tcp_subr.c revision 1.120
1 /* $NetBSD: tcp_subr.c,v 1.120 2001/11/04 13:42:27 matt 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, 2000, 2001 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 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69 /* 70 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. All advertising materials mentioning features or use of this software 82 * must display the following acknowledgement: 83 * This product includes software developed by the University of 84 * California, Berkeley and its contributors. 85 * 4. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 102 */ 103 104 #include "opt_inet.h" 105 #include "opt_ipsec.h" 106 #include "opt_tcp_compat_42.h" 107 #include "opt_inet_csum.h" 108 #include "rnd.h" 109 110 #include <sys/param.h> 111 #include <sys/proc.h> 112 #include <sys/systm.h> 113 #include <sys/malloc.h> 114 #include <sys/mbuf.h> 115 #include <sys/socket.h> 116 #include <sys/socketvar.h> 117 #include <sys/protosw.h> 118 #include <sys/errno.h> 119 #include <sys/kernel.h> 120 #include <sys/pool.h> 121 #if NRND > 0 122 #include <sys/md5.h> 123 #include <sys/rnd.h> 124 #endif 125 126 #include <net/route.h> 127 #include <net/if.h> 128 129 #include <netinet/in.h> 130 #include <netinet/in_systm.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/ip_var.h> 134 #include <netinet/ip_icmp.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/in6_var.h> 144 #include <netinet6/ip6protosw.h> 145 #include <netinet/icmp6.h> 146 #endif 147 148 #include <netinet/tcp.h> 149 #include <netinet/tcp_fsm.h> 150 #include <netinet/tcp_seq.h> 151 #include <netinet/tcp_timer.h> 152 #include <netinet/tcp_var.h> 153 #include <netinet/tcpip.h> 154 155 #ifdef IPSEC 156 #include <netinet6/ipsec.h> 157 #endif /*IPSEC*/ 158 159 #ifdef INET6 160 struct in6pcb tcb6; 161 #endif 162 163 /* patchable/settable parameters for tcp */ 164 int tcp_mssdflt = TCP_MSS; 165 int tcp_rttdflt = TCPTV_SRTTDFLT / PR_SLOWHZ; 166 int tcp_do_rfc1323 = 1; /* window scaling / timestamps (obsolete) */ 167 #if NRND > 0 168 int tcp_do_rfc1948 = 0; /* ISS by cryptographic hash */ 169 #endif 170 int tcp_do_sack = 1; /* selective acknowledgement */ 171 int tcp_do_win_scale = 1; /* RFC1323 window scaling */ 172 int tcp_do_timestamps = 1; /* RFC1323 timestamps */ 173 int tcp_do_newreno = 0; /* Use the New Reno algorithms */ 174 int tcp_ack_on_push = 0; /* set to enable immediate ACK-on-PUSH */ 175 int tcp_init_win = 1; 176 int tcp_mss_ifmtu = 0; 177 #ifdef TCP_COMPAT_42 178 int tcp_compat_42 = 1; 179 #else 180 int tcp_compat_42 = 0; 181 #endif 182 int tcp_rst_ppslim = 100; /* 100pps */ 183 184 /* tcb hash */ 185 #ifndef TCBHASHSIZE 186 #define TCBHASHSIZE 128 187 #endif 188 int tcbhashsize = TCBHASHSIZE; 189 190 /* syn hash parameters */ 191 #define TCP_SYN_HASH_SIZE 293 192 #define TCP_SYN_BUCKET_SIZE 35 193 int tcp_syn_cache_size = TCP_SYN_HASH_SIZE; 194 int tcp_syn_cache_limit = TCP_SYN_HASH_SIZE*TCP_SYN_BUCKET_SIZE; 195 int tcp_syn_bucket_limit = 3*TCP_SYN_BUCKET_SIZE; 196 struct syn_cache_head tcp_syn_cache[TCP_SYN_HASH_SIZE]; 197 198 int tcp_freeq __P((struct tcpcb *)); 199 200 #ifdef INET 201 void tcp_mtudisc_callback __P((struct in_addr)); 202 #endif 203 #ifdef INET6 204 void tcp6_mtudisc_callback __P((struct in6_addr *)); 205 #endif 206 207 void tcp_mtudisc __P((struct inpcb *, int)); 208 #ifdef INET6 209 void tcp6_mtudisc __P((struct in6pcb *, int)); 210 #endif 211 212 struct pool tcpcb_pool; 213 214 #ifdef TCP_CSUM_COUNTERS 215 #include <sys/device.h> 216 217 struct evcnt tcp_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 218 NULL, "tcp", "hwcsum bad"); 219 struct evcnt tcp_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 220 NULL, "tcp", "hwcsum ok"); 221 struct evcnt tcp_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 222 NULL, "tcp", "hwcsum data"); 223 struct evcnt tcp_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 224 NULL, "tcp", "swcsum"); 225 #endif /* TCP_CSUM_COUNTERS */ 226 227 /* 228 * Tcp initialization 229 */ 230 void 231 tcp_init() 232 { 233 int hlen; 234 235 pool_init(&tcpcb_pool, sizeof(struct tcpcb), 0, 0, 0, "tcpcbpl", 236 0, NULL, NULL, M_PCB); 237 in_pcbinit(&tcbtable, tcbhashsize, tcbhashsize); 238 #ifdef INET6 239 tcb6.in6p_next = tcb6.in6p_prev = &tcb6; 240 #endif 241 242 hlen = sizeof(struct ip) + sizeof(struct tcphdr); 243 #ifdef INET6 244 if (sizeof(struct ip) < sizeof(struct ip6_hdr)) 245 hlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 246 #endif 247 if (max_protohdr < hlen) 248 max_protohdr = hlen; 249 if (max_linkhdr + hlen > MHLEN) 250 panic("tcp_init"); 251 252 #ifdef INET 253 icmp_mtudisc_callback_register(tcp_mtudisc_callback); 254 #endif 255 #ifdef INET6 256 icmp6_mtudisc_callback_register(tcp6_mtudisc_callback); 257 #endif 258 259 /* Initialize timer state. */ 260 tcp_timer_init(); 261 262 /* Initialize the compressed state engine. */ 263 syn_cache_init(); 264 265 #ifdef TCP_CSUM_COUNTERS 266 evcnt_attach_static(&tcp_hwcsum_bad); 267 evcnt_attach_static(&tcp_hwcsum_ok); 268 evcnt_attach_static(&tcp_hwcsum_data); 269 evcnt_attach_static(&tcp_swcsum); 270 #endif /* TCP_CSUM_COUNTERS */ 271 } 272 273 /* 274 * Create template to be used to send tcp packets on a connection. 275 * Call after host entry created, allocates an mbuf and fills 276 * in a skeletal tcp/ip header, minimizing the amount of work 277 * necessary when the connection is used. 278 */ 279 struct mbuf * 280 tcp_template(tp) 281 struct tcpcb *tp; 282 { 283 struct inpcb *inp = tp->t_inpcb; 284 #ifdef INET6 285 struct in6pcb *in6p = tp->t_in6pcb; 286 #endif 287 struct tcphdr *n; 288 struct mbuf *m; 289 int hlen; 290 291 switch (tp->t_family) { 292 case AF_INET: 293 hlen = sizeof(struct ip); 294 if (inp) 295 break; 296 #ifdef INET6 297 if (in6p) { 298 /* mapped addr case */ 299 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 300 && IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 301 break; 302 } 303 #endif 304 return NULL; /*EINVAL*/ 305 #ifdef INET6 306 case AF_INET6: 307 hlen = sizeof(struct ip6_hdr); 308 if (in6p) { 309 /* more sainty check? */ 310 break; 311 } 312 return NULL; /*EINVAL*/ 313 #endif 314 default: 315 hlen = 0; /*pacify gcc*/ 316 return NULL; /*EAFNOSUPPORT*/ 317 } 318 #ifdef DIAGNOSTIC 319 if (hlen + sizeof(struct tcphdr) > MCLBYTES) 320 panic("mclbytes