tcp_input.c revision 1.258
1 /* $NetBSD: tcp_input.c,v 1.258 2006/12/09 05:33:04 dyoung 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 1.1 (NRL) 17 January 1995 34 * 35 * NRL grants permission for redistribution and use in source and binary 36 * forms, with or without modification, of the software and documentation 37 * created at NRL provided that the following conditions are met: 38 * 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. All advertising materials mentioning features or use of this software 45 * must display the following acknowledgements: 46 * This product includes software developed by the University of 47 * California, Berkeley and its contributors. 48 * This product includes software developed at the Information 49 * Technology Division, US Naval Research Laboratory. 50 * 4. Neither the name of the NRL nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS 55 * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 56 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A 57 * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NRL OR 58 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 59 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 60 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 61 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 62 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 63 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 64 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 65 * 66 * The views and conclusions contained in the software and documentation 67 * are those of the authors and should not be interpreted as representing 68 * official policies, either expressed or implied, of the US Naval 69 * Research Laboratory (NRL). 70 */ 71 72 /*- 73 * Copyright (c) 1997, 1998, 1999, 2001, 2005, 2006 The NetBSD Foundation, Inc. 74 * All rights reserved. 75 * 76 * This code is derived from software contributed to The NetBSD Foundation 77 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 78 * Facility, NASA Ames Research Center. 79 * This code is derived from software contributed to The NetBSD Foundation 80 * by Charles M. Hannum. 81 * This code is derived from software contributed to The NetBSD Foundation 82 * by Rui Paulo. 83 * 84 * Redistribution and use in source and binary forms, with or without 85 * modification, are permitted provided that the following conditions 86 * are met: 87 * 1. Redistributions of source code must retain the above copyright 88 * notice, this list of conditions and the following disclaimer. 89 * 2. Redistributions in binary form must reproduce the above copyright 90 * notice, this list of conditions and the following disclaimer in the 91 * documentation and/or other materials provided with the distribution. 92 * 3. All advertising materials mentioning features or use of this software 93 * must display the following acknowledgement: 94 * This product includes software developed by the NetBSD 95 * Foundation, Inc. and its contributors. 96 * 4. Neither the name of The NetBSD Foundation nor the names of its 97 * contributors may be used to endorse or promote products derived 98 * from this software without specific prior written permission. 99 * 100 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 101 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 102 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 103 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 104 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 105 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 106 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 107 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 108 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 109 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 110 * POSSIBILITY OF SUCH DAMAGE. 111 */ 112 113 /* 114 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995 115 * The Regents of the University of California. All rights reserved. 116 * 117 * Redistribution and use in source and binary forms, with or without 118 * modification, are permitted provided that the following conditions 119 * are met: 120 * 1. Redistributions of source code must retain the above copyright 121 * notice, this list of conditions and the following disclaimer. 122 * 2. Redistributions in binary form must reproduce the above copyright 123 * notice, this list of conditions and the following disclaimer in the 124 * documentation and/or other materials provided with the distribution. 125 * 3. Neither the name of the University nor the names of its contributors 126 * may be used to endorse or promote products derived from this software 127 * without specific prior written permission. 128 * 129 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 130 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 131 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 132 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 133 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 134 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 135 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 136 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 137 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 138 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 139 * SUCH DAMAGE. 140 * 141 * @(#)tcp_input.c 8.12 (Berkeley) 5/24/95 142 */ 143 144 /* 145 * TODO list for SYN cache stuff: 146 * 147 * Find room for a "state" field, which is needed to keep a 148 * compressed state for TIME_WAIT TCBs. It's been noted already 149 * that this is fairly important for very high-volume web and 150 * mail servers, which use a large number of short-lived 151 * connections. 152 */ 153 154 #include <sys/cdefs.h> 155 __KERNEL_RCSID(0, "$NetBSD: tcp_input.c,v 1.258 2006/12/09 05:33:04 dyoung Exp $"); 156 157 #include "opt_inet.h" 158 #include "opt_ipsec.h" 159 #include "opt_inet_csum.h" 160 #include "opt_tcp_debug.h" 161 162 #include <sys/param.h> 163 #include <sys/systm.h> 164 #include <sys/malloc.h> 165 #include <sys/mbuf.h> 166 #include <sys/protosw.h> 167 #include <sys/socket.h> 168 #include <sys/socketvar.h> 169 #include <sys/errno.h> 170 #include <sys/syslog.h> 171 #include <sys/pool.h> 172 #include <sys/domain.h> 173 #include <sys/kernel.h> 174 #ifdef TCP_SIGNATURE 175 #include <sys/md5.h> 176 #endif 177 178 #include <net/if.h> 179 #include <net/route.h> 180 #include <net/if_types.h> 181 182 #include <netinet/in.h> 183 #include <netinet/in_systm.h> 184 #include <netinet/ip.h> 185 #include <netinet/in_pcb.h> 186 #include <netinet/in_var.h> 187 #include <netinet/ip_var.h> 188 #include <netinet/in_offload.h> 189 190 #ifdef INET6 191 #ifndef INET 192 #include <netinet/in.h> 193 #endif 194 #include <netinet/ip6.h> 195 #include <netinet6/ip6_var.h> 196 #include <netinet6/in6_pcb.h> 197 #include <netinet6/ip6_var.h> 198 #include <netinet6/in6_var.h> 199 #include <netinet/icmp6.h> 200 #include <netinet6/nd6.h> 201 #ifdef TCP_SIGNATURE 202 #include <netinet6/scope6_var.h> 203 #endif 204 #endif 205 206 #ifndef INET6 207 /* always need ip6.h for IP6_EXTHDR_GET */ 208 #include <netinet/ip6.h> 209 #endif 210 211 #include <netinet/tcp.h> 212 #include <netinet/tcp_fsm.h> 213 #include <netinet/tcp_seq.h> 214 #include <netinet/tcp_timer.h> 215 #include <netinet/tcp_var.h> 216 #include <netinet/tcpip.h> 217 #include <netinet/tcp_congctl.h> 218 #include <netinet/tcp_debug.h> 219 220 #include <machine/stdarg.h> 221 222 #ifdef IPSEC 223 #include <netinet6/ipsec.h> 224 #include <netkey/key.h> 225 #endif /*IPSEC*/ 226 #ifdef INET6 227 #include "faith.h" 228 #if defined(NFAITH) && NFAITH > 0 229 #include <net/if_faith.h> 230 #endif 231 #endif /* IPSEC */ 232 233 #ifdef FAST_IPSEC 234 #include <netipsec/ipsec.h> 235 #include <netipsec/ipsec_var.h> /* XXX ipsecstat namespace */ 236 #include <netipsec/key.h> 237 #ifdef INET6 238 #include <netipsec/ipsec6.h> 239 #endif 240 #endif /* FAST_IPSEC*/ 241 242 int tcprexmtthresh = 3; 243 int tcp_log_refused; 244 245 static int tcp_rst_ppslim_count = 0; 246 static struct timeval tcp_rst_ppslim_last; 247 static int tcp_ackdrop_ppslim_count = 0; 248 static struct timeval tcp_ackdrop_ppslim_last; 249 250 #define TCP_PAWS_IDLE (24U * 24 * 60 * 60 * PR_SLOWHZ) 251 252 /* for modulo comparisons of timestamps */ 253 #define TSTMP_LT(a,b) ((int)((a)-(b)) < 0) 254 #define TSTMP_GEQ(a,b) ((int)((a)-(b)) >= 0) 255 256 /* 257 * Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. 258 */ 259 #ifdef INET6 260 #define ND6_HINT(tp) \ 261 do { \ 262 if (tp && tp->t_in6pcb && tp->t_family == AF_INET6 && \ 263 tp->t_in6pcb->in6p_route.ro_rt) { \ 264 nd6_nud_hint(tp->t_in6pcb->in6p_route.ro_rt, NULL, 0); \ 265 } \ 266 } while (/*CONSTCOND*/ 0) 267 #else 268 #define ND6_HINT(tp) 269 #endif 270 271 /* 272 * Macro to compute ACK transmission behavior. Delay the ACK unless 273 * we have already delayed an ACK (must send an ACK every two segments). 274 * We also ACK immediately if we received a PUSH and the ACK-on-PUSH 275 * option is enabled. 276 */ 277 #define TCP_SETUP_ACK(tp, th) \ 278 do { \ 279 if ((tp)->t_flags & TF_DELACK || \ 280 (tcp_ack_on_push && (th)->th_flags & TH_PUSH)) \ 281 tp->t_flags |= TF_ACKNOW; \ 282 else \ 283 TCP_SET_DELACK(tp); \ 284 } while (/*CONSTCOND*/ 0) 285 286 #define ICMP_CHECK(tp, th, acked) \ 287 do { \ 288 /* \ 289 * If we had a pending ICMP message that \ 290 * refers to data that have just been \ 291 * acknowledged, disregard the recorded ICMP \ 292 * message. \ 293 */ \ 294 if (((tp)->t_flags & TF_PMTUD_PEND) && \ 295 SEQ_GT((th)->th_ack, (tp)->t_pmtud_th_seq)) \ 296 (tp)->t_flags &= ~TF_PMTUD_PEND; \ 297 \ 298 /* \ 299 * Keep track of the largest chunk of data \ 300 * acknowledged since last PMTU update \ 301 */ \ 302 if ((tp)->t_pmtud_mss_acked < (acked)) \ 303 (tp)->t_pmtud_mss_acked = (acked); \ 304 } while (/*CONSTCOND*/ 0) 305 306 /* 307 * Convert TCP protocol fields to host order for easier processing. 308 */ 309 #define TCP_FIELDS_TO_HOST(th) \ 310 do { \ 311 NTOHL((th)->th_seq); \ 312 NTOHL((th)->th_ack); \ 313 NTOHS((th)->th_win); \ 314 NTOHS((th)->th_urp); \ 315 } while (/*CONSTCOND*/ 0) 316 317 /* 318 * ... and reverse the above. 