tcp_usrreq.c revision 1.206
1 /* $NetBSD: tcp_usrreq.c,v 1.206 2015/04/24 22:32:37 rtr Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1997, 1998, 2005, 2006 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 38 * Facility, NASA Ames Research Center. 39 * This code is derived from software contributed to The NetBSD Foundation 40 * by Charles M. Hannum. 41 * This code is derived from software contributed to The NetBSD Foundation 42 * by Rui Paulo. 43 * 44 * Redistribution and use in source and binary forms, with or without 45 * modification, are permitted provided that the following conditions 46 * are met: 47 * 1. Redistributions of source code must retain the above copyright 48 * notice, this list of conditions and the following disclaimer. 49 * 2. Redistributions in binary form must reproduce the above copyright 50 * notice, this list of conditions and the following disclaimer in the 51 * documentation and/or other materials provided with the distribution. 52 * 53 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 54 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 55 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 56 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 57 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 58 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 59 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 60 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 61 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 62 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 63 * POSSIBILITY OF SUCH DAMAGE. 64 */ 65 66 /* 67 * Copyright (c) 1982, 1986, 1988, 1993, 1995 68 * The Regents of the University of California. All rights reserved. 69 * 70 * Redistribution and use in source and binary forms, with or without 71 * modification, are permitted provided that the following conditions 72 * are met: 73 * 1. Redistributions of source code must retain the above copyright 74 * notice, this list of conditions and the following disclaimer. 75 * 2. Redistributions in binary form must reproduce the above copyright 76 * notice, this list of conditions and the following disclaimer in the 77 * documentation and/or other materials provided with the distribution. 78 * 3. Neither the name of the University nor the names of its contributors 79 * may be used to endorse or promote products derived from this software 80 * without specific prior written permission. 81 * 82 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 83 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 84 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 85 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 86 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 87 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 88 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 89 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 90 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 91 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 92 * SUCH DAMAGE. 93 * 94 * @(#)tcp_usrreq.c 8.5 (Berkeley) 6/21/95 95 */ 96 97 /* 98 * TCP protocol interface to socket abstraction. 99 */ 100 101 #include <sys/cdefs.h> 102 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.206 2015/04/24 22:32:37 rtr Exp $"); 103 104 #include "opt_inet.h" 105 #include "opt_tcp_debug.h" 106 #include "opt_mbuftrace.h" 107 108 #include <sys/param.h> 109 #include <sys/systm.h> 110 #include <sys/kernel.h> 111 #include <sys/mbuf.h> 112 #include <sys/socket.h> 113 #include <sys/socketvar.h> 114 #include <sys/protosw.h> 115 #include <sys/errno.h> 116 #include <sys/stat.h> 117 #include <sys/proc.h> 118 #include <sys/domain.h> 119 #include <sys/sysctl.h> 120 #include <sys/kauth.h> 121 #include <sys/kernel.h> 122 #include <sys/uidinfo.h> 123 124 #include <net/if.h> 125 #include <net/route.h> 126 127 #include <netinet/in.h> 128 #include <netinet/in_systm.h> 129 #include <netinet/in_var.h> 130 #include <netinet/ip.h> 131 #include <netinet/in_pcb.h> 132 #include <netinet/ip_var.h> 133 #include <netinet/in_offload.h> 134 135 #ifdef INET6 136 #ifndef INET 137 #include <netinet/in.h> 138 #endif 139 #include <netinet/ip6.h> 140 #include <netinet6/in6_pcb.h> 141 #include <netinet6/ip6_var.h> 142 #include <netinet6/scope6_var.h> 143 #endif 144 145 #include <netinet/tcp.h> 146 #include <netinet/tcp_fsm.h> 147 #include <netinet/tcp_seq.h> 148 #include <netinet/tcp_timer.h> 149 #include <netinet/tcp_var.h> 150 #include <netinet/tcp_private.h> 151 #include <netinet/tcp_congctl.h> 152 #include <netinet/tcpip.h> 153 #include <netinet/tcp_debug.h> 154 #include <netinet/tcp_vtw.h> 155 156 #include "opt_tcp_space.h" 157 158 static int 159 tcp_debug_capture(struct tcpcb *tp, int req) 160 { 161 #ifdef KPROF 162 tcp_acounts[tp->t_state][req]++; 163 #endif 164 #ifdef TCP_DEBUG 165 return tp->t_state; 166 #endif 167 return 0; 168 } 169 170 static inline void 171 tcp_debug_trace(struct socket *so, struct tcpcb *tp, int ostate, int req) 172 { 173 #ifdef TCP_DEBUG 174 if (tp && (so->so_options & SO_DEBUG)) 175 tcp_trace(TA_USER, ostate, tp, NULL, req); 176 #endif 177 } 178 179 static int 180 tcp_getpcb(struct socket *so, struct inpcb **inp, 181 struct in6pcb **in6p, struct tcpcb **tp) 182 { 183 184 KASSERT(solocked(so)); 185 186 /* 187 * When a TCP is attached to a socket, then there will be 188 * a (struct inpcb) pointed at by the socket, and this 189 * structure will point at a subsidary (struct tcpcb). 190 */ 191 switch (so->so_proto->pr_domain->dom_family) { 192 #ifdef INET 193 case PF_INET: 194 *inp = sotoinpcb(so); 195 if (*inp == NULL) 196 return EINVAL; 197 *tp = intotcpcb(*inp); 198 break; 199 #endif 200 #ifdef INET6 201 case PF_INET6: 202 *in6p = sotoin6pcb(so); 203 if (*in6p == NULL) 204 return EINVAL; 205 *tp = in6totcpcb(*in6p); 206 break; 207 #endif 208 default: 209 return EAFNOSUPPORT; 210 } 211 212 KASSERT(tp != NULL); 213 214 return 0; 215 } 216 217 /* 218 * Process a TCP user request for TCP tb. If this is a send request 219 * then m is the mbuf chain of send data. If this is a timer expiration 220 * (called from the software clock routine), then timertype tells which timer. 221 */ 222 static int 223 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam, 224 struct mbuf *control, struct lwp *l) 225 { 226 KASSERT(req != PRU_ATTACH); 227 KASSERT(req != PRU_DETACH); 228 KASSERT(req != PRU_ACCEPT); 229 KASSERT(req != PRU_BIND); 230 KASSERT(req != PRU_LISTEN); 231 KASSERT(req != PRU_CONNECT); 232 KASSERT(req != PRU_CONNECT2); 233 KASSERT(req != PRU_DISCONNECT); 234 KASSERT(req != PRU_SHUTDOWN); 235 KASSERT(req != PRU_ABORT); 236 KASSERT(req != PRU_CONTROL); 237 KASSERT(req != PRU_SENSE); 238 KASSERT(req != PRU_PEERADDR); 239 KASSERT(req != PRU_SOCKADDR); 240 KASSERT(req != PRU_RCVD); 241 KASSERT(req != PRU_RCVOOB); 242 KASSERT(req != PRU_SEND); 243 KASSERT(req != PRU_SENDOOB); 244 KASSERT(req != PRU_PURGEIF); 245 246 KASSERT(solocked(so)); 247 248 panic("tcp_usrreq"); 249 250 return 0; 251 } 252 253 static void 254 change_keepalive(struct socket *so, struct tcpcb *tp) 255 { 256 tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl; 257 TCP_TIMER_DISARM(tp, TCPT_KEEP); 258 TCP_TIMER_DISARM(tp, TCPT_2MSL); 259 260 if (tp->t_state == TCPS_SYN_RECEIVED || 261 tp->t_state == TCPS_SYN_SENT) { 262 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 263 } else if (so->so_options & SO_KEEPALIVE && 264 tp->t_state <= TCPS_CLOSE_WAIT) { 265 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl); 266 } else { 267 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 268 } 269 270 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 271 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 272 } 273 274 /* 275 * Export TCP internal state information via a struct tcp_info, based on the 276 * Linux 2.6 API. Not ABI compatible as our constants are mapped differently 277 * (TCP state machine, etc). We export all information using FreeBSD-native 278 * constants -- for example, the numeric values for tcpi_state will differ 279 * from Linux. 