 if (error) 788 goto release; 789 } 790 #endif 791 tp->t_state = TCPS_LISTEN; 792 793 release: 794 tcp_debug_trace(so, tp, ostate, PRU_LISTEN); 795 splx(s); 796 797 return error; 798 } 799 800 static int 801 tcp_connect(struct socket *so, struct mbuf *nam, struct lwp *l) 802 { 803 struct inpcb *inp = NULL; 804 struct in6pcb *in6p = NULL; 805 struct tcpcb *tp = NULL; 806 int s; 807 int error = 0; 808 int ostate = 0; 809 810 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 811 return error; 812 813 ostate = tcp_debug_capture(tp, PRU_CONNECT); 814 815 /* 816 * Initiate connection to peer. 817 * Create a template for use in transmissions on this connection. 818 * Enter SYN_SENT state, and mark socket as connecting. 819 * Start keep-alive timer, and seed output sequence space. 820 * Send initial segment on connection. 821 */ 822 s = splsoftnet(); 823 #ifdef INET 824 if (inp) { 825 if (inp->inp_lport == 0) { 826 error = in_pcbbind(inp, NULL, l); 827 if (error) 828 goto release; 829 } 830 error = in_pcbconnect(inp, nam, l); 831 } 832 #endif 833 #ifdef INET6 834 if (in6p) { 835 if (in6p->in6p_lport == 0) { 836 error = in6_pcbbind(in6p, NULL, l); 837 if (error) 838 goto release; 839 } 840 error = in6_pcbconnect(in6p, nam, l); 841 if (!error) { 842 /* mapped addr case */ 843 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 844 tp->t_family = AF_INET; 845 else 846 tp->t_family = AF_INET6; 847 } 848 } 849 #endif 850 if (error) 851 goto release; 852 tp->t_template = tcp_template(tp); 853 if (tp->t_template == 0) { 854 #ifdef INET 855 if (inp) 856 in_pcbdisconnect(inp); 857 #endif 858 #ifdef INET6 859 if (in6p) 860 in6_pcbdisconnect(in6p); 861 #endif 862 error = ENOBUFS; 863 goto release; 864 } 865 /* 866 * Compute window scaling to request. 867 * XXX: This should be moved to tcp_output(). 868 */ 869 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 870 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 871 tp->request_r_scale++; 872 soisconnecting(so); 873 TCP_STATINC(TCP_STAT_CONNATTEMPT); 874 tp->t_state = TCPS_SYN_SENT; 875 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 876 tp->iss = tcp_new_iss(tp, 0); 877 tcp_sendseqinit(tp); 878 error = tcp_output(tp); 879 880 release: 881 tcp_debug_trace(so, tp, ostate, PRU_CONNECT); 882 splx(s); 883 884 return error; 885 } 886 887 static int 888 tcp_connect2(struct socket *so, struct socket *so2) 889 { 890 struct inpcb *inp = NULL; 891 struct in6pcb *in6p = NULL; 892 struct tcpcb *tp = NULL; 893 int error = 0; 894 int ostate = 0; 895 896 KASSERT(solocked(so)); 897 898 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 899 return error; 900 901 ostate = tcp_debug_capture(tp, PRU_CONNECT2); 902 903 tcp_debug_trace(so, tp, ostate, PRU_CONNECT2); 904 905 return EOPNOTSUPP; 906 } 907 908 static int 909 tcp_disconnect(struct socket *so) 910 { 911 struct inpcb *inp = NULL; 912 struct in6pcb *in6p = NULL; 913 struct tcpcb *tp = NULL; 914 int error = 0; 915 int ostate = 0; 916 int s; 917 918 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 919 return error; 920 921 ostate = tcp_debug_capture(tp, PRU_DISCONNECT); 922 923 /* 924 * Initiate disconnect from peer. 925 * If connection never passed embryonic stage, just drop; 926 * else if don't need to let data drain, then can just drop anyways, 927 * else have to begin TCP shutdown process: mark socket disconnecting, 928 * drain unread data, state switch to reflect user close, and 929 * send segment (e.g. FIN) to peer. Socket will be really disconnected 930 * when peer sends FIN and acks ours. 931 * 932 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 933 */ 934 s = splsoftnet(); 935 tp = tcp_disconnect1(tp); 936 tcp_debug_trace(so, tp, ostate, PRU_DISCONNECT); 937 splx(s); 938 939 return error; 940 } 941 942 static int 943 tcp_shutdown(struct socket *so) 944 { 945 struct inpcb *inp = NULL; 946 struct in6pcb *in6p = NULL; 947 struct tcpcb *tp = NULL; 948 int error = 0; 949 int ostate = 0; 950 int s; 951 952 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 953 return error; 954 955 ostate = tcp_debug_capture(tp, PRU_SHUTDOWN); 956 /* 957 * Mark the connection as being incapable of further output. 958 */ 959 s = splsoftnet(); 960 socantsendmore(so); 961 tp = tcp_usrclosed(tp); 962 if (tp) 963 error = tcp_output(tp); 964 tcp_debug_trace(so, tp, ostate, PRU_SHUTDOWN); 965 splx(s); 966 967 return error; 968 } 969 970 static int 971 tcp_abort(struct socket *so) 972 { 973 struct inpcb *inp = NULL; 974 struct in6pcb *in6p = NULL; 975 struct tcpcb *tp = NULL; 976 int error = 0; 977 int ostate = 0; 978 int s; 979 980 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 981 return error; 982 983 ostate = tcp_debug_capture(tp, PRU_ABORT); 984 985 /* 986 * Abort the TCP. 987 */ 988 s = splsoftnet(); 989 tp = tcp_drop(tp, ECONNABORTED); 990 tcp_debug_trace(so, tp, ostate, PRU_ABORT); 991 splx(s); 992 993 return error; 994 } 995 996 static int 997 tcp_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp) 998 { 999 switch (so->so_proto->pr_domain->dom_family) { 1000 #ifdef INET 1001 case PF_INET: 1002 return in_control(so, cmd, nam, ifp); 1003 #endif 1004 #ifdef INET6 1005 case PF_INET6: 1006 return in6_control(so, cmd, nam, ifp); 1007 #endif 1008 default: 1009 return EAFNOSUPPORT; 1010 } 1011 } 1012 1013 static int 1014 tcp_stat(struct socket *so, struct stat *ub) 1015 { 1016 KASSERT(solocked(so)); 1017 1018 /* stat: don't bother with a blocksize. */ 1019 return 0; 1020 } 1021 1022 static int 1023 tcp_peeraddr(struct socket *so, struct mbuf *nam) 1024 { 1025 struct inpcb *inp = NULL; 1026 struct in6pcb *in6p = NULL; 1027 struct tcpcb *tp = NULL; 1028 int ostate = 0; 1029 int error = 0; 1030 int s; 1031 1032 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1033 return error; 1034 1035 ostate = tcp_debug_capture(tp, PRU_PEERADDR); 1036 1037 s = splsoftnet(); 1038 #ifdef INET 1039 if (inp) 1040 in_setpeeraddr(inp, nam); 1041 #endif 1042 #ifdef INET6 1043 if (in6p) 1044 in6_setpeeraddr(in6p, nam); 1045 #endif 1046 tcp_debug_trace(so, tp, ostate, PRU_PEERADDR); 1047 splx(s); 1048 1049 return 0; 1050 } 1051 1052 static int 1053 tcp_sockaddr(struct socket *so, struct mbuf *nam) 1054 { 1055 struct inpcb *inp = NULL; 1056 struct in6pcb *in6p = NULL; 1057 struct tcpcb *tp = NULL; 1058 int ostate = 0; 1059 int error = 0; 1060 int s; 1061 1062 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1063 return error; 1064 1065 ostate = tcp_debug_capture(tp, PRU_SOCKADDR); 1066 1067 s = splsoftnet(); 1068 #ifdef INET 1069 if (inp) 1070 in_setsockaddr(inp, nam); 1071 #endif 1072 #ifdef INET6 1073 if (in6p) 1074 in6_setsockaddr(in6p, nam); 1075 #endif 1076 tcp_debug_trace(so, tp, ostate, PRU_SOCKADDR); 1077 splx(s); 1078 1079 return 0; 1080 } 1081 1082 static int 1083 tcp_rcvd(struct socket *so, int flags, struct lwp *l) 1084 { 1085 struct inpcb *inp = NULL; 1086 struct in6pcb *in6p = NULL; 1087 struct tcpcb *tp = NULL; 1088 int ostate = 0; 1089 int error = 0; 1090 int s; 1091 1092 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1093 return error; 1094 1095 ostate = tcp_debug_capture(tp, PRU_RCVD); 1096 1097 /* 1098 * After a receive, possibly send window update to peer. 1099 * 1100 * soreceive() calls this function when a user receives 1101 * ancillary data on a listening socket. We don't call 1102 * tcp_output in such a case, since there is no header 1103 * template for a listening socket and hence the kernel 1104 * will panic. 