319 */ 320 #define TCP_FIELDS_TO_NET(th) \ 321 do { \ 322 HTONL((th)->th_seq); \ 323 HTONL((th)->th_ack); \ 324 HTONS((th)->th_win); \ 325 HTONS((th)->th_urp); \ 326 } while (/*CONSTCOND*/ 0) 327 328 #ifdef TCP_CSUM_COUNTERS 329 #include <sys/device.h> 330 331 #if defined(INET) 332 extern struct evcnt tcp_hwcsum_ok; 333 extern struct evcnt tcp_hwcsum_bad; 334 extern struct evcnt tcp_hwcsum_data; 335 extern struct evcnt tcp_swcsum; 336 #endif /* defined(INET) */ 337 #if defined(INET6) 338 extern struct evcnt tcp6_hwcsum_ok; 339 extern struct evcnt tcp6_hwcsum_bad; 340 extern struct evcnt tcp6_hwcsum_data; 341 extern struct evcnt tcp6_swcsum; 342 #endif /* defined(INET6) */ 343 344 #define TCP_CSUM_COUNTER_INCR(ev) (ev)->ev_count++ 345 346 #else 347 348 #define TCP_CSUM_COUNTER_INCR(ev) /* nothing */ 349 350 #endif /* TCP_CSUM_COUNTERS */ 351 352 #ifdef TCP_REASS_COUNTERS 353 #include <sys/device.h> 354 355 extern struct evcnt tcp_reass_; 356 extern struct evcnt tcp_reass_empty; 357 extern struct evcnt tcp_reass_iteration[8]; 358 extern struct evcnt tcp_reass_prependfirst; 359 extern struct evcnt tcp_reass_prepend; 360 extern struct evcnt tcp_reass_insert; 361 extern struct evcnt tcp_reass_inserttail; 362 extern struct evcnt tcp_reass_append; 363 extern struct evcnt tcp_reass_appendtail; 364 extern struct evcnt tcp_reass_overlaptail; 365 extern struct evcnt tcp_reass_overlapfront; 366 extern struct evcnt tcp_reass_segdup; 367 extern struct evcnt tcp_reass_fragdup; 368 369 #define TCP_REASS_COUNTER_INCR(ev) (ev)->ev_count++ 370 371 #else 372 373 #define TCP_REASS_COUNTER_INCR(ev) /* nothing */ 374 375 #endif /* TCP_REASS_COUNTERS */ 376 377 static int tcp_reass(struct tcpcb *, const struct tcphdr *, struct mbuf *, 378 int *); 379 static int tcp_dooptions(struct tcpcb *, const u_char *, int, 380 const struct tcphdr *, struct mbuf *, int, struct tcp_opt_info *); 381 382 #ifdef INET 383 static void tcp4_log_refused(const struct ip *, const struct tcphdr *); 384 #endif 385 #ifdef INET6 386 static void tcp6_log_refused(const struct ip6_hdr *, const struct tcphdr *); 387 #endif 388 389 #define TRAVERSE(x) while ((x)->m_next) (x) = (x)->m_next 390 391 #if defined(MBUFTRACE) 392 struct mowner tcp_reass_mowner = MOWNER_INIT("tcp", "reass"); 393 #endif /* defined(MBUFTRACE) */ 394 395 static POOL_INIT(tcpipqent_pool, sizeof(struct ipqent), 0, 0, 0, "tcpipqepl", 396 NULL); 397 398 struct ipqent * 399 tcpipqent_alloc() 400 { 401 struct ipqent *ipqe; 402 int s; 403 404 s = splvm(); 405 ipqe = pool_get(&tcpipqent_pool, PR_NOWAIT); 406 splx(s); 407 408 return ipqe; 409 } 410 411 void 412 tcpipqent_free(struct ipqent *ipqe) 413 { 414 int s; 415 416 s = splvm(); 417 pool_put(&tcpipqent_pool, ipqe); 418 splx(s); 419 } 420 421 static int 422 tcp_reass(struct tcpcb *tp, const struct tcphdr *th, struct mbuf *m, int *tlen) 423 { 424 struct ipqent *p, *q, *nq, *tiqe = NULL; 425 struct socket *so = NULL; 426 int pkt_flags; 427 tcp_seq pkt_seq; 428 unsigned pkt_len; 429 u_long rcvpartdupbyte = 0; 430 u_long rcvoobyte; 431 #ifdef TCP_REASS_COUNTERS 432 u_int count = 0; 433 #endif 434 435 if (tp->t_inpcb) 436 so = tp->t_inpcb->inp_socket; 437 #ifdef INET6 438 else if (tp->t_in6pcb) 439 so = tp->t_in6pcb->in6p_socket; 440 #endif 441 442 TCP_REASS_LOCK_CHECK(tp); 443 444 /* 445 * Call with th==0 after become established to 446 * force pre-ESTABLISHED data up to user socket. 447 */ 448 if (th == 0) 449 goto present; 450 451 m_claimm(m, &tcp_reass_mowner); 452 453 rcvoobyte = *tlen; 454 /* 455 * Copy these to local variables because the tcpiphdr 456 * gets munged while we are collapsing mbufs. 457 */ 458 pkt_seq = th->th_seq; 459 pkt_len = *tlen; 460 pkt_flags = th->th_flags; 461 462 TCP_REASS_COUNTER_INCR(&tcp_reass_); 463 464 if ((p = TAILQ_LAST(&tp->segq, ipqehead)) != NULL) { 465 /* 466 * When we miss a packet, the vast majority of time we get 467 * packets that follow it in order. So optimize for that. 468 */ 469 if (pkt_seq == p->ipqe_seq + p->ipqe_len) { 470 p->ipqe_len += pkt_len; 471 p->ipqe_flags |= pkt_flags; 472 m_cat(p->ipre_mlast, m); 473 TRAVERSE(p->ipre_mlast); 474 m = NULL; 475 tiqe = p; 476 TAILQ_REMOVE(&tp->timeq, p, ipqe_timeq); 477 TCP_REASS_COUNTER_INCR(&tcp_reass_appendtail); 478 goto skip_replacement; 479 } 480 /* 481 * While we're here, if the pkt is completely beyond 482 * anything we have, just insert it at the tail. 483 */ 484 if (SEQ_GT(pkt_seq, p->ipqe_seq + p->ipqe_len)) { 485 TCP_REASS_COUNTER_INCR(&tcp_reass_inserttail); 486 goto insert_it; 487 } 488 } 489 490 q = TAILQ_FIRST(&tp->segq); 491 492 if (q != NULL) { 493 /* 494 * If this segment immediately precedes the first out-of-order 495 * block, simply slap the segment in front of it and (mostly) 496 * skip the complicated logic. 497 */ 498 if (pkt_seq + pkt_len == q->ipqe_seq) { 499 q->ipqe_seq = pkt_seq; 500 q->ipqe_len += pkt_len; 501 q->ipqe_flags |= pkt_flags; 502 m_cat(m, q->ipqe_m); 503 q->ipqe_m = m; 504 q->ipre_mlast = m; /* last mbuf may have changed */ 505 TRAVERSE(q->ipre_mlast); 506 tiqe = q; 507 TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq); 508 TCP_REASS_COUNTER_INCR(&tcp_reass_prependfirst); 509 goto skip_replacement; 510 } 511 } else { 512 TCP_REASS_COUNTER_INCR(&tcp_reass_empty); 513 } 514 515 /* 516 * Find a segment which begins after this one does. 517 */ 518 for (p = NULL; q != NULL; q = nq) { 519 nq = TAILQ_NEXT(q, ipqe_q); 520 #ifdef TCP_REASS_COUNTERS 521 count++; 522 #endif 523 /* 524 * If the received segment is just right after this 525 * fragment, merge the two together and then check 526 * for further overlaps. 527 */ 528 if (q->ipqe_seq + q->ipqe_len == pkt_seq) { 529 #ifdef TCPREASS_DEBUG 530 printf("tcp_reass[%p]: concat %u:%u(%u) to %u:%u(%u)\n", 531 tp, pkt_seq, pkt_seq + pkt_len, pkt_len, 532 q->ipqe_seq, q->ipqe_seq + q->ipqe_len, q->ipqe_len); 533 #endif 534 pkt_len += q->ipqe_len; 535 pkt_flags |= q->ipqe_flags; 536 pkt_seq = q->ipqe_seq; 537 m_cat(q->ipre_mlast, m); 538 TRAVERSE(q->ipre_mlast); 539 m = q->ipqe_m; 540 TCP_REASS_COUNTER_INCR(&tcp_reass_append); 541 goto free_ipqe; 542 } 543 /* 544 * If the received segment is completely past this 545 * fragment, we need to go the next fragment. 546 */ 547 if (SEQ_LT(q->ipqe_seq + q->ipqe_len, pkt_seq)) { 548 p = q; 549 continue; 550 } 551 /* 552 * If the fragment is past the received segment, 553 * it (or any following) can't be concatenated. 554 */ 555 if (SEQ_GT(q->ipqe_seq, pkt_seq + pkt_len)) { 556 TCP_REASS_COUNTER_INCR(&tcp_reass_insert); 557 break; 558 } 559 560 /* 561 * We've received all the data in this segment before. 562 * mark it as a duplicate and return. 563 */ 564 if (SEQ_LEQ(q->ipqe_seq, pkt_seq) && 565 SEQ_GEQ(q->ipqe_seq + q->ipqe_len, pkt_seq + pkt_len)) { 566 tcpstat.tcps_rcvduppack++; 567 tcpstat.tcps_rcvdupbyte += pkt_len; 568 tcp_new_dsack(tp, pkt_seq, pkt_len); 569 m_freem(m); 570 if (tiqe != NULL) { 571 tcpipqent_free(tiqe); 572 } 573 TCP_REASS_COUNTER_INCR(&tcp_reass_segdup); 574 return (0); 575 } 576 /* 577 * Received segment completely overlaps this fragment 578 * so we drop the fragment (this keeps the temporal 579 * ordering of segments correct). 580 */ 581 if (SEQ_GEQ(q->ipqe_seq, pkt_seq) && 582 SEQ_LEQ(q->ipqe_seq + q->ipqe_len, pkt_seq + pkt_len)) { 583 rcvpartdupbyte += q->ipqe_len; 584 m_freem(q->ipqe_m); 585 TCP_REASS_COUNTER_INCR(&tcp_reass_fragdup); 586 goto free_ipqe; 587 } 588 /* 589 * RX'ed segment extends past the end of the 590 * fragment. Drop the overlapping bytes. Then 591 * merge the fragment and segment then treat as 592 * a longer received packet. 593 */ 594 if (SEQ_LT(q->ipqe_seq, pkt_seq) && 595 SEQ_GT(q->ipqe_seq + q->ipqe_len, pkt_seq)) { 596 int overlap = q->ipqe_seq + q->ipqe_len - pkt_seq; 597 #ifdef TCPREASS_DEBUG 598 printf("tcp_reass[%p]: trim starting %d bytes of %u:%u(%u)\n", 599 tp, overlap, 600 pkt_seq, pkt_seq + pkt_len, pkt_len); 601 #endif 602 m_adj(m, overlap); 603 rcvpartdupbyte += overlap; 604 m_cat(q->ipre_mlast, m); 605 TRAVERSE(q->ipre_mlast); 606 m = q->ipqe_m; 607 pkt_seq = q->ipqe_seq; 608 pkt_len += q->ipqe_len - overlap; 609 rcvoobyte -= overlap; 610 TCP_REASS_COUNTER_INCR(&tcp_reass_overlaptail); 611 goto free_ipqe; 612 } 613 /* 614 * RX'ed segment extends past the front of the 615 * fragment. Drop the overlapping bytes on the 616 * received packet. The packet will then be 617 * contatentated with this fragment a bit later. 618 */ 619 if (SEQ_GT(q->ipqe_seq, pkt_seq) && 620 SEQ_LT(q->ipqe_seq, pkt_seq + pkt_len)) { 621 int overlap = pkt_seq + pkt_len - q->ipqe_seq; 622 #ifdef TCPREASS_DEBUG 623 printf("tcp_reass[%p]: trim trailing %d bytes of %u:%u(%u)\n", 624 tp, overlap, 625 pkt_seq, pkt_seq + pkt_len, pkt_len); 626 #endif 627 m_adj(m, -overlap); 628 pkt_len -= overlap; 629 rcvpartdupbyte += overlap; 630 TCP_REASS_COUNTER_INCR(&tcp_reass_overlapfront); 631 rcvoobyte -= overlap; 632 } 633 /* 634 * If the received segment immediates precedes this 635 * fragment then tack the fragment onto this segment 636 * and reinsert the data. 637 */ 638 if (q->ipqe_seq == pkt_seq + pkt_len) { 639 #ifdef TCPREASS_DEBUG 640 printf("tcp_reass[%p]: append %u:%u(%u) to %u:%u(%u)\n", 641 tp, q->ipqe_seq, q->ipqe_seq + q->ipqe_len, q->ipqe_len, 642 pkt_seq, pkt_seq + pkt_len, pkt_len); 643 #endif 644 pkt_len += q->ipqe_len; 645 pkt_flags |= q->ipqe_flags; 646 m_cat(m, q->ipqe_m); 647 TAILQ_REMOVE(&tp->segq, q, ipqe_q); 648 TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq); 649 tp->t_segqlen--; 650 KASSERT(tp->t_segqlen >= 0); 651 KASSERT(tp->t_segqlen != 0 || 652 (TAILQ_EMPTY(&tp->segq) && 653 TAILQ_EMPTY(&tp->timeq))); 654 if (tiqe == NULL) { 655 tiqe = q; 656 } else { 657 tcpipqent_free(q); 658 } 659 TCP_REASS_COUNTER_INCR(&tcp_reass_prepend); 660 break; 661 } 662 /* 663 * If the fragment is before the segment, remember it. 664 * When this loop is terminated, p will contain the 665 * pointer to fragment that is right before the received 666 * segment. 667 */ 668 if (SEQ_LEQ(q->ipqe_seq, pkt_seq)) 669 p = q; 670 671 continue; 672 673 /* 674 * This is a common operation. It also will allow 675 * to save doing a malloc/free in most instances. 676 */ 677 free_ipqe: 678 TAILQ_REMOVE(&tp->segq, q, ipqe_q); 679 TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq); 680 tp->t_segqlen--; 681 KASSERT(tp->t_segqlen >= 0); 682 KASSERT(tp->t_segqlen != 0 || 683 (TAILQ_EMPTY(&tp->segq) && TAILQ_EMPTY(&tp->timeq))); 684 if (tiqe == NULL) { 685 tiqe = q; 686 } else { 687 tcpipqent_free(q); 688 } 689 } 690 691 #ifdef TCP_REASS_COUNTERS 692 if (count > 7) 693 TCP_REASS_COUNTER_INCR(&tcp_reass_iteration[0]); 694 else if (count > 0) 695 TCP_REASS_COUNTER_INCR(&tcp_reass_iteration[count]); 696 #endif 697 698 insert_it: 699 700 /* 701 * Allocate a new queue entry since the received segment did not 702 * collapse onto any other out-of-order block; thus we are allocating 703 * a new block. If it had collapsed, tiqe would not be NULL and 704 * we would be reusing it. 705 * XXX If we can't, just drop the packet. XXX 706 */ 707 if (tiqe == NULL) { 708 tiqe = tcpipqent_alloc(); 709 if (tiqe == NULL) { 710 tcpstat.tcps_rcvmemdrop++; 711 m_freem(m); 712 return (0); 713 } 714 } 715 716 /* 717 * Update the counters. 