280 */ 281 static void 282 tcp_fill_info(struct tcpcb *tp, struct tcp_info *ti) 283 { 284 285 bzero(ti, sizeof(*ti)); 286 287 ti->tcpi_state = tp->t_state; 288 if ((tp->t_flags & TF_REQ_TSTMP) && (tp->t_flags & TF_RCVD_TSTMP)) 289 ti->tcpi_options |= TCPI_OPT_TIMESTAMPS; 290 if (tp->t_flags & TF_SACK_PERMIT) 291 ti->tcpi_options |= TCPI_OPT_SACK; 292 if ((tp->t_flags & TF_REQ_SCALE) && (tp->t_flags & TF_RCVD_SCALE)) { 293 ti->tcpi_options |= TCPI_OPT_WSCALE; 294 ti->tcpi_snd_wscale = tp->snd_scale; 295 ti->tcpi_rcv_wscale = tp->rcv_scale; 296 } 297 if (tp->t_flags & TF_ECN_PERMIT) { 298 ti->tcpi_options |= TCPI_OPT_ECN; 299 } 300 301 ti->tcpi_rto = tp->t_rxtcur * tick; 302 ti->tcpi_last_data_recv = (long)(hardclock_ticks - 303 (int)tp->t_rcvtime) * tick; 304 ti->tcpi_rtt = ((u_int64_t)tp->t_srtt * tick) >> TCP_RTT_SHIFT; 305 ti->tcpi_rttvar = ((u_int64_t)tp->t_rttvar * tick) >> TCP_RTTVAR_SHIFT; 306 307 ti->tcpi_snd_ssthresh = tp->snd_ssthresh; 308 /* Linux API wants these in # of segments, apparently */ 309 ti->tcpi_snd_cwnd = tp->snd_cwnd / tp->t_segsz; 310 ti->tcpi_snd_wnd = tp->snd_wnd / tp->t_segsz; 311 312 /* 313 * FreeBSD-specific extension fields for tcp_info. 314 */ 315 ti->tcpi_rcv_space = tp->rcv_wnd; 316 ti->tcpi_rcv_nxt = tp->rcv_nxt; 317 ti->tcpi_snd_bwnd = 0; /* Unused, kept for compat. */ 318 ti->tcpi_snd_nxt = tp->snd_nxt; 319 ti->tcpi_snd_mss = tp->t_segsz; 320 ti->tcpi_rcv_mss = tp->t_segsz; 321 #ifdef TF_TOE 322 if (tp->t_flags & TF_TOE) 323 ti->tcpi_options |= TCPI_OPT_TOE; 324 #endif 325 /* From the redundant department of redundancies... */ 326 ti->__tcpi_retransmits = ti->__tcpi_retrans = 327 ti->tcpi_snd_rexmitpack = tp->t_sndrexmitpack; 328 329 ti->tcpi_rcv_ooopack = tp->t_rcvoopack; 330 ti->tcpi_snd_zerowin = tp->t_sndzerowin; 331 } 332 333 int 334 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt) 335 { 336 int error = 0, s; 337 struct inpcb *inp; 338 #ifdef INET6 339 struct in6pcb *in6p; 340 #endif 341 struct tcpcb *tp; 342 struct tcp_info ti; 343 u_int ui; 344 int family; /* family of the socket */ 345 int level, optname, optval; 346 347 level = sopt->sopt_level; 348 optname = sopt->sopt_name; 349 350 family = so->so_proto->pr_domain->dom_family; 351 352 s = splsoftnet(); 353 switch (family) { 354 #ifdef INET 355 case PF_INET: 356 inp = sotoinpcb(so); 357 #ifdef INET6 358 in6p = NULL; 359 #endif 360 break; 361 #endif 362 #ifdef INET6 363 case PF_INET6: 364 inp = NULL; 365 in6p = sotoin6pcb(so); 366 break; 367 #endif 368 default: 369 splx(s); 370 panic("%s: af %d", __func__, family); 371 } 372 #ifndef INET6 373 if (inp == NULL) 374 #else 375 if (inp == NULL && in6p == NULL) 376 #endif 377 { 378 splx(s); 379 return (ECONNRESET); 380 } 381 if (level != IPPROTO_TCP) { 382 switch (family) { 383 #ifdef INET 384 case PF_INET: 385 error = ip_ctloutput(op, so, sopt); 386 break; 387 #endif 388 #ifdef INET6 389 case PF_INET6: 390 error = ip6_ctloutput(op, so, sopt); 391 break; 392 #endif 393 } 394 splx(s); 395 return (error); 396 } 397 if (inp) 398 tp = intotcpcb(inp); 399 #ifdef INET6 400 else if (in6p) 401 tp = in6totcpcb(in6p); 402 #endif 403 else 404 tp = NULL; 405 406 switch (op) { 407 case PRCO_SETOPT: 408 switch (optname) { 409 #ifdef TCP_SIGNATURE 410 case TCP_MD5SIG: 411 error = sockopt_getint(sopt, &optval); 412 if (error) 413 break; 414 if (optval > 0) 415 tp->t_flags |= TF_SIGNATURE; 416 else 417 tp->t_flags &= ~TF_SIGNATURE; 418 break; 419 #endif /* TCP_SIGNATURE */ 420 421 case TCP_NODELAY: 422 error = sockopt_getint(sopt, &optval); 423 if (error) 424 break; 425 if (optval) 426 tp->t_flags |= TF_NODELAY; 427 else 428 tp->t_flags &= ~TF_NODELAY; 429 break; 430 431 case TCP_MAXSEG: 432 error = sockopt_getint(sopt, &optval); 433 if (error) 434 break; 435 if (optval > 0 && optval <= tp->t_peermss) 436 tp->t_peermss = optval; /* limit on send size */ 437 else 438 error = EINVAL; 439 break; 440 #ifdef notyet 441 case TCP_CONGCTL: 442 /* XXX string overflow XXX */ 443 error = tcp_congctl_select(tp, sopt->sopt_data); 444 break; 445 #endif 446 447 case TCP_KEEPIDLE: 448 error = sockopt_get(sopt, &ui, sizeof(ui)); 449 if (error) 450 break; 451 if (ui > 0) { 452 tp->t_keepidle = ui; 453 change_keepalive(so, tp); 454 } else 455 error = EINVAL; 456 break; 457 458 case TCP_KEEPINTVL: 459 error = sockopt_get(sopt, &ui, sizeof(ui)); 460 if (error) 461 break; 462 if (ui > 0) { 463 tp->t_keepintvl = ui; 464 change_keepalive(so, tp); 465 } else 466 error = EINVAL; 467 break; 468 469 case TCP_KEEPCNT: 470 error = sockopt_get(sopt, &ui, sizeof(ui)); 471 if (error) 472 break; 473 if (ui > 0) { 474 tp->t_keepcnt = ui; 475 change_keepalive(so, tp); 476 } else 477 error = EINVAL; 478 break; 479 480 case TCP_KEEPINIT: 481 error = sockopt_get(sopt, &ui, sizeof(ui)); 482 if (error) 483 break; 484 if (ui > 0) { 485 tp->t_keepinit = ui; 486 change_keepalive(so, tp); 487 } else 488 error = EINVAL; 489 break; 490 491 default: 492 error = ENOPROTOOPT; 493 break; 494 } 495 break; 496 497 case PRCO_GETOPT: 498 switch (optname) { 499 #ifdef TCP_SIGNATURE 500 case TCP_MD5SIG: 501 optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0; 502 error = sockopt_set(sopt, &optval, sizeof(optval)); 503 break; 504 #endif 505 case TCP_NODELAY: 506 optval = tp->t_flags & TF_NODELAY; 507 error = sockopt_set(sopt, &optval, sizeof(optval)); 508 break; 509 case TCP_MAXSEG: 510 optval = tp->t_peermss; 511 error = sockopt_set(sopt, &optval, sizeof(optval)); 512 break; 513 case TCP_INFO: 514 tcp_fill_info(tp, &ti); 515 error = sockopt_set(sopt, &ti, sizeof ti); 516 break; 517 #ifdef notyet 518 case TCP_CONGCTL: 519 break; 520 #endif 521 default: 522 error = ENOPROTOOPT; 523 break; 524 } 525 break; 526 } 527 splx(s); 528 return (error); 529 } 530 531 #ifndef TCP_SENDSPACE 532 #define TCP_SENDSPACE 1024*32 533 #endif 534 int tcp_sendspace = TCP_SENDSPACE; 535 #ifndef TCP_RECVSPACE 536 #define TCP_RECVSPACE 1024*32 537 #endif 538 int tcp_recvspace = TCP_RECVSPACE; 539 540 /* 541 * tcp_attach: attach TCP protocol to socket, allocating internet protocol 542 * control block, TCP control block, buffer space and entering LISTEN state 543 * if to accept connections. 544 */ 545 static int 546 tcp_attach(struct socket *so, int proto) 547 { 548 struct tcpcb *tp; 549 struct inpcb *inp; 550 #ifdef INET6 551 struct in6pcb *in6p; 552 #endif 553 int s, error, family; 554 555 /* Assign the lock (must happen even if we will error out). */ 556 s = splsoftnet(); 557 sosetlock(so); 558 KASSERT(solocked(so)); 559 560 family = so->so_proto->pr_domain->dom_family; 561 switch (family) { 562 #ifdef INET 563 case PF_INET: 564 inp = sotoinpcb(so); 565 #ifdef INET6 566 in6p = NULL; 567 #endif 568 break; 569 #endif 570 #ifdef INET6 571 case PF_INET6: 572 inp = NULL; 573 in6p = sotoin6pcb(so); 574 break; 575 #endif 576 default: 577 error = EAFNOSUPPORT; 578 goto out; 579 } 580 581 KASSERT(inp == NULL); 582 #ifdef INET6 583 KASSERT(in6p == NULL); 584 #endif 585 586 #ifdef MBUFTRACE 587 so->so_mowner = &tcp_sock_mowner; 588 so->so_rcv.sb_mowner = &tcp_sock_rx_mowner; 589 so->so_snd.sb_mowner = &tcp_sock_tx_mowner; 590 #endif 591 if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) { 592 error = soreserve(so, tcp_sendspace, tcp_recvspace); 593 if (error) 594 goto out; 595 } 596 597 so->so_rcv.sb_flags |= SB_AUTOSIZE; 598 so->so_snd.sb_flags |= SB_AUTOSIZE; 599 600 switch (family) { 601 #ifdef INET 602 case PF_INET: 603 error = in_pcballoc(so, &tcbtable); 604 if (error) 605 goto out; 606 inp = sotoinpcb(so); 607 #ifdef INET6 608 in6p = NULL; 609 #endif 610 break; 611 #endif 612 #ifdef INET6 613 case PF_INET6: 614 error = in6_pcballoc(so, &tcbtable); 615 if (error) 616 goto out; 617 inp = NULL; 618 in6p = sotoin6pcb(so); 619 break; 620 #endif 621 default: 622 error = EAFNOSUPPORT; 623 goto out; 624 } 625 if (inp) 626 tp = tcp_newtcpcb(family, (void *)inp); 627 #ifdef INET6 628 else if (in6p) 629 tp = tcp_newtcpcb(family, (void *)in6p); 630 #endif 631 else 632 tp = NULL; 633 634 if (tp == NULL) { 635 int nofd = so->so_state & SS_NOFDREF; /* XXX */ 636 637 so->so_state &= ~SS_NOFDREF; /* don't free the socket yet */ 638 #ifdef INET 639 if (inp) 640 in_pcbdetach(inp); 641 #endif 642 #ifdef INET6 643 if (in6p) 644 in6_pcbdetach(in6p); 645 #endif 646 so->so_state |= nofd; 647 error = ENOBUFS; 648 goto out; 649 } 650 tp->t_state = TCPS_CLOSED; 651 if ((so->so_options & SO_LINGER) && so->so_linger == 0) { 652 so->so_linger = TCP_LINGERTIME; 653 } 654 out: 655 KASSERT(solocked(so)); 656 splx(s); 657 return error; 658 } 659 660 static void 661 tcp_detach(struct socket *so) 662 { 663 struct inpcb *inp = NULL; 664 struct in6pcb *in6p = NULL; 665 struct tcpcb *tp = NULL; 666 int s; 667 668 if (tcp_getpcb(so, &inp, &in6p, &tp) != 0) 669 return; 670 671 s = splsoftnet(); 672 (void)tcp_disconnect1(tp); 673 splx(s); 674 } 675 676 static int 677 tcp_accept(struct socket *so, struct sockaddr *nam) 678 { 679 struct inpcb *inp = NULL; 680 struct in6pcb *in6p = NULL; 681 struct tcpcb *tp = NULL; 682 int ostate = 0; 683 int error = 0; 684 int s; 685 686 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 687 return error; 688 689 ostate = tcp_debug_capture(tp, PRU_ACCEPT); 690 691 /* 692 * Accept a connection. Essentially all the work is 693 * done at higher levels; just return the address 694 * of the peer, storing through addr. 695 */ 696 s = splsoftnet(); 697 #ifdef INET 698 if (inp) { 699 in_setpeeraddr(inp, (struct sockaddr_in *)nam); 700 } 701 #endif 702 #ifdef INET6 703 if (in6p) { 704 in6_setpeeraddr(in6p, (struct sockaddr_in6 *)nam); 705 } 706 #endif 707 tcp_debug_trace(so, tp, ostate, PRU_ACCEPT); 708 splx(s); 709 710 return 0; 711 } 712 713 static int 714 tcp_bind(struct socket *so, struct sockaddr *nam, struct lwp *l) 715 { 716 struct inpcb *inp = NULL; 717 struct in6pcb *in6p = NULL; 718 struct sockaddr_in *sin = (struct sockaddr_in *)nam; 719 #ifdef INET6 720 struct sockaddr_in6 *sin6 = (struct sockaddr_in6 *)nam; 721 #endif /* INET6 */ 722 struct tcpcb *tp = NULL; 723 int s; 724 int error = 0; 725 int ostate = 0; 726 727 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 728 return error; 729 730 ostate = tcp_debug_capture(tp, PRU_BIND); 731 732 /* 733 * Give the socket an address. 