1105 */ 1106 s = splsoftnet(); 1107 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 1108 (void) tcp_output(tp); 1109 splx(s); 1110 1111 tcp_debug_trace(so, tp, ostate, PRU_RCVD); 1112 1113 return 0; 1114 } 1115 1116 static int 1117 tcp_recvoob(struct socket *so, struct mbuf *m, int flags) 1118 { 1119 struct inpcb *inp = NULL; 1120 struct in6pcb *in6p = NULL; 1121 struct tcpcb *tp = NULL; 1122 int ostate = 0; 1123 int error = 0; 1124 int s; 1125 1126 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1127 return error; 1128 1129 ostate = tcp_debug_capture(tp, PRU_RCVOOB); 1130 1131 s = splsoftnet(); 1132 if ((so->so_oobmark == 0 && 1133 (so->so_state & SS_RCVATMARK) == 0) || 1134 so->so_options & SO_OOBINLINE || 1135 tp->t_oobflags & TCPOOB_HADDATA) { 1136 splx(s); 1137 return EINVAL; 1138 } 1139 1140 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 1141 splx(s); 1142 return EWOULDBLOCK; 1143 } 1144 1145 m->m_len = 1; 1146 *mtod(m, char *) = tp->t_iobc; 1147 if ((flags & MSG_PEEK) == 0) 1148 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 1149 1150 tcp_debug_trace(so, tp, ostate, PRU_RCVOOB); 1151 splx(s); 1152 1153 return 0; 1154 } 1155 1156 static int 1157 tcp_send(struct socket *so, struct mbuf *m, struct mbuf *nam, 1158 struct mbuf *control, struct lwp *l) 1159 { 1160 struct inpcb *inp = NULL; 1161 struct in6pcb *in6p = NULL; 1162 struct tcpcb *tp = NULL; 1163 int ostate = 0; 1164 int error = 0; 1165 int s; 1166 1167 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1168 return error; 1169 1170 ostate = tcp_debug_capture(tp, PRU_SEND); 1171 1172 /* 1173 * Do a send by putting data in output queue and updating urgent 1174 * marker if URG set. Possibly send more data. 1175 */ 1176 s = splsoftnet(); 1177 if (control && control->m_len) { 1178 m_freem(control); 1179 m_freem(m); 1180 tcp_debug_trace(so, tp, ostate, PRU_SEND); 1181 splx(s); 1182 return EINVAL; 1183 } 1184 1185 sbappendstream(&so->so_snd, m); 1186 error = tcp_output(tp); 1187 tcp_debug_trace(so, tp, ostate, PRU_SEND); 1188 splx(s); 1189 1190 return error; 1191 } 1192 1193 static int 1194 tcp_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control) 1195 { 1196 struct inpcb *inp = NULL; 1197 struct in6pcb *in6p = NULL; 1198 struct tcpcb *tp = NULL; 1199 int ostate = 0; 1200 int error = 0; 1201 int s; 1202 1203 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) { 1204 m_freem(m); 1205 m_freem(control); 1206 return error; 1207 } 1208 1209 ostate = tcp_debug_capture(tp, PRU_SENDOOB); 1210 1211 s = splsoftnet(); 1212 if (sbspace(&so->so_snd) < -512) { 1213 m_freem(m); 1214 m_freem(control); 1215 splx(s); 1216 return ENOBUFS; 1217 } 1218 /* 1219 * According to RFC961 (Assigned Protocols), 1220 * the urgent pointer points to the last octet 1221 * of urgent data. We continue, however, 1222 * to consider it to indicate the first octet 1223 * of data past the urgent section. 1224 * Otherwise, snd_up should be one lower. 1225 */ 1226 sbappendstream(&so->so_snd, m); 1227 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 1228 tp->t_force = 1; 1229 error = tcp_output(tp); 1230 tp->t_force = 0; 1231 tcp_debug_trace(so, tp, ostate, PRU_SENDOOB); 1232 splx(s); 1233 m_freem(control); 1234 1235 return error; 1236 } 1237 1238 static int 1239 tcp_purgeif(struct socket *so, struct ifnet *ifp) 1240 { 1241 int s; 1242 1243 s = splsoftnet(); 1244 mutex_enter(softnet_lock); 1245 switch (so->so_proto->pr_domain->dom_family) { 1246 #ifdef INET 1247 case PF_INET: 1248 in_pcbpurgeif0(&tcbtable, ifp); 1249 in_purgeif(ifp); 1250 in_pcbpurgeif(&tcbtable, ifp); 1251 break; 1252 #endif 1253 #ifdef INET6 1254 case PF_INET6: 1255 in6_pcbpurgeif0(&tcbtable, ifp); 1256 in6_purgeif(ifp); 1257 in6_pcbpurgeif(&tcbtable, ifp); 1258 break; 1259 #endif 1260 default: 1261 mutex_exit(softnet_lock); 1262 splx(s); 1263 return EAFNOSUPPORT; 1264 } 1265 mutex_exit(softnet_lock); 1266 splx(s); 1267 1268 return 0; 1269 } 1270 1271 /* 1272 * Initiate (or continue) disconnect. 1273 * If embryonic state, just send reset (once). 1274 * If in ``let data drain'' option and linger null, just drop. 1275 * Otherwise (hard), mark socket disconnecting and drop 1276 * current input data; switch states based on user close, and 1277 * send segment to peer (with FIN). 1278 */ 1279 struct tcpcb * 1280 tcp_disconnect1(struct tcpcb *tp) 1281 { 1282 struct socket *so; 1283 1284 if (tp->t_inpcb) 1285 so = tp->t_inpcb->inp_socket; 1286 #ifdef INET6 1287 else if (tp->t_in6pcb) 1288 so = tp->t_in6pcb->in6p_socket; 1289 #endif 1290 else 1291 so = NULL; 1292 1293 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 1294 tp = tcp_close(tp); 1295 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 1296 tp = tcp_drop(tp, 0); 1297 else { 1298 soisdisconnecting(so); 1299 sbflush(&so->so_rcv); 1300 tp = tcp_usrclosed(tp); 1301 if (tp) 1302 (void) tcp_output(tp); 1303 } 1304 return (tp); 1305 } 1306 1307 /* 1308 * User issued close, and wish to trail through shutdown states: 1309 * if never received SYN, just forget it. If got a SYN from peer, 1310 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 1311 * If already got a FIN from peer, then almost done; go to LAST_ACK 1312 * state. In all other cases, have already sent FIN to peer (e.g. 1313 * after PRU_SHUTDOWN), and just have to play tedious game waiting 1314 * for peer to send FIN or not respond to keep-alives, etc. 1315 * We can let the user exit from the close as soon as the FIN is acked. 