too small for t_template"); 321 #endif 322 m = tp->t_template; 323 if (m && m->m_len == hlen + sizeof(struct tcphdr)) 324 ; 325 else { 326 if (m) 327 m_freem(m); 328 m = tp->t_template = NULL; 329 MGETHDR(m, M_DONTWAIT, MT_HEADER); 330 if (m && hlen + sizeof(struct tcphdr) > MHLEN) { 331 MCLGET(m, M_DONTWAIT); 332 if ((m->m_flags & M_EXT) == 0) { 333 m_free(m); 334 m = NULL; 335 } 336 } 337 if (m == NULL) 338 return NULL; 339 m->m_pkthdr.len = m->m_len = hlen + sizeof(struct tcphdr); 340 } 341 342 bzero(mtod(m, caddr_t), m->m_len); 343 344 n = (struct tcphdr *)(mtod(m, caddr_t) + hlen); 345 346 switch (tp->t_family) { 347 case AF_INET: 348 { 349 struct ipovly *ipov; 350 mtod(m, struct ip *)->ip_v = 4; 351 ipov = mtod(m, struct ipovly *); 352 ipov->ih_pr = IPPROTO_TCP; 353 ipov->ih_len = htons(sizeof(struct tcphdr)); 354 if (inp) { 355 ipov->ih_src = inp->inp_laddr; 356 ipov->ih_dst = inp->inp_faddr; 357 } 358 #ifdef INET6 359 else if (in6p) { 360 /* mapped addr case */ 361 bcopy(&in6p->in6p_laddr.s6_addr32[3], &ipov->ih_src, 362 sizeof(ipov->ih_src)); 363 bcopy(&in6p->in6p_faddr.s6_addr32[3], &ipov->ih_dst, 364 sizeof(ipov->ih_dst)); 365 } 366 #endif 367 /* 368 * Compute the pseudo-header portion of the checksum 369 * now. We incrementally add in the TCP option and 370 * payload lengths later, and then compute the TCP 371 * checksum right before the packet is sent off onto 372 * the wire. 373 */ 374 n->th_sum = in_cksum_phdr(ipov->ih_src.s_addr, 375 ipov->ih_dst.s_addr, 376 htons(sizeof(struct tcphdr) + IPPROTO_TCP)); 377 break; 378 } 379 #ifdef INET6 380 case AF_INET6: 381 { 382 struct ip6_hdr *ip6; 383 mtod(m, struct ip *)->ip_v = 6; 384 ip6 = mtod(m, struct ip6_hdr *); 385 ip6->ip6_nxt = IPPROTO_TCP; 386 ip6->ip6_plen = htons(sizeof(struct tcphdr)); 387 ip6->ip6_src = in6p->in6p_laddr; 388 ip6->ip6_dst = in6p->in6p_faddr; 389 ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK; 390 if (ip6_auto_flowlabel) { 391 ip6->ip6_flow &= ~IPV6_FLOWLABEL_MASK; 392 ip6->ip6_flow |= 393 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 394 } 395 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 396 ip6->ip6_vfc |= IPV6_VERSION; 397 398 /* 399 * Compute the pseudo-header portion of the checksum 400 * now. We incrementally add in the TCP option and 401 * payload lengths later, and then compute the TCP 402 * checksum right before the packet is sent off onto 403 * the wire. 404 */ 405 n->th_sum = in6_cksum_phdr(&in6p->in6p_laddr, 406 &in6p->in6p_faddr, htonl(sizeof(struct tcphdr)), 407 htonl(IPPROTO_TCP)); 408 break; 409 } 410 #endif 411 } 412 if (inp) { 413 n->th_sport = inp->inp_lport; 414 n->th_dport = inp->inp_fport; 415 } 416 #ifdef INET6 417 else if (in6p) { 418 n->th_sport = in6p->in6p_lport; 419 n->th_dport = in6p->in6p_fport; 420 } 421 #endif 422 n->th_seq = 0; 423 n->th_ack = 0; 424 n->th_x2 = 0; 425 n->th_off = 5; 426 n->th_flags = 0; 427 n->th_win = 0; 428 n->th_urp = 0; 429 return (m); 430 } 431 432 /* 433 * Send a single message to the TCP at address specified by 434 * the given TCP/IP header. If m == 0, then we make a copy 435 * of the tcpiphdr at ti and send directly to the addressed host. 436 * This is used to force keep alive messages out using the TCP 437 * template for a connection tp->t_template. If flags are given 438 * then we send a message back to the TCP which originated the 439 * segment ti, and discard the mbuf containing it and any other 440 * attached mbufs. 441 * 442 * In any case the ack and sequence number of the transmitted 443 * segment are as specified by the parameters. 444 */ 445 int 446 tcp_respond(tp, template, m, th0, ack, seq, flags) 447 struct tcpcb *tp; 448 struct mbuf *template; 449 struct mbuf *m; 450 struct tcphdr *th0; 451 tcp_seq ack, seq; 452 int flags; 453 { 454 struct route *ro; 455 int error, tlen, win = 0; 456 int hlen; 457 struct ip *ip; 458 #ifdef INET6 459 struct ip6_hdr *ip6; 460 #endif 461 int family; /* family on packet, not inpcb/in6pcb! */ 462 struct tcphdr *th; 463 464 if (tp != NULL && (flags & TH_RST) == 0) { 465 #ifdef DIAGNOSTIC 466 if (tp->t_inpcb && tp->t_in6pcb) 467 panic("tcp_respond: both t_inpcb and t_in6pcb are set"); 468 #endif 469 #ifdef INET 470 if (tp->t_inpcb) 471 win = sbspace(&tp->t_inpcb->inp_socket->so_rcv); 472 #endif 473 #ifdef INET6 474 if (tp->t_in6pcb) 475 win = sbspace(&tp->t_in6pcb->in6p_socket->so_rcv); 476 #endif 477 } 478 479 ip = NULL; 480 #ifdef INET6 481 ip6 = NULL; 482 #endif 483 if (m == 0) { 484 if (!template) 485 return EINVAL; 486 487 /* get family information from template */ 488 switch (mtod(template, struct ip *)->ip_v) { 489 case 4: 490 family = AF_INET; 491 hlen = sizeof(struct ip); 492 break; 493 #ifdef INET6 494 case 6: 495 family = AF_INET6; 496 hlen = sizeof(struct ip6_hdr); 497 break; 498 #endif 499 default: 500 return EAFNOSUPPORT; 501 } 502 503 MGETHDR(m, M_DONTWAIT, MT_HEADER); 504 if (m) { 505 MCLGET(m, M_DONTWAIT); 506 if ((m->m_flags & M_EXT) == 0) { 507 m_free(m); 508 m = NULL; 509 } 510 } 511 if (m == NULL) 512 return (ENOBUFS); 513 514 if (tcp_compat_42) 515 tlen = 1; 516 else 517 tlen = 0; 518 519 m->m_data += max_linkhdr; 520 bcopy(mtod(template, caddr_t), mtod(m, caddr_t), 521 template->m_len); 522 switch (family) { 523 case AF_INET: 524 ip = mtod(m, struct ip *); 525 th = (struct tcphdr *)(ip + 1); 526 break; 527 #ifdef INET6 528 case AF_INET6: 529 ip6 = mtod(m, struct ip6_hdr *); 530 th = (struct tcphdr *)(ip6 + 1); 531 break; 532 #endif 533 #if 0 534 default: 535 /* noone will visit here */ 536 m_freem(m); 537 return EAFNOSUPPORT; 538 #endif 539 } 540 flags = TH_ACK; 541 } else { 542 543 if ((m->m_flags & M_PKTHDR) == 0) { 544 #if 0 545 printf("non PKTHDR to tcp_respond\n"); 546 #endif 547 m_freem(m); 548 return EINVAL; 549 } 550 #ifdef DIAGNOSTIC 551 if (!th0) 552 panic("th0 == NULL in tcp_respond"); 553 #endif 554 555 /* get family information from m */ 556 switch (mtod(m, struct ip *)->ip_v) { 557 case 4: 558 family = AF_INET; 559 hlen = sizeof(struct ip); 560 ip = mtod(m, struct ip *); 561 break; 562 #ifdef INET6 563 case 6: 564 family = AF_INET6; 565 hlen = sizeof(struct ip6_hdr); 566 ip6 = mtod(m, struct ip6_hdr *); 567 break; 568 #endif 569 default: 570 m_freem(m); 571 return EAFNOSUPPORT; 572 } 573 if ((flags & TH_SYN) == 0 || sizeof(*th0) > (th0->th_off << 2)) 574 tlen = sizeof(*th0); 575 