718 */ 719 tcpstat.tcps_rcvoopack++; 720 tcpstat.tcps_rcvoobyte += rcvoobyte; 721 if (rcvpartdupbyte) { 722 tcpstat.tcps_rcvpartduppack++; 723 tcpstat.tcps_rcvpartdupbyte += rcvpartdupbyte; 724 } 725 726 /* 727 * Insert the new fragment queue entry into both queues. 728 */ 729 tiqe->ipqe_m = m; 730 tiqe->ipre_mlast = m; 731 tiqe->ipqe_seq = pkt_seq; 732 tiqe->ipqe_len = pkt_len; 733 tiqe->ipqe_flags = pkt_flags; 734 if (p == NULL) { 735 TAILQ_INSERT_HEAD(&tp->segq, tiqe, ipqe_q); 736 #ifdef TCPREASS_DEBUG 737 if (tiqe->ipqe_seq != tp->rcv_nxt) 738 printf("tcp_reass[%p]: insert %u:%u(%u) at front\n", 739 tp, pkt_seq, pkt_seq + pkt_len, pkt_len); 740 #endif 741 } else { 742 TAILQ_INSERT_AFTER(&tp->segq, p, tiqe, ipqe_q); 743 #ifdef TCPREASS_DEBUG 744 printf("tcp_reass[%p]: insert %u:%u(%u) after %u:%u(%u)\n", 745 tp, pkt_seq, pkt_seq + pkt_len, pkt_len, 746 p->ipqe_seq, p->ipqe_seq + p->ipqe_len, p->ipqe_len); 747 #endif 748 } 749 tp->t_segqlen++; 750 751 skip_replacement: 752 753 TAILQ_INSERT_HEAD(&tp->timeq, tiqe, ipqe_timeq); 754 755 present: 756 /* 757 * Present data to user, advancing rcv_nxt through 758 * completed sequence space. 759 */ 760 if (TCPS_HAVEESTABLISHED(tp->t_state) == 0) 761 return (0); 762 q = TAILQ_FIRST(&tp->segq); 763 if (q == NULL || q->ipqe_seq != tp->rcv_nxt) 764 return (0); 765 if (tp->t_state == TCPS_SYN_RECEIVED && q->ipqe_len) 766 return (0); 767 768 tp->rcv_nxt += q->ipqe_len; 769 pkt_flags = q->ipqe_flags & TH_FIN; 770 ND6_HINT(tp); 771 772 TAILQ_REMOVE(&tp->segq, q, ipqe_q); 773 TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq); 774 tp->t_segqlen--; 775 KASSERT(tp->t_segqlen >= 0); 776 KASSERT(tp->t_segqlen != 0 || 777 (TAILQ_EMPTY(&tp->segq) && TAILQ_EMPTY(&tp->timeq))); 778 if (so->so_state & SS_CANTRCVMORE) 779 m_freem(q->ipqe_m); 780 else 781 sbappendstream(&so->so_rcv, q->ipqe_m); 782 tcpipqent_free(q); 783 sorwakeup(so); 784 return (pkt_flags); 785 } 786 787 #ifdef INET6 788 int 789 tcp6_input(struct mbuf **mp, int *offp, int proto) 790 { 791 struct mbuf *m = *mp; 792 793 /* 794 * draft-itojun-ipv6-tcp-to-anycast 795 * better place to put this in? 796 */ 797 if (m->m_flags & M_ANYCAST6) { 798 struct ip6_hdr *ip6; 799 if (m->m_len < sizeof(struct ip6_hdr)) { 800 if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) { 801 tcpstat.tcps_rcvshort++; 802 return IPPROTO_DONE; 803 } 804 } 805 ip6 = mtod(m, struct ip6_hdr *); 806 icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR, 807 (caddr_t)&ip6->ip6_dst - (caddr_t)ip6); 808 return IPPROTO_DONE; 809 } 810 811 tcp_input(m, *offp, proto); 812 return IPPROTO_DONE; 813 } 814 #endif 815 816 #ifdef INET 817 static void 818 tcp4_log_refused(const struct ip *ip, const struct tcphdr *th) 819 { 820 char src[4*sizeof "123"]; 821 char dst[4*sizeof "123"]; 822 823 if (ip) { 824 strlcpy(src, inet_ntoa(ip->ip_src), sizeof(src)); 825 strlcpy(dst, inet_ntoa(ip->ip_dst), sizeof(dst)); 826 } 827 else { 828 strlcpy(src, "(unknown)", sizeof(src)); 829 strlcpy(dst, "(unknown)", sizeof(dst)); 830 } 831 log(LOG_INFO, 832 "Connection attempt to TCP %s:%d from %s:%d\n", 833 dst, ntohs(th->th_dport), 834 src, ntohs(th->th_sport)); 835 } 836 #endif 837 838 #ifdef INET6 839 static void 840 tcp6_log_refused(const struct ip6_hdr *ip6, const struct tcphdr *th) 841 { 842 char src[INET6_ADDRSTRLEN]; 843 char dst[INET6_ADDRSTRLEN]; 844 845 if (ip6) { 846 strlcpy(src, ip6_sprintf(&ip6->ip6_src), sizeof(src)); 847 strlcpy(dst, ip6_sprintf(&ip6->ip6_dst), sizeof(dst)); 848 } 849 else { 850 strlcpy(src, "(unknown v6)", sizeof(src)); 851 strlcpy(dst, "(unknown v6)", sizeof(dst)); 852 } 853 log(LOG_INFO, 854 "Connection attempt to TCP [%s]:%d from [%s]:%d\n", 855 dst, ntohs(th->th_dport), 856 src, ntohs(th->th_sport)); 857 } 858 #endif 859 860 /* 861 * Checksum extended TCP header and data. 862 */ 863 int 864 tcp_input_checksum(int af, struct mbuf *m, const struct tcphdr *th, 865 int toff, int off, int tlen) 866 { 867 868 /* 869 * XXX it's better to record and check if this mbuf is 870 * already checked. 871 */ 872 873 switch (af) { 874 #ifdef INET 875 case AF_INET: 876 switch (m->m_pkthdr.csum_flags & 877 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_TCPv4) | 878 M_CSUM_TCP_UDP_BAD | M_CSUM_DATA)) { 879 case M_CSUM_TCPv4|M_CSUM_TCP_UDP_BAD: 880 TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_bad); 881 goto badcsum; 882 883 case M_CSUM_TCPv4|M_CSUM_DATA: { 884 u_int32_t hw_csum = m->m_pkthdr.csum_data; 885 886 TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_data); 887 if (m->m_pkthdr.csum_flags & M_CSUM_NO_PSEUDOHDR) { 888 const struct ip *ip = 889 mtod(m, const struct ip *); 890 891 hw_csum = in_cksum_phdr(ip->ip_src.s_addr, 892 ip->ip_dst.s_addr, 893 htons(hw_csum + tlen + off + IPPROTO_TCP)); 894 } 895 if ((hw_csum ^ 0xffff) != 0) 896 goto badcsum; 897 break; 898 } 899 900 case M_CSUM_TCPv4: 901 /* Checksum was okay. */ 902 TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_ok); 903 break; 904 905 default: 906 /* 907 * Must compute it ourselves. Maybe skip checksum 908 * on loopback interfaces. 909 */ 910 if (__predict_true(!(m->m_pkthdr.rcvif->if_flags & 911 IFF_LOOPBACK) || 912 tcp_do_loopback_cksum)) { 913 TCP_CSUM_COUNTER_INCR(&tcp_swcsum); 914 if (in4_cksum(m, IPPROTO_TCP, toff, 915 tlen + off) != 0) 916 goto badcsum; 917 } 918 break; 919 } 920 break; 921 #endif /* INET4 */ 922 923 #ifdef INET6 924 case AF_INET6: 925 switch (m->m_pkthdr.csum_flags & 926 ((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_TCPv6) | 927 M_CSUM_TCP_UDP_BAD | M_CSUM_DATA)) { 928 case M_CSUM_TCPv6|M_CSUM_TCP_UDP_BAD: 929 TCP_CSUM_COUNTER_INCR(&tcp6_hwcsum_bad); 930 goto badcsum; 931 932 #if 0 /* notyet */ 933 case M_CSUM_TCPv6|M_CSUM_DATA: 934 #endif 935 936 case M_CSUM_TCPv6: 937 /* Checksum was okay. */ 938 TCP_CSUM_COUNTER_INCR(&tcp6_hwcsum_ok); 939 break; 940 941 default: 942 /* 943 * Must compute it ourselves. Maybe skip checksum 944 * on loopback interfaces. 945 */ 946 if (__predict_true((m->m_flags & M_LOOP) == 0 || 947 tcp_do_loopback_cksum)) { 948 TCP_CSUM_COUNTER_INCR(&tcp6_swcsum); 949 if (in6_cksum(m, IPPROTO_TCP, toff, 950 tlen + off) != 0) 951 goto badcsum; 952 } 953 } 954 break; 955 #endif /* INET6 */ 956 } 957 958 return 0; 959 960 badcsum: 961 tcpstat.tcps_rcvbadsum++; 962 return -1; 963 } 964 965 /* 966 * TCP input routine, follows pages 65-76 of RFC 793 very closely. 967 */ 968 void 969 tcp_input(struct mbuf *m, ...) 970 { 971 struct tcphdr *th; 972 struct ip *ip; 973 struct inpcb *inp; 974 #ifdef INET6 975 struct ip6_hdr *ip6; 976 struct in6pcb *in6p; 977 #endif 978 u_int8_t *optp = NULL; 979 int optlen = 0; 980 int len, tlen, toff, hdroptlen = 0; 981 struct tcpcb *tp = 0; 982 int tiflags; 983 struct socket *so = NULL; 984 int todrop, dupseg, acked, ourfinisacked, needoutput = 0; 985 #ifdef TCP_DEBUG 986 short ostate = 0; 987 #endif 988 u_long tiwin; 989 struct tcp_opt_info opti; 990 int off, iphlen; 991 va_list ap; 992 int af; /* af on the wire */ 993 struct mbuf *tcp_saveti = NULL; 994 uint32_t ts_rtt; 995 uint8_t iptos; 996 997 MCLAIM(m, &tcp_rx_mowner); 998 va_start(ap, m); 999 toff = va_arg(ap, int); 1000 (void)va_arg(ap, int); /* ignore value, advance ap */ 1001 va_end(ap); 1002 1003 tcpstat.tcps_rcvtotal++; 1004 1005 bzero(&opti, sizeof(opti)); 1006 opti.ts_present = 0; 1007 opti.maxseg = 0; 1008 1009 /* 1010 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN. 1011 * 1012 * TCP is, by definition, unicast, so we reject all 1013 * multicast outright. 1014 * 1015 * Note, there are additional src/dst address checks in 1016 * the AF-specific code below. 1017 */ 1018 if (m->m_flags & (M_BCAST|M_MCAST)) { 1019 /* XXX stat */ 1020 goto drop; 1021 } 1022 #ifdef INET6 1023 if (m->m_flags & M_ANYCAST6) { 1024 /* XXX stat */ 1025 goto drop; 1026 } 1027 #endif 1028 1029 /* 1030 * Get IP and TCP header. 1031 * Note: IP leaves IP header in first mbuf. 1032 */ 1033 ip = mtod(m, struct ip *); 1034 #ifdef INET6 1035 ip6 = NULL; 1036 #endif 1037 switch (ip->ip_v) { 1038 #ifdef INET 1039 case 4: 1040 af = AF_INET; 1041 iphlen = sizeof(struct ip); 1042 ip = mtod(m, struct ip *); 1043 IP6_EXTHDR_GET(th, struct tcphdr *, m, toff, 1044 sizeof(struct tcphdr)); 1045 if (th == NULL) { 1046 tcpstat.tcps_rcvshort++; 1047 return; 1048 } 1049 /* We do the checksum after PCB lookup... */ 1050 len = ntohs(ip->ip_len); 1051 tlen = len - toff; 1052 iptos = ip->ip_tos; 1053 break; 1054 #endif 1055 #ifdef INET6 1056 case 6: 1057 ip = NULL; 1058 iphlen = sizeof(struct ip6_hdr); 1059 af = AF_INET6; 1060 ip6 = mtod(m, struct ip6_hdr *); 1061 IP6_EXTHDR_GET(th, struct tcphdr *, m, toff, 1062 sizeof(struct tcphdr)); 1063 if (th == NULL) { 1064 tcpstat.tcps_rcvshort++; 1065 return; 1066 } 1067 1068 /* Be proactive about malicious use of IPv4 mapped address */ 1069 if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) || 1070 IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) { 1071 /* XXX stat */ 1072 goto drop; 1073 } 1074 1075 /* 1076 * Be proactive about unspecified IPv6 address in source. 1077 * As we use all-zero to indicate unbounded/unconnected pcb, 1078 * unspecified IPv6 address can be used to confuse us. 1079 * 1080 * Note that packets with unspecified IPv6 destination is 1081 * already dropped in ip6_input. 1082 */ 1083 if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) { 1084 /* XXX stat */ 1085 goto drop; 1086 } 1087 1088 /* 1089 * Make sure destination address is not multicast. 1090 * Source address checked in ip6_input(). 1091 */ 1092 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) { 1093 /* XXX stat */ 1094 goto drop; 1095 } 1096 1097 /* We do the checksum after PCB lookup... */ 1098 len = m->m_pkthdr.len; 1099 tlen = len - toff; 1100 iptos = (ntohl(ip6->ip6_flow) >> 20) & 0xff; 1101 break; 1102 #endif 1103 default: 1104 m_freem(m); 1105 return; 1106 } 1107 1108 KASSERT(TCP_HDR_ALIGNED_P(th)); 1109 1110 /* 1111 * Check that TCP offset makes sense, 1112 * pull out TCP options and adjust length. XXX 1113 */ 1114 off = th->th_off << 2; 1115 if (off < sizeof (struct tcphdr) || off > tlen) { 1116 tcpstat.tcps_rcvbadoff++; 1117 goto drop; 1118 } 1119 tlen -= off; 1120 1121 /* 1122 * tcp_input() has been modified to use tlen to mean the TCP data 1123 * length throughout the function. Other functions can use 1124 * m->m_pkthdr.len as the basis for calculating the TCP data length. 