734 */ 735 s = splsoftnet(); 736 switch (so->so_proto->pr_domain->dom_family) { 737 #ifdef INET 738 case PF_INET: 739 error = in_pcbbind(inp, sin, l); 740 break; 741 #endif 742 #ifdef INET6 743 case PF_INET6: 744 error = in6_pcbbind(in6p, sin6, l); 745 if (!error) { 746 /* mapped addr case */ 747 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr)) 748 tp->t_family = AF_INET; 749 else 750 tp->t_family = AF_INET6; 751 } 752 break; 753 #endif 754 } 755 tcp_debug_trace(so, tp, ostate, PRU_BIND); 756 splx(s); 757 758 return error; 759 } 760 761 static int 762 tcp_listen(struct socket *so, struct lwp *l) 763 { 764 struct inpcb *inp = NULL; 765 struct in6pcb *in6p = NULL; 766 struct tcpcb *tp = NULL; 767 int error = 0; 768 int ostate = 0; 769 int s; 770 771 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 772 return error; 773 774 ostate = tcp_debug_capture(tp, PRU_LISTEN); 775 776 /* 777 * Prepare to accept connections. 778 */ 779 s = splsoftnet(); 780 #ifdef INET 781 if (inp && inp->inp_lport == 0) { 782 error = in_pcbbind(inp, NULL, l); 783 if (error) 784 goto release; 785 } 786 #endif 787 #ifdef INET6 788 if (in6p && in6p->in6p_lport == 0) { 789 error = in6_pcbbind(in6p, NULL, l); 790 if (error) 791 goto release; 792 } 793 #endif 794 tp->t_state = TCPS_LISTEN; 795 796 release: 797 tcp_debug_trace(so, tp, ostate, PRU_LISTEN); 798 splx(s); 799 800 return error; 801 } 802 803 static int 804 tcp_connect(struct socket *so, struct mbuf *nam, struct lwp *l) 805 { 806 struct inpcb *inp = NULL; 807 struct in6pcb *in6p = NULL; 808 struct tcpcb *tp = NULL; 809 int s; 810 int error = 0; 811 int ostate = 0; 812 813 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 814 return error; 815 816 ostate = tcp_debug_capture(tp, PRU_CONNECT); 817 818 /* 819 * Initiate connection to peer. 820 * Create a template for use in transmissions on this connection. 821 * Enter SYN_SENT state, and mark socket as connecting. 822 * Start keep-alive timer, and seed output sequence space. 823 * Send initial segment on connection. 824 */ 825 s = splsoftnet(); 826 #ifdef INET 827 if (inp) { 828 if (inp->inp_lport == 0) { 829 error = in_pcbbind(inp, NULL, l); 830 if (error) 831 goto release; 832 } 833 error = in_pcbconnect(inp, nam, l); 834 } 835 #endif 836 #ifdef INET6 837 if (in6p) { 838 if (in6p->in6p_lport == 0) { 839 error = in6_pcbbind(in6p, NULL, l); 840 if (error) 841 goto release; 842 } 843 error = in6_pcbconnect(in6p, nam, l); 844 if (!error) { 845 /* mapped addr case */ 846 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 847 tp->t_family = AF_INET; 848 else 849 tp->t_family = AF_INET6; 850 } 851 } 852 #endif 853 if (error) 854 goto release; 855 tp->t_template = tcp_template(tp); 856 if (tp->t_template == 0) { 857 #ifdef INET 858 if (inp) 859 in_pcbdisconnect(inp); 860 #endif 861 #ifdef INET6 862 if (in6p) 863 in6_pcbdisconnect(in6p); 864 #endif 865 error = ENOBUFS; 866 goto release; 867 } 868 /* 869 * Compute window scaling to request. 870 * XXX: This should be moved to tcp_output(). 871 */ 872 while (tp->request_r_scale < TCP_MAX_WINSHIFT && 873 (TCP_MAXWIN << tp->request_r_scale) < sb_max) 874 tp->request_r_scale++; 875 soisconnecting(so); 876 TCP_STATINC(TCP_STAT_CONNATTEMPT); 877 tp->t_state = TCPS_SYN_SENT; 878 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit); 879 tp->iss = tcp_new_iss(tp, 0); 880 tcp_sendseqinit(tp); 881 error = tcp_output(tp); 882 883 release: 884 tcp_debug_trace(so, tp, ostate, PRU_CONNECT); 885 splx(s); 886 887 return error; 888 } 889 890 static int 891 tcp_connect2(struct socket *so, struct socket *so2) 892 { 893 struct inpcb *inp = NULL; 894 struct in6pcb *in6p = NULL; 895 struct tcpcb *tp = NULL; 896 int error = 0; 897 int ostate = 0; 898 899 KASSERT(solocked(so)); 900 901 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 902 return error; 903 904 ostate = tcp_debug_capture(tp, PRU_CONNECT2); 905 906 tcp_debug_trace(so, tp, ostate, PRU_CONNECT2); 907 908 return EOPNOTSUPP; 909 } 910 911 static int 912 tcp_disconnect(struct socket *so) 913 { 914 struct inpcb *inp = NULL; 915 struct in6pcb *in6p = NULL; 916 struct tcpcb *tp = NULL; 917 int error = 0; 918 int ostate = 0; 919 int s; 920 921 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 922 return error; 923 924 ostate = tcp_debug_capture(tp, PRU_DISCONNECT); 925 926 /* 927 * Initiate disconnect from peer. 928 * If connection never passed embryonic stage, just drop; 929 * else if don't need to let data drain, then can just drop anyways, 930 * else have to begin TCP shutdown process: mark socket disconnecting, 931 * drain unread data, state switch to reflect user close, and 932 * send segment (e.g. FIN) to peer. Socket will be really disconnected 933 * when peer sends FIN and acks ours. 934 * 935 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB. 936 */ 937 s = splsoftnet(); 938 tp = tcp_disconnect1(tp); 939 tcp_debug_trace(so, tp, ostate, PRU_DISCONNECT); 940 splx(s); 941 942 return error; 943 } 944 945 static int 946 tcp_shutdown(struct socket *so) 947 { 948 struct inpcb *inp = NULL; 949 struct in6pcb *in6p = NULL; 950 struct tcpcb *tp = NULL; 951 int error = 0; 952 int ostate = 0; 953 int s; 954 955 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 956 return error; 957 958 ostate = tcp_debug_capture(tp, PRU_SHUTDOWN); 959 /* 960 * Mark the connection as being incapable of further output. 961 */ 962 s = splsoftnet(); 963 socantsendmore(so); 964 tp = tcp_usrclosed(tp); 965 if (tp) 966 error = tcp_output(tp); 967 tcp_debug_trace(so, tp, ostate, PRU_SHUTDOWN); 968 splx(s); 969 970 return error; 971 } 972 973 static int 974 tcp_abort(struct socket *so) 975 { 976 struct inpcb *inp = NULL; 977 struct in6pcb *in6p = NULL; 978 struct tcpcb *tp = NULL; 979 int error = 0; 980 int ostate = 0; 981 int s; 982 983 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 984 return error; 985 986 ostate = tcp_debug_capture(tp, PRU_ABORT); 987 988 /* 989 * Abort the TCP. 990 */ 991 s = splsoftnet(); 992 tp = tcp_drop(tp, ECONNABORTED); 993 tcp_debug_trace(so, tp, ostate, PRU_ABORT); 994 splx(s); 995 996 return error; 997 } 998 999 static int 1000 tcp_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp) 1001 { 1002 switch (so->so_proto->pr_domain->dom_family) { 1003 #ifdef INET 1004 case PF_INET: 1005 return in_control(so, cmd, nam, ifp); 1006 #endif 1007 #ifdef INET6 1008 case PF_INET6: 1009 return in6_control(so, cmd, nam, ifp); 1010 #endif 1011 default: 1012 return EAFNOSUPPORT; 1013 } 1014 } 1015 1016 static int 1017 tcp_stat(struct socket *so, struct stat *ub) 1018 { 1019 KASSERT(solocked(so)); 1020 1021 /* stat: don't bother with a blocksize. */ 1022 return 0; 1023 } 1024 1025 static int 1026 tcp_peeraddr(struct socket *so, struct sockaddr *nam) 1027 { 1028 struct inpcb *inp = NULL; 1029 struct in6pcb *in6p = NULL; 1030 struct tcpcb *tp = NULL; 1031 int ostate = 0; 1032 int error = 0; 1033 int s; 1034 1035 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1036 return error; 1037 1038 ostate = tcp_debug_capture(tp, PRU_PEERADDR); 1039 1040 s = splsoftnet(); 1041 #ifdef INET 1042 if (inp) { 1043 in_setpeeraddr(inp, (struct sockaddr_in *)nam); 1044 } 1045 #endif 1046 #ifdef INET6 1047 if (in6p) { 1048 in6_setpeeraddr(in6p, (struct sockaddr_in6 *)nam); 1049 } 1050 #endif 1051 tcp_debug_trace(so, tp, ostate, PRU_PEERADDR); 1052 splx(s); 1053 1054 return 0; 1055 } 1056 1057 static int 1058 tcp_sockaddr(struct socket *so, struct sockaddr *nam) 1059 { 1060 struct inpcb *inp = NULL; 1061 struct in6pcb *in6p = NULL; 1062 struct tcpcb *tp = NULL; 1063 int ostate = 0; 1064 int error = 0; 1065 int s; 1066 1067 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1068 return error; 1069 1070 ostate = tcp_debug_capture(tp, PRU_SOCKADDR); 1071 1072 s = splsoftnet(); 1073 #ifdef INET 1074 if (inp) { 1075 in_setsockaddr(inp, (struct sockaddr_in *)nam); 1076 } 1077 #endif 1078 #ifdef INET6 1079 if (in6p) { 1080 in6_setsockaddr(in6p, (struct sockaddr_in6 *)nam); 1081 } 1082 #endif 1083 tcp_debug_trace(so, tp, ostate, PRU_SOCKADDR); 1084 splx(s); 1085 1086 return 0; 1087 } 1088 1089 static int 1090 tcp_rcvd(struct socket *so, int flags, struct lwp *l) 1091 { 1092 struct inpcb *inp = NULL; 1093 struct in6pcb *in6p = NULL; 1094 struct tcpcb *tp = NULL; 1095 int ostate = 0; 1096 int error = 0; 1097 int s; 1098 1099 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1100 return error; 1101 1102 ostate = tcp_debug_capture(tp, PRU_RCVD); 1103 1104 /* 1105 * After a receive, possibly send window update to peer. 1106 * 1107 * soreceive() calls this function when a user receives 1108 * ancillary data on a listening socket. We don't call 1109 * tcp_output in such a case, since there is no header 1110 * template for a listening socket and hence the kernel 1111 * will panic. 