else 576 tlen = th0->th_off << 2; 577 578 if (m->m_len > hlen + tlen && (m->m_flags & M_EXT) == 0 && 579 mtod(m, caddr_t) + hlen == (caddr_t)th0) { 580 m->m_len = hlen + tlen; 581 m_freem(m->m_next); 582 m->m_next = NULL; 583 } else { 584 struct mbuf *n; 585 586 #ifdef DIAGNOSTIC 587 if (max_linkhdr + hlen + tlen > MCLBYTES) { 588 m_freem(m); 589 return EMSGSIZE; 590 } 591 #endif 592 MGETHDR(n, M_DONTWAIT, MT_HEADER); 593 if (n && max_linkhdr + hlen + tlen > MHLEN) { 594 MCLGET(n, M_DONTWAIT); 595 if ((n->m_flags & M_EXT) == 0) { 596 m_freem(n); 597 n = NULL; 598 } 599 } 600 if (!n) { 601 m_freem(m); 602 return ENOBUFS; 603 } 604 605 n->m_data += max_linkhdr; 606 n->m_len = hlen + tlen; 607 m_copyback(n, 0, hlen, mtod(m, caddr_t)); 608 m_copyback(n, hlen, tlen, (caddr_t)th0); 609 610 m_freem(m); 611 m = n; 612 n = NULL; 613 } 614 615 #define xchg(a,b,type) { type t; t=a; a=b; b=t; } 616 switch (family) { 617 case AF_INET: 618 ip = mtod(m, struct ip *); 619 th = (struct tcphdr *)(ip + 1); 620 ip->ip_p = IPPROTO_TCP; 621 xchg(ip->ip_dst, ip->ip_src, struct in_addr); 622 ip->ip_p = IPPROTO_TCP; 623 break; 624 #ifdef INET6 625 case AF_INET6: 626 ip6 = mtod(m, struct ip6_hdr *); 627 th = (struct tcphdr *)(ip6 + 1); 628 ip6->ip6_nxt = IPPROTO_TCP; 629 xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); 630 ip6->ip6_nxt = IPPROTO_TCP; 631 break; 632 #endif 633 #if 0 634 default: 635 /* noone will visit here */ 636 m_freem(m); 637 return EAFNOSUPPORT; 638 #endif 639 } 640 xchg(th->th_dport, th->th_sport, u_int16_t); 641 #undef xchg 642 tlen = 0; /*be friendly with the following code*/ 643 } 644 th->th_seq = htonl(seq); 645 th->th_ack = htonl(ack); 646 th->th_x2 = 0; 647 if ((flags & TH_SYN) == 0) { 648 if (tp) 649 win >>= tp->rcv_scale; 650 if (win > TCP_MAXWIN) 651 win = TCP_MAXWIN; 652 th->th_win = htons((u_int16_t)win); 653 th->th_off = sizeof (struct tcphdr) >> 2; 654 tlen += sizeof(*th); 655 } else 656 tlen += th->th_off << 2; 657 m->m_len = hlen + tlen; 658 m->m_pkthdr.len = hlen + tlen; 659 m->m_pkthdr.rcvif = (struct ifnet *) 0; 660 th->th_flags = flags; 661 th->th_urp = 0; 662 663 switch (family) { 664 #ifdef INET 665 case AF_INET: 666 { 667 struct ipovly *ipov = (struct ipovly *)ip; 668 bzero(ipov->ih_x1, sizeof ipov->ih_x1); 669 ipov->ih_len = htons((u_int16_t)tlen); 670 671 th->th_sum = 0; 672 th->th_sum = in_cksum(m, hlen + tlen); 673 ip->ip_len = hlen + tlen; /*will be flipped on output*/ 674 ip->ip_ttl = ip_defttl; 675 break; 676 } 677 #endif 678 #ifdef INET6 679 case AF_INET6: 680 { 681 th->th_sum = 0; 682 th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), 683 tlen); 684 ip6->ip6_plen = ntohs(tlen); 685 if (tp && tp->t_in6pcb) { 686 struct ifnet *oifp; 687 ro = (struct route *)&tp->t_in6pcb->in6p_route; 688 oifp = ro->ro_rt ? ro->ro_rt->rt_ifp : NULL; 689 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, oifp); 690 } else 691 ip6->ip6_hlim = ip6_defhlim; 692 ip6->ip6_flow &= ~IPV6_FLOWINFO_MASK; 693 if (ip6_auto_flowlabel) { 694 ip6->ip6_flow |= 695 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 696 } 697 break; 698 } 699 #endif 700 } 701 702 #ifdef IPSEC 703 (void)ipsec_setsocket(m, NULL); 704 #endif /*IPSEC*/ 705 706 if (tp != NULL && tp->t_inpcb != NULL) { 707 ro = &tp->t_inpcb->inp_route; 708 #ifdef IPSEC 709 if (ipsec_setsocket(m, tp->t_inpcb->inp_socket) != 0) { 710 m_freem(m); 711 return ENOBUFS; 712 } 713 #endif 714 #ifdef DIAGNOSTIC 715 if (family != AF_INET) 716 panic("tcp_respond: address family mismatch"); 717 if (!in_hosteq(ip->ip_dst, tp->t_inpcb->inp_faddr)) { 718 panic("tcp_respond: ip_dst %x != inp_faddr %x", 719 ntohl(ip->ip_dst.s_addr), 720 ntohl(tp->t_inpcb->inp_faddr.s_addr)); 721 } 722 #endif 723 } 724 #ifdef INET6 725 else if (tp != NULL && tp->t_in6pcb != NULL) { 726 ro = (struct route *)&tp->t_in6pcb->in6p_route; 727 #ifdef IPSEC 728 if (ipsec_setsocket(m, tp->t_in6pcb->in6p_socket) != 0) { 729 m_freem(m); 730 return ENOBUFS; 731 } 732 #endif 733 #ifdef DIAGNOSTIC 734 if (family == AF_INET) { 735 if (!IN6_IS_ADDR_V4MAPPED(&tp->t_in6pcb->in6p_faddr)) 736 panic("tcp_respond: not mapped addr"); 737 if (bcmp(&ip->ip_dst, 738 &tp->t_in6pcb->in6p_faddr.s6_addr32[3], 739 sizeof(ip->ip_dst)) != 0) { 740 panic("tcp_respond: ip_dst != in6p_faddr"); 741 } 742 } else if (family == AF_INET6) { 743 if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, &tp->t_in6pcb->in6p_faddr)) 744 panic("tcp_respond: ip6_dst != in6p_faddr"); 745 } else 746 panic("tcp_respond: address family mismatch"); 747 #endif 748 } 749 #endif 750 else 751 ro = NULL; 752 753 switch (family) { 754 #ifdef INET 755 case AF_INET: 756 error = ip_output(m, NULL, ro, 757 (ip_mtudisc ? IP_MTUDISC : 0), 758 NULL); 759 break; 760 #endif 761 #ifdef INET6 762 case AF_INET6: 763 error = ip6_output(m, NULL, (struct route_in6 *)ro, 0, NULL, 764 NULL); 765 break; 766 #endif 767 default: 768 error = EAFNOSUPPORT; 769 break; 770 } 771 772 return (error); 773 } 774 775 /* 776 * Create a new TCP control block, making an 777 * empty reassembly queue and hooking it to the argument 778 * protocol control block. 779 */ 780 struct tcpcb * 781 tcp_newtcpcb(family, aux) 782 int family; /* selects inpcb, or in6pcb */ 783 void *aux; 784 { 785 struct tcpcb *tp; 786 int i; 787 788 switch (family) { 789 case PF_INET: 790 break; 791 #ifdef INET6 792 case PF_INET6: 793 break; 794 #endif 795 default: 796 return NULL; 797 } 798 799 tp = pool_get(&tcpcb_pool, PR_NOWAIT); 800 if (tp == NULL) 801 return (NULL); 802 bzero((caddr_t)tp, sizeof(struct tcpcb)); 803 LIST_INIT(&tp->segq); 804 LIST_INIT(&tp->timeq); 805 tp->t_family = family; /* may be overridden later on */ 806 tp->t_peermss = tcp_mssdflt; 807 tp->t_ourmss = tcp_mssdflt; 808 tp->t_segsz = tcp_mssdflt; 809 LIST_INIT(&tp->t_sc); 810 811 callout_init(&tp->t_delack_ch); 812 for (i = 0; i < TCPT_NTIMERS; i++) 813 TCP_TIMER_INIT(tp, i); 814 815 tp->t_flags = 0; 816 if (tcp_do_rfc1323 && tcp_do_win_scale) 817 tp->t_flags |= TF_REQ_SCALE; 818 if (tcp_do_rfc1323 && tcp_do_timestamps) 819 tp->t_flags |= TF_REQ_TSTMP; 820 if (tcp_do_sack == 2) 821 tp->t_flags |= TF_WILL_SACK; 822 else if (tcp_do_sack == 1) 823 tp->t_flags |= TF_WILL_SACK|TF_IGNR_RXSACK; 824 tp->t_flags |= TF_CANT_TXSACK; 825 switch (family) { 826 case PF_INET: 827 tp->t_inpcb = (struct inpcb *)aux; 828 break; 829 #ifdef