1125 * rja 1126 */ 1127 1128 if (off > sizeof (struct tcphdr)) { 1129 IP6_EXTHDR_GET(th, struct tcphdr *, m, toff, off); 1130 if (th == NULL) { 1131 tcpstat.tcps_rcvshort++; 1132 return; 1133 } 1134 /* 1135 * NOTE: ip/ip6 will not be affected by m_pulldown() 1136 * (as they're before toff) and we don't need to update those. 1137 */ 1138 KASSERT(TCP_HDR_ALIGNED_P(th)); 1139 optlen = off - sizeof (struct tcphdr); 1140 optp = ((u_int8_t *)th) + sizeof(struct tcphdr); 1141 /* 1142 * Do quick retrieval of timestamp options ("options 1143 * prediction?"). If timestamp is the only option and it's 1144 * formatted as recommended in RFC 1323 appendix A, we 1145 * quickly get the values now and not bother calling 1146 * tcp_dooptions(), etc. 1147 */ 1148 if ((optlen == TCPOLEN_TSTAMP_APPA || 1149 (optlen > TCPOLEN_TSTAMP_APPA && 1150 optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) && 1151 *(u_int32_t *)optp == htonl(TCPOPT_TSTAMP_HDR) && 1152 (th->th_flags & TH_SYN) == 0) { 1153 opti.ts_present = 1; 1154 opti.ts_val = ntohl(*(u_int32_t *)(optp + 4)); 1155 opti.ts_ecr = ntohl(*(u_int32_t *)(optp + 8)); 1156 optp = NULL; /* we've parsed the options */ 1157 } 1158 } 1159 tiflags = th->th_flags; 1160 1161 /* 1162 * Locate pcb for segment. 1163 */ 1164 findpcb: 1165 inp = NULL; 1166 #ifdef INET6 1167 in6p = NULL; 1168 #endif 1169 switch (af) { 1170 #ifdef INET 1171 case AF_INET: 1172 inp = in_pcblookup_connect(&tcbtable, ip->ip_src, th->th_sport, 1173 ip->ip_dst, th->th_dport); 1174 if (inp == 0) { 1175 ++tcpstat.tcps_pcbhashmiss; 1176 inp = in_pcblookup_bind(&tcbtable, ip->ip_dst, th->th_dport); 1177 } 1178 #ifdef INET6 1179 if (inp == 0) { 1180 struct in6_addr s, d; 1181 1182 /* mapped addr case */ 1183 bzero(&s, sizeof(s)); 1184 s.s6_addr16[5] = htons(0xffff); 1185 bcopy(&ip->ip_src, &s.s6_addr32[3], sizeof(ip->ip_src)); 1186 bzero(&d, sizeof(d)); 1187 d.s6_addr16[5] = htons(0xffff); 1188 bcopy(&ip->ip_dst, &d.s6_addr32[3], sizeof(ip->ip_dst)); 1189 in6p = in6_pcblookup_connect(&tcbtable, &s, 1190 th->th_sport, &d, th->th_dport, 0); 1191 if (in6p == 0) { 1192 ++tcpstat.tcps_pcbhashmiss; 1193 in6p = in6_pcblookup_bind(&tcbtable, &d, 1194 th->th_dport, 0); 1195 } 1196 } 1197 #endif 1198 #ifndef INET6 1199 if (inp == 0) 1200 #else 1201 if (inp == 0 && in6p == 0) 1202 #endif 1203 { 1204 ++tcpstat.tcps_noport; 1205 if (tcp_log_refused && 1206 (tiflags & (TH_RST|TH_ACK|TH_SYN)) == TH_SYN) { 1207 tcp4_log_refused(ip, th); 1208 } 1209 TCP_FIELDS_TO_HOST(th); 1210 goto dropwithreset_ratelim; 1211 } 1212 #if defined(IPSEC) || defined(FAST_IPSEC) 1213 if (inp && (inp->inp_socket->so_options & SO_ACCEPTCONN) == 0 && 1214 ipsec4_in_reject(m, inp)) { 1215 ipsecstat.in_polvio++; 1216 goto drop; 1217 } 1218 #ifdef INET6 1219 else if (in6p && 1220 (in6p->in6p_socket->so_options & SO_ACCEPTCONN) == 0 && 1221 ipsec4_in_reject_so(m, in6p->in6p_socket)) { 1222 ipsecstat.in_polvio++; 1223 goto drop; 1224 } 1225 #endif 1226 #endif /*IPSEC*/ 1227 break; 1228 #endif /*INET*/ 1229 #ifdef INET6 1230 case AF_INET6: 1231 { 1232 int faith; 1233 1234 #if defined(NFAITH) && NFAITH > 0 1235 faith = faithprefix(&ip6->ip6_dst); 1236 #else 1237 faith = 0; 1238 #endif 1239 in6p = in6_pcblookup_connect(&tcbtable, &ip6->ip6_src, 1240 th->th_sport, &ip6->ip6_dst, th->th_dport, faith); 1241 if (in6p == NULL) { 1242 ++tcpstat.tcps_pcbhashmiss; 1243 in6p = in6_pcblookup_bind(&tcbtable, &ip6->ip6_dst, 1244 th->th_dport, faith); 1245 } 1246 if (in6p == NULL) { 1247 ++tcpstat.tcps_noport; 1248 if (tcp_log_refused && 1249 (tiflags & (TH_RST|TH_ACK|TH_SYN)) == TH_SYN) { 1250 tcp6_log_refused(ip6, th); 1251 } 1252 TCP_FIELDS_TO_HOST(th); 1253 goto dropwithreset_ratelim; 1254 } 1255 #if defined(IPSEC) || defined(FAST_IPSEC) 1256 if ((in6p->in6p_socket->so_options & SO_ACCEPTCONN) == 0 && 1257 ipsec6_in_reject(m, in6p)) { 1258 ipsec6stat.in_polvio++; 1259 goto drop; 1260 } 1261 #endif /*IPSEC*/ 1262 break; 1263 } 1264 #endif 1265 } 1266 1267 /* 1268 * If the state is CLOSED (i.e., TCB does not exist) then 1269 * all data in the incoming segment is discarded. 1270 * If the TCB exists but is in CLOSED state, it is embryonic, 1271 * but should either do a listen or a connect soon. 1272 */ 1273 tp = NULL; 1274 so = NULL; 1275 if (inp) { 1276 tp = intotcpcb(inp); 1277 so = inp->inp_socket; 1278 } 1279 #ifdef INET6 1280 else if (in6p) { 1281 tp = in6totcpcb(in6p); 1282 so = in6p->in6p_socket; 1283 } 1284 #endif 1285 if (tp == 0) { 1286 TCP_FIELDS_TO_HOST(th); 1287 goto dropwithreset_ratelim; 1288 } 1289 if (tp->t_state == TCPS_CLOSED) 1290 goto drop; 1291 1292 /* 1293 * Checksum extended TCP header and data. 1294 */ 1295 if (tcp_input_checksum(af, m, th, toff, off, tlen)) 1296 goto badcsum; 1297 1298 TCP_FIELDS_TO_HOST(th); 1299 1300 /* Unscale the window into a 32-bit value. */ 1301 if ((tiflags & TH_SYN) == 0) 1302 tiwin = th->th_win << tp->snd_scale; 1303 else 1304 tiwin = th->th_win; 1305 1306 #ifdef INET6 1307 /* save packet options if user wanted */ 1308 if (in6p && (in6p->in6p_flags & IN6P_CONTROLOPTS)) { 1309 if (in6p->in6p_options) { 1310 m_freem(in6p->in6p_options); 1311 in6p->in6p_options = 0; 1312 } 1313 KASSERT(ip6 != NULL); 1314 ip6_savecontrol(in6p, &in6p->in6p_options, ip6, m); 1315 } 1316 #endif 1317 1318 if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) { 1319 union syn_cache_sa src; 1320 union syn_cache_sa dst; 1321 1322 bzero(&src, sizeof(src)); 1323 bzero(&dst, sizeof(dst)); 1324 switch (af) { 1325 #ifdef INET 1326 case AF_INET: 1327 src.sin.sin_len = sizeof(struct sockaddr_in); 1328 src.sin.sin_family = AF_INET; 1329 src.sin.sin_addr = ip->ip_src; 1330 src.sin.sin_port = th->th_sport; 1331 1332 dst.sin.sin_len = sizeof(struct sockaddr_in); 1333 dst.sin.sin_family = AF_INET; 1334 dst.sin.sin_addr = ip->ip_dst; 1335 dst.sin.sin_port = th->th_dport; 1336 break; 1337 #endif 1338 #ifdef INET6 1339 case AF_INET6: 1340 src.sin6.sin6_len = sizeof(struct sockaddr_in6); 1341 src.sin6.sin6_family = AF_INET6; 1342 src.sin6.sin6_addr = ip6->ip6_src; 1343 src.sin6.sin6_port = th->th_sport; 1344 1345 dst.sin6.sin6_len = sizeof(struct sockaddr_in6); 1346 dst.sin6.sin6_family = AF_INET6; 1347 dst.sin6.sin6_addr = ip6->ip6_dst; 1348 dst.sin6.sin6_port = th->th_dport; 1349 break; 1350 #endif /* INET6 */ 1351 default: 1352 goto badsyn; /*sanity*/ 1353 } 1354 1355 if (so->so_options & SO_DEBUG) { 1356 #ifdef TCP_DEBUG 1357 ostate = tp->t_state; 1358 #endif 1359 1360 tcp_saveti = NULL; 1361 if (iphlen + sizeof(struct tcphdr) > MHLEN) 1362 goto nosave; 1363 1364 if (m->m_len > iphlen && (m->m_flags & M_EXT) == 0) { 1365 tcp_saveti = m_copym(m, 0, iphlen, M_DONTWAIT); 1366 if (!tcp_saveti) 1367 goto nosave; 1368 } else { 1369 MGETHDR(tcp_saveti, M_DONTWAIT, MT_HEADER); 1370 if (!tcp_saveti) 1371 goto nosave; 1372 MCLAIM(m, &tcp_mowner); 1373 tcp_saveti->m_len = iphlen; 1374 m_copydata(m, 0, iphlen, 1375 mtod(tcp_saveti, caddr_t)); 1376 } 1377 1378 if (M_TRAILINGSPACE(tcp_saveti) < sizeof(struct tcphdr)) { 1379 m_freem(tcp_saveti); 1380 tcp_saveti = NULL; 1381 } else { 1382 tcp_saveti->m_len += sizeof(struct tcphdr); 1383 bcopy(th, mtod(tcp_saveti, caddr_t) + iphlen, 1384 sizeof(struct tcphdr)); 1385 } 1386 nosave:; 1387 } 1388 if (so->so_options & SO_ACCEPTCONN) { 1389 if ((tiflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) { 1390 if (tiflags & TH_RST) { 1391 syn_cache_reset(&src.sa, &dst.sa, th); 1392 } else if ((tiflags & (TH_ACK|TH_SYN)) == 1393 (TH_ACK|TH_SYN)) { 1394 /* 1395 * Received a SYN,ACK. This should 1396 * never happen while we are in 1397 * LISTEN. Send an RST. 1398 */ 1399 goto badsyn; 1400 } else if (tiflags & TH_ACK) { 1401 so = syn_cache_get(&src.sa, &dst.sa, 1402 th, toff, tlen, so, m); 1403 if (so == NULL) { 1404 /* 1405 * We don't have a SYN for 1406 * this ACK; send an RST. 1407 */ 1408 goto badsyn; 1409 } else if (so == 1410 (struct socket *)(-1)) { 1411 /* 1412 * We were unable to create 1413 * the connection. If the 1414 * 3-way handshake was 1415 * completed, and RST has 1416 * been sent to the peer. 1417 * Since the mbuf might be 1418 * in use for the reply, 1419 * do not free it. 1420 */ 1421 m = NULL; 1422 } else { 1423 /* 1424 * We have created a 1425 * full-blown connection. 1426 */ 1427 tp = NULL; 1428 inp = NULL; 1429 #ifdef INET6 1430 in6p = NULL; 1431 #endif 1432 switch (so->so_proto->pr_domain->dom_family) { 1433 #ifdef INET 1434 case AF_INET: 1435 inp = sotoinpcb(so); 1436 tp = intotcpcb(inp); 1437 break; 1438 #endif 1439 #ifdef INET6 1440 case AF_INET6: 1441 in6p = sotoin6pcb(so); 1442 tp = in6totcpcb(in6p); 1443 break; 1444 #endif 1445 } 1446 if (tp == NULL) 1447 goto badsyn; /*XXX*/ 1448 tiwin <<= tp->snd_scale; 1449 goto after_listen; 1450 } 1451 } else { 1452 /* 1453 * None of RST, SYN or ACK was set. 1454 * This is an invalid packet for a 1455 * TCB in LISTEN state. Send a RST. 1456 */ 1457 goto badsyn; 1458 } 1459 } else { 1460 /* 1461 * Received a SYN. 1462 * 1463 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN 1464 */ 1465 if (m->m_flags & (M_BCAST|M_MCAST)) 1466 goto drop; 1467 1468 switch (af) { 1469 #ifdef INET6 1470 case AF_INET6: 1471 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) 1472 goto drop; 1473 break; 1474 #endif /* INET6 */ 1475 case AF_INET: 1476 if (IN_MULTICAST(ip->ip_dst.s_addr) || 1477 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) 1478 goto drop; 1479 break; 1480 } 1481 1482 #ifdef INET6 1483 /* 1484 * If deprecated address is forbidden, we do 1485 * not accept SYN to deprecated interface 1486 * address to prevent any new inbound 1487 * connection from getting established. 1488 * When we do not accept SYN, we send a TCP 1489 * RST, with deprecated source address (instead 1490 * of dropping it). We compromise it as it is 1491 * much better for peer to send a RST, and 1492 * RST will be the final packet for the 1493 * exchange. 1494 * 1495 * If we do not forbid deprecated addresses, we 1496 * accept the SYN packet. RFC2462 does not 1497 * suggest dropping SYN in this case. 1498 * If we decipher RFC2462 5.5.4, it says like 1499 * this: 1500 * 1. use of deprecated addr with existing 1501 * communication is okay - "SHOULD continue 1502 * to be used" 1503 * 2. use of it with new communication: 1504 * (2a) "SHOULD NOT be used if alternate 1505 * address with sufficient scope is 1506 * available" 1507 * (2b) nothing mentioned otherwise. 1508 * Here we fall into (2b) case as we have no 1509 * choice in our source address selection - we 1510 * must obey the peer. 1511 * 1512 * The wording in RFC2462 is confusing, and 1513 * there are multiple description text for 1514 * deprecated address handling - worse, they 1515 * are not exactly the same. I believe 5.5.4 1516 * is the best one, so we follow 5.5.4. 