1112 */ 1113 s = splsoftnet(); 1114 if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0) 1115 (void) tcp_output(tp); 1116 splx(s); 1117 1118 tcp_debug_trace(so, tp, ostate, PRU_RCVD); 1119 1120 return 0; 1121 } 1122 1123 static int 1124 tcp_recvoob(struct socket *so, struct mbuf *m, int flags) 1125 { 1126 struct inpcb *inp = NULL; 1127 struct in6pcb *in6p = NULL; 1128 struct tcpcb *tp = NULL; 1129 int ostate = 0; 1130 int error = 0; 1131 int s; 1132 1133 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1134 return error; 1135 1136 ostate = tcp_debug_capture(tp, PRU_RCVOOB); 1137 1138 s = splsoftnet(); 1139 if ((so->so_oobmark == 0 && 1140 (so->so_state & SS_RCVATMARK) == 0) || 1141 so->so_options & SO_OOBINLINE || 1142 tp->t_oobflags & TCPOOB_HADDATA) { 1143 splx(s); 1144 return EINVAL; 1145 } 1146 1147 if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) { 1148 splx(s); 1149 return EWOULDBLOCK; 1150 } 1151 1152 m->m_len = 1; 1153 *mtod(m, char *) = tp->t_iobc; 1154 if ((flags & MSG_PEEK) == 0) 1155 tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA); 1156 1157 tcp_debug_trace(so, tp, ostate, PRU_RCVOOB); 1158 splx(s); 1159 1160 return 0; 1161 } 1162 1163 static int 1164 tcp_send(struct socket *so, struct mbuf *m, struct mbuf *nam, 1165 struct mbuf *control, struct lwp *l) 1166 { 1167 struct inpcb *inp = NULL; 1168 struct in6pcb *in6p = NULL; 1169 struct tcpcb *tp = NULL; 1170 int ostate = 0; 1171 int error = 0; 1172 int s; 1173 1174 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1175 return error; 1176 1177 ostate = tcp_debug_capture(tp, PRU_SEND); 1178 1179 /* 1180 * Do a send by putting data in output queue and updating urgent 1181 * marker if URG set. Possibly send more data. 1182 */ 1183 s = splsoftnet(); 1184 if (control && control->m_len) { 1185 m_freem(control); 1186 m_freem(m); 1187 tcp_debug_trace(so, tp, ostate, PRU_SEND); 1188 splx(s); 1189 return EINVAL; 1190 } 1191 1192 sbappendstream(&so->so_snd, m); 1193 error = tcp_output(tp); 1194 tcp_debug_trace(so, tp, ostate, PRU_SEND); 1195 splx(s); 1196 1197 return error; 1198 } 1199 1200 static int 1201 tcp_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control) 1202 { 1203 struct inpcb *inp = NULL; 1204 struct in6pcb *in6p = NULL; 1205 struct tcpcb *tp = NULL; 1206 int ostate = 0; 1207 int error = 0; 1208 int s; 1209 1210 if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0) 1211 return error; 1212 1213 ostate = tcp_debug_capture(tp, PRU_SENDOOB); 1214 1215 s = splsoftnet(); 1216 if (sbspace(&so->so_snd) < -512) { 1217 m_freem(m); 1218 splx(s); 1219 return ENOBUFS; 1220 } 1221 /* 1222 * According to RFC961 (Assigned Protocols), 1223 * the urgent pointer points to the last octet 1224 * of urgent data. We continue, however, 1225 * to consider it to indicate the first octet 1226 * of data past the urgent section. 1227 * Otherwise, snd_up should be one lower. 1228 */ 1229 sbappendstream(&so->so_snd, m); 1230 tp->snd_up = tp->snd_una + so->so_snd.sb_cc; 1231 tp->t_force = 1; 1232 error = tcp_output(tp); 1233 tp->t_force = 0; 1234 tcp_debug_trace(so, tp, ostate, PRU_SENDOOB); 1235 splx(s); 1236 1237 return error; 1238 } 1239 1240 static int 1241 tcp_purgeif(struct socket *so, struct ifnet *ifp) 1242 { 1243 int s; 1244 1245 s = splsoftnet(); 1246 mutex_enter(softnet_lock); 1247 switch (so->so_proto->pr_domain->dom_family) { 1248 #ifdef INET 1249 case PF_INET: 1250 in_pcbpurgeif0(&tcbtable, ifp); 1251 in_purgeif(ifp); 1252 in_pcbpurgeif(&tcbtable, ifp); 1253 break; 1254 #endif 1255 #ifdef INET6 1256 case PF_INET6: 1257 in6_pcbpurgeif0(&tcbtable, ifp); 1258 in6_purgeif(ifp); 1259 in6_pcbpurgeif(&tcbtable, ifp); 1260 break; 1261 #endif 1262 default: 1263 mutex_exit(softnet_lock); 1264 splx(s); 1265 return EAFNOSUPPORT; 1266 } 1267 mutex_exit(softnet_lock); 1268 splx(s); 1269 1270 return 0; 1271 } 1272 1273 /* 1274 * Initiate (or continue) disconnect. 1275 * If embryonic state, just send reset (once). 1276 * If in ``let data drain'' option and linger null, just drop. 1277 * Otherwise (hard), mark socket disconnecting and drop 1278 * current input data; switch states based on user close, and 1279 * send segment to peer (with FIN). 1280 */ 1281 struct tcpcb * 1282 tcp_disconnect1(struct tcpcb *tp) 1283 { 1284 struct socket *so; 1285 1286 if (tp->t_inpcb) 1287 so = tp->t_inpcb->inp_socket; 1288 #ifdef INET6 1289 else if (tp->t_in6pcb) 1290 so = tp->t_in6pcb->in6p_socket; 1291 #endif 1292 else 1293 so = NULL; 1294 1295 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 1296 tp = tcp_close(tp); 1297 else if ((so->so_options & SO_LINGER) && so->so_linger == 0) 1298 tp = tcp_drop(tp, 0); 1299 else { 1300 soisdisconnecting(so); 1301 sbflush(&so->so_rcv); 1302 tp = tcp_usrclosed(tp); 1303 if (tp) 1304 (void) tcp_output(tp); 1305 } 1306 return (tp); 1307 } 1308 1309 /* 1310 * User issued close, and wish to trail through shutdown states: 1311 * if never received SYN, just forget it. If got a SYN from peer, 1312 * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN. 1313 * If already got a FIN from peer, then almost done; go to LAST_ACK 1314 * state. In all other cases, have already sent FIN to peer (e.g. 1315 * after PRU_SHUTDOWN), and just have to play tedious game waiting 1316 * for peer to send FIN or not respond to keep-alives, etc. 1317 * We can let the user exit from the close as soon as the FIN is acked. 1318 */ 1319 struct tcpcb * 1320 tcp_usrclosed(struct tcpcb *tp) 1321 { 1322 1323 switch (tp->t_state) { 1324 1325 case TCPS_CLOSED: 1326 case TCPS_LISTEN: 1327 case TCPS_SYN_SENT: 1328 tp->t_state = TCPS_CLOSED; 1329 tp = tcp_close(tp); 1330 break; 1331 1332 case TCPS_SYN_RECEIVED: 1333 case TCPS_ESTABLISHED: 1334 tp->t_state = TCPS_FIN_WAIT_1; 1335 break; 1336 1337 case TCPS_CLOSE_WAIT: 1338 tp->t_state = TCPS_LAST_ACK; 1339 break; 1340 } 1341 if (tp && tp->t_state >= TCPS_FIN_WAIT_2) { 1342 struct socket *so; 1343 if (tp->t_inpcb) 1344 so = tp->t_inpcb->inp_socket; 1345 #ifdef INET6 1346 else if (tp->t_in6pcb) 1347 so = tp->t_in6pcb->in6p_socket; 1348 #endif 1349 else 1350 so = NULL; 1351 if (so) 1352 soisdisconnected(so); 1353 /* 1354 * If we are in FIN_WAIT_2, we arrived here because the 1355 * application did a shutdown of the send side. Like the 1356 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after 1357 * a full close, we start a timer to make sure sockets are 1358 * not left in FIN_WAIT_2 forever. 1359 */ 1360 if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0)) 1361 TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle); 1362 else if (tp->t_state == TCPS_TIME_WAIT 1363 && ((tp->t_inpcb 1364 && (tcp4_vtw_enable & 1) 1365 && vtw_add(AF_INET, tp)) 1366 || 1367 (tp->t_in6pcb 1368 && (tcp6_vtw_enable & 1) 1369 && vtw_add(AF_INET6, tp)))) { 1370 tp = 0; 1371 } 1372 } 1373 return (tp); 1374 } 1375 1376 /* 1377 * sysctl helper routine for net.inet.ip.mssdflt. it can't be less 1378 * than 32. 1379 */ 1380 static int 1381 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS) 1382 { 1383 int error, mssdflt; 1384 struct sysctlnode node; 1385 1386 mssdflt = tcp_mssdflt; 1387 node = *rnode; 1388 node.sysctl_data = &mssdflt; 1389 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1390 if (error || newp == NULL) 1391 return (error); 1392 1393 if (mssdflt < 32) 1394 return (EINVAL); 1395 tcp_mssdflt = mssdflt; 1396 1397 mutex_enter(softnet_lock); 1398 tcp_tcpcb_template(); 1399 mutex_exit(softnet_lock); 1400 1401 return (0); 1402 } 1403 1404 /* 1405 * sysctl helper for TCP CB template update 1406 */ 1407 static int 1408 sysctl_update_tcpcb_template(SYSCTLFN_ARGS) 1409 { 1410 int t, error; 1411 struct sysctlnode node; 1412 1413 /* follow procedures in sysctl(9) manpage */ 1414 t = *(int *)rnode->sysctl_data; 1415 node = *rnode; 1416 node.sysctl_data = &t; 1417 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1418 if (error || newp == NULL) 1419 return error; 1420 1421 if (t < 0) 1422 return EINVAL; 1423 1424 *(int *)rnode->sysctl_data = t; 1425 1426 mutex_enter(softnet_lock); 1427 tcp_tcpcb_template(); 1428 mutex_exit(softnet_lock); 1429 1430 return 0; 1431 } 1432 1433 /* 1434 * sysctl helper routine for setting port related values under 1435 * net.inet.ip and net.inet6.ip6. does basic range checking and does 1436 * additional checks for each type. this code has placed in 1437 * tcp_input.c since INET and INET6 both use the same tcp code. 1438 * 1439 * this helper is not static so that both inet and inet6 can use it. 1440 */ 1441 int 1442 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS) 1443 { 1444 int error, tmp; 1445 int apmin, apmax; 1446 #ifndef IPNOPRIVPORTS 1447 int lpmin, lpmax; 1448 #endif /* IPNOPRIVPORTS */ 1449 struct sysctlnode node; 1450 1451 if (namelen != 0) 1452 return (EINVAL); 1453 1454 switch (name[-3]) { 1455 #ifdef INET 1456 case PF_INET: 1457 apmin = anonportmin; 1458 apmax = anonportmax; 1459 #ifndef IPNOPRIVPORTS 1460 lpmin = lowportmin; 1461 lpmax = lowportmax; 1462 #endif /* IPNOPRIVPORTS */ 1463 break; 1464 #endif /* INET */ 1465 #ifdef INET6 1466 case PF_INET6: 1467 apmin = ip6_anonportmin; 1468 apmax = ip6_anonportmax; 1469 #ifndef IPNOPRIVPORTS 1470 lpmin = ip6_lowportmin; 1471 lpmax = ip6_lowportmax; 1472 #endif /* IPNOPRIVPORTS */ 1473 break; 1474 #endif /* INET6 */ 1475 default: 