INET6 830 case PF_INET6: 831 tp->t_in6pcb = (struct in6pcb *)aux; 832 break; 833 #endif 834 } 835 /* 836 * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no 837 * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives 838 * reasonable initial retransmit time. 839 */ 840 tp->t_srtt = TCPTV_SRTTBASE; 841 tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << (TCP_RTTVAR_SHIFT + 2 - 1); 842 tp->t_rttmin = TCPTV_MIN; 843 TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), 844 TCPTV_MIN, TCPTV_REXMTMAX); 845 tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; 846 tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; 847 if (family == AF_INET) { 848 struct inpcb *inp = (struct inpcb *)aux; 849 inp->inp_ip.ip_ttl = ip_defttl; 850 inp->inp_ppcb = (caddr_t)tp; 851 } 852 #ifdef INET6 853 else if (family == AF_INET6) { 854 struct in6pcb *in6p = (struct in6pcb *)aux; 855 in6p->in6p_ip6.ip6_hlim = in6_selecthlim(in6p, 856 in6p->in6p_route.ro_rt ? in6p->in6p_route.ro_rt->rt_ifp 857 : NULL); 858 in6p->in6p_ppcb = (caddr_t)tp; 859 } 860 #endif 861 862 /* 863 * Initialize our timebase. When we send timestamps, we take 864 * the delta from tcp_now -- this means each connection always 865 * gets a timebase of 0, which makes it, among other things, 866 * more difficult to determine how long a system has been up, 867 * and thus how many TCP sequence increments have occurred. 868 */ 869 tp->ts_timebase = tcp_now; 870 871 return (tp); 872 } 873 874 /* 875 * Drop a TCP connection, reporting 876 * the specified error. If connection is synchronized, 877 * then send a RST to peer. 878 */ 879 struct tcpcb * 880 tcp_drop(tp, errno) 881 struct tcpcb *tp; 882 int errno; 883 { 884 struct socket *so = NULL; 885 886 #ifdef DIAGNOSTIC 887 if (tp->t_inpcb && tp->t_in6pcb) 888 panic("tcp_drop: both t_inpcb and t_in6pcb are set"); 889 #endif 890 #ifdef INET 891 if (tp->t_inpcb) 892 so = tp->t_inpcb->inp_socket; 893 #endif 894 #ifdef INET6 895 if (tp->t_in6pcb) 896 so = tp->t_in6pcb->in6p_socket; 897 #endif 898 if (!so) 899 return NULL; 900 901 if (TCPS_HAVERCVDSYN(tp->t_state)) { 902 tp->t_state = TCPS_CLOSED; 903 (void) tcp_output(tp); 904 tcpstat.tcps_drops++; 905 } else 906 tcpstat.tcps_conndrops++; 907 if (errno == ETIMEDOUT && tp->t_softerror) 908 errno = tp->t_softerror; 909 so->so_error = errno; 910 return (tcp_close(tp)); 911 } 912 913 /* 914 * Close a TCP control block: 915 * discard all space held by the tcp 916 * discard internet protocol block 917 * wake up any sleepers 918 */ 919 struct tcpcb * 920 tcp_close(tp) 921 struct tcpcb *tp; 922 { 923 struct inpcb *inp; 924 #ifdef INET6 925 struct in6pcb *in6p; 926 #endif 927 struct socket *so; 928 #ifdef RTV_RTT 929 struct rtentry *rt; 930 #endif 931 struct route *ro; 932 933 inp = tp->t_inpcb; 934 #ifdef INET6 935 in6p = tp->t_in6pcb; 936 #endif 937 so = NULL; 938 ro = NULL; 939 if (inp) { 940 so = inp->inp_socket; 941 ro = &inp->inp_route; 942 } 943 #ifdef INET6 944 else if (in6p) { 945 so = in6p->in6p_socket; 946 ro = (struct route *)&in6p->in6p_route; 947 } 948 #endif 949 950 #ifdef RTV_RTT 951 /* 952 * If we sent enough data to get some meaningful characteristics, 953 * save them in the routing entry. 'Enough' is arbitrarily 954 * defined as the sendpipesize (default 4K) * 16. This would 955 * give us 16 rtt samples assuming we only get one sample per 956 * window (the usual case on a long haul net). 16 samples is 957 * enough for the srtt filter to converge to within 5% of the correct 958 * value; fewer samples and we could save a very bogus rtt. 959 * 960 * Don't update the default route's characteristics and don't 961 * update anything that the user "locked". 962 */ 963 if (SEQ_LT(tp->iss + so->so_snd.sb_hiwat * 16, tp->snd_max) && 964 ro && (rt = ro->ro_rt) && 965 !in_nullhost(satosin(rt_key(rt))->sin_addr)) { 966 u_long i = 0; 967 968 if ((rt->rt_rmx.rmx_locks & RTV_RTT) == 0) { 969 i = tp->t_srtt * 970 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 971 if (rt->rt_rmx.rmx_rtt && i) 972 /* 973 * filter this update to half the old & half 974 * the new values, converting scale. 975 * See route.h and tcp_var.h for a 976 * description of the scaling constants. 977 */ 978 rt->rt_rmx.rmx_rtt = 979 (rt->rt_rmx.rmx_rtt + i) / 2; 980 else 981 rt->rt_rmx.rmx_rtt = i; 982 } 983 if ((rt->rt_rmx.rmx_locks & RTV_RTTVAR) == 0) { 984 i = tp->t_rttvar * 985 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTTVAR_SHIFT + 2)); 986 if (rt->rt_rmx.rmx_rttvar && i) 987 rt->rt_rmx.rmx_rttvar = 988 (rt->rt_rmx.rmx_rttvar + i) / 2; 989 else 990 rt->rt_rmx.rmx_rttvar = i; 991 } 992 /* 993 * update the pipelimit (ssthresh) if it has been updated 994 * already or if a pipesize was specified & the threshhold 995 * got below half the pipesize. I.e., wait for bad news 996 * before we start updating, then update on both good 997 * and bad news. 998 */ 999 if (((rt->rt_rmx.rmx_locks & RTV_SSTHRESH) == 0 && 1000 (i = tp->snd_ssthresh) && rt->rt_rmx.rmx_ssthresh) || 1001 i < (rt->rt_rmx.rmx_sendpipe / 2)) { 1002 /* 1003 * convert the limit from user data bytes to 1004 * packets then to packet data bytes. 1005 */ 1006 i = (i + tp->t_segsz / 2) / tp->t_segsz; 1007 if (i < 2) 1008 i = 2; 1009 i *= (u_long)(tp->t_segsz + sizeof (struct tcpiphdr)); 1010 if (rt->rt_rmx.rmx_ssthresh) 1011 rt->rt_rmx.rmx_ssthresh = 1012 (rt->rt_rmx.rmx_ssthresh + i) / 2; 1013 else 1014 rt->rt_rmx.rmx_ssthresh = i; 1015 } 1016 } 1017 #endif /* RTV_RTT */ 1018 /* free the reassembly queue, if any */ 1019 TCP_REASS_LOCK(tp); 1020 (void) tcp_freeq(tp); 1021 TCP_REASS_UNLOCK(tp); 1022 1023 tcp_canceltimers(tp); 1024 TCP_CLEAR_DELACK(tp); 1025 syn_cache_cleanup(tp); 1026 1027 if (tp->t_template) { 1028 m_free(tp->t_template); 1029 tp->t_template = NULL; 1030 } 1031 pool_put(&tcpcb_pool, tp); 1032 if (inp) { 1033 inp->inp_ppcb = 0; 1034 soisdisconnected(so); 1035 in_pcbdetach(inp); 1036 } 1037 #ifdef INET6 1038 else if (in6p) { 1039 in6p->in6p_ppcb = 0; 1040 soisdisconnected(so); 1041 in6_pcbdetach(in6p); 1042 } 1043 #endif 1044 tcpstat.tcps_closed++; 1045 return ((struct tcpcb *)0); 1046 } 1047 1048 int 1049 tcp_freeq(tp) 1050 struct tcpcb *tp; 1051 { 1052 struct ipqent *qe; 1053 int rv = 0; 1054 #ifdef TCPREASS_DEBUG 1055 int i = 0; 1056 #endif 1057 1058 TCP_REASS_LOCK_CHECK(tp); 1059 1060 while ((qe = tp->segq.lh_first) != NULL) { 1061 #ifdef TCPREASS_DEBUG 1062 printf("tcp_freeq[%p,%d]: %u:%u(%u) 0x%02x\n", 1063 tp, i++, qe->ipqe_seq, qe->ipqe_seq + qe->ipqe_len, 1064 qe->ipqe_len, qe->ipqe_flags & (TH_SYN|TH_FIN|TH_RST)); 1065 #endif 1066 LIST_REMOVE(qe, ipqe_q); 1067 LIST_REMOVE(qe, ipqe_timeq); 1068 m_freem(qe->ipqe_m); 1069 pool_put(&ipqent_pool, qe); 1070 rv = 1; 1071 } 1072 return (rv); 1073 } 1074 1075 /* 1076 * Protocol drain routine. Called when memory is in short supply. 