1517 */ 1518 if (af == AF_INET6 && !ip6_use_deprecated) { 1519 struct in6_ifaddr *ia6; 1520 if ((ia6 = in6ifa_ifpwithaddr(m->m_pkthdr.rcvif, 1521 &ip6->ip6_dst)) && 1522 (ia6->ia6_flags & IN6_IFF_DEPRECATED)) { 1523 tp = NULL; 1524 goto dropwithreset; 1525 } 1526 } 1527 #endif 1528 1529 #ifdef IPSEC 1530 switch (af) { 1531 #ifdef INET 1532 case AF_INET: 1533 if (ipsec4_in_reject_so(m, so)) { 1534 ipsecstat.in_polvio++; 1535 tp = NULL; 1536 goto dropwithreset; 1537 } 1538 break; 1539 #endif 1540 #ifdef INET6 1541 case AF_INET6: 1542 if (ipsec6_in_reject_so(m, so)) { 1543 ipsec6stat.in_polvio++; 1544 tp = NULL; 1545 goto dropwithreset; 1546 } 1547 break; 1548 #endif 1549 } 1550 #endif 1551 1552 /* 1553 * LISTEN socket received a SYN 1554 * from itself? This can't possibly 1555 * be valid; drop the packet. 1556 */ 1557 if (th->th_sport == th->th_dport) { 1558 int i; 1559 1560 switch (af) { 1561 #ifdef INET 1562 case AF_INET: 1563 i = in_hosteq(ip->ip_src, ip->ip_dst); 1564 break; 1565 #endif 1566 #ifdef INET6 1567 case AF_INET6: 1568 i = IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, &ip6->ip6_dst); 1569 break; 1570 #endif 1571 default: 1572 i = 1; 1573 } 1574 if (i) { 1575 tcpstat.tcps_badsyn++; 1576 goto drop; 1577 } 1578 } 1579 1580 /* 1581 * SYN looks ok; create compressed TCP 1582 * state for it. 1583 */ 1584 if (so->so_qlen <= so->so_qlimit && 1585 syn_cache_add(&src.sa, &dst.sa, th, tlen, 1586 so, m, optp, optlen, &opti)) 1587 m = NULL; 1588 } 1589 goto drop; 1590 } 1591 } 1592 1593 after_listen: 1594 #ifdef DIAGNOSTIC 1595 /* 1596 * Should not happen now that all embryonic connections 1597 * are handled with compressed state. 1598 */ 1599 if (tp->t_state == TCPS_LISTEN) 1600 panic("tcp_input: TCPS_LISTEN"); 1601 #endif 1602 1603 /* 1604 * Segment received on connection. 1605 * Reset idle time and keep-alive timer. 1606 */ 1607 tp->t_rcvtime = tcp_now; 1608 if (TCPS_HAVEESTABLISHED(tp->t_state)) 1609 TCP_TIMER_ARM(tp, TCPT_KEEP, tcp_keepidle); 1610 1611 /* 1612 * Process options. 1613 */ 1614 #ifdef TCP_SIGNATURE 1615 if (optp || (tp->t_flags & TF_SIGNATURE)) 1616 #else 1617 if (optp) 1618 #endif 1619 if (tcp_dooptions(tp, optp, optlen, th, m, toff, &opti) < 0) 1620 goto drop; 1621 1622 if (TCP_SACK_ENABLED(tp)) { 1623 tcp_del_sackholes(tp, th); 1624 } 1625 1626 if (TCP_ECN_ALLOWED(tp)) { 1627 switch (iptos & IPTOS_ECN_MASK) { 1628 case IPTOS_ECN_CE: 1629 tp->t_flags |= TF_ECN_SND_ECE; 1630 tcpstat.tcps_ecn_ce++; 1631 break; 1632 case IPTOS_ECN_ECT0: 1633 tcpstat.tcps_ecn_ect++; 1634 break; 1635 case IPTOS_ECN_ECT1: 1636 /* XXX: ignore for now -- rpaulo */ 1637 break; 1638 } 1639 1640 if (tiflags & TH_CWR) 1641 tp->t_flags &= ~TF_ECN_SND_ECE; 1642 1643 /* 1644 * Congestion experienced. 1645 * Ignore if we are already trying to recover. 1646 */ 1647 if ((tiflags & TH_ECE) && SEQ_GEQ(tp->snd_una, tp->snd_recover)) 1648 tp->t_congctl->cong_exp(tp); 1649 } 1650 1651 if (opti.ts_present && opti.ts_ecr) { 1652 /* 1653 * Calculate the RTT from the returned time stamp and the 1654 * connection's time base. If the time stamp is later than 1655 * the current time, or is extremely old, fall back to non-1323 1656 * RTT calculation. Since ts_ecr is unsigned, we can test both 1657 * at the same time. 1658 */ 1659 ts_rtt = TCP_TIMESTAMP(tp) - opti.ts_ecr + 1; 1660 if (ts_rtt > TCP_PAWS_IDLE) 1661 ts_rtt = 0; 1662 } else { 1663 ts_rtt = 0; 1664 } 1665 1666 /* 1667 * Header prediction: check for the two common cases 1668 * of a uni-directional data xfer. If the packet has 1669 * no control flags, is in-sequence, the window didn't 1670 * change and we're not retransmitting, it's a 1671 * candidate. If the length is zero and the ack moved 1672 * forward, we're the sender side of the xfer. Just 1673 * free the data acked & wake any higher level process 1674 * that was blocked waiting for space. If the length 1675 * is non-zero and the ack didn't move, we're the 1676 * receiver side. If we're getting packets in-order 1677 * (the reassembly queue is empty), add the data to 1678 * the socket buffer and note that we need a delayed ack. 1679 */ 1680 if (tp->t_state == TCPS_ESTABLISHED && 1681 (tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ECE|TH_CWR|TH_ACK)) 1682 == TH_ACK && 1683 (!opti.ts_present || TSTMP_GEQ(opti.ts_val, tp->ts_recent)) && 1684 th->th_seq == tp->rcv_nxt && 1685 tiwin && tiwin == tp->snd_wnd && 1686 tp->snd_nxt == tp->snd_max) { 1687 1688 /* 1689 * If last ACK falls within this segment's sequence numbers, 1690 * record the timestamp. 1691 * NOTE: 1692 * 1) That the test incorporates suggestions from the latest 1693 * proposal of the tcplw (at) cray.com list (Braden 1993/04/26). 1694 * 2) That updating only on newer timestamps interferes with 1695 * our earlier PAWS tests, so this check should be solely 1696 * predicated on the sequence space of this segment. 1697 * 3) That we modify the segment boundary check to be 1698 * Last.ACK.Sent <= SEG.SEQ + SEG.Len 1699 * instead of RFC1323's 1700 * Last.ACK.Sent < SEG.SEQ + SEG.Len, 1701 * This modified check allows us to overcome RFC1323's 1702 * limitations as described in Stevens TCP/IP Illustrated 1703 * Vol. 2 p.869. In such cases, we can still calculate the 1704 * RTT correctly when RCV.NXT == Last.ACK.Sent. 1705 */ 1706 if (opti.ts_present && 1707 SEQ_LEQ(th->th_seq, tp->last_ack_sent) && 1708 SEQ_LEQ(tp->last_ack_sent, th->th_seq + tlen + 1709 ((tiflags & (TH_SYN|TH_FIN)) != 0))) { 1710 tp->ts_recent_age = tcp_now; 1711 tp->ts_recent = opti.ts_val; 1712 } 1713 1714 if (tlen == 0) { 1715 /* Ack prediction. */ 1716 if (SEQ_GT(th->th_ack, tp->snd_una) && 1717 SEQ_LEQ(th->th_ack, tp->snd_max) && 1718 tp->snd_cwnd >= tp->snd_wnd && 1719 tp->t_partialacks < 0) { 1720 /* 1721 * this is a pure ack for outstanding data. 1722 */ 1723 ++tcpstat.tcps_predack; 1724 if (ts_rtt) 1725 tcp_xmit_timer(tp, ts_rtt); 1726 else if (tp->t_rtttime && 1727 SEQ_GT(th->th_ack, tp->t_rtseq)) 1728 tcp_xmit_timer(tp, 1729 tcp_now - tp->t_rtttime); 1730 acked = th->th_ack - tp->snd_una; 1731 tcpstat.tcps_rcvackpack++; 1732 tcpstat.tcps_rcvackbyte += acked; 1733 ND6_HINT(tp); 1734 1735 if (acked > (tp->t_lastoff - tp->t_inoff)) 1736 tp->t_lastm = NULL; 1737 sbdrop(&so->so_snd, acked); 1738 tp->t_lastoff -= acked; 1739 1740 ICMP_CHECK(tp, th, acked); 1741 1742 tp->snd_una = th->th_ack; 1743 tp->snd_fack = tp->snd_una; 1744 if (SEQ_LT(tp->snd_high, tp->snd_una)) 1745 tp->snd_high = tp->snd_una; 1746 m_freem(m); 1747 1748 /* 1749 * If all outstanding data are acked, stop 1750 * retransmit timer, otherwise restart timer 1751 * using current (possibly backed-off) value. 1752 * If process is waiting for space, 1753 * wakeup/selwakeup/signal. If data 1754 * are ready to send, let tcp_output 1755 * decide between more output or persist. 1756 */ 1757 if (tp->snd_una == tp->snd_max) 1758 TCP_TIMER_DISARM(tp, TCPT_REXMT); 1759 else if (TCP_TIMER_ISARMED(tp, 1760 TCPT_PERSIST) == 0) 1761 TCP_TIMER_ARM(tp, TCPT_REXMT, 1762 tp->t_rxtcur); 1763 1764 sowwakeup(so); 1765 if (so->so_snd.sb_cc) 1766 (void) tcp_output(tp); 1767 if (tcp_saveti) 1768 m_freem(tcp_saveti); 1769 return; 1770 } 1771 } else if (th->th_ack == tp->snd_una && 1772 TAILQ_FIRST(&tp->segq) == NULL && 1773 tlen <= sbspace(&so->so_rcv)) { 1774 /* 1775 * this is a pure, in-sequence data packet 1776 * with nothing on the reassembly queue and 1777 * we have enough buffer space to take it. 1778 */ 1779 ++tcpstat.tcps_preddat; 1780 tp->rcv_nxt += tlen; 1781 tcpstat.tcps_rcvpack++; 1782 tcpstat.tcps_rcvbyte += tlen; 1783 ND6_HINT(tp); 1784 /* 1785 * Drop TCP, IP headers and TCP options then add data 1786 * to socket buffer. 1787 */ 1788 if (so->so_state & SS_CANTRCVMORE) 1789 m_freem(m); 1790 else { 1791 m_adj(m, toff + off); 1792 sbappendstream(&so->so_rcv, m); 1793 } 1794 sorwakeup(so); 1795 TCP_SETUP_ACK(tp, th); 1796 if (tp->t_flags & TF_ACKNOW) 1797 (void) tcp_output(tp); 1798 if (tcp_saveti) 1799 m_freem(tcp_saveti); 1800 return; 1801 } 1802 } 1803 1804 /* 1805 * Compute mbuf offset to TCP data segment. 1806 */ 1807 hdroptlen = toff + off; 1808 1809 /* 1810 * Calculate amount of space in receive window, 1811 * and then do TCP input processing. 1812 * Receive window is amount of space in rcv queue, 1813 * but not less than advertised window. 1814 */ 1815 { int win; 1816 1817 win = sbspace(&so->so_rcv); 1818 if (win < 0) 1819 win = 0; 1820 tp->rcv_wnd = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt)); 1821 } 1822 1823 switch (tp->t_state) { 1824 /* 1825 * If the state is SYN_SENT: 1826 * if seg contains an ACK, but not for our SYN, drop the input. 1827 * if seg contains a RST, then drop the connection. 1828 * if seg does not contain SYN, then drop it. 1829 * Otherwise this is an acceptable SYN segment 1830 * initialize tp->rcv_nxt and tp->irs 1831 * if seg contains ack then advance tp->snd_una 1832 * if seg contains a ECE and ECN support is enabled, the stream 1833 * is ECN capable. 1834 * if SYN has been acked change to ESTABLISHED else SYN_RCVD state 1835 * arrange for segment to be acked (eventually) 1836 * continue processing rest of data/controls, beginning with URG 1837 */ 1838 case TCPS_SYN_SENT: 1839 if ((tiflags & TH_ACK) && 1840 (SEQ_LEQ(th->th_ack, tp->iss) || 1841 SEQ_GT(th->th_ack, tp->snd_max))) 1842 goto dropwithreset; 1843 if (tiflags & TH_RST) { 1844 if (tiflags & TH_ACK) 1845 tp = tcp_drop(tp, ECONNREFUSED); 1846 goto drop; 1847 } 1848 if ((tiflags & TH_SYN) == 0) 1849 goto drop; 1850 if (tiflags & TH_ACK) { 1851 tp->snd_una = th->th_ack; 1852 if (SEQ_LT(tp->snd_nxt, tp->snd_una)) 1853 tp->snd_nxt = tp->snd_una; 1854 if (SEQ_LT(tp->snd_high, tp->snd_una)) 1855 tp->snd_high = tp->snd_una; 1856 TCP_TIMER_DISARM(tp, TCPT_REXMT); 1857 1858 if ((tiflags & TH_ECE) && tcp_do_ecn) { 1859 tp->t_flags |= TF_ECN_PERMIT; 1860 tcpstat.tcps_ecn_shs++; 1861 } 1862 1863 } 1864 tp->irs = th->th_seq; 1865 tcp_rcvseqinit(tp); 1866 tp->t_flags |= TF_ACKNOW; 1867 tcp_mss_from_peer(tp, opti.maxseg); 1868 1869 /* 1870 * Initialize the initial congestion window. If we 1871 * had to retransmit the SYN, we must initialize cwnd 1872 * to 1 segment (i.e. the Loss Window). 