1476 return (EINVAL); 1477 } 1478 1479 /* 1480 * insert temporary copy into node, perform lookup on 1481 * temporary, then restore pointer 1482 */ 1483 node = *rnode; 1484 tmp = *(int*)rnode->sysctl_data; 1485 node.sysctl_data = &tmp; 1486 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1487 if (error || newp == NULL) 1488 return (error); 1489 1490 /* 1491 * simple port range check 1492 */ 1493 if (tmp < 0 || tmp > 65535) 1494 return (EINVAL); 1495 1496 /* 1497 * per-node range checks 1498 */ 1499 switch (rnode->sysctl_num) { 1500 case IPCTL_ANONPORTMIN: 1501 case IPV6CTL_ANONPORTMIN: 1502 if (tmp >= apmax) 1503 return (EINVAL); 1504 #ifndef IPNOPRIVPORTS 1505 if (tmp < IPPORT_RESERVED) 1506 return (EINVAL); 1507 #endif /* IPNOPRIVPORTS */ 1508 break; 1509 1510 case IPCTL_ANONPORTMAX: 1511 case IPV6CTL_ANONPORTMAX: 1512 if (apmin >= tmp) 1513 return (EINVAL); 1514 #ifndef IPNOPRIVPORTS 1515 if (tmp < IPPORT_RESERVED) 1516 return (EINVAL); 1517 #endif /* IPNOPRIVPORTS */ 1518 break; 1519 1520 #ifndef IPNOPRIVPORTS 1521 case IPCTL_LOWPORTMIN: 1522 case IPV6CTL_LOWPORTMIN: 1523 if (tmp >= lpmax || 1524 tmp > IPPORT_RESERVEDMAX || 1525 tmp < IPPORT_RESERVEDMIN) 1526 return (EINVAL); 1527 break; 1528 1529 case IPCTL_LOWPORTMAX: 1530 case IPV6CTL_LOWPORTMAX: 1531 if (lpmin >= tmp || 1532 tmp > IPPORT_RESERVEDMAX || 1533 tmp < IPPORT_RESERVEDMIN) 1534 return (EINVAL); 1535 break; 1536 #endif /* IPNOPRIVPORTS */ 1537 1538 default: 1539 return (EINVAL); 1540 } 1541 1542 *(int*)rnode->sysctl_data = tmp; 1543 1544 return (0); 1545 } 1546 1547 static inline int 1548 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp) 1549 { 1550 if (oldp) { 1551 size_t sz; 1552 uid_t uid; 1553 int error; 1554 1555 if (sockp->so_cred == NULL) 1556 return EPERM; 1557 1558 uid = kauth_cred_geteuid(sockp->so_cred); 1559 sz = MIN(sizeof(uid), *oldlenp); 1560 if ((error = copyout(&uid, oldp, sz)) != 0) 1561 return error; 1562 } 1563 *oldlenp = sizeof(uid_t); 1564 return 0; 1565 } 1566 1567 static inline int 1568 inet4_ident_core(struct in_addr raddr, u_int rport, 1569 struct in_addr laddr, u_int lport, 1570 void *oldp, size_t *oldlenp, 1571 struct lwp *l, int dodrop) 1572 { 1573 struct inpcb *inp; 1574 struct socket *sockp; 1575 1576 inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0); 1577 1578 if (inp == NULL || (sockp = inp->inp_socket) == NULL) 1579 return ESRCH; 1580 1581 if (dodrop) { 1582 struct tcpcb *tp; 1583 int error; 1584 1585 if (inp == NULL || (tp = intotcpcb(inp)) == NULL || 1586 (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0) 1587 return ESRCH; 1588 1589 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1590 KAUTH_REQ_NETWORK_SOCKET_DROP, inp->inp_socket, tp, NULL); 1591 if (error) 1592 return (error); 1593 1594 (void)tcp_drop(tp, ECONNABORTED); 1595 return 0; 1596 } 1597 else 1598 return copyout_uid(sockp, oldp, oldlenp); 1599 } 1600 1601 #ifdef INET6 1602 static inline int 1603 inet6_ident_core(struct in6_addr *raddr, u_int rport, 1604 struct in6_addr *laddr, u_int lport, 1605 void *oldp, size_t *oldlenp, 1606 struct lwp *l, int dodrop) 1607 { 1608 struct in6pcb *in6p; 1609 struct socket *sockp; 1610 1611 in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0, 0); 1612 1613 if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL) 1614 return ESRCH; 1615 1616 if (dodrop) { 1617 struct tcpcb *tp; 1618 int error; 1619 1620 if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL || 1621 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0) 1622 return ESRCH; 1623 1624 error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1625 KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL); 1626 if (error) 1627 return (error); 1628 1629 (void)tcp_drop(tp, ECONNABORTED); 1630 return 0; 1631 } 1632 else 1633 return copyout_uid(sockp, oldp, oldlenp); 1634 } 1635 #endif 1636 1637 /* 1638 * sysctl helper routine for the net.inet.tcp.drop and 1639 * net.inet6.tcp6.drop nodes. 1640 */ 1641 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident 1642 1643 /* 1644 * sysctl helper routine for the net.inet.tcp.ident and 1645 * net.inet6.tcp6.ident nodes. contains backwards compat code for the 1646 * old way of looking up the ident information for ipv4 which involves 1647 * stuffing the port/addr pairs into the mib lookup. 1648 */ 1649 static int 1650 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS) 1651 { 1652 #ifdef INET 1653 struct sockaddr_in *si4[2]; 1654 #endif /* INET */ 1655 #ifdef INET6 1656 struct sockaddr_in6 *si6[2]; 1657 #endif /* INET6 */ 1658 struct sockaddr_storage sa[2]; 1659 int error, pf, dodrop; 1660 1661 dodrop = name[-1] == TCPCTL_DROP; 1662 if (dodrop) { 1663 if (oldp != NULL || *oldlenp != 0) 1664 return EINVAL; 1665 if (newp == NULL) 1666 return EPERM; 1667 if (newlen < sizeof(sa)) 1668 return ENOMEM; 1669 } 1670 if (namelen != 4 && namelen != 0) 1671 return EINVAL; 1672 if (name[-2] != IPPROTO_TCP) 1673 return EINVAL; 1674 pf = name[-3]; 1675 1676 /* old style lookup, ipv4 only */ 1677 if (namelen == 4) { 1678 #ifdef INET 1679 struct in_addr laddr, raddr; 1680 u_int lport, rport; 1681 1682 if (pf != PF_INET) 1683 return EPROTONOSUPPORT; 1684 raddr.s_addr = (uint32_t)name[0]; 1685 rport = (u_int)name[1]; 1686 laddr.s_addr = (uint32_t)name[2]; 1687 lport = (u_int)name[3]; 1688 1689 mutex_enter(softnet_lock); 1690 error = inet4_ident_core(raddr, rport, laddr, lport, 1691 oldp, oldlenp, l, dodrop); 1692 mutex_exit(softnet_lock); 1693 return error; 1694 #else /* INET */ 1695 return EINVAL; 1696 #endif /* INET */ 1697 } 1698 1699 if (newp == NULL || newlen != sizeof(sa)) 1700 return EINVAL; 1701 error = copyin(newp, &sa, newlen); 1702 if (error) 1703 return error; 1704 1705 /* 1706 * requested families must match 1707 */ 1708 if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family) 1709 return EINVAL; 1710 1711 switch (pf) { 1712 #ifdef INET6 1713 case PF_INET6: 1714 si6[0] = (struct sockaddr_in6*)&sa[0]; 1715 si6[1] = (struct sockaddr_in6*)&sa[1]; 1716 if (si6[0]->sin6_len != sizeof(*si6[0]) || 1717 si6[1]->sin6_len != sizeof(*si6[1])) 1718 return EINVAL; 1719 1720 if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) && 1721 !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) { 1722 error = sa6_embedscope(si6[0], ip6_use_defzone); 1723 if (error) 1724 return error; 1725 error = sa6_embedscope(si6[1], ip6_use_defzone); 1726 if (error) 1727 return error; 1728 1729 mutex_enter(softnet_lock); 1730 error = inet6_ident_core(&si6[0]->sin6_addr, 1731 si6[0]->sin6_port, &si6[1]->sin6_addr, 1732 si6[1]->sin6_port, oldp, oldlenp, l, dodrop); 1733 mutex_exit(softnet_lock); 1734 return error; 1735 } 1736 1737 if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) != 1738 IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) 1739 return EINVAL; 1740 1741 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]); 1742 in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]); 1743 /*FALLTHROUGH*/ 1744 #endif /* INET6 */ 1745 #ifdef INET 1746 case PF_INET: 1747 si4[0] = (struct sockaddr_in*)&sa[0]; 1748 si4[1] = (struct sockaddr_in*)&sa[1]; 1749 if (si4[0]->sin_len != sizeof(*si4[0]) || 1750 si4[0]->sin_len != sizeof(*si4[1])) 1751 return EINVAL; 1752 1753 mutex_enter(softnet_lock); 1754 error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port, 1755 si4[1]->sin_addr, si4[1]->sin_port, 1756 oldp, oldlenp, l, dodrop); 1757 mutex_exit(softnet_lock); 1758 return error; 1759 #endif /* INET */ 1760 default: 1761 return EPROTONOSUPPORT; 1762 } 1763 } 1764 1765 /* 1766 * sysctl helper for the inet and inet6 pcblists. handles tcp/udp and 1767 * inet/inet6, as well as raw pcbs for each. specifically not 1768 * declared static so that raw sockets and udp/udp6 can use it as 1769 * well. 1770 */ 1771 int 1772 sysctl_inpcblist(SYSCTLFN_ARGS) 1773 { 1774 #ifdef INET 1775 struct sockaddr_in *in; 1776 const struct inpcb *inp; 1777 #endif 1778 #ifdef INET6 1779 struct sockaddr_in6 *in6; 1780 const struct in6pcb *in6p; 1781 #endif 1782 struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data); 1783 const struct inpcb_hdr *inph; 1784 struct tcpcb *tp; 1785 struct kinfo_pcb pcb; 1786 char *dp; 1787 size_t len, needed, elem_size, out_size; 1788 int error, elem_count, pf, proto, pf2; 1789 1790 if (namelen != 4) 1791 return (EINVAL); 1792 1793 if (oldp != NULL) { 1794 len = *oldlenp; 1795 elem_size = name[2]; 1796 elem_count = name[3]; 1797 if (elem_size != sizeof(pcb)) 1798 return EINVAL; 1799 } else { 1800 len = 0; 1801 elem_count = INT_MAX; 1802 elem_size = sizeof(pcb); 1803 } 1804 error = 0; 1805 dp = oldp; 1806 out_size = elem_size; 1807 needed = 0; 1808 1809 if (namelen == 1 && name[0] == CTL_QUERY) 1810 return (sysctl_query(SYSCTLFN_CALL(rnode))); 1811 1812 if (name - oname != 4) 1813 return (EINVAL); 1814 1815 pf = oname[1]; 1816 proto = oname[2]; 1817 pf2 = (oldp != NULL) ? pf : 0; 1818 1819 mutex_enter(softnet_lock); 1820 1821 TAILQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) { 1822 #ifdef INET 1823 inp = (const struct inpcb *)inph; 1824 #endif 1825 #ifdef INET6 1826 in6p = (const struct