1316 */ 1317 struct tcpcb * 1318 tcp_usrclosed(struct tcpcb *tp) 1319 { 1320 1321 switch (tp->t_state) { 1322 1323 case TCPS_CLOSED: 1324 case TCPS_LISTEN: 1325 case TCPS_SYN_SENT: 1326 tp->t_state = TCPS_CLOSED; 1327 tp = tcp_close(tp); 1328 break; 1329 1330 case TCPS_SYN_RECEIVED: 1331 case TCPS_ESTABLISHED: 1332 tp->t_state = TCPS_FIN_WAIT_1; 1333 break; 1334 1335 case TCPS_CLOSE_WAIT: 1336 tp->t_state = TCPS_LAST_ACK; 1337 break; 1338 } 1339 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1340 struct socket *so; 1341 if (tp->t_inpcb) 1342 so = tp->t_inpcb->inp_socket; 1343 #ifdef INET6 1344 else if (tp->t_in6pcb) 1345 so = tp->t_in6pcb->in6p_socket; 1346 #endif 1347 else 1348 so = NULL; 1349 if (so) 1350 soisdisconnected(so); 1351 /* 1352 * If we are in FIN_WAIT_2, we arrived here because the 1353 * application did a shutdown of the send side. Like the 1354 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1355 * a full close, we start a timer to make sure sockets are 1356 * not left in FIN_WAIT_2 forever. 1357 */ 1358 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1359 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1360 else if (tp->t_state == TCPS_TIME_WAIT 1361 && ((tp->t_inpcb 1362 && (tcp4_vtw_enable & 1) 1363 && vtw_add(AF_INET, tp)) 1364 || 1365 (tp->t_in6pcb 1366 && (tcp6_vtw_enable & 1) 1367 && vtw_add(AF_INET6, tp)))) { 1368 tp = 0; 1369 } 1370 } 1371 return (tp); 1372 } 1373 1374 /* 1375 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1376 * than 32. 1377 */ 1378 static int 1379 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1380 { 1381 int error, mssdflt; 1382 struct sysctlnode node; 1383 1384 mssdflt = tcp_mssdflt; 1385 node = *rnode; 1386 node.sysctl_data = &mssdflt; 1387 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1388 if (error || newp == NULL) 1389 return (error); 1390 1391 if (mssdflt < 32) 1392 return (EINVAL); 1393 tcp_mssdflt = mssdflt; 1394 1395 mutex_enter(softnet_lock); 1396 tcp_tcpcb_template(); 1397 mutex_exit(softnet_lock); 1398 1399 return (0); 1400 } 1401 1402 /* 1403 * sysctl helper for TCP CB template update 1404 */ 1405 static int 1406 sysctl_update_tcpcb_template(SYSCTLFN_ARGS) 1407 { 1408 int t, error; 1409 struct sysctlnode node; 1410 1411 /* follow procedures in sysctl(9) manpage */ 1412 t = *(int *)rnode->sysctl_data; 1413 node = *rnode; 1414 node.sysctl_data = &t; 1415 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1416 if (error || newp == NULL) 1417 return error; 1418 1419 if (t < 0) 1420 return EINVAL; 1421 1422 *(int *)rnode->sysctl_data = t; 1423 1424 mutex_enter(softnet_lock); 1425 tcp_tcpcb_template(); 1426 mutex_exit(softnet_lock); 1427 1428 return 0; 1429 } 1430 1431 /* 1432 * sysctl helper routine for setting port related values under 1433 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1434 * additional checks for each type. this code has placed in 1435 * tcp_input.c since INET and INET6 both use the same tcp code. 1436 * 1437 * this helper is not static so that both inet and inet6 can use it. 1438 */ 1439 int 1440 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1441 { 1442 int error, tmp; 1443 int apmin, apmax; 1444 #ifndef IPNOPRIVPORTS 1445 int lpmin, lpmax; 1446 #endif /* IPNOPRIVPORTS */ 1447 struct sysctlnode node; 1448 1449 if (namelen != 0) 1450 return (EINVAL); 1451 1452 switch (name[-3]) { 1453 #ifdef INET 1454 case PF_INET: 1455 apmin = anonportmin; 1456 apmax = anonportmax; 1457 #ifndef IPNOPRIVPORTS 1458 lpmin = lowportmin; 1459 lpmax = lowportmax; 1460 #endif /* IPNOPRIVPORTS */ 1461 break; 1462 #endif /* INET */ 1463 #ifdef INET6 1464 case PF_INET6: 1465 apmin = ip6_anonportmin; 1466 apmax = ip6_anonportmax; 1467 #ifndef IPNOPRIVPORTS 1468 lpmin = ip6_lowportmin; 1469 lpmax = ip6_lowportmax; 1470 #endif /* IPNOPRIVPORTS */ 1471 break; 1472 #endif /* INET6 */ 1473 default: 1474 return (EINVAL); 1475 } 1476 1477 /* 1478 * insert temporary copy into node, perform lookup on 1479 * temporary, then restore pointer 1480 */ 1481 node = *rnode; 1482 tmp = *(int*)rnode->sysctl_data; 1483 node.sysctl_data = &tmp; 1484 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1485 if (error || newp == NULL) 1486 return (error); 1487 1488 /* 1489 * simple port range check 1490 */ 1491 if (tmp < 0 || tmp > 65535) 1492 return (EINVAL); 1493 1494 /* 1495 * per-node range checks 1496 */ 1497 switch (rnode->sysctl_num) { 1498 case IPCTL_ANONPORTMIN: 1499 case IPV6CTL_ANONPORTMIN: 1500 if (tmp >= apmax) 1501 return (EINVAL); 1502 #ifndef IPNOPRIVPORTS 1503 if (tmp < IPPORT_RESERVED) 1504 return (EINVAL); 1505 #endif /* IPNOPRIVPORTS */ 1506 break; 1507 1508 case IPCTL_ANONPORTMAX: 1509 case IPV6CTL_ANONPORTMAX: 1510 if (apmin >= tmp) 1511 return (EINVAL); 1512 #ifndef IPNOPRIVPORTS 1513 if (tmp < IPPORT_RESERVED) 1514 return (EINVAL); 1515 #endif /* IPNOPRIVPORTS */ 1516 break; 1517 1518 #ifndef IPNOPRIVPORTS 1519 case IPCTL_LOWPORTMIN: 1520 case IPV6CTL_LOWPORTMIN: 1521 if (tmp >= lpmax || 1522 tmp > IPPORT_RESERVEDMAX || 1523 tmp < IPPORT_RESERVEDMIN) 1524 return (EINVAL); 1525 break; 1526 1527 case IPCTL_LOWPORTMAX: 1528 case IPV6CTL_LOWPORTMAX: 1529 if (lpmin >= tmp || 1530 tmp > IPPORT_RESERVEDMAX || 1531 tmp < IPPORT_RESERVEDMIN) 1532 return (EINVAL); 1533 break; 1534 #endif /* IPNOPRIVPORTS */ 1535 1536 default: 1537 return (EINVAL); 1538 } 1539 1540 *(int*)rnode->sysctl_data = tmp; 1541 1542 return (0); 1543 } 1544 1545 static inline int 1546 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1547 { 1548 if (oldp) { 1549 size_t sz; 1550 uid_t uid; 1551 int error; 1552 1553 if (sockp->so_cred == NULL) 1554 return EPERM; 1555 1556 uid = kauth_cred_geteuid(sockp->so_cred); 1557 sz = MIN(sizeof(uid), *oldlenp); 1558 if ((error = copyout(&uid, oldp, sz)) != 0) 1559 return error; 1560 } 1561 *oldlenp = sizeof(uid_t); 1562 return 0; 1563 } 1564 1565 static inline int 1566 inet4_ident_core(struct in_addr raddr, u_int rport, 1567 struct in_addr laddr, u_int lport, 1568 void *oldp, size_t *oldlenp, 1569 struct lwp *l, int dodrop) 1570 { 1571 struct inpcb *inp; 1572 struct socket *sockp; 1573 1574 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1575 1576 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1577 return ESRCH; 1578 1579 if (dodrop) { 1580 struct tcpcb *tp; 1581 int error; 1582 1583 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1584 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1585 return ESRCH; 1586 1587 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1588 KAUTH_REQ_NETWORK_SOCKET_DROP, inp->inp_socket, tp, NULL); 1589 if (error) 1590 return (error); 1591 1592 (void)tcp_drop(tp, ECONNABORTED); 1593 return 0; 1594 } 1595 else 1596 return copyout_uid(sockp, oldp, oldlenp); 1597 } 1598 1599 #ifdef INET6 1600 static inline int 1601 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1602 struct in6_addr *laddr, u_int lport, 1603 void *oldp, size_t *oldlenp, 1604 struct lwp *l, int dodrop) 1605 { 1606 struct in6pcb *in6p; 1607 struct socket *sockp; 1608 1609 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0, 0); 1610 1611 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1612 return ESRCH; 1613 1614 if (dodrop) { 1615 struct tcpcb *tp; 1616 int error; 1617 1618 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1619 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1620 return ESRCH; 1621 1622 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1623 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1624 if (error) 1625 return (error); 1626 1627 (void)tcp_drop(tp, ECONNABORTED); 1628 return 0; 1629 } 1630 else 1631 return copyout_uid(sockp, oldp, oldlenp); 1632 } 1633 #endif 1634 1635 /* 1636 * sysctl helper routine for the net.inet.tcp.drop and 1637 * net.inet6.tcp6.drop nodes. 