1873 */ 1874 if (tp->t_flags & TF_SYN_REXMT) 1875 tp->snd_cwnd = tp->t_peermss; 1876 else { 1877 int ss = tcp_init_win; 1878 #ifdef INET 1879 if (inp != NULL && in_localaddr(inp->inp_faddr)) 1880 ss = tcp_init_win_local; 1881 #endif 1882 #ifdef INET6 1883 if (in6p != NULL && in6_localaddr(&in6p->in6p_faddr)) 1884 ss = tcp_init_win_local; 1885 #endif 1886 tp->snd_cwnd = TCP_INITIAL_WINDOW(ss, tp->t_peermss); 1887 } 1888 1889 tcp_rmx_rtt(tp); 1890 if (tiflags & TH_ACK) { 1891 tcpstat.tcps_connects++; 1892 soisconnected(so); 1893 tcp_established(tp); 1894 /* Do window scaling on this connection? */ 1895 if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == 1896 (TF_RCVD_SCALE|TF_REQ_SCALE)) { 1897 tp->snd_scale = tp->requested_s_scale; 1898 tp->rcv_scale = tp->request_r_scale; 1899 } 1900 TCP_REASS_LOCK(tp); 1901 (void) tcp_reass(tp, NULL, (struct mbuf *)0, &tlen); 1902 TCP_REASS_UNLOCK(tp); 1903 /* 1904 * if we didn't have to retransmit the SYN, 1905 * use its rtt as our initial srtt & rtt var. 1906 */ 1907 if (tp->t_rtttime) 1908 tcp_xmit_timer(tp, tcp_now - tp->t_rtttime); 1909 } else 1910 tp->t_state = TCPS_SYN_RECEIVED; 1911 1912 /* 1913 * Advance th->th_seq to correspond to first data byte. 1914 * If data, trim to stay within window, 1915 * dropping FIN if necessary. 1916 */ 1917 th->th_seq++; 1918 if (tlen > tp->rcv_wnd) { 1919 todrop = tlen - tp->rcv_wnd; 1920 m_adj(m, -todrop); 1921 tlen = tp->rcv_wnd; 1922 tiflags &= ~TH_FIN; 1923 tcpstat.tcps_rcvpackafterwin++; 1924 tcpstat.tcps_rcvbyteafterwin += todrop; 1925 } 1926 tp->snd_wl1 = th->th_seq - 1; 1927 tp->rcv_up = th->th_seq; 1928 goto step6; 1929 1930 /* 1931 * If the state is SYN_RECEIVED: 1932 * If seg contains an ACK, but not for our SYN, drop the input 1933 * and generate an RST. See page 36, rfc793 1934 */ 1935 case TCPS_SYN_RECEIVED: 1936 if ((tiflags & TH_ACK) && 1937 (SEQ_LEQ(th->th_ack, tp->iss) || 1938 SEQ_GT(th->th_ack, tp->snd_max))) 1939 goto dropwithreset; 1940 break; 1941 } 1942 1943 /* 1944 * States other than LISTEN or SYN_SENT. 1945 * First check timestamp, if present. 1946 * Then check that at least some bytes of segment are within 1947 * receive window. If segment begins before rcv_nxt, 1948 * drop leading data (and SYN); if nothing left, just ack. 1949 * 1950 * RFC 1323 PAWS: If we have a timestamp reply on this segment 1951 * and it's less than ts_recent, drop it. 1952 */ 1953 if (opti.ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent && 1954 TSTMP_LT(opti.ts_val, tp->ts_recent)) { 1955 1956 /* Check to see if ts_recent is over 24 days old. */ 1957 if (tcp_now - tp->ts_recent_age > TCP_PAWS_IDLE) { 1958 /* 1959 * Invalidate ts_recent. If this segment updates 1960 * ts_recent, the age will be reset later and ts_recent 1961 * will get a valid value. If it does not, setting 1962 * ts_recent to zero will at least satisfy the 1963 * requirement that zero be placed in the timestamp 1964 * echo reply when ts_recent isn't valid. The 1965 * age isn't reset until we get a valid ts_recent 1966 * because we don't want out-of-order segments to be 1967 * dropped when ts_recent is old. 1968 */ 1969 tp->ts_recent = 0; 1970 } else { 1971 tcpstat.tcps_rcvduppack++; 1972 tcpstat.tcps_rcvdupbyte += tlen; 1973 tcpstat.tcps_pawsdrop++; 1974 tcp_new_dsack(tp, th->th_seq, tlen); 1975 goto dropafterack; 1976 } 1977 } 1978 1979 todrop = tp->rcv_nxt - th->th_seq; 1980 dupseg = FALSE; 1981 if (todrop > 0) { 1982 if (tiflags & TH_SYN) { 1983 tiflags &= ~TH_SYN; 1984 th->th_seq++; 1985 if (th->th_urp > 1) 1986 th->th_urp--; 1987 else { 1988 tiflags &= ~TH_URG; 1989 th->th_urp = 0; 1990 } 1991 todrop--; 1992 } 1993 if (todrop > tlen || 1994 (todrop == tlen && (tiflags & TH_FIN) == 0)) { 1995 /* 1996 * Any valid FIN or RST must be to the left of the 1997 * window. At this point the FIN or RST must be a 1998 * duplicate or out of sequence; drop it. 1999 */ 2000 if (tiflags & TH_RST) 2001 goto drop; 2002 tiflags &= ~(TH_FIN|TH_RST); 2003 /* 2004 * Send an ACK to resynchronize and drop any data. 2005 * But keep on processing for RST or ACK. 2006 */ 2007 tp->t_flags |= TF_ACKNOW; 2008 todrop = tlen; 2009 dupseg = TRUE; 2010 tcpstat.tcps_rcvdupbyte += todrop; 2011 tcpstat.tcps_rcvduppack++; 2012 } else if ((tiflags & TH_RST) && 2013 th->th_seq != tp->last_ack_sent) { 2014 /* 2015 * Test for reset before adjusting the sequence 2016 * number for overlapping data. 2017 */ 2018 goto dropafterack_ratelim; 2019 } else { 2020 tcpstat.tcps_rcvpartduppack++; 2021 tcpstat.tcps_rcvpartdupbyte += todrop; 2022 } 2023 tcp_new_dsack(tp, th->th_seq, todrop); 2024 hdroptlen += todrop; /*drop from head afterwards*/ 2025 th->th_seq += todrop; 2026 tlen -= todrop; 2027 if (th->th_urp > todrop) 2028 th->th_urp -= todrop; 2029 else { 2030 tiflags &= ~TH_URG; 2031 th->th_urp = 0; 2032 } 2033 } 2034 2035 /* 2036 * If new data are received on a connection after the 2037 * user processes are gone, then RST the other end. 2038 */ 2039 if ((so->so_state & SS_NOFDREF) && 2040 tp->t_state > TCPS_CLOSE_WAIT && tlen) { 2041 tp = tcp_close(tp); 2042 tcpstat.tcps_rcvafterclose++; 2043 goto dropwithreset; 2044 } 2045 2046 /* 2047 * If segment ends after window, drop trailing data 2048 * (and PUSH and FIN); if nothing left, just ACK. 2049 */ 2050 todrop = (th->th_seq + tlen) - (tp->rcv_nxt+tp->rcv_wnd); 2051 if (todrop > 0) { 2052 tcpstat.tcps_rcvpackafterwin++; 2053 if (todrop >= tlen) { 2054 /* 2055 * The segment actually starts after the window. 2056 * th->th_seq + tlen - tp->rcv_nxt - tp->rcv_wnd >= tlen 2057 * th->th_seq - tp->rcv_nxt - tp->rcv_wnd >= 0 2058 * th->th_seq >= tp->rcv_nxt + tp->rcv_wnd 2059 */ 2060 tcpstat.tcps_rcvbyteafterwin += tlen; 2061 /* 2062 * If a new connection request is received 2063 * while in TIME_WAIT, drop the old connection 2064 * and start over if the sequence numbers 2065 * are above the previous ones. 2066 * 2067 * NOTE: We will checksum the packet again, and 2068 * so we need to put the header fields back into 2069 * network order! 2070 * XXX This kind of sucks, but we don't expect 2071 * XXX this to happen very often, so maybe it 2072 * XXX doesn't matter so much. 2073 */ 2074 if (tiflags & TH_SYN && 2075 tp->t_state == TCPS_TIME_WAIT && 2076 SEQ_GT(th->th_seq, tp->rcv_nxt)) { 2077 tp = tcp_close(tp); 2078 TCP_FIELDS_TO_NET(th); 2079 goto findpcb; 2080 } 2081 /* 2082 * If window is closed can only take segments at 2083 * window edge, and have to drop data and PUSH from 2084 * incoming segments. Continue processing, but 2085 * remember to ack. Otherwise, drop segment 2086 * and (if not RST) ack. 2087 */ 2088 if (tp->rcv_wnd == 0 && th->th_seq == tp->rcv_nxt) { 2089 tp->t_flags |= TF_ACKNOW; 2090 tcpstat.tcps_rcvwinprobe++; 2091 } else 2092 goto dropafterack; 2093 } else 2094 tcpstat.tcps_rcvbyteafterwin += todrop; 2095 m_adj(m, -todrop); 2096 tlen -= todrop; 2097 tiflags &= ~(TH_PUSH|TH_FIN); 2098 } 2099 2100 /* 2101 * If last ACK falls within this segment's sequence numbers, 2102 * and the timestamp is newer, record it. 2103 */ 2104 if (opti.ts_present && TSTMP_GEQ(opti.ts_val, tp->ts_recent) && 2105 SEQ_LEQ(th->th_seq, tp->last_ack_sent) && 2106 SEQ_LT(tp->last_ack_sent, th->th_seq + tlen + 2107 ((tiflags & (TH_SYN|TH_FIN)) != 0))) { 2108 tp->ts_recent_age = tcp_now; 2109 tp->ts_recent = opti.ts_val; 2110 } 2111 2112 /* 2113 * If the RST bit is set examine the state: 2114 * SYN_RECEIVED STATE: 2115 * If passive open, return to LISTEN state. 2116 * If active open, inform user that connection was refused. 2117 * ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES: 2118 * Inform user that connection was reset, and close tcb. 2119 * CLOSING, LAST_ACK, TIME_WAIT STATES 2120 * Close the tcb. 2121 */ 2122 if (tiflags & TH_RST) { 2123 if (th->th_seq != tp->last_ack_sent) 2124 goto dropafterack_ratelim; 2125 2126 switch (tp->t_state) { 2127 case TCPS_SYN_RECEIVED: 2128 so->so_error = ECONNREFUSED; 2129 goto close; 2130 2131 case TCPS_ESTABLISHED: 2132 case TCPS_FIN_WAIT_1: 2133 case TCPS_FIN_WAIT_2: 2134 case TCPS_CLOSE_WAIT: 2135 so->so_error = ECONNRESET; 2136 close: 2137 tp->t_state = TCPS_CLOSED; 2138 tcpstat.tcps_drops++; 2139 tp = tcp_close(tp); 2140 goto drop; 2141 2142 case TCPS_CLOSING: 2143 case TCPS_LAST_ACK: 2144 case TCPS_TIME_WAIT: 2145 tp = tcp_close(tp); 2146 goto drop; 2147 } 2148 } 2149 2150 /* 2151 * Since we've covered the SYN-SENT and SYN-RECEIVED states above 2152 * we must be in a synchronized state. RFC791 states (under RST 2153 * generation) that any unacceptable segment (an out-of-order SYN 2154 * qualifies) received in a synchronized state must elicit only an 2155 * empty acknowledgment segment ... and the connection remains in 2156 * the same state. 2157 */ 2158 if (tiflags & TH_SYN) { 2159 if (tp->rcv_nxt == th->th_seq) { 2160 tcp_respond(tp, m, m, th, (tcp_seq)0, th->th_ack - 1, 2161 TH_ACK); 2162 if (tcp_saveti) 2163 m_freem(tcp_saveti); 2164 return; 2165 } 2166 2167 goto dropafterack_ratelim; 2168 } 2169 2170 /* 2171 * If the ACK bit is off we drop the segment and return. 2172 */ 2173 if ((tiflags & TH_ACK) == 0) { 2174 if (tp->t_flags & TF_ACKNOW) 2175 goto dropafterack; 2176 else 2177 goto drop; 2178 } 2179 2180 /* 2181 * Ack processing. 2182 */ 2183 switch (tp->t_state) { 2184 2185 /* 2186 * In SYN_RECEIVED state if the ack ACKs our SYN then enter 2187 * ESTABLISHED state and continue processing, otherwise 2188 * send an RST. 2189 */ 2190 case TCPS_SYN_RECEIVED: 2191 if (SEQ_GT(tp->snd_una, th->th_ack) || 2192 SEQ_GT(th->th_ack, tp->snd_max)) 2193 goto dropwithreset; 2194 tcpstat.tcps_connects++; 2195 soisconnected(so); 2196 tcp_established(tp); 2197 /* Do window scaling? */ 2198 if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == 2199 (TF_RCVD_SCALE|TF_REQ_SCALE)) { 2200 tp->snd_scale = tp->requested_s_scale; 2201 tp->rcv_scale = tp->request_r_scale; 2202 } 2203 TCP_REASS_LOCK(tp); 2204 (void) tcp_reass(tp, NULL, (struct mbuf *)0, &tlen); 2205 TCP_REASS_UNLOCK(tp); 2206 tp->snd_wl1 = th->th_seq - 1; 2207 /* fall into ... */ 2208 2209 /* 2210 * In ESTABLISHED state: drop duplicate ACKs; ACK out of range 2211 * ACKs. If the ack is in the range 2212 * tp->snd_una < th->th_ack <= tp->snd_max 2213 * then advance tp->snd_una to th->th_ack and drop 2214 * data from the retransmission queue. If this ACK reflects 2215 * more up to date window information we update our window information. 2216 */ 2217 case TCPS_ESTABLISHED: 2218 case TCPS_FIN_WAIT_1: 2219 case TCPS_FIN_WAIT_2: 2220 case TCPS_CLOSE_WAIT: 2221 case TCPS_CLOSING: 2222 case TCPS_LAST_ACK: 2223 case TCPS_TIME_WAIT: 2224 2225 if (SEQ_LEQ(th->th_ack, tp->snd_una)) { 2226 if (tlen == 0 && !dupseg && tiwin == tp->snd_wnd) { 2227 tcpstat.tcps_rcvdupack++; 2228 /* 2229 * If we have outstanding data (other than 2230 * a window probe), this is a completely 2231 * duplicate ack (ie, window info didn't 2232 * change), the ack is the biggest we've 2233 * seen and we've seen exactly our rexmt 2234 * threshhold of them, assume a packet 2235 * has been dropped and retransmit it. 2236 * Kludge snd_nxt & the congestion 2237 * window so we send only this one 2238 * packet. 