in6pcb *)inph; 1827 #endif 1828 1829 if (inph->inph_af != pf) 1830 continue; 1831 1832 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET, 1833 KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL, 1834 NULL) != 0) 1835 continue; 1836 1837 memset(&pcb, 0, sizeof(pcb)); 1838 1839 pcb.ki_family = pf; 1840 pcb.ki_type = proto; 1841 1842 switch (pf2) { 1843 case 0: 1844 /* just probing for size */ 1845 break; 1846 #ifdef INET 1847 case PF_INET: 1848 pcb.ki_family = inp->inp_socket->so_proto-> 1849 pr_domain->dom_family; 1850 pcb.ki_type = inp->inp_socket->so_proto-> 1851 pr_type; 1852 pcb.ki_protocol = inp->inp_socket->so_proto-> 1853 pr_protocol; 1854 pcb.ki_pflags = inp->inp_flags; 1855 1856 pcb.ki_sostate = inp->inp_socket->so_state; 1857 pcb.ki_prstate = inp->inp_state; 1858 if (proto == IPPROTO_TCP) { 1859 tp = intotcpcb(inp); 1860 pcb.ki_tstate = tp->t_state; 1861 pcb.ki_tflags = tp->t_flags; 1862 } 1863 1864 pcb.ki_pcbaddr = PTRTOUINT64(inp); 1865 pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb); 1866 pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket); 1867 1868 pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc; 1869 pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc; 1870 1871 in = satosin(&pcb.ki_src); 1872 in->sin_len = sizeof(*in); 1873 in->sin_family = pf; 1874 in->sin_port = inp->inp_lport; 1875 in->sin_addr = inp->inp_laddr; 1876 if (pcb.ki_prstate >= INP_CONNECTED) { 1877 in = satosin(&pcb.ki_dst); 1878 in->sin_len = sizeof(*in); 1879 in->sin_family = pf; 1880 in->sin_port = inp->inp_fport; 1881 in->sin_addr = inp->inp_faddr; 1882 } 1883 break; 1884 #endif 1885 #ifdef INET6 1886 case PF_INET6: 1887 pcb.ki_family = in6p->in6p_socket->so_proto-> 1888 pr_domain->dom_family; 1889 pcb.ki_type = in6p->in6p_socket->so_proto->pr_type; 1890 pcb.ki_protocol = in6p->in6p_socket->so_proto-> 1891 pr_protocol; 1892 pcb.ki_pflags = in6p->in6p_flags; 1893 1894 pcb.ki_sostate = in6p->in6p_socket->so_state; 1895 pcb.ki_prstate = in6p->in6p_state; 1896 if (proto == IPPROTO_TCP) { 1897 tp = in6totcpcb(in6p); 1898 pcb.ki_tstate = tp->t_state; 1899 pcb.ki_tflags = tp->t_flags; 1900 } 1901 1902 pcb.ki_pcbaddr = PTRTOUINT64(in6p); 1903 pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb); 1904 pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket); 1905 1906 pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc; 1907 pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc; 1908 1909 in6 = satosin6(&pcb.ki_src); 1910 in6->sin6_len = sizeof(*in6); 1911 in6->sin6_family = pf; 1912 in6->sin6_port = in6p->in6p_lport; 1913 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1914 in6->sin6_addr = in6p->in6p_laddr; 1915 in6->sin6_scope_id = 0; /* XXX? */ 1916 1917 if (pcb.ki_prstate >= IN6P_CONNECTED) { 1918 in6 = satosin6(&pcb.ki_dst); 1919 in6->sin6_len = sizeof(*in6); 1920 in6->sin6_family = pf; 1921 in6->sin6_port = in6p->in6p_fport; 1922 in6->sin6_flowinfo = in6p->in6p_flowinfo; 1923 in6->sin6_addr = in6p->in6p_faddr; 1924 in6->sin6_scope_id = 0; /* XXX? */ 1925 } 1926 break; 1927 #endif 1928 } 1929 1930 if (len >= elem_size && elem_count > 0) { 1931 error = copyout(&pcb, dp, out_size); 1932 if (error) { 1933 mutex_exit(softnet_lock); 1934 return (error); 1935 } 1936 dp += elem_size; 1937 len -= elem_size; 1938 } 1939 needed += elem_size; 1940 if (elem_count > 0 && elem_count != INT_MAX) 1941 elem_count--; 1942 } 1943 1944 *oldlenp = needed; 1945 if (oldp == NULL) 1946 *oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb); 1947 1948 mutex_exit(softnet_lock); 1949 1950 return (error); 1951 } 1952 1953 static int 1954 sysctl_tcp_congctl(SYSCTLFN_ARGS) 1955 { 1956 struct sysctlnode node; 1957 int error; 1958 char newname[TCPCC_MAXLEN]; 1959 1960 strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1); 1961 1962 node = *rnode; 1963 node.sysctl_data = newname; 1964 node.sysctl_size = sizeof(newname); 1965 1966 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1967 1968 if (error || 1969 newp == NULL || 1970 strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0) 1971 return error; 1972 1973 mutex_enter(softnet_lock); 1974 error = tcp_congctl_select(NULL, newname); 1975 mutex_exit(softnet_lock); 1976 1977 return error; 1978 } 1979 1980 static int 1981 sysctl_tcp_init_win(SYSCTLFN_ARGS) 1982 { 1983 int error; 1984 u_int iw; 1985 struct sysctlnode node; 1986 1987 iw = *(u_int *)rnode->sysctl_data; 1988 node = *rnode; 1989 node.sysctl_data = &iw; 1990 node.sysctl_size = sizeof(iw); 1991 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 1992 if (error || newp == NULL) 1993 return error; 1994 1995 if (iw >= __arraycount(tcp_init_win_max)) 1996 return EINVAL; 1997 *(u_int *)rnode->sysctl_data = iw; 1998 return 0; 1999 } 2000 2001 static int 2002 sysctl_tcp_keep(SYSCTLFN_ARGS) 2003 { 2004 int error; 2005 u_int tmp; 2006 struct sysctlnode node; 2007 2008 node = *rnode; 2009 tmp = *(u_int *)rnode->sysctl_data; 2010 node.sysctl_data = &tmp; 2011 2012 error = sysctl_lookup(SYSCTLFN_CALL(&node)); 2013 if (error || newp == NULL) 2014 return error; 2015 2016 mutex_enter(softnet_lock); 2017 2018 *(u_int *)rnode->sysctl_data = tmp; 2019 tcp_tcpcb_template(); /* update the template */ 2020 2021 mutex_exit(softnet_lock); 2022 return 0; 2023 } 2024 2025 static int 2026 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS) 2027 { 2028 2029 return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS)); 2030 } 2031 2032 /* 2033 * this (second stage) setup routine is a replacement for tcp_sysctl() 2034 * (which is currently used for ipv4 and ipv6) 2035 */ 2036 static void 2037 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname, 2038 const char *tcpname) 2039 { 2040 const struct sysctlnode *sack_node; 2041 const struct sysctlnode *abc_node; 2042 const struct sysctlnode *ecn_node; 2043 const struct sysctlnode *congctl_node; 2044 const struct sysctlnode *mslt_node; 2045 const struct sysctlnode *vtw_node; 2046 #ifdef TCP_DEBUG 2047 extern struct tcp_debug tcp_debug[TCP_NDEBUG]; 2048 extern int tcp_debx; 2049 #endif 2050 2051 sysctl_createv(clog, 0, NULL, NULL, 2052 CTLFLAG_PERMANENT, 2053 CTLTYPE_NODE, pfname, NULL, 2054 NULL, 0, NULL, 0, 2055 CTL_NET, pf, CTL_EOL); 2056 sysctl_createv(clog, 0, NULL, NULL, 2057 CTLFLAG_PERMANENT, 2058 CTLTYPE_NODE, tcpname, 2059 SYSCTL_DESCR("TCP related settings"), 2060 NULL, 0, NULL, 0, 2061 CTL_NET, pf, IPPROTO_TCP, CTL_EOL); 2062 2063 sysctl_createv(clog, 0, NULL, NULL, 2064 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2065 CTLTYPE_INT, "rfc1323", 2066 SYSCTL_DESCR("Enable RFC1323 TCP extensions"), 2067 sysctl_update_tcpcb_template, 0, &tcp_do_rfc1323, 0, 2068 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL); 2069 sysctl_createv(clog, 0, NULL, NULL, 2070 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2071 CTLTYPE_INT, "sendspace", 2072 SYSCTL_DESCR("Default TCP send buffer size"), 2073 NULL, 0, &tcp_sendspace, 0, 2074 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL); 2075 sysctl_createv(clog, 0, NULL, NULL, 2076 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2077 CTLTYPE_INT, "recvspace", 2078 SYSCTL_DESCR("Default TCP receive buffer size"), 2079 NULL, 0, &tcp_recvspace, 0, 2080 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL); 2081 sysctl_createv(clog, 0, NULL, NULL, 2082 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2083 CTLTYPE_INT, "mssdflt", 2084 SYSCTL_DESCR("Default maximum segment size"), 2085 sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0, 2086 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL); 2087 sysctl_createv(clog, 0, NULL, NULL, 2088 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2089 CTLTYPE_INT, "minmss", 2090 SYSCTL_DESCR("Lower limit for TCP maximum segment size"), 2091 NULL, 0, &tcp_minmss, 0, 2092 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2093 sysctl_createv(clog, 0, NULL, NULL, 2094 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2095 CTLTYPE_INT, "msl", 2096 SYSCTL_DESCR("Maximum Segment Life"), 2097 NULL, 0, &tcp_msl, 0, 2098 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL); 2099 sysctl_createv(clog, 0, NULL, NULL, 2100 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2101 CTLTYPE_INT, "syn_cache_limit", 2102 SYSCTL_DESCR("Maximum number of entries in the TCP " 2103 "compressed state engine"), 2104 NULL, 0, &tcp_syn_cache_limit, 0, 2105 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT, 2106 CTL_EOL); 2107 sysctl_createv(clog, 0, NULL, NULL, 2108 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2109 CTLTYPE_INT, "syn_bucket_limit", 2110 SYSCTL_DESCR("Maximum number of entries per hash " 2111 "bucket in the TCP compressed state " 2112 "engine"), 2113 NULL, 0, &tcp_syn_bucket_limit, 0, 2114 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT, 2115 CTL_EOL); 2116 #if 0 /* obsoleted */ 2117 sysctl_createv(clog, 0, NULL, NULL, 2118 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2119 CTLTYPE_INT, "syn_cache_interval", 2120 SYSCTL_DESCR("TCP compressed state engine's timer interval"), 2121 NULL, 0, &tcp_syn_cache_interval, 0, 2122 