1077 */ 1078 void 1079 tcp_drain() 1080 { 1081 struct inpcb *inp; 1082 struct tcpcb *tp; 1083 1084 /* 1085 * Free the sequence queue of all TCP connections. 1086 */ 1087 inp = tcbtable.inpt_queue.cqh_first; 1088 if (inp) /* XXX */ 1089 for (; inp != (struct inpcb *)&tcbtable.inpt_queue; 1090 inp = inp->inp_queue.cqe_next) { 1091 if ((tp = intotcpcb(inp)) != NULL) { 1092 /* 1093 * We may be called from a device's interrupt 1094 * context. If the tcpcb is already busy, 1095 * just bail out now. 1096 */ 1097 if (tcp_reass_lock_try(tp) == 0) 1098 continue; 1099 if (tcp_freeq(tp)) 1100 tcpstat.tcps_connsdrained++; 1101 TCP_REASS_UNLOCK(tp); 1102 } 1103 } 1104 } 1105 1106 /* 1107 * Notify a tcp user of an asynchronous error; 1108 * store error as soft error, but wake up user 1109 * (for now, won't do anything until can select for soft error). 1110 */ 1111 void 1112 tcp_notify(inp, error) 1113 struct inpcb *inp; 1114 int error; 1115 { 1116 struct tcpcb *tp = (struct tcpcb *)inp->inp_ppcb; 1117 struct socket *so = inp->inp_socket; 1118 1119 /* 1120 * Ignore some errors if we are hooked up. 1121 * If connection hasn't completed, has retransmitted several times, 1122 * and receives a second error, give up now. This is better 1123 * than waiting a long time to establish a connection that 1124 * can never complete. 1125 */ 1126 if (tp->t_state == TCPS_ESTABLISHED && 1127 (error == EHOSTUNREACH || error == ENETUNREACH || 1128 error == EHOSTDOWN)) { 1129 return; 1130 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1131 tp->t_rxtshift > 3 && tp->t_softerror) 1132 so->so_error = error; 1133 else 1134 tp->t_softerror = error; 1135 wakeup((caddr_t) &so->so_timeo); 1136 sorwakeup(so); 1137 sowwakeup(so); 1138 } 1139 1140 #ifdef INET6 1141 void 1142 tcp6_notify(in6p, error) 1143 struct in6pcb *in6p; 1144 int error; 1145 { 1146 struct tcpcb *tp = (struct tcpcb *)in6p->in6p_ppcb; 1147 struct socket *so = in6p->in6p_socket; 1148 1149 /* 1150 * Ignore some errors if we are hooked up. 1151 * If connection hasn't completed, has retransmitted several times, 1152 * and receives a second error, give up now. This is better 1153 * than waiting a long time to establish a connection that 1154 * can never complete. 1155 */ 1156 if (tp->t_state == TCPS_ESTABLISHED && 1157 (error == EHOSTUNREACH || error == ENETUNREACH || 1158 error == EHOSTDOWN)) { 1159 return; 1160 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1161 tp->t_rxtshift > 3 && tp->t_softerror) 1162 so->so_error = error; 1163 else 1164 tp->t_softerror = error; 1165 wakeup((caddr_t) &so->so_timeo); 1166 sorwakeup(so); 1167 sowwakeup(so); 1168 } 1169 #endif 1170 1171 #ifdef INET6 1172 void 1173 tcp6_ctlinput(cmd, sa, d) 1174 int cmd; 1175 struct sockaddr *sa; 1176 void *d; 1177 { 1178 struct tcphdr th; 1179 void (*notify) __P((struct in6pcb *, int)) = tcp6_notify; 1180 int nmatch; 1181 struct ip6_hdr *ip6; 1182 const struct sockaddr_in6 *sa6_src = NULL; 1183 struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa; 1184 struct mbuf *m; 1185 int off; 1186 1187 if (sa->sa_family != AF_INET6 || 1188 sa->sa_len != sizeof(struct sockaddr_in6)) 1189 return; 1190 if ((unsigned)cmd >= PRC_NCMDS) 1191 return; 1192 else if (cmd == PRC_QUENCH) { 1193 /* XXX there's no PRC_QUENCH in IPv6 */ 1194 notify = tcp6_quench; 1195 } else if (PRC_IS_REDIRECT(cmd)) 1196 notify = in6_rtchange, d = NULL; 1197 else if (cmd == PRC_MSGSIZE) 1198 ; /* special code is present, see below */ 1199 else if (cmd == PRC_HOSTDEAD) 1200 d = NULL; 1201 else if (inet6ctlerrmap[cmd] == 0) 1202 return; 1203 1204 /* if the parameter is from icmp6, decode it. */ 1205 if (d != NULL) { 1206 struct ip6ctlparam *ip6cp = (struct ip6ctlparam *)d; 1207 m = ip6cp->ip6c_m; 1208 ip6 = ip6cp->ip6c_ip6; 1209 off = ip6cp->ip6c_off; 1210 sa6_src = ip6cp->ip6c_src; 1211 } else { 1212 m = NULL; 1213 ip6 = NULL; 1214 sa6_src = &sa6_any; 1215 } 1216 1217 if (ip6) { 1218 /* 1219 * XXX: We assume that when ip6 is non NULL, 1220 * M and OFF are valid. 1221 */ 1222 1223 /* check if we can safely examine src and dst ports */ 1224 if (m->m_pkthdr.len < off + sizeof(th)) { 1225 if (cmd == PRC_MSGSIZE) 1226 icmp6_mtudisc_update((struct ip6ctlparam *)d, 0); 1227 return; 1228 } 1229 1230 bzero(&th, sizeof(th)); 1231 m_copydata(m, off, sizeof(th), (caddr_t)&th); 1232 1233 if (cmd == PRC_MSGSIZE) { 1234 int valid = 0; 1235 1236 /* 1237 * Check to see if we have a valid TCP connection 1238 * corresponding to the address in the ICMPv6 message 1239 * payload. 1240 */ 1241 if (in6_pcblookup_connect(&tcb6, &sa6->sin6_addr, 1242 th.th_dport, (struct in6_addr *)&sa6_src->sin6_addr, 1243 th.th_sport, 0)) 1244 valid++; 1245 1246 /* 1247 * Depending on the value of "valid" and routing table 1248 * size (mtudisc_{hi,lo}wat), we will: 1249 * - recalcurate the new MTU and create the 1250 * corresponding routing entry, or 1251 * - ignore the MTU change notification. 1252 */ 1253 icmp6_mtudisc_update((struct ip6ctlparam *)d, valid); 1254 1255 /* 1256 * no need to call in6_pcbnotify, it should have been 1257 * called via callback if necessary 1258 */ 1259 return; 1260 } 1261 1262 nmatch = in6_pcbnotify(&tcb6, sa, th.th_dport, 1263 (struct sockaddr *)sa6_src, th.th_sport, cmd, NULL, notify); 1264 if (nmatch == 0 && syn_cache_count && 1265 (inet6ctlerrmap[cmd] == EHOSTUNREACH || 1266 inet6ctlerrmap[cmd] == ENETUNREACH || 1267 inet6ctlerrmap[cmd] == EHOSTDOWN)) 1268 syn_cache_unreach((struct sockaddr *)sa6_src, 1269 sa, &th); 1270 } else { 1271 (void) in6_pcbnotify(&tcb6, sa, 0, (struct sockaddr *)sa6_src, 1272 0, cmd, NULL, notify); 1273 } 1274 } 1275 #endif 1276 1277 #ifdef INET 1278 /* assumes that ip header and tcp header are contiguous on mbuf */ 1279 void * 1280 tcp_ctlinput(cmd, sa, v) 1281 int cmd; 1282 struct sockaddr *sa; 1283 void *v; 1284 { 1285 struct ip *ip = v; 1286 struct tcphdr *th; 1287 struct icmp *icp; 1288 extern const int inetctlerrmap[]; 1289 void (*notify) __P((struct inpcb *, int)) = tcp_notify; 1290 int errno; 1291 int nmatch; 1292 1293 if (sa->sa_family != AF_INET || 1294 sa->sa_len != sizeof(struct sockaddr_in)) 1295 return NULL; 1296 if ((unsigned)cmd >= PRC_NCMDS) 1297 return NULL; 1298 errno = inetctlerrmap[cmd]; 1299 if (cmd == PRC_QUENCH) 1300 notify = tcp_quench; 1301 else if (PRC_IS_REDIRECT(cmd)) 1302 notify = in_rtchange, ip = 0; 1303 else if (cmd == PRC_MSGSIZE && ip_mtudisc && ip && ip->ip_v == 4) { 1304 /* 1305 * Check to see if we have a valid TCP connection 1306 * corresponding to the address in the ICMP message 1307 * payload. 