2239 */ 2240 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 || 2241 th->th_ack != tp->snd_una) 2242 tp->t_dupacks = 0; 2243 else if (tp->t_partialacks < 0 && 2244 ((!TCP_SACK_ENABLED(tp) && 2245 ++tp->t_dupacks == tcprexmtthresh) || 2246 TCP_FACK_FASTRECOV(tp))) { 2247 /* 2248 * Do the fast retransmit, and adjust 2249 * congestion control paramenters. 2250 */ 2251 if (tp->t_congctl->fast_retransmit(tp, th)) { 2252 /* False fast retransmit */ 2253 break; 2254 } else 2255 goto drop; 2256 } else if (tp->t_dupacks > tcprexmtthresh) { 2257 tp->snd_cwnd += tp->t_segsz; 2258 (void) tcp_output(tp); 2259 goto drop; 2260 } 2261 } else { 2262 /* 2263 * If the ack appears to be very old, only 2264 * allow data that is in-sequence. This 2265 * makes it somewhat more difficult to insert 2266 * forged data by guessing sequence numbers. 2267 * Sent an ack to try to update the send 2268 * sequence number on the other side. 2269 */ 2270 if (tlen && th->th_seq != tp->rcv_nxt && 2271 SEQ_LT(th->th_ack, 2272 tp->snd_una - tp->max_sndwnd)) 2273 goto dropafterack; 2274 } 2275 break; 2276 } 2277 /* 2278 * If the congestion window was inflated to account 2279 * for the other side's cached packets, retract it. 2280 */ 2281 /* XXX: make SACK have his own congestion control 2282 * struct -- rpaulo */ 2283 if (TCP_SACK_ENABLED(tp)) 2284 tcp_sack_newack(tp, th); 2285 else 2286 tp->t_congctl->fast_retransmit_newack(tp, th); 2287 if (SEQ_GT(th->th_ack, tp->snd_max)) { 2288 tcpstat.tcps_rcvacktoomuch++; 2289 goto dropafterack; 2290 } 2291 acked = th->th_ack - tp->snd_una; 2292 tcpstat.tcps_rcvackpack++; 2293 tcpstat.tcps_rcvackbyte += acked; 2294 2295 /* 2296 * If we have a timestamp reply, update smoothed 2297 * round trip time. If no timestamp is present but 2298 * transmit timer is running and timed sequence 2299 * number was acked, update smoothed round trip time. 2300 * Since we now have an rtt measurement, cancel the 2301 * timer backoff (cf., Phil Karn's retransmit alg.). 2302 * Recompute the initial retransmit timer. 2303 */ 2304 if (ts_rtt) 2305 tcp_xmit_timer(tp, ts_rtt); 2306 else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq)) 2307 tcp_xmit_timer(tp, tcp_now - tp->t_rtttime); 2308 2309 /* 2310 * If all outstanding data is acked, stop retransmit 2311 * timer and remember to restart (more output or persist). 2312 * If there is more data to be acked, restart retransmit 2313 * timer, using current (possibly backed-off) value. 2314 */ 2315 if (th->th_ack == tp->snd_max) { 2316 TCP_TIMER_DISARM(tp, TCPT_REXMT); 2317 needoutput = 1; 2318 } else if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) 2319 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 2320 2321 /* 2322 * New data has been acked, adjust the congestion window. 2323 */ 2324 tp->t_congctl->newack(tp, th); 2325 2326 ND6_HINT(tp); 2327 if (acked > so->so_snd.sb_cc) { 2328 tp->snd_wnd -= so->so_snd.sb_cc; 2329 sbdrop(&so->so_snd, (int)so->so_snd.sb_cc); 2330 ourfinisacked = 1; 2331 } else { 2332 if (acked > (tp->t_lastoff - tp->t_inoff)) 2333 tp->t_lastm = NULL; 2334 sbdrop(&so->so_snd, acked); 2335 tp->t_lastoff -= acked; 2336 tp->snd_wnd -= acked; 2337 ourfinisacked = 0; 2338 } 2339 sowwakeup(so); 2340 2341 ICMP_CHECK(tp, th, acked); 2342 2343 tp->snd_una = th->th_ack; 2344 if (SEQ_GT(tp->snd_una, tp->snd_fack)) 2345 tp->snd_fack = tp->snd_una; 2346 if (SEQ_LT(tp->snd_nxt, tp->snd_una)) 2347 tp->snd_nxt = tp->snd_una; 2348 if (SEQ_LT(tp->snd_high, tp->snd_una)) 2349 tp->snd_high = tp->snd_una; 2350 2351 switch (tp->t_state) { 2352 2353 /* 2354 * In FIN_WAIT_1 STATE in addition to the processing 2355 * for the ESTABLISHED state if our FIN is now acknowledged 2356 * then enter FIN_WAIT_2. 2357 */ 2358 case TCPS_FIN_WAIT_1: 2359 if (ourfinisacked) { 2360 /* 2361 * If we can't receive any more 2362 * data, then closing user can proceed. 2363 * Starting the timer is contrary to the 2364 * specification, but if we don't get a FIN 2365 * we'll hang forever. 2366 */ 2367 if (so->so_state & SS_CANTRCVMORE) { 2368 soisdisconnected(so); 2369 if (tcp_maxidle > 0) 2370 TCP_TIMER_ARM(tp, TCPT_2MSL, 2371 tcp_maxidle); 2372 } 2373 tp->t_state = TCPS_FIN_WAIT_2; 2374 } 2375 break; 2376 2377 /* 2378 * In CLOSING STATE in addition to the processing for 2379 * the ESTABLISHED state if the ACK acknowledges our FIN 2380 * then enter the TIME-WAIT state, otherwise ignore 2381 * the segment. 2382 */ 2383 case TCPS_CLOSING: 2384 if (ourfinisacked) { 2385 tp->t_state = TCPS_TIME_WAIT; 2386 tcp_canceltimers(tp); 2387 TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * TCPTV_MSL); 2388 soisdisconnected(so); 2389 } 2390 break; 2391 2392 /* 2393 * In LAST_ACK, we may still be waiting for data to drain 2394 * and/or to be acked, as well as for the ack of our FIN. 2395 * If our FIN is now acknowledged, delete the TCB, 2396 * enter the closed state and return. 2397 */ 2398 case TCPS_LAST_ACK: 2399 if (ourfinisacked) { 2400 tp = tcp_close(tp); 2401 goto drop; 2402 } 2403 break; 2404 2405 /* 2406 * In TIME_WAIT state the only thing that should arrive 2407 * is a retransmission of the remote FIN. Acknowledge 2408 * it and restart the finack timer. 2409 */ 2410 case TCPS_TIME_WAIT: 2411 TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * TCPTV_MSL); 2412 goto dropafterack; 2413 } 2414 } 2415 2416 step6: 2417 /* 2418 * Update window information. 2419 * Don't look at window if no ACK: TAC's send garbage on first SYN. 2420 */ 2421 if ((tiflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, th->th_seq) || 2422 (tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) || 2423 (tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) { 2424 /* keep track of pure window updates */ 2425 if (tlen == 0 && 2426 tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd) 2427 tcpstat.tcps_rcvwinupd++; 2428 tp->snd_wnd = tiwin; 2429 tp->snd_wl1 = th->th_seq; 2430 tp->snd_wl2 = th->th_ack; 2431 if (tp->snd_wnd > tp->max_sndwnd) 2432 tp->max_sndwnd = tp->snd_wnd; 2433 needoutput = 1; 2434 } 2435 2436 /* 2437 * Process segments with URG. 2438 */ 2439 if ((tiflags & TH_URG) && th->th_urp && 2440 TCPS_HAVERCVDFIN(tp->t_state) == 0) { 2441 /* 2442 * This is a kludge, but if we receive and accept 2443 * random urgent pointers, we'll crash in 2444 * soreceive. It's hard to imagine someone 2445 * actually wanting to send this much urgent data. 2446 */ 2447 if (th->th_urp + so->so_rcv.sb_cc > sb_max) { 2448 th->th_urp = 0; /* XXX */ 2449 tiflags &= ~TH_URG; /* XXX */ 2450 goto dodata; /* XXX */ 2451 } 2452 /* 2453 * If this segment advances the known urgent pointer, 2454 * then mark the data stream. This should not happen 2455 * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since 2456 * a FIN has been received from the remote side. 2457 * In these states we ignore the URG. 2458 * 2459 * According to RFC961 (Assigned Protocols), 2460 * the urgent pointer points to the last octet 2461 * of urgent data. We continue, however, 2462 * to consider it to indicate the first octet 2463 * of data past the urgent section as the original 2464 * spec states (in one of two places). 2465 */ 2466 if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) { 2467 tp->rcv_up = th->th_seq + th->th_urp; 2468 so->so_oobmark = so->so_rcv.sb_cc + 2469 (tp->rcv_up - tp->rcv_nxt) - 1; 2470 if (so->so_oobmark == 0) 2471 so->so_state |= SS_RCVATMARK; 2472 sohasoutofband(so); 2473 tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA); 2474 } 2475 /* 2476 * Remove out of band data so doesn't get presented to user. 2477 * This can happen independent of advancing the URG pointer, 2478 * but if two URG's are pending at once, some out-of-band 2479 * data may creep in... ick. 2480 */ 2481 if (th->th_urp <= (u_int16_t) tlen 2482 #ifdef SO_OOBINLINE 2483 && (so->so_options & SO_OOBINLINE) == 0 2484 #endif 2485 ) 2486 tcp_pulloutofband(so, th, m, hdroptlen); 2487 } else 2488 /* 2489 * If no out of band data is expected, 2490 * pull receive urgent pointer along 2491 * with the receive window. 2492 */ 2493 if (SEQ_GT(tp->rcv_nxt, tp->rcv_up)) 2494 tp->rcv_up = tp->rcv_nxt; 2495 dodata: /* XXX */ 2496 2497 /* 2498 * Process the segment text, merging it into the TCP sequencing queue, 2499 * and arranging for acknowledgement of receipt if necessary. 2500 * This process logically involves adjusting tp->rcv_wnd as data 2501 * is presented to the user (this happens in tcp_usrreq.c, 2502 * case PRU_RCVD). If a FIN has already been received on this 2503 * connection then we just ignore the text. 2504 */ 2505 if ((tlen || (tiflags & TH_FIN)) && 2506 TCPS_HAVERCVDFIN(tp->t_state) == 0) { 2507 /* 2508 * Insert segment ti into reassembly queue of tcp with 2509 * control block tp. Return TH_FIN if reassembly now includes 2510 * a segment with FIN. The macro form does the common case 2511 * inline (segment is the next to be received on an 2512 * established connection, and the queue is empty), 2513 * avoiding linkage into and removal from the queue and 2514 * repetition of various conversions. 2515 * Set DELACK for segments received in order, but ack 2516 * immediately when segments are out of order 2517 * (so fast retransmit can work). 2518 */ 2519 /* NOTE: this was TCP_REASS() macro, but used only once */ 2520 TCP_REASS_LOCK(tp); 2521 if (th->th_seq == tp->rcv_nxt && 2522 TAILQ_FIRST(&tp->segq) == NULL && 2523 tp->t_state == TCPS_ESTABLISHED) { 2524 TCP_SETUP_ACK(tp, th); 2525 tp->rcv_nxt += tlen; 2526 tiflags = th->th_flags & TH_FIN; 2527 tcpstat.tcps_rcvpack++; 2528 tcpstat.tcps_rcvbyte += tlen; 2529 ND6_HINT(tp); 2530 if (so->so_state & SS_CANTRCVMORE) 2531 m_freem(m); 2532 else { 2533 m_adj(m, hdroptlen); 2534 sbappendstream(&(so)->so_rcv, m); 2535 } 2536 sorwakeup(so); 2537 } else { 2538 m_adj(m, hdroptlen); 2539 tiflags = tcp_reass(tp, th, m, &tlen); 2540 tp->t_flags |= TF_ACKNOW; 2541 } 2542 TCP_REASS_UNLOCK(tp); 2543 2544 /* 2545 * Note the amount of data that peer has sent into 2546 * our window, in order to estimate the sender's 2547 * buffer size. 2548 */ 2549 len = so->so_rcv.sb_hiwat - (tp->rcv_adv - tp->rcv_nxt); 2550 } else { 2551 m_freem(m); 2552 m = NULL; 2553 tiflags &= ~TH_FIN; 2554 } 2555 2556 /* 2557 * If FIN is received ACK the FIN and let the user know 2558 * that the connection is closing. Ignore a FIN received before 2559 * the connection is fully established. 