1638 */ 1639 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1640 1641 /* 1642 * sysctl helper routine for the net.inet.tcp.ident and 1643 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1644 * old way of looking up the ident information for ipv4 which involves 1645 * stuffing the port/addr pairs into the mib lookup. 1646 */ 1647 static int 1648 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1649 { 1650 #ifdef INET 1651 struct sockaddr_in *si4[2]; 1652 #endif /* INET */ 1653 #ifdef INET6 1654 struct sockaddr_in6 *si6[2]; 1655 #endif /* INET6 */ 1656 struct sockaddr_storage sa[2]; 1657 int error, pf, dodrop; 1658 1659 dodrop = name[-1] == TCPCTL_DROP; 1660 if (dodrop) { 1661 if (oldp != NULL || *oldlenp != 0) 1662 return EINVAL; 1663 if (newp == NULL) 1664 return EPERM; 1665 if (newlen < sizeof(sa)) 1666 return ENOMEM; 1667 } 1668 if (namelen != 4 && namelen != 0) 1669 return EINVAL; 1670 if (name[-2] != IPPROTO_TCP) 1671 return EINVAL; 1672 pf = name[-3]; 1673 1674 /* old style lookup, ipv4 only */ 1675 if (namelen == 4) { 1676 #ifdef INET 1677 struct in_addr laddr, raddr; 1678 u_int lport, rport; 1679 1680 if (pf != PF_INET) 1681 return EPROTONOSUPPORT; 1682 raddr.s_addr = (uint32_t)name[0]; 1683 rport = (u_int)name[1]; 1684 laddr.s_addr = (uint32_t)name[2]; 1685 lport = (u_int)name[3]; 1686 1687 mutex_enter(softnet_lock); 1688 error = inet4_ident_core(raddr, rport, laddr, lport, 1689 oldp, oldlenp, l, dodrop); 1690 mutex_exit(softnet_lock); 1691 return error; 1692 #else /* INET */ 1693 return EINVAL; 1694 #endif /* INET */ 1695 } 1696 1697 if (newp == NULL || newlen != sizeof(sa)) 1698 return EINVAL; 1699 error = copyin(newp, &sa, newlen); 1700 if (error) 1701 return error; 1702 1703 /* 1704 * requested families must match 1705 */ 1706 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1707 return EINVAL; 1708 1709 switch (pf) { 1710 #ifdef INET6 1711 case PF_INET6: 1712 si6[0] = (struct sockaddr_in6*)&sa[0]; 1713 si6[1] = (struct sockaddr_in6*)&sa[1]; 1714 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1715 si6[1]->sin6_len != sizeof(*si6[1])) 1716 return EINVAL; 1717 1718 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1719 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1720 error = sa6_embedscope(si6[0], ip6_use_defzone); 1721 if (error) 1722 return error; 1723 error = sa6_embedscope(si6[1], ip6_use_defzone); 1724 if (error) 1725 return error; 1726 1727 mutex_enter(softnet_lock); 1728 error = inet6_ident_core(&si6[0]->sin6_addr, 1729 si6[0]->sin6_port, &si6[1]->sin6_addr, 1730 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1731 mutex_exit(softnet_lock); 1732 return error; 1733 } 1734 1735 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1736 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1737 return EINVAL; 1738 1739 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1740 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1741 /*FALLTHROUGH*/ 1742 #endif /* INET6 */ 1743 #ifdef INET 1744 case PF_INET: 1745 si4[0] = (struct sockaddr_in*)&sa[0]; 1746 si4[1] = (struct sockaddr_in*)&sa[1]; 1747 if (si4[0]->sin_len != sizeof(*si4[0]) || 1748 si4[0]->sin_len != sizeof(*si4[1])) 1749 return EINVAL; 1750 1751 mutex_enter(softnet_lock); 1752 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1753 si4[1]->sin_addr, si4[1]->sin_port, 1754 oldp, oldlenp, l, dodrop); 1755 mutex_exit(softnet_lock); 1756 return error; 1757 #endif /* INET */ 1758 default: 1759 return EPROTONOSUPPORT; 1760 } 1761 } 1762 1763 /* 1764 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1765 * inet/inet6, as well as raw pcbs for each. specifically not 1766 * declared static so that raw sockets and udp/udp6 can use it as 1767 * well. 1768 */ 1769 int 1770 sysctl_inpcblist(SYSCTLFN_ARGS) 1771 { 1772 #ifdef INET 1773 struct sockaddr_in *in; 1774 const struct inpcb *inp; 1775 #endif 1776 #ifdef INET6 1777 struct sockaddr_in6 *in6; 1778 const struct in6pcb *in6p; 1779 #endif 1780 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1781 const struct inpcb_hdr *inph; 1782 struct tcpcb *tp; 1783 struct kinfo_pcb pcb; 1784 char *dp; 1785 size_t len, needed, elem_size, out_size; 1786 int error, elem_count, pf, proto, pf2; 1787 1788 if (namelen != 4) 1789 return (EINVAL); 1790 1791 if (oldp != NULL) { 1792 len = *oldlenp; 1793 elem_size = name[2]; 1794 elem_count = name[3]; 1795 if (elem_size != sizeof(pcb)) 1796 return EINVAL; 1797 } else { 1798 len = 0; 1799 elem_count = INT_MAX; 1800 elem_size = sizeof(pcb); 1801 } 1802 error = 0; 1803 dp = oldp; 1804 out_size = elem_size; 1805 needed = 0; 1806 1807 if (namelen == 1 && name[0] == CTL_QUERY) 1808 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1809 1810 if (name - oname != 4) 1811 return (EINVAL); 1812 1813 pf = oname[1]; 1814 proto = oname[2]; 1815 pf2 = (oldp != NULL) ? pf : 0; 1816 1817 mutex_enter(softnet_lock); 1818 1819 TAILQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1820 #ifdef INET 1821 inp = (const struct inpcb *)inph; 1822 #endif 1823 #ifdef INET6 1824 in6p = (const struct in6pcb *)inph; 1825 #endif 1826 1827 if (inph->inph_af != pf) 1828 continue; 1829 1830 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1831 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1832 NULL) != 0) 1833 continue; 1834 1835 memset(&pcb, 0, sizeof(pcb)); 1836 1837 pcb.ki_family = pf; 1838 pcb.ki_type = proto; 1839 1840 switch (pf2) { 1841 case 0: 1842 /* just probing for size */ 1843 break; 1844 #ifdef INET 1845 case PF_INET: 1846 pcb.ki_family = inp->inp_socket->so_proto-> 1847 pr_domain->dom_family; 1848 pcb.ki_type = inp->inp_socket->so_proto-> 1849 pr_type; 1850 pcb.ki_protocol = inp->inp_socket->so_proto-> 1851 pr_protocol; 1852 pcb.ki_pflags = inp->inp_flags; 