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER, 2123 CTL_EOL); 2124 #endif 2125 sysctl_createv(clog, 0, NULL, NULL, 2126 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2127 CTLTYPE_INT, "init_win", 2128 SYSCTL_DESCR("Initial TCP congestion window"), 2129 sysctl_tcp_init_win, 0, &tcp_init_win, 0, 2130 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL); 2131 sysctl_createv(clog, 0, NULL, NULL, 2132 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2133 CTLTYPE_INT, "mss_ifmtu", 2134 SYSCTL_DESCR("Use interface MTU for calculating MSS"), 2135 NULL, 0, &tcp_mss_ifmtu, 0, 2136 CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL); 2137 sysctl_createv(clog, 0, NULL, &sack_node, 2138 CTLFLAG_PERMANENT, 2139 CTLTYPE_NODE, "sack", 2140 SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"), 2141 NULL, 0, NULL, 0, 2142 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL); 2143 2144 /* Congctl subtree */ 2145 sysctl_createv(clog, 0, NULL, &congctl_node, 2146 CTLFLAG_PERMANENT, 2147 CTLTYPE_NODE, "congctl", 2148 SYSCTL_DESCR("TCP Congestion Control"), 2149 NULL, 0, NULL, 0, 2150 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2151 sysctl_createv(clog, 0, &congctl_node, NULL, 2152 CTLFLAG_PERMANENT, 2153 CTLTYPE_STRING, "available", 2154 SYSCTL_DESCR("Available Congestion Control Mechanisms"), 2155 NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL); 2156 sysctl_createv(clog, 0, &congctl_node, NULL, 2157 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2158 CTLTYPE_STRING, "selected", 2159 SYSCTL_DESCR("Selected Congestion Control Mechanism"), 2160 sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN, 2161 CTL_CREATE, CTL_EOL); 2162 2163 sysctl_createv(clog, 0, NULL, NULL, 2164 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2165 CTLTYPE_INT, "win_scale", 2166 SYSCTL_DESCR("Use RFC1323 window scale options"), 2167 sysctl_update_tcpcb_template, 0, &tcp_do_win_scale, 0, 2168 CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL); 2169 sysctl_createv(clog, 0, NULL, NULL, 2170 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2171 CTLTYPE_INT, "timestamps", 2172 SYSCTL_DESCR("Use RFC1323 time stamp options"), 2173 sysctl_update_tcpcb_template, 0, &tcp_do_timestamps, 0, 2174 CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL); 2175 sysctl_createv(clog, 0, NULL, NULL, 2176 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2177 CTLTYPE_INT, "compat_42", 2178 SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"), 2179 NULL, 0, &tcp_compat_42, 0, 2180 CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL); 2181 sysctl_createv(clog, 0, NULL, NULL, 2182 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2183 CTLTYPE_INT, "cwm", 2184 SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window " 2185 "Monitoring"), 2186 NULL, 0, &tcp_cwm, 0, 2187 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL); 2188 sysctl_createv(clog, 0, NULL, NULL, 2189 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2190 CTLTYPE_INT, "cwm_burstsize", 2191 SYSCTL_DESCR("Congestion Window Monitoring allowed " 2192 "burst count in packets"), 2193 NULL, 0, &tcp_cwm_burstsize, 0, 2194 CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE, 2195 CTL_EOL); 2196 sysctl_createv(clog, 0, NULL, NULL, 2197 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2198 CTLTYPE_INT, "ack_on_push", 2199 SYSCTL_DESCR("Immediately return ACK when PSH is " 2200 "received"), 2201 NULL, 0, &tcp_ack_on_push, 0, 2202 CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL); 2203 sysctl_createv(clog, 0, NULL, NULL, 2204 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2205 CTLTYPE_INT, "keepidle", 2206 SYSCTL_DESCR("Allowed connection idle ticks before a " 2207 "keepalive probe is sent"), 2208 sysctl_tcp_keep, 0, &tcp_keepidle, 0, 2209 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL); 2210 sysctl_createv(clog, 0, NULL, NULL, 2211 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2212 CTLTYPE_INT, "keepintvl", 2213 SYSCTL_DESCR("Ticks before next keepalive probe is sent"), 2214 sysctl_tcp_keep, 0, &tcp_keepintvl, 0, 2215 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL); 2216 sysctl_createv(clog, 0, NULL, NULL, 2217 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2218 CTLTYPE_INT, "keepcnt", 2219 SYSCTL_DESCR("Number of keepalive probes to send"), 2220 sysctl_tcp_keep, 0, &tcp_keepcnt, 0, 2221 CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL); 2222 sysctl_createv(clog, 0, NULL, NULL, 2223 CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE, 2224 CTLTYPE_INT, "slowhz", 2225 SYSCTL_DESCR("Keepalive ticks per second"), 2226 NULL, PR_SLOWHZ, NULL, 0, 2227 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL); 2228 sysctl_createv(clog, 0, NULL, NULL, 2229 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2230 CTLTYPE_INT, "log_refused", 2231 SYSCTL_DESCR("Log refused TCP connections"), 2232 NULL, 0, &tcp_log_refused, 0, 2233 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL); 2234 #if 0 /* obsoleted */ 2235 sysctl_createv(clog, 0, NULL, NULL, 2236 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2237 CTLTYPE_INT, "rstratelimit", NULL, 2238 NULL, 0, &tcp_rst_ratelim, 0, 2239 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL); 2240 #endif 2241 sysctl_createv(clog, 0, NULL, NULL, 2242 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2243 CTLTYPE_INT, "rstppslimit", 2244 SYSCTL_DESCR("Maximum number of RST packets to send " 2245 "per second"), 2246 NULL, 0, &tcp_rst_ppslim, 0, 2247 CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL); 2248 sysctl_createv(clog, 0, NULL, NULL, 2249 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2250 CTLTYPE_INT, "delack_ticks", 2251 SYSCTL_DESCR("Number of ticks to delay sending an ACK"), 2252 NULL, 0, &tcp_delack_ticks, 0, 2253 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL); 2254 sysctl_createv(clog, 0, NULL, NULL, 2255 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2256 CTLTYPE_INT, "init_win_local", 2257 SYSCTL_DESCR("Initial TCP window size (in segments)"), 2258 sysctl_tcp_init_win, 0, &tcp_init_win_local, 0, 2259 CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL, 2260 CTL_EOL); 2261 sysctl_createv(clog, 0, NULL, NULL, 2262 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2263 CTLTYPE_STRUCT, "ident", 2264 SYSCTL_DESCR("RFC1413 Identification Protocol lookups"), 2265 sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t), 2266 CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL); 2267 sysctl_createv(clog, 0, NULL, NULL, 2268 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2269 CTLTYPE_INT, "do_loopback_cksum", 2270 SYSCTL_DESCR("Perform TCP checksum on loopback"), 2271 NULL, 0, &tcp_do_loopback_cksum, 0, 2272 CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM, 2273 CTL_EOL); 2274 sysctl_createv(clog, 0, NULL, NULL, 2275 CTLFLAG_PERMANENT, 2276 CTLTYPE_STRUCT, "pcblist", 2277 SYSCTL_DESCR("TCP protocol control block list"), 2278 sysctl_inpcblist, 0, &tcbtable, 0, 2279 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2280 CTL_EOL); 2281 sysctl_createv(clog, 0, NULL, NULL, 2282 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2283 CTLTYPE_INT, "keepinit", 2284 SYSCTL_DESCR("Ticks before initial tcp connection times out"), 2285 sysctl_tcp_keep, 0, &tcp_keepinit, 0, 2286 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2287 2288 /* TCP socket buffers auto-sizing nodes */ 2289 sysctl_createv(clog, 0, NULL, NULL, 2290 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2291 CTLTYPE_INT, "recvbuf_auto", 2292 SYSCTL_DESCR("Enable automatic receive " 2293 "buffer sizing (experimental)"), 2294 NULL, 0, &tcp_do_autorcvbuf, 0, 2295 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2296 sysctl_createv(clog, 0, NULL, NULL, 2297 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2298 CTLTYPE_INT, "recvbuf_inc", 2299 SYSCTL_DESCR("Incrementor step size of " 2300 "automatic receive buffer"), 2301 NULL, 0, &tcp_autorcvbuf_inc, 0, 2302 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2303 sysctl_createv(clog, 0, NULL, NULL, 2304 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2305 CTLTYPE_INT, "recvbuf_max", 2306 SYSCTL_DESCR("Max size of automatic receive buffer"), 2307 NULL, 0, &tcp_autorcvbuf_max, 0, 2308 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2309 2310 sysctl_createv(clog, 0, NULL, NULL, 2311 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2312 CTLTYPE_INT, "sendbuf_auto", 2313 SYSCTL_DESCR("Enable automatic send " 2314 "buffer sizing (experimental)"), 2315 NULL, 0, &tcp_do_autosndbuf, 0, 2316 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2317 sysctl_createv(clog, 0, NULL, NULL, 2318 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2319 CTLTYPE_INT, "sendbuf_inc", 2320 SYSCTL_DESCR("Incrementor step size of " 2321 "automatic send buffer"), 2322 NULL, 0, &tcp_autosndbuf_inc, 0, 2323 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2324 sysctl_createv(clog, 0, NULL, NULL, 2325 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2326 