1308 * 1309 * Boundary check is made in icmp_input(), with ICMP_ADVLENMIN. 1310 */ 1311 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1312 if (in_pcblookup_connect(&tcbtable, 1313 ip->ip_dst, th->th_dport, 1314 ip->ip_src, th->th_sport) == NULL) 1315 return NULL; 1316 1317 /* 1318 * Now that we've validated that we are actually communicating 1319 * with the host indicated in the ICMP message, locate the 1320 * ICMP header, recalculate the new MTU, and create the 1321 * corresponding routing entry. 1322 */ 1323 icp = (struct icmp *)((caddr_t)ip - 1324 offsetof(struct icmp, icmp_ip)); 1325 icmp_mtudisc(icp, ip->ip_dst); 1326 1327 return NULL; 1328 } else if (cmd == PRC_HOSTDEAD) 1329 ip = 0; 1330 else if (errno == 0) 1331 return NULL; 1332 if (ip && ip->ip_v == 4 && sa->sa_family == AF_INET) { 1333 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1334 nmatch = in_pcbnotify(&tcbtable, satosin(sa)->sin_addr, 1335 th->th_dport, ip->ip_src, th->th_sport, errno, notify); 1336 if (nmatch == 0 && syn_cache_count && 1337 (inetctlerrmap[cmd] == EHOSTUNREACH || 1338 inetctlerrmap[cmd] == ENETUNREACH || 1339 inetctlerrmap[cmd] == EHOSTDOWN)) { 1340 struct sockaddr_in sin; 1341 bzero(&sin, sizeof(sin)); 1342 sin.sin_len = sizeof(sin); 1343 sin.sin_family = AF_INET; 1344 sin.sin_port = th->th_sport; 1345 sin.sin_addr = ip->ip_src; 1346 syn_cache_unreach((struct sockaddr *)&sin, sa, th); 1347 } 1348 1349 /* XXX mapped address case */ 1350 } else 1351 in_pcbnotifyall(&tcbtable, satosin(sa)->sin_addr, errno, 1352 notify); 1353 return NULL; 1354 } 1355 1356 /* 1357 * When a source quence is received, we are being notifed of congestion. 1358 * Close the congestion window down to the Loss Window (one segment). 1359 * We will gradually open it again as we proceed. 1360 */ 1361 void 1362 tcp_quench(inp, errno) 1363 struct inpcb *inp; 1364 int errno; 1365 { 1366 struct tcpcb *tp = intotcpcb(inp); 1367 1368 if (tp) 1369 tp->snd_cwnd = tp->t_segsz; 1370 } 1371 #endif 1372 1373 #ifdef INET6 1374 void 1375 tcp6_quench(in6p, errno) 1376 struct in6pcb *in6p; 1377 int errno; 1378 { 1379 struct tcpcb *tp = in6totcpcb(in6p); 1380 1381 if (tp) 1382 tp->snd_cwnd = tp->t_segsz; 1383 } 1384 #endif 1385 1386 #ifdef INET 1387 /* 1388 * Path MTU Discovery handlers. 1389 */ 1390 void 1391 tcp_mtudisc_callback(faddr) 1392 struct in_addr faddr; 1393 { 1394 1395 in_pcbnotifyall(&tcbtable, faddr, EMSGSIZE, tcp_mtudisc); 1396 } 1397 1398 /* 1399 * On receipt of path MTU corrections, flush old route and replace it 1400 * with the new one. Retransmit all unacknowledged packets, to ensure 1401 * that all packets will be received. 1402 */ 1403 void 1404 tcp_mtudisc(inp, errno) 1405 struct inpcb *inp; 1406 int errno; 1407 { 1408 struct tcpcb *tp = intotcpcb(inp); 1409 struct rtentry *rt = in_pcbrtentry(inp); 1410 1411 if (tp != 0) { 1412 if (rt != 0) { 1413 /* 1414 * If this was not a host route, remove and realloc. 1415 */ 1416 if ((rt->rt_flags & RTF_HOST) == 0) { 1417 in_rtchange(inp, errno); 1418 if ((rt = in_pcbrtentry(inp)) == 0) 1419 return; 1420 } 1421 1422 /* 1423 * Slow start out of the error condition. We 1424 * use the MTU because we know it's smaller 1425 * than the previously transmitted segment. 1426 * 1427 * Note: This is more conservative than the 1428 * suggestion in draft-floyd-incr-init-win-03. 1429 */ 1430 if (rt->rt_rmx.rmx_mtu != 0) 1431 tp->snd_cwnd = 1432 TCP_INITIAL_WINDOW(tcp_init_win, 1433 rt->rt_rmx.rmx_mtu); 1434 } 1435 1436 /* 1437 * Resend unacknowledged packets. 1438 */ 1439 tp->snd_nxt = tp->snd_una; 1440 tcp_output(tp); 1441 } 1442 } 1443 #endif 1444 1445 #ifdef INET6 1446 /* 1447 * Path MTU Discovery handlers. 1448 */ 1449 void 1450 tcp6_mtudisc_callback(faddr) 1451 struct in6_addr *faddr; 1452 { 1453 struct sockaddr_in6 sin6; 1454 1455 bzero(&sin6, sizeof(sin6)); 1456 sin6.sin6_family = AF_INET6; 1457 sin6.sin6_len = sizeof(struct sockaddr_in6); 1458 sin6.sin6_addr = *faddr; 1459 (void) in6_pcbnotify(&tcb6, (struct sockaddr *)&sin6, 0, 1460 (struct sockaddr *)&sa6_any, 0, PRC_MSGSIZE, NULL, tcp6_mtudisc); 1461 } 1462 1463 void 1464 tcp6_mtudisc(in6p, errno) 1465 struct in6pcb *in6p; 1466 int errno; 1467 { 1468 struct tcpcb *tp = in6totcpcb(in6p); 1469 struct rtentry *rt = in6_pcbrtentry(in6p); 1470 1471 if (tp != 0) { 1472 if (rt != 0) { 1473 /* 1474 * If this was not a host route, remove and realloc. 1475 */ 1476 if ((rt->rt_flags & RTF_HOST) == 0) { 1477 in6_rtchange(in6p, errno); 1478 if ((rt = in6_pcbrtentry(in6p)) == 0) 1479 return; 1480 } 1481 1482 /* 1483 * Slow start out of the error condition. We 1484 * use the MTU because we know it's smaller 1485 * than the previously transmitted segment. 1486 * 1487 * Note: This is more conservative than the 1488 * suggestion in draft-floyd-incr-init-win-03. 1489 */ 1490 if (rt->rt_rmx.rmx_mtu != 0) 1491 tp->snd_cwnd = 1492 TCP_INITIAL_WINDOW(tcp_init_win, 1493 rt->rt_rmx.rmx_mtu); 1494 } 1495 1496 /* 1497 * Resend unacknowledged packets. 1498 */ 1499 tp->snd_nxt = tp->snd_una; 1500 tcp_output(tp); 1501 } 1502 } 1503 #endif /* INET6 */ 1504 1505 /* 1506 * Compute the MSS to advertise to the peer. Called only during 1507 * the 3-way handshake. If we are the server (peer initiated 1508 * connection), we are called with a pointer to the interface 1509 * on which the SYN packet arrived. If we are the client (we 1510 * initiated connection), we are called with a pointer to the 1511 * interface out which this connection should go. 1512 * 1513 * NOTE: Do not subtract IP option/extension header size nor IPsec 1514 * header size from MSS advertisement. MSS option must hold the maximum 1515 * segment size we can accept, so it must always be: 1516 * max(if mtu) - ip header - tcp header 1517 */ 1518 u_long 1519 tcp_mss_to_advertise(ifp, af) 1520 const struct ifnet *ifp; 1521 int af; 1522 { 1523 extern u_long in_maxmtu; 1524 u_long mss = 0; 1525 u_long hdrsiz; 1526 1527 /* 1528 * In order to avoid defeating path MTU discovery on the peer, 1529 * we advertise the max MTU of all attached networks as our MSS, 1530 * per RFC 1191, section 3.1. 