1853 1854 pcb.ki_sostate = inp->inp_socket->so_state; 1855 pcb.ki_prstate = inp->inp_state; 1856 if (proto == IPPROTO_TCP) { 1857 tp = intotcpcb(inp); 1858 pcb.ki_tstate = tp->t_state; 1859 pcb.ki_tflags = tp->t_flags; 1860 } 1861 1862 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1863 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1864 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1865 1866 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1867 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1868 1869 in = satosin(&pcb.ki_src); 1870 in->sin_len = sizeof(*in); 1871 in->sin_family = pf; 1872 in->sin_port = inp->inp_lport; 1873 in->sin_addr = inp->inp_laddr; 1874 if (pcb.ki_prstate >= INP_CONNECTED) { 1875 in = satosin(&pcb.ki_dst); 1876 in->sin_len = sizeof(*in); 1877 in->sin_family = pf; 1878 in->sin_port = inp->inp_fport; 1879 in->sin_addr = inp->inp_faddr; 1880 } 1881 break; 1882 #endif 1883 #ifdef INET6 1884 case PF_INET6: 1885 pcb.ki_family = in6p->in6p_socket->so_proto-> 1886 pr_domain->dom_family; 1887 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1888 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1889 pr_protocol; 1890 pcb.ki_pflags = in6p->in6p_flags; 1891 1892 pcb.ki_sostate = in6p->in6p_socket->so_state; 1893 pcb.ki_prstate = in6p->in6p_state; 1894 if (proto == IPPROTO_TCP) { 1895 tp = in6totcpcb(in6p); 1896 pcb.ki_tstate = tp->t_state; 1897 pcb.ki_tflags = tp->t_flags; 1898 } 1899 1900 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1901 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1902 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1903 1904 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1905 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1906 1907 in6 = satosin6(&pcb.ki_src); 1908 in6->sin6_len = sizeof(*in6); 1909 in6->sin6_family = pf; 1910 in6->sin6_port = in6p->in6p_lport; 1911 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1912 in6->sin6_addr = in6p->in6p_laddr; 1913 in6->sin6_scope_id = 0; /* XXX? */ 1914 1915 if (pcb.ki_prstate >= IN6P_CONNECTED) { 1916 in6 = satosin6(&pcb.ki_dst); 1917 in6->sin6_len = sizeof(*in6); 1918 in6->sin6_family = pf; 1919 in6->sin6_port = in6p->in6p_fport; 1920 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1921 in6->sin6_addr = in6p->in6p_faddr; 1922 in6->sin6_scope_id = 0; /* XXX? */ 1923 } 1924 break; 1925 #endif 1926 } 1927 1928 if (len >= elem_size && elem_count > 0) { 1929 error = copyout(&pcb, dp, out_size); 1930 if (error) { 1931 mutex_exit(softnet_lock); 1932 return (error); 1933 } 1934 dp += elem_size; 1935 len -= elem_size; 1936 } 1937 needed += elem_size; 1938 if (elem_count > 0 && elem_count != INT_MAX) 1939 elem_count--; 1940 } 1941 1942 *oldlenp = needed; 1943 if (oldp == NULL) 1944 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 1945 1946 mutex_exit(softnet_lock); 1947 1948 return (error); 1949 } 1950 1951 static int 1952 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1953 { 1954 struct sysctlnode node; 1955 int error; 1956 char newname[TCPCC_MAXLEN]; 1957 1958 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1959 1960 node = *rnode; 1961 node.sysctl_data = newname; 1962 node.sysctl_size = sizeof(newname); 1963 1964 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1965 1966 if (error || 1967 newp == NULL || 1968 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1969 return error; 1970 1971 mutex_enter(softnet_lock); 1972 error = tcp_congctl_select(NULL, newname); 1973 mutex_exit(softnet_lock); 1974 1975 return error; 1976 } 1977 1978 static int 1979 sysctl_tcp_init_win(SYSCTLFN_ARGS) 1980 { 1981 int error; 1982 u_int iw; 1983 struct sysctlnode node; 1984 1985 iw = *(u_int *)rnode->sysctl_data; 1986 node = *rnode; 1987 node.sysctl_data = &iw; 1988 node.sysctl_size = sizeof(iw); 1989 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1990 if (error || newp == NULL) 1991 return error; 1992 1993 if (iw >= __arraycount(tcp_init_win_max)) 1994 return EINVAL; 1995 *(u_int *)rnode->sysctl_data = iw; 1996 return 0; 1997 } 1998 1999 static int 2000 sysctl_tcp_keep(SYSCTLFN_ARGS) 2001 { 2002 int error; 2003 u_int tmp; 2004 struct sysctlnode node; 2005 2006 node = *rnode; 2007 tmp = *(u_int *)rnode->sysctl_data; 2008 node.sysctl_data = &tmp; 2009 2010 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2011 if (error || newp == NULL) 2012 return error; 2013 2014 mutex_enter(softnet_lock); 2015 2016 *(u_int *)rnode->sysctl_data = tmp; 2017 tcp_tcpcb_template(); /* update the template */ 2018 2019 mutex_exit(softnet_lock); 2020 return 0; 2021 } 2022 2023 static int 2024 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 2025 { 2026 2027 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 2028 } 2029 2030 /* 2031 * this (second stage) setup routine is a replacement for tcp_sysctl() 2032 * (which is currently used for ipv4 and ipv6) 2033 */ 2034 static void 2035 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 2036 const char *tcpname) 2037 { 2038 const struct sysctlnode *sack_node; 2039 const struct sysctlnode *abc_node; 2040 const struct sysctlnode *ecn_node; 2041 const struct sysctlnode *congctl_node; 2042 const struct sysctlnode *mslt_node; 2043 const struct sysctlnode *vtw_node; 2044 #ifdef TCP_DEBUG 2045 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 2046 extern int tcp_debx; 2047 #endif 2048 2049 sysctl_createv(clog, 0, NULL, NULL, 2050 CTLFLAG_PERMANENT, 2051 CTLTYPE_NODE, pfname, NULL, 2052 NULL, 0, NULL, 0, 2053 CTL_NET, pf, CTL_EOL); 2054 sysctl_createv(clog, 0, NULL, NULL, 2055 CTLFLAG_PERMANENT, 2056 CTLTYPE_NODE, tcpname, 2057 SYSCTL_DESCR("TCP related settings"), 2058 NULL, 0, NULL, 0, 2059 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 2060 2061 sysctl_createv(clog, 0, NULL, NULL, 2062 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2063 CTLTYPE_INT, "rfc1323", 2064 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 2065 sysctl_update_tcpcb_template, 0, &tcp_do_rfc1323, 0, 2066 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 2067 sysctl_createv(clog, 0, NULL, NULL, 2068 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2069 CTLTYPE_INT, "sendspace", 2070 SYSCTL_DESCR("Default TCP send buffer size"), 2071 NULL, 0, &tcp_sendspace, 0, 2072 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 2073 sysctl_createv(clog, 0, NULL, NULL, 2074 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2075 CTLTYPE_INT, "recvspace", 2076 SYSCTL_DESCR("Default TCP receive buffer size"), 2077 NULL, 0, &tcp_recvspace, 0, 2078 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 2079 