CTLTYPE_INT, "sendbuf_max", 2327 SYSCTL_DESCR("Max size of automatic send buffer"), 2328 NULL, 0, &tcp_autosndbuf_max, 0, 2329 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2330 2331 /* ECN subtree */ 2332 sysctl_createv(clog, 0, NULL, &ecn_node, 2333 CTLFLAG_PERMANENT, 2334 CTLTYPE_NODE, "ecn", 2335 SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"), 2336 NULL, 0, NULL, 0, 2337 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2338 sysctl_createv(clog, 0, &ecn_node, NULL, 2339 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2340 CTLTYPE_INT, "enable", 2341 SYSCTL_DESCR("Enable TCP Explicit Congestion " 2342 "Notification"), 2343 NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL); 2344 sysctl_createv(clog, 0, &ecn_node, NULL, 2345 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2346 CTLTYPE_INT, "maxretries", 2347 SYSCTL_DESCR("Number of times to retry ECN setup " 2348 "before disabling ECN on the connection"), 2349 NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL); 2350 2351 /* SACK gets its own little subtree. */ 2352 sysctl_createv(clog, 0, NULL, &sack_node, 2353 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2354 CTLTYPE_INT, "enable", 2355 SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"), 2356 NULL, 0, &tcp_do_sack, 0, 2357 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2358 sysctl_createv(clog, 0, NULL, &sack_node, 2359 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2360 CTLTYPE_INT, "maxholes", 2361 SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"), 2362 NULL, 0, &tcp_sack_tp_maxholes, 0, 2363 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2364 sysctl_createv(clog, 0, NULL, &sack_node, 2365 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2366 CTLTYPE_INT, "globalmaxholes", 2367 SYSCTL_DESCR("Global maximum number of TCP SACK holes"), 2368 NULL, 0, &tcp_sack_globalmaxholes, 0, 2369 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2370 sysctl_createv(clog, 0, NULL, &sack_node, 2371 CTLFLAG_PERMANENT, 2372 CTLTYPE_INT, "globalholes", 2373 SYSCTL_DESCR("Global number of TCP SACK holes"), 2374 NULL, 0, &tcp_sack_globalholes, 0, 2375 CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL); 2376 2377 sysctl_createv(clog, 0, NULL, NULL, 2378 CTLFLAG_PERMANENT, 2379 CTLTYPE_STRUCT, "stats", 2380 SYSCTL_DESCR("TCP statistics"), 2381 sysctl_net_inet_tcp_stats, 0, NULL, 0, 2382 CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS, 2383 CTL_EOL); 2384 sysctl_createv(clog, 0, NULL, NULL, 2385 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2386 CTLTYPE_INT, "local_by_rtt", 2387 SYSCTL_DESCR("Use RTT estimator to decide which hosts " 2388 "are local"), 2389 NULL, 0, &tcp_rttlocal, 0, 2390 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2391 #ifdef TCP_DEBUG 2392 sysctl_createv(clog, 0, NULL, NULL, 2393 CTLFLAG_PERMANENT, 2394 CTLTYPE_STRUCT, "debug", 2395 SYSCTL_DESCR("TCP sockets debug information"), 2396 NULL, 0, &tcp_debug, sizeof(tcp_debug), 2397 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG, 2398 CTL_EOL); 2399 sysctl_createv(clog, 0, NULL, NULL, 2400 CTLFLAG_PERMANENT, 2401 CTLTYPE_INT, "debx", 2402 SYSCTL_DESCR("Number of TCP debug sockets messages"), 2403 NULL, 0, &tcp_debx, sizeof(tcp_debx), 2404 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX, 2405 CTL_EOL); 2406 #endif 2407 sysctl_createv(clog, 0, NULL, NULL, 2408 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2409 CTLTYPE_STRUCT, "drop", 2410 SYSCTL_DESCR("TCP drop connection"), 2411 sysctl_net_inet_tcp_drop, 0, NULL, 0, 2412 CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL); 2413 sysctl_createv(clog, 0, NULL, NULL, 2414 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2415 CTLTYPE_INT, "iss_hash", 2416 SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic " 2417 "hash computation"), 2418 NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948), 2419 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, 2420 CTL_EOL); 2421 2422 /* ABC subtree */ 2423 2424 sysctl_createv(clog, 0, NULL, &abc_node, 2425 CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc", 2426 SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"), 2427 NULL, 0, NULL, 0, 2428 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2429 sysctl_createv(clog, 0, &abc_node, NULL, 2430 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2431 CTLTYPE_INT, "enable", 2432 SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"), 2433 NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL); 2434 sysctl_createv(clog, 0, &abc_node, NULL, 2435 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2436 CTLTYPE_INT, "aggressive", 2437 SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"), 2438 NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL); 2439 2440 /* MSL tuning subtree */ 2441 2442 sysctl_createv(clog, 0, NULL, &mslt_node, 2443 CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt", 2444 SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"), 2445 NULL, 0, NULL, 0, 2446 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2447 sysctl_createv(clog, 0, &mslt_node, NULL, 2448 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2449 CTLTYPE_INT, "enable", 2450 SYSCTL_DESCR("Enable TIME_WAIT truncation"), 2451 NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL); 2452 sysctl_createv(clog, 0, &mslt_node, NULL, 2453 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2454 CTLTYPE_INT, "loopback", 2455 SYSCTL_DESCR("MSL value to use for loopback connections"), 2456 NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL); 2457 sysctl_createv(clog, 0, &mslt_node, NULL, 2458 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2459 CTLTYPE_INT, "local", 2460 SYSCTL_DESCR("MSL value to use for local connections"), 2461 NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL); 2462 sysctl_createv(clog, 0, &mslt_node, NULL, 2463 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2464 CTLTYPE_INT, "remote", 2465 SYSCTL_DESCR("MSL value to use for remote connections"), 2466 NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL); 2467 sysctl_createv(clog, 0, &mslt_node, NULL, 2468 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2469 CTLTYPE_INT, "remote_threshold", 2470 SYSCTL_DESCR("RTT estimate value to promote local to remote"), 2471 NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL); 2472 2473 /* vestigial TIME_WAIT tuning subtree */ 2474 2475 sysctl_createv(clog, 0, NULL, &vtw_node, 2476 CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw", 2477 SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"), 2478 NULL, 0, NULL, 0, 2479 CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL); 2480 sysctl_createv(clog, 0, &vtw_node, NULL, 2481 CTLFLAG_PERMANENT|CTLFLAG_READWRITE, 2482 CTLTYPE_INT, "enable", 2483 SYSCTL_DESCR("Enable Vestigial TIME_WAIT"), 2484 sysctl_tcp_vtw_enable, 0, 2485 (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable, 2486 0, CTL_CREATE, CTL_EOL); 2487 sysctl_createv(clog, 0, &vtw_node, NULL, 2488 CTLFLAG_PERMANENT|CTLFLAG_READONLY, 2489 CTLTYPE_INT, "entries", 2490 SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"), 2491 NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL); 2492 } 2493 2494 void 2495 tcp_usrreq_init(void) 2496 { 2497 2498 #ifdef INET 2499 sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp"); 2500 #endif 2501 #ifdef INET6 2502 sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6"); 2503 #endif 2504 } 2505 2506 PR_WRAP_USRREQS(tcp) 2507 #define tcp_attach tcp_attach_wrapper 2508 #define tcp_detach tcp_detach_wrapper 2509 #define tcp_accept tcp_accept_wrapper 2510 #define tcp_bind tcp_bind_wrapper 2511 #define tcp_listen tcp_listen_wrapper 2512 #define tcp_connect tcp_connect_wrapper 2513 #define tcp_connect2 tcp_connect2_wrapper 2514 #define tcp_disconnect tcp_disconnect_wrapper 2515 #define tcp_shutdown tcp_shutdown_wrapper 2516 #define tcp_abort tcp_abort_wrapper 2517 #define tcp_ioctl tcp_ioctl_wrapper 2518 #define tcp_stat tcp_stat_wrapper 2519 #define tcp_peeraddr tcp_peeraddr_wrapper 2520 #define tcp_sockaddr tcp_sockaddr_wrapper 2521 #define tcp_rcvd tcp_rcvd_wrapper 2522 #define tcp_recvoob tcp_recvoob_wrapper 2523 #define tcp_send tcp_send_wrapper 2524 #define tcp_sendoob tcp_sendoob_wrapper 2525 #define tcp_purgeif tcp_purgeif_wrapper 2526 #define tcp_usrreq tcp_usrreq_wrapper 2527 2528 const struct pr_usrreqs tcp_usrreqs = { 2529 .pr_attach = tcp_attach, 2530 .pr_detach = tcp_detach, 2531 .pr_accept = tcp_accept, 2532 .pr_bind = tcp_bind, 2533 .pr_listen = tcp_listen, 2534 .pr_connect = tcp_connect, 2535 .pr_connect2 = tcp_connect2, 2536 .pr_disconnect = tcp_disconnect, 2537 .pr_shutdown = tcp_shutdown, 2538 .pr_abort = tcp_abort, 2539 .pr_ioctl = tcp_ioctl, 2540 .pr_stat = tcp_stat, 2541 .pr_peeraddr = tcp_peeraddr, 2542 .pr_sockaddr = tcp_sockaddr, 2543 .pr_rcvd = tcp_rcvd, 2544 .pr_recvoob = tcp_recvoob, 2545 .pr_send = tcp_send, 2546 .pr_sendoob = tcp_sendoob, 2547 .pr_purgeif = tcp_purgeif, 2548 .pr_generic = tcp_usrreq, 2549 }; 2550
Indexes created Sat Sep 20 22:09:52 GMT 2025