1531 * 1532 * We provide the option to advertise just the MTU of 1533 * the interface on which we hope this connection will 1534 * be receiving. If we are responding to a SYN, we 1535 * will have a pretty good idea about this, but when 1536 * initiating a connection there is a bit more doubt. 1537 * 1538 * We also need to ensure that loopback has a large enough 1539 * MSS, as the loopback MTU is never included in in_maxmtu. 1540 */ 1541 1542 if (ifp != NULL) 1543 mss = ifp->if_mtu; 1544 1545 if (tcp_mss_ifmtu == 0) 1546 switch (af) { 1547 case AF_INET: 1548 mss = max(in_maxmtu, mss); 1549 break; 1550 #ifdef INET6 1551 case AF_INET6: 1552 mss = max(in6_maxmtu, mss); 1553 break; 1554 #endif 1555 } 1556 1557 switch (af) { 1558 case AF_INET: 1559 hdrsiz = sizeof(struct ip); 1560 break; 1561 #ifdef INET6 1562 case AF_INET6: 1563 hdrsiz = sizeof(struct ip6_hdr); 1564 break; 1565 #endif 1566 default: 1567 hdrsiz = 0; 1568 break; 1569 } 1570 hdrsiz += sizeof(struct tcphdr); 1571 if (mss > hdrsiz) 1572 mss -= hdrsiz; 1573 1574 mss = max(tcp_mssdflt, mss); 1575 return (mss); 1576 } 1577 1578 /* 1579 * Set connection variables based on the peer's advertised MSS. 1580 * We are passed the TCPCB for the actual connection. If we 1581 * are the server, we are called by the compressed state engine 1582 * when the 3-way handshake is complete. If we are the client, 1583 * we are called when we receive the SYN,ACK from the server. 1584 * 1585 * NOTE: Our advertised MSS value must be initialized in the TCPCB 1586 * before this routine is called! 1587 */ 1588 void 1589 tcp_mss_from_peer(tp, offer) 1590 struct tcpcb *tp; 1591 int offer; 1592 { 1593 struct socket *so; 1594 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1595 struct rtentry *rt; 1596 #endif 1597 u_long bufsize; 1598 int mss; 1599 1600 #ifdef DIAGNOSTIC 1601 if (tp->t_inpcb && tp->t_in6pcb) 1602 panic("tcp_mss_from_peer: both t_inpcb and t_in6pcb are set"); 1603 #endif 1604 so = NULL; 1605 rt = NULL; 1606 #ifdef INET 1607 if (tp->t_inpcb) { 1608 so = tp->t_inpcb->inp_socket; 1609 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1610 rt = in_pcbrtentry(tp->t_inpcb); 1611 #endif 1612 } 1613 #endif 1614 #ifdef INET6 1615 if (tp->t_in6pcb) { 1616 so = tp->t_in6pcb->in6p_socket; 1617 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1618 rt = in6_pcbrtentry(tp->t_in6pcb); 1619 #endif 1620 } 1621 #endif 1622 1623 /* 1624 * As per RFC1122, use the default MSS value, unless they 1625 * sent us an offer. Do not accept offers less than 32 bytes. 1626 */ 1627 mss = tcp_mssdflt; 1628 if (offer) 1629 mss = offer; 1630 mss = max(mss, 32); /* sanity */ 1631 tp->t_peermss = mss; 1632 mss -= tcp_optlen(tp); 1633 #ifdef INET 1634 if (tp->t_inpcb) 1635 mss -= ip_optlen(tp->t_inpcb); 1636 #endif 1637 #ifdef INET6 1638 if (tp->t_in6pcb) 1639 mss -= ip6_optlen(tp->t_in6pcb); 1640 #endif 1641 1642 /* 1643 * If there's a pipesize, change the socket buffer to that size. 1644 * Make the socket buffer an integral number of MSS units. If 1645 * the MSS is larger than the socket buffer, artificially decrease 1646 * the MSS. 1647 */ 1648 #ifdef RTV_SPIPE 1649 if (rt != NULL && rt->rt_rmx.rmx_sendpipe != 0) 1650 bufsize = rt->rt_rmx.rmx_sendpipe; 1651 else 1652 #endif 1653 bufsize = so->so_snd.sb_hiwat; 1654 if (bufsize < mss) 1655 mss = bufsize; 1656 else { 1657 bufsize = roundup(bufsize, mss); 1658 if (bufsize > sb_max) 1659 bufsize = sb_max; 1660 (void) sbreserve(&so->so_snd, bufsize); 1661 } 1662 tp->t_segsz = mss; 1663 1664 #ifdef RTV_SSTHRESH 1665 if (rt != NULL && rt->rt_rmx.rmx_ssthresh) { 1666 /* 1667 * There's some sort of gateway or interface buffer 1668 * limit on the path. Use this to set the slow 1669 * start threshold, but set the threshold to no less 1670 * than 2 * MSS. 1671 */ 1672 tp->snd_ssthresh = max(2 * mss, rt->rt_rmx.rmx_ssthresh); 1673 } 1674 #endif 1675 } 1676 1677 /* 1678 * Processing necessary when a TCP connection is established. 1679 */ 1680 void 1681 tcp_established(tp) 1682 struct tcpcb *tp; 1683 { 1684 struct socket *so; 1685 #ifdef RTV_RPIPE 1686 struct rtentry *rt; 1687 #endif 1688 u_long bufsize; 1689 1690 #ifdef DIAGNOSTIC 1691 if (tp->t_inpcb && tp->t_in6pcb) 1692 panic("tcp_established: both t_inpcb and t_in6pcb are set"); 1693 #endif 1694 so = NULL; 1695 rt = NULL; 1696 #ifdef INET 1697 if (tp->t_inpcb) { 1698 so = tp->t_inpcb->inp_socket; 1699 #if defined(RTV_RPIPE) 1700 rt = in_pcbrtentry(tp->t_inpcb); 1701 #endif 1702 } 1703 #endif 1704 #ifdef INET6 1705 if (tp->t_in6pcb) { 1706 so = tp->t_in6pcb->in6p_socket; 1707 #if defined(RTV_RPIPE) 1708 rt = in6_pcbrtentry(tp->t_in6pcb); 1709 #endif 1710 } 1711 #endif 1712 1713 tp->t_state = TCPS_ESTABLISHED; 1714 TCP_TIMER_ARM(tp, TCPT_KEEP, tcp_keepidle); 1715 1716 #ifdef RTV_RPIPE 1717 if (rt != NULL && rt->rt_rmx.rmx_recvpipe != 0) 1718 bufsize = rt->rt_rmx.rmx_recvpipe; 1719 else 1720 #endif 1721 bufsize = so->so_rcv.sb_hiwat; 1722 if (bufsize > tp->t_ourmss) { 1723 bufsize = roundup(bufsize, tp->t_ourmss); 1724 if (bufsize > sb_max) 1725 bufsize = sb_max; 1726 (void) sbreserve(&so->so_rcv, bufsize); 1727 } 1728 } 1729 1730 /* 1731 * Check if there's an initial rtt or rttvar. Convert from the 1732 * route-table units to scaled multiples of the slow timeout timer. 1733 * Called only during the 3-way handshake. 