sysctl_createv(clog, 0, NULL, NULL, 2080 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2081 CTLTYPE_INT, "mssdflt", 2082 SYSCTL_DESCR("Default maximum segment size"), 2083 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 2084 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 2085 sysctl_createv(clog, 0, NULL, NULL, 2086 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2087 CTLTYPE_INT, "minmss", 2088 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 2089 NULL, 0, &tcp_minmss, 0, 2090 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2091 sysctl_createv(clog, 0, NULL, NULL, 2092 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2093 CTLTYPE_INT, "msl", 2094 SYSCTL_DESCR("Maximum Segment Life"), 2095 NULL, 0, &tcp_msl, 0, 2096 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 2097 sysctl_createv(clog, 0, NULL, NULL, 2098 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2099 CTLTYPE_INT, "syn_cache_limit", 2100 SYSCTL_DESCR("Maximum number of entries in the TCP " 2101 "compressed state engine"), 2102 NULL, 0, &tcp_syn_cache_limit, 0, 2103 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 2104 CTL_EOL); 2105 sysctl_createv(clog, 0, NULL, NULL, 2106 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2107 CTLTYPE_INT, "syn_bucket_limit", 2108 SYSCTL_DESCR("Maximum number of entries per hash " 2109 "bucket in the TCP compressed state " 2110 "engine"), 2111 NULL, 0, &tcp_syn_bucket_limit, 0, 2112 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 2113 CTL_EOL); 2114 #if 0 /* obsoleted */ 2115 sysctl_createv(clog, 0, NULL, NULL, 2116 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2117 CTLTYPE_INT, "syn_cache_interval", 2118 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 2119 NULL, 0, &tcp_syn_cache_interval, 0, 2120 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 2121 CTL_EOL); 2122 #endif 2123 sysctl_createv(clog, 0, NULL, NULL, 2124 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2125 CTLTYPE_INT, "init_win", 2126 SYSCTL_DESCR("Initial TCP congestion window"), 2127 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 2128 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 2129 sysctl_createv(clog, 0, NULL, NULL, 2130 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2131 CTLTYPE_INT, "mss_ifmtu", 2132 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 2133 NULL, 0, &tcp_mss_ifmtu, 0, 2134 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 2135 sysctl_createv(clog, 0, NULL, &sack_node, 2136 CTLFLAG_PERMANENT, 2137 CTLTYPE_NODE, "sack", 2138 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 2139 NULL, 0, NULL, 0, 2140 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 2141 2142 /* Congctl subtree */ 2143 sysctl_createv(clog, 0, NULL, &congctl_node, 2144 CTLFLAG_PERMANENT, 2145 CTLTYPE_NODE, "congctl", 2146 SYSCTL_DESCR("TCP Congestion Control"), 2147 NULL, 0, NULL, 0, 2148 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2149 sysctl_createv(clog, 0, &congctl_node, NULL, 2150 CTLFLAG_PERMANENT, 2151 CTLTYPE_STRING, "available", 2152 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 2153 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 2154 sysctl_createv(clog, 0, &congctl_node, NULL, 2155 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2156 CTLTYPE_STRING, "selected", 2157 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 2158 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 2159 CTL_CREATE, CTL_EOL); 2160 2161 sysctl_createv(clog, 0, NULL, NULL, 2162 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2163 CTLTYPE_INT, "win_scale", 2164 SYSCTL_DESCR("Use RFC1323 window scale options"), 2165 sysctl_update_tcpcb_template, 0, &tcp_do_win_scale, 0, 2166 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 2167 sysctl_createv(clog, 0, NULL, NULL, 2168 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2169 CTLTYPE_INT, "timestamps", 2170 SYSCTL_DESCR("Use RFC1323 time stamp options"), 2171 sysctl_update_tcpcb_template, 0, &tcp_do_timestamps, 0, 2172 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 2173 sysctl_createv(clog, 0, NULL, NULL, 2174 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2175 CTLTYPE_INT, "compat_42", 2176 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 2177 NULL, 0, &tcp_compat_42, 0, 2178 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 2179 sysctl_createv(clog, 0, NULL, NULL, 2180 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2181 CTLTYPE_INT, "cwm", 2182 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 2183 "Monitoring"), 2184 NULL, 0, &tcp_cwm, 0, 2185 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 2186 sysctl_createv(clog, 0, NULL, NULL, 2187 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2188 CTLTYPE_INT, "cwm_burstsize", 2189 SYSCTL_DESCR("Congestion Window Monitoring allowed " 2190 "burst count in packets"), 2191 NULL, 0, &tcp_cwm_burstsize, 0, 2192 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 2193 CTL_EOL); 2194 sysctl_createv(clog, 0, NULL, NULL, 2195 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2196 CTLTYPE_INT, "ack_on_push", 2197 SYSCTL_DESCR("Immediately return ACK when PSH is " 2198 "received"), 2199 NULL, 0, &tcp_ack_on_push, 0, 2200 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 2201 sysctl_createv(clog, 0, NULL, NULL, 2202 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2203 CTLTYPE_INT, "keepidle", 2204 SYSCTL_DESCR("Allowed connection idle ticks before a " 2205 "keepalive probe is sent"), 2206 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 2207 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 2208 sysctl_createv(clog, 0, NULL, NULL, 2209 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2210 CTLTYPE_INT, "keepintvl", 2211 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 2212 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 2213 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 2214 sysctl_createv(clog, 0, NULL, NULL, 2215 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2216 CTLTYPE_INT, "keepcnt", 2217 SYSCTL_DESCR("Number of keepalive probes to send"), 2218 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 2219 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 2220 sysctl_createv(clog, 0, NULL, NULL, 2221 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 2222 CTLTYPE_INT, "slowhz", 2223 SYSCTL_DESCR("Keepalive ticks per second"), 2224 NULL, PR_SLOWHZ, NULL, 0, 2225 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 2226 sysctl_createv(clog, 0, NULL, NULL, 2227 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2228 CTLTYPE_INT, "log_refused", 2229 SYSCTL_DESCR("Log refused TCP connections"), 2230 NULL, 0, &tcp_log_refused, 0, 2231 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 2232 #if 0 /* obsoleted */ 2233 sysctl_createv(clog, 0, NULL, NULL, 2234 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2235 CTLTYPE_INT, "rstratelimit", NULL, 2236 NULL, 0, &tcp_rst_ratelim, 0, 2237 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 2238 #endif 2239 sysctl_createv(clog, 0, NULL, NULL, 2240 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2241 CTLTYPE_INT, "rstppslimit", 2242 SYSCTL_DESCR("Maximum number of RST packets to send " 2243 "per second"), 2244 NULL, 0, &tcp_rst_ppslim, 0, 2245 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 2246 sysctl_createv(clog, 0, NULL, NULL, 2247 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2248 CTLTYPE_INT, "delack_ticks", 2249 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 2250 NULL, 0, &tcp_delack_ticks, 0, 2251 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 2252 sysctl_createv(clog, 0, NULL, NULL, 2253 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2254 CTLTYPE_INT, "init_win_local", 2255 SYSCTL_DESCR("Initial TCP window size (in segments)"), 2256 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 2257 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 2258 CTL_EOL); 2259 sysctl_createv(clog, 0, NULL, NULL, 2260 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2261 CTLTYPE_STRUCT, "ident", 2262 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 2263 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 2264 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 2265 sysctl_createv(clog, 0, NULL, NULL, 2266 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2267 CTLTYPE_INT, "do_loopback_cksum", 2268 SYSCTL_DESCR("Perform TCP checksum on loopback"), 2269 NULL, 0, &tcp_do_loopback_cksum, 0, 2270 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 2271 CTL_EOL); 2272 sysctl_createv(clog, 0, NULL, NULL, 2273 CTLFLAG_PERMANENT, 2274 CTLTYPE_STRUCT, "pcblist", 2275 SYSCTL_DESCR("TCP protocol control block list"), 2276 sysctl_inpcblist, 0, &tcbtable, 0, 2277 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2278 CTL_EOL); 2279 sysctl_createv(clog, 0, NULL, NULL, 2280 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2281 CTLTYPE_INT, "keepinit", 2282 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 2283 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 2284 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2285 2286 /* TCP socket buffers auto-sizing nodes */ 2287 sysctl_createv(clog, 0, NULL, NULL, 2288 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2289 CTLTYPE_INT, "recvbuf_auto", 2290 SYSCTL_DESCR("Enable automatic receive " 2291 "buffer sizing (experimental)"), 2292 NULL, 0, &tcp_do_autorcvbuf, 0, 2293 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2294 sysctl_createv(clog, 0, NULL, NULL, 2295 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2296 CTLTYPE_INT, "recvbuf_inc", 2297 SYSCTL_DESCR("Incrementor step size of " 2298 "automatic receive buffer"), 2299 NULL, 0, &tcp_autorcvbuf_inc, 0, 2300 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2301 sysctl_createv(clog, 0, NULL, NULL, 2302 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2303 CTLTYPE_INT, "recvbuf_max", 2304 SYSCTL_DESCR("Max size of automatic receive buffer"), 2305 NULL, 0, &tcp_autorcvbuf_max, 0, 2306 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2307 2308 sysctl_createv(clog, 0, NULL, NULL, 2309 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2310 CTLTYPE_INT, "sendbuf_auto", 2311 SYSCTL_DESCR("Enable automatic send " 2312 "buffer sizing (experimental)"), 2313 NULL, 0, &tcp_do_autosndbuf, 0, 2314 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2315 sysctl_createv(clog, 0, NULL, NULL, 2316 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2317 CTLTYPE_INT, "sendbuf_inc", 2318 SYSCTL_DESCR("Incrementor step size of " 2319 "automatic send buffer"), 2320 NULL, 0, &tcp_autosndbuf_inc, 0, 2321 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2322 sysctl_createv(clog, 0, NULL, NULL, 2323 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2324 CTLTYPE_INT, "sendbuf_max", 2325 SYSCTL_DESCR("Max size of automatic send buffer"), 2326 NULL, 0, &tcp_autosndbuf_max, 0, 2327 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2328 2329 /* ECN subtree */ 2330 sysctl_createv(clog, 0, NULL, &ecn_node, 2331 CTLFLAG_PERMANENT, 2332 CTLTYPE_NODE, "ecn", 2333 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 2334 NULL, 0, NULL, 0, 2335 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2336 sysctl_createv(clog, 0, &ecn_node, NULL, 2337 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2338 CTLTYPE_INT, "enable", 2339 SYSCTL_DESCR("Enable TCP Explicit Congestion " 2340 "Notification"), 2341 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 2342 sysctl_createv(clog, 0, &ecn_node, NULL, 2343 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2344 CTLTYPE_INT, "maxretries", 2345 SYSCTL_DESCR("Number of times to retry ECN setup " 2346 "before disabling ECN on the connection"), 2347 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 2348 2349 /* SACK gets it's own little subtree. */ 2350 sysctl_createv(clog, 0, NULL, &sack_node, 2351 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2352 CTLTYPE_INT, "enable", 2353 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 2354 NULL, 0, &tcp_do_sack, 0, 2355 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2356 sysctl_createv(clog, 0, NULL, &sack_node, 2357 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2358 CTLTYPE_INT, "maxholes", 2359 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 2360 NULL, 0, &tcp_sack_tp_maxholes, 0, 2361 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2362 sysctl_createv(clog, 0, NULL, &sack_node, 2363 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2364 CTLTYPE_INT, "globalmaxholes", 2365 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 2366 NULL, 0, &tcp_sack_globalmaxholes, 0, 2367 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2368 sysctl_createv(clog, 0, NULL, &sack_node, 2369 CTLFLAG_PERMANENT, 2370 CTLTYPE_INT, "globalholes", 2371 SYSCTL_DESCR("Global number of TCP SACK holes"), 2372 NULL, 0, &tcp_sack_globalholes, 0, 2373 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2374 2375 sysctl_createv(clog, 0, NULL, NULL, 2376 CTLFLAG_PERMANENT, 2377 CTLTYPE_STRUCT, "stats", 2378 SYSCTL_DESCR("TCP statistics"), 2379 sysctl_net_inet_tcp_stats, 0, NULL, 0, 2380 