1734 */ 1735 void 1736 tcp_rmx_rtt(tp) 1737 struct tcpcb *tp; 1738 { 1739 #ifdef RTV_RTT 1740 struct rtentry *rt = NULL; 1741 int rtt; 1742 1743 #ifdef DIAGNOSTIC 1744 if (tp->t_inpcb && tp->t_in6pcb) 1745 panic("tcp_rmx_rtt: both t_inpcb and t_in6pcb are set"); 1746 #endif 1747 #ifdef INET 1748 if (tp->t_inpcb) 1749 rt = in_pcbrtentry(tp->t_inpcb); 1750 #endif 1751 #ifdef INET6 1752 if (tp->t_in6pcb) 1753 rt = in6_pcbrtentry(tp->t_in6pcb); 1754 #endif 1755 if (rt == NULL) 1756 return; 1757 1758 if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt)) { 1759 /* 1760 * XXX The lock bit for MTU indicates that the value 1761 * is also a minimum value; this is subject to time. 1762 */ 1763 if (rt->rt_rmx.rmx_locks & RTV_RTT) 1764 TCPT_RANGESET(tp->t_rttmin, 1765 rtt / (RTM_RTTUNIT / PR_SLOWHZ), 1766 TCPTV_MIN, TCPTV_REXMTMAX); 1767 tp->t_srtt = rtt / 1768 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1769 if (rt->rt_rmx.rmx_rttvar) { 1770 tp->t_rttvar = rt->rt_rmx.rmx_rttvar / 1771 ((RTM_RTTUNIT / PR_SLOWHZ) >> 1772 (TCP_RTTVAR_SHIFT + 2)); 1773 } else { 1774 /* Default variation is +- 1 rtt */ 1775 tp->t_rttvar = 1776 tp->t_srtt >> (TCP_RTT_SHIFT - TCP_RTTVAR_SHIFT); 1777 } 1778 TCPT_RANGESET(tp->t_rxtcur, 1779 ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2), 1780 tp->t_rttmin, TCPTV_REXMTMAX); 1781 } 1782 #endif 1783 } 1784 1785 tcp_seq tcp_iss_seq = 0; /* tcp initial seq # */ 1786 #if NRND > 0 1787 u_int8_t tcp_iss_secret[16]; /* 128 bits; should be plenty */ 1788 #endif 1789 1790 /* 1791 * Get a new sequence value given a tcp control block 1792 */ 1793 tcp_seq 1794 tcp_new_iss(struct tcpcb *tp, tcp_seq addin) 1795 { 1796 1797 #ifdef INET 1798 if (tp->t_inpcb != NULL) { 1799 return (tcp_new_iss1(&tp->t_inpcb->inp_laddr, 1800 &tp->t_inpcb->inp_faddr, tp->t_inpcb->inp_lport, 1801 tp->t_inpcb->inp_fport, sizeof(tp->t_inpcb->inp_laddr), 1802 addin)); 1803 } 1804 #endif 1805 #ifdef INET6 1806 if (tp->t_in6pcb != NULL) { 1807 return (tcp_new_iss1(&tp->t_in6pcb->in6p_laddr, 1808 &tp->t_in6pcb->in6p_faddr, tp->t_in6pcb->in6p_lport, 1809 tp->t_in6pcb->in6p_fport, sizeof(tp->t_in6pcb->in6p_laddr), 1810 addin)); 1811 } 1812 #endif 1813 /* Not possible. */ 1814 panic("tcp_new_iss"); 1815 } 1816 1817 /* 1818 * This routine actually generates a new TCP initial sequence number. 1819 */ 1820 tcp_seq 1821 tcp_new_iss1(void *laddr, void *faddr, u_int16_t lport, u_int16_t fport, 1822 size_t addrsz, tcp_seq addin) 1823 { 1824 tcp_seq tcp_iss; 1825 1826 #if NRND > 0 1827 static int beenhere; 1828 1829 /* 1830 * If we haven't been here before, initialize our cryptographic 1831 * hash secret. 1832 */ 1833 if (beenhere == 0) { 1834 rnd_extract_data(tcp_iss_secret, sizeof(tcp_iss_secret), 1835 RND_EXTRACT_ANY); 1836 beenhere = 1; 1837 } 1838 1839 if (tcp_do_rfc1948) { 1840 MD5_CTX ctx; 1841 u_int8_t hash[16]; /* XXX MD5 knowledge */ 1842 1843 /* 1844 * Compute the base value of the ISS. It is a hash 1845 * of (saddr, sport, daddr, dport, secret). 1846 */ 1847 MD5Init(&ctx); 1848 1849 MD5Update(&ctx, (u_char *) laddr, addrsz); 1850 MD5Update(&ctx, (u_char *) &lport, sizeof(lport)); 1851 1852 MD5Update(&ctx, (u_char *) faddr, addrsz); 1853 MD5Update(&ctx, (u_char *) &fport, sizeof(fport)); 1854 1855 MD5Update(&ctx, tcp_iss_secret, sizeof(tcp_iss_secret)); 1856 1857 MD5Final(hash, &ctx); 1858 1859 memcpy(&tcp_iss, hash, sizeof(tcp_iss)); 1860 1861 /* 1862 * Now increment our "timer", and add it in to 1863 * the computed value. 1864 * 1865 * XXX Use `addin'? 1866 * XXX TCP_ISSINCR too large to use? 1867 */ 1868 tcp_iss_seq += TCP_ISSINCR; 1869 #ifdef TCPISS_DEBUG 1870 printf("ISS hash 0x%08x, ", tcp_iss); 1871 #endif 1872 tcp_iss += tcp_iss_seq + addin; 1873 #ifdef TCPISS_DEBUG 1874 printf("new ISS 0x%08x\n", tcp_iss); 1875 #endif 1876 } else 1877 #endif /* NRND > 0 */ 1878 { 1879 /* 1880 * Randomize. 1881 */ 1882 #if NRND > 0 1883 rnd_extract_data(&tcp_iss, sizeof(tcp_iss), RND_EXTRACT_ANY); 1884 #else 1885 tcp_iss = random(); 1886 #endif 1887 1888 /* 1889 * If we were asked to add some amount to a known value, 1890 * we will take a random value obtained above, mask off 1891 * the upper bits, and add in the known value. We also 1892 * add in a constant to ensure that we are at least a 1893 * certain distance from the original value. 1894 * 1895 * This is used when an old connection is in timed wait 1896 * and we have a new one coming in, for instance. 1897 */ 1898 if (addin != 0) { 1899 #ifdef TCPISS_DEBUG 1900 printf("Random %08x, ", tcp_iss); 1901 #endif 1902 tcp_iss &= TCP_ISS_RANDOM_MASK; 1903 tcp_iss += addin + TCP_ISSINCR; 1904 #ifdef TCPISS_DEBUG 1905 printf("Old ISS %08x, ISS %08x\n", addin, tcp_iss); 1906 #endif 1907 } else { 1908 tcp_iss &= TCP_ISS_RANDOM_MASK; 1909 tcp_iss += tcp_iss_seq; 1910 tcp_iss_seq += TCP_ISSINCR; 1911 #ifdef TCPISS_DEBUG 1912 printf("ISS %08x\n", tcp_iss); 1913 #endif 1914 } 1915 } 1916 1917 if (tcp_compat_42) { 1918 /* 1919 * Limit it to the positive range for really old TCP 1920 * implementations. 1921 */ 1922 if (tcp_iss >= 0x80000000) 1923 tcp_iss &= 0x7fffffff; /* XXX */ 1924 } 1925 1926 return (tcp_iss); 1927 } 1928 1929 #ifdef IPSEC 1930 /* compute ESP/AH header size for TCP, including outer IP header. */ 1931 size_t 1932 ipsec4_hdrsiz_tcp(tp) 1933 struct tcpcb *tp; 1934 { 1935 struct inpcb *inp; 1936 size_t hdrsiz; 1937 1938 /* XXX mapped addr case (tp->t_in6pcb) */ 1939 if (!tp || !tp->t_template || !(inp = tp->t_inpcb)) 1940 return 0; 1941 switch (tp->t_family) { 1942 case AF_INET: 1943 /* XXX: should use currect direction. */ 1944 hdrsiz = ipsec4_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, inp); 1945 break; 1946 default: 1947 hdrsiz = 0; 1948 break; 1949 } 1950 1951 return hdrsiz; 1952 } 1953 1954 #ifdef INET6 1955 size_t 1956 ipsec6_hdrsiz_tcp(tp) 1957 struct tcpcb *tp; 1958 { 1959 struct in6pcb *in6p; 1960 size_t hdrsiz; 1961 1962 if (!tp || !tp->t_template || !(in6p = tp->t_in6pcb)) 1963 return 0; 1964 switch (tp->t_family) { 1965 case AF_INET6: 1966 /* XXX: should use currect direction. */ 1967 hdrsiz = ipsec6_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, in6p); 1968 break; 1969 case AF_INET: 1970 /* mapped address case - tricky */ 1971 default: 1972 hdrsiz = 0; 1973 break; 1974 } 1975 1976 return hdrsiz; 1977 } 1978 #endif 1979 #endif /*IPSEC*/ 1980 1981 /* 1982 * Determine the length of the TCP options for this connection. 1983 * 1984 * XXX: What do we do for SACK, when we add that? Just reserve 1985 * all of the space? Otherwise we can't exactly be incrementing 1986 * cwnd by an amount that varies depending on the amount we last 1987 * had to SACK! 1988 */ 1989 1990 u_int 1991 tcp_optlen(tp) 1992 struct tcpcb *tp; 1993 { 1994 if ((tp->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP|TF_NOOPT)) == 1995 (TF_REQ_TSTMP | TF_RCVD_TSTMP)) 1996 return TCPOLEN_TSTAMP_APPA; 1997 else 1998 return 0; 1999 } 2000
Indexes created Sun Sep 21 20:09:37 GMT 2025