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 2381 CTL_EOL); 2382 sysctl_createv(clog, 0, NULL, NULL, 2383 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2384 CTLTYPE_INT, "local_by_rtt", 2385 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 2386 "are local"), 2387 NULL, 0, &tcp_rttlocal, 0, 2388 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2389 #ifdef TCP_DEBUG 2390 sysctl_createv(clog, 0, NULL, NULL, 2391 CTLFLAG_PERMANENT, 2392 CTLTYPE_STRUCT, "debug", 2393 SYSCTL_DESCR("TCP sockets debug information"), 2394 NULL, 0, &tcp_debug, sizeof(tcp_debug), 2395 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 2396 CTL_EOL); 2397 sysctl_createv(clog, 0, NULL, NULL, 2398 CTLFLAG_PERMANENT, 2399 CTLTYPE_INT, "debx", 2400 SYSCTL_DESCR("Number of TCP debug sockets messages"), 2401 NULL, 0, &tcp_debx, sizeof(tcp_debx), 2402 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 2403 CTL_EOL); 2404 #endif 2405 sysctl_createv(clog, 0, NULL, NULL, 2406 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2407 CTLTYPE_STRUCT, "drop", 2408 SYSCTL_DESCR("TCP drop connection"), 2409 sysctl_net_inet_tcp_drop, 0, NULL, 0, 2410 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 2411 sysctl_createv(clog, 0, NULL, NULL, 2412 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2413 CTLTYPE_INT, "iss_hash", 2414 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 2415 "hash computation"), 2416 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 2417 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2418 CTL_EOL); 2419 2420 /* ABC subtree */ 2421 2422 sysctl_createv(clog, 0, NULL, &abc_node, 2423 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 2424 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 2425 NULL, 0, NULL, 0, 2426 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2427 sysctl_createv(clog, 0, &abc_node, NULL, 2428 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2429 CTLTYPE_INT, "enable", 2430 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 2431 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 2432 sysctl_createv(clog, 0, &abc_node, NULL, 2433 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2434 CTLTYPE_INT, "aggressive", 2435 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 2436 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 2437 2438 /* MSL tuning subtree */ 2439 2440 sysctl_createv(clog, 0, NULL, &mslt_node, 2441 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 2442 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 2443 NULL, 0, NULL, 0, 2444 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2445 sysctl_createv(clog, 0, &mslt_node, NULL, 2446 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2447 CTLTYPE_INT, "enable", 2448 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 2449 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 2450 sysctl_createv(clog, 0, &mslt_node, NULL, 2451 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2452 CTLTYPE_INT, "loopback", 2453 SYSCTL_DESCR("MSL value to use for loopback connections"), 2454 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 2455 sysctl_createv(clog, 0, &mslt_node, NULL, 2456 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2457 CTLTYPE_INT, "local", 2458 SYSCTL_DESCR("MSL value to use for local connections"), 2459 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 2460 sysctl_createv(clog, 0, &mslt_node, NULL, 2461 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2462 CTLTYPE_INT, "remote", 2463 SYSCTL_DESCR("MSL value to use for remote connections"), 2464 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 2465 sysctl_createv(clog, 0, &mslt_node, NULL, 2466 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2467 CTLTYPE_INT, "remote_threshold", 2468 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 2469 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 2470 2471 /* vestigial TIME_WAIT tuning subtree */ 2472 2473 sysctl_createv(clog, 0, NULL, &vtw_node, 2474 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 2475 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 2476 NULL, 0, NULL, 0, 2477 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2478 sysctl_createv(clog, 0, &vtw_node, NULL, 2479 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2480 CTLTYPE_INT, "enable", 2481 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 2482 sysctl_tcp_vtw_enable, 0, 2483 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 2484 0, CTL_CREATE, CTL_EOL); 2485 sysctl_createv(clog, 0, &vtw_node, NULL, 2486 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 2487 CTLTYPE_INT, "entries", 2488 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 2489 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 2490 } 2491 2492 void 2493 tcp_usrreq_init(void) 2494 { 2495 2496 #ifdef INET 2497 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 2498 #endif 2499 #ifdef INET6 2500 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 2501 #endif 2502 } 2503 2504 PR_WRAP_USRREQS(tcp) 2505 #define tcp_attach tcp_attach_wrapper 2506 #define tcp_detach tcp_detach_wrapper 2507 #define tcp_accept tcp_accept_wrapper 2508 #define tcp_bind tcp_bind_wrapper 2509 #define tcp_listen tcp_listen_wrapper 2510 #define tcp_connect tcp_connect_wrapper 2511 #define tcp_connect2 tcp_connect2_wrapper 2512 #define tcp_disconnect tcp_disconnect_wrapper 2513 #define tcp_shutdown tcp_shutdown_wrapper 2514 #define tcp_abort tcp_abort_wrapper 2515 #define tcp_ioctl tcp_ioctl_wrapper 2516 #define tcp_stat tcp_stat_wrapper 2517 #define tcp_peeraddr tcp_peeraddr_wrapper 2518 #define tcp_sockaddr tcp_sockaddr_wrapper 2519 #define tcp_rcvd tcp_rcvd_wrapper 2520 #define tcp_recvoob tcp_recvoob_wrapper 2521 #define tcp_send tcp_send_wrapper 2522 #define tcp_sendoob tcp_sendoob_wrapper 2523 #define tcp_purgeif tcp_purgeif_wrapper 2524 #define tcp_usrreq tcp_usrreq_wrapper 2525 2526 const struct pr_usrreqs tcp_usrreqs = { 2527 .pr_attach = tcp_attach, 2528 .pr_detach = tcp_detach, 2529 .pr_accept = tcp_accept, 2530 .pr_bind = tcp_bind, 2531 .pr_listen = tcp_listen, 2532 .pr_connect = tcp_connect, 2533 .pr_connect2 = tcp_connect2, 2534 .pr_disconnect = tcp_disconnect, 2535 .pr_shutdown = tcp_shutdown, 2536 .pr_abort = tcp_abort, 2537 .pr_ioctl = tcp_ioctl, 2538 .pr_stat = tcp_stat, 2539 .pr_peeraddr = tcp_peeraddr, 2540 .pr_sockaddr = tcp_sockaddr, 2541 .pr_rcvd = tcp_rcvd, 2542 .pr_recvoob = tcp_recvoob, 2543 .pr_send = tcp_send, 2544 .pr_sendoob = tcp_sendoob, 2545 .pr_purgeif = tcp_purgeif, 2546 .pr_generic = tcp_usrreq, 2547 }; 2548
Indexes created Mon Sep 22 13:09:51 GMT 2025