2560 */ 2561 if ((tiflags & TH_FIN) && TCPS_HAVEESTABLISHED(tp->t_state)) { 2562 if (TCPS_HAVERCVDFIN(tp->t_state) == 0) { 2563 socantrcvmore(so); 2564 tp->t_flags |= TF_ACKNOW; 2565 tp->rcv_nxt++; 2566 } 2567 switch (tp->t_state) { 2568 2569 /* 2570 * In ESTABLISHED STATE enter the CLOSE_WAIT state. 2571 */ 2572 case TCPS_ESTABLISHED: 2573 tp->t_state = TCPS_CLOSE_WAIT; 2574 break; 2575 2576 /* 2577 * If still in FIN_WAIT_1 STATE FIN has not been acked so 2578 * enter the CLOSING state. 2579 */ 2580 case TCPS_FIN_WAIT_1: 2581 tp->t_state = TCPS_CLOSING; 2582 break; 2583 2584 /* 2585 * In FIN_WAIT_2 state enter the TIME_WAIT state, 2586 * starting the time-wait timer, turning off the other 2587 * standard timers. 2588 */ 2589 case TCPS_FIN_WAIT_2: 2590 tp->t_state = TCPS_TIME_WAIT; 2591 tcp_canceltimers(tp); 2592 TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * TCPTV_MSL); 2593 soisdisconnected(so); 2594 break; 2595 2596 /* 2597 * In TIME_WAIT state restart the 2 MSL time_wait timer. 2598 */ 2599 case TCPS_TIME_WAIT: 2600 TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * TCPTV_MSL); 2601 break; 2602 } 2603 } 2604 #ifdef TCP_DEBUG 2605 if (so->so_options & SO_DEBUG) 2606 tcp_trace(TA_INPUT, ostate, tp, tcp_saveti, 0); 2607 #endif 2608 2609 /* 2610 * Return any desired output. 2611 */ 2612 if (needoutput || (tp->t_flags & TF_ACKNOW)) { 2613 (void) tcp_output(tp); 2614 } 2615 if (tcp_saveti) 2616 m_freem(tcp_saveti); 2617 return; 2618 2619 badsyn: 2620 /* 2621 * Received a bad SYN. Increment counters and dropwithreset. 2622 */ 2623 tcpstat.tcps_badsyn++; 2624 tp = NULL; 2625 goto dropwithreset; 2626 2627 dropafterack: 2628 /* 2629 * Generate an ACK dropping incoming segment if it occupies 2630 * sequence space, where the ACK reflects our state. 2631 */ 2632 if (tiflags & TH_RST) 2633 goto drop; 2634 goto dropafterack2; 2635 2636 dropafterack_ratelim: 2637 /* 2638 * We may want to rate-limit ACKs against SYN/RST attack. 2639 */ 2640 if (ppsratecheck(&tcp_ackdrop_ppslim_last, &tcp_ackdrop_ppslim_count, 2641 tcp_ackdrop_ppslim) == 0) { 2642 /* XXX stat */ 2643 goto drop; 2644 } 2645 /* ...fall into dropafterack2... */ 2646 2647 dropafterack2: 2648 m_freem(m); 2649 tp->t_flags |= TF_ACKNOW; 2650 (void) tcp_output(tp); 2651 if (tcp_saveti) 2652 m_freem(tcp_saveti); 2653 return; 2654 2655 dropwithreset_ratelim: 2656 /* 2657 * We may want to rate-limit RSTs in certain situations, 2658 * particularly if we are sending an RST in response to 2659 * an attempt to connect to or otherwise communicate with 2660 * a port for which we have no socket. 2661 */ 2662 if (ppsratecheck(&tcp_rst_ppslim_last, &tcp_rst_ppslim_count, 2663 tcp_rst_ppslim) == 0) { 2664 /* XXX stat */ 2665 goto drop; 2666 } 2667 /* ...fall into dropwithreset... */ 2668 2669 dropwithreset: 2670 /* 2671 * Generate a RST, dropping incoming segment. 2672 * Make ACK acceptable to originator of segment. 2673 */ 2674 if (tiflags & TH_RST) 2675 goto drop; 2676 2677 switch (af) { 2678 #ifdef INET6 2679 case AF_INET6: 2680 /* For following calls to tcp_respond */ 2681 if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) 2682 goto drop; 2683 break; 2684 #endif /* INET6 */ 2685 case AF_INET: 2686 if (IN_MULTICAST(ip->ip_dst.s_addr) || 2687 in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif)) 2688 goto drop; 2689 } 2690 2691 if (tiflags & TH_ACK) 2692 (void)tcp_respond(tp, m, m, th, (tcp_seq)0, th->th_ack, TH_RST); 2693 else { 2694 if (tiflags & TH_SYN) 2695 tlen++; 2696 (void)tcp_respond(tp, m, m, th, th->th_seq + tlen, (tcp_seq)0, 2697 TH_RST|TH_ACK); 2698 } 2699 if (tcp_saveti) 2700 m_freem(tcp_saveti); 2701 return; 2702 2703 badcsum: 2704 drop: 2705 /* 2706 * Drop space held by incoming segment and return. 2707 */ 2708 if (tp) { 2709 if (tp->t_inpcb) 2710 so = tp->t_inpcb->inp_socket; 2711 #ifdef INET6 2712 else if (tp->t_in6pcb) 2713 so = tp->t_in6pcb->in6p_socket; 2714 #endif 2715 else 2716 so = NULL; 2717 #ifdef TCP_DEBUG 2718 if (so && (so->so_options & SO_DEBUG) != 0) 2719 tcp_trace(TA_DROP, ostate, tp, tcp_saveti, 0); 2720 #endif 2721 } 2722 if (tcp_saveti) 2723 m_freem(tcp_saveti); 2724 m_freem(m); 2725 return; 2726 } 2727 2728 #ifdef TCP_SIGNATURE 2729 int 2730 tcp_signature_apply(void *fstate, caddr_t data, u_int len) 2731 { 2732 2733 MD5Update(fstate, (u_char *)data, len); 2734 return (0); 2735 } 2736 2737 struct secasvar * 2738 tcp_signature_getsav(struct mbuf *m, struct tcphdr *th) 2739 { 2740 struct secasvar *sav; 2741 #ifdef FAST_IPSEC 2742 union sockaddr_union dst; 2743 #endif 2744 struct ip *ip; 2745 struct ip6_hdr *ip6; 2746 2747 ip = mtod(m, struct ip *); 2748 switch (ip->ip_v) { 2749 case 4: 2750 ip = mtod(m, struct ip *); 2751 ip6 = NULL; 2752 break; 2753 case 6: 2754 ip = NULL; 2755 ip6 = mtod(m, struct ip6_hdr *); 2756 break; 2757 default: 2758 return (NULL); 2759 } 2760 2761 #ifdef FAST_IPSEC 2762 /* Extract the destination from the IP header in the mbuf. */ 2763 bzero(&dst, sizeof(union sockaddr_union)); 2764 dst.sa.sa_len = sizeof(struct sockaddr_in); 2765 dst.sa.sa_family = AF_INET; 2766 dst.sin.sin_addr = ip->ip_dst; 2767 2768 /* 2769 * Look up an SADB entry which matches the address of the peer. 2770 */ 2771 sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI)); 2772 #else 2773 if (ip) 2774 sav = key_allocsa(AF_INET, (caddr_t)&ip->ip_src, 2775 (caddr_t)&ip->ip_dst, IPPROTO_TCP, 2776 htonl(TCP_SIG_SPI), 0, 0); 2777 else 2778 sav = key_allocsa(AF_INET6, (caddr_t)&ip6->ip6_src, 2779 (caddr_t)&ip6->ip6_dst, IPPROTO_TCP, 2780 htonl(TCP_SIG_SPI), 0, 0); 2781 #endif 2782 2783 return (sav); /* freesav must be performed by caller */ 2784 } 2785 2786 int 2787 tcp_signature(struct mbuf *m, struct tcphdr *th, int thoff, 2788 struct secasvar *sav, char *sig) 2789 { 2790 MD5_CTX ctx; 2791 struct ip *ip; 2792 struct ipovly *ipovly; 2793 struct ip6_hdr *ip6; 2794 struct ippseudo ippseudo; 2795 struct ip6_hdr_pseudo ip6pseudo; 2796 struct tcphdr th0; 2797 int l, tcphdrlen; 2798 2799 if (sav == NULL) 2800 return (-1); 2801 2802 tcphdrlen = th->th_off * 4; 2803 2804 switch (mtod(m, struct ip *)->ip_v) { 2805 case 4: 2806 ip = mtod(m, struct ip *); 2807 ip6 = NULL; 2808 break; 2809 case 6: 2810 ip = NULL; 2811 ip6 = mtod(m, struct ip6_hdr *); 2812 break; 2813 default: 2814 return (-1); 2815 } 2816 2817 MD5Init(&ctx); 2818 2819 if (ip) { 2820 memset(&ippseudo, 0, sizeof(ippseudo)); 2821 ipovly = (struct ipovly *)ip; 2822 ippseudo.ippseudo_src = ipovly->ih_src; 2823 ippseudo.ippseudo_dst = ipovly->ih_dst; 2824 ippseudo.ippseudo_pad = 0; 2825 ippseudo.ippseudo_p = IPPROTO_TCP; 2826 ippseudo.ippseudo_len = htons(m->m_pkthdr.len - thoff); 2827 MD5Update(&ctx, (char *)&ippseudo, sizeof(ippseudo)); 2828 } else { 2829 memset(&ip6pseudo, 0, sizeof(ip6pseudo)); 2830 ip6pseudo.ip6ph_src = ip6->ip6_src; 2831 in6_clearscope(&ip6pseudo.ip6ph_src); 2832 ip6pseudo.ip6ph_dst = ip6->ip6_dst; 2833 in6_clearscope(&ip6pseudo.ip6ph_dst); 2834 ip6pseudo.ip6ph_len = htons(m->m_pkthdr.len - thoff); 2835 ip6pseudo.ip6ph_nxt = IPPROTO_TCP; 2836 MD5Update(&ctx, (char *)&ip6pseudo, sizeof(ip6pseudo)); 2837 } 2838 2839 th0 = *th; 2840 th0.th_sum = 0; 2841 MD5Update(&ctx, (char *)&th0, sizeof(th0)); 2842 2843 l = m->m_pkthdr.len - thoff - tcphdrlen; 2844 if (l > 0) 2845 m_apply(m, thoff + tcphdrlen, 2846 m->m_pkthdr.len - thoff - tcphdrlen, 2847 tcp_signature_apply, &ctx); 2848 2849 MD5Update(&ctx, _KEYBUF(sav->key_auth), _KEYLEN(sav->key_auth)); 2850 MD5Final(sig, &ctx); 2851 2852 return (0); 2853 } 2854 #endif 2855 2856 static int 2857 tcp_dooptions(struct tcpcb *tp, const u_char *cp, int cnt, 2858 const struct tcphdr *th, 2859 struct mbuf *m, int toff, struct tcp_opt_info *oi) 2860 { 2861 u_int16_t mss; 2862 int opt, optlen = 0; 2863 #ifdef TCP_SIGNATURE 2864 caddr_t sigp = NULL; 2865 char sigbuf[TCP_SIGLEN]; 2866 struct secasvar *sav = NULL; 2867 #endif 2868 2869 for (; cp && cnt > 0; cnt -= optlen, cp += optlen) { 2870 opt = cp[0]; 2871 if (opt == TCPOPT_EOL) 2872 break; 2873 if (opt == TCPOPT_NOP) 2874 optlen = 1; 2875 else { 2876 if (cnt < 2) 2877 break; 2878 optlen = cp[1]; 2879 if (optlen < 2 || optlen > cnt) 2880 break; 2881 } 2882 switch (opt) { 2883 2884 default: 2885 continue; 2886 2887 case TCPOPT_MAXSEG: 2888 if (optlen != TCPOLEN_MAXSEG) 2889 continue; 2890 if (!(th->th_flags & TH_SYN)) 2891 continue; 2892 if (TCPS_HAVERCVDSYN(tp->t_state)) 2893 continue; 2894 bcopy(cp + 2, &mss, sizeof(mss)); 2895 oi->maxseg = ntohs(mss); 2896 break; 2897 2898 case TCPOPT_WINDOW: 2899 if (optlen != TCPOLEN_WINDOW) 2900 continue; 2901 if (!(th->th_flags & TH_SYN)) 2902 continue; 2903 if (TCPS_HAVERCVDSYN(tp->t_state)) 2904 continue; 2905 tp->t_flags |= TF_RCVD_SCALE; 2906 tp->requested_s_scale = cp[2]; 2907 if (tp->requested_s_scale > TCP_MAX_WINSHIFT) { 2908 #if 0 /*XXX*/ 2909 char *p; 2910 2911 if (ip) 2912 p = ntohl(ip->ip_src); 2913 #ifdef INET6 2914 else if (ip6) 2915 p = ip6_sprintf(&ip6->ip6_src); 2916 #endif 2917 else 2918 p = "(unknown)"; 2919 log(LOG_ERR, "TCP: invalid wscale %d from %s, " 2920 "assuming %d\n", 2921 tp->requested_s_scale, p, 2922 TCP_MAX_WINSHIFT); 2923 #else 2924 log(LOG_ERR, "TCP: invalid wscale %d, " 2925 "assuming %d\n", 2926 tp->requested_s_scale, 2927 TCP_MAX_WINSHIFT); 2928 #endif 2929 tp->requested_s_scale = TCP_MAX_WINSHIFT; 2930 } 2931 break; 2932 2933 case TCPOPT_TIMESTAMP: 2934 if (optlen != TCPOLEN_TIMESTAMP) 2935 continue; 2936 oi->ts_present = 1; 2937 bcopy(cp + 2, &oi->ts_val, sizeof(oi->ts_val)); 2938 NTOHL(oi->ts_val); 2939 bcopy(cp + 6, &oi->ts_ecr, sizeof(oi->ts_ecr)); 2940 NTOHL(oi->ts_ecr); 2941 2942 if (!(th->th_flags & TH_SYN)) 2943 continue; 2944 if (TCPS_HAVERCVDSYN(tp->t_state)) 2945 continue; 2946 /* 2947 * A timestamp received in a SYN makes 2948 * it ok to send timestamp requests and replies. 2949 */ 2950 tp->t_flags |= TF_RCVD_TSTMP; 2951 tp->ts_recent = oi->ts_val; 2952 tp->ts_recent_age = tcp_now; 2953 break; 2954 2955 case TCPOPT_SACK_PERMITTED: 2956 if (optlen != TCPOLEN_SACK_PERMITTED) 2957 continue; 2958 if (!(th->th_flags & TH_SYN)) 2959 continue; 2960 if (TCPS_HAVERCVDSYN(tp->t_state)) 2961 continue; 2962 if (tcp_do_sack) { 2963 tp->t_flags |= TF_SACK_PERMIT; 2964 tp->t_flags |= TF_WILL_SACK; 2965 } 2966 break; 2967 2968 case TCPOPT_SACK: 2969 tcp_sack_option(tp, th, cp, optlen); 2970 break; 2971 #ifdef TCP_SIGNATURE 2972 case TCPOPT_SIGNATURE: 2973 if (optlen != TCPOLEN_SIGNATURE) 2974 continue; 2975 if (sigp && bcmp(sigp, cp + 2, TCP_SIGLEN)) 2976 return (-1); 2977 2978 sigp = sigbuf; 2979 memcpy(sigbuf, cp + 2, TCP_SIGLEN); 2980 memset(cp + 2, 0, TCP_SIGLEN); 2981 tp->t_flags |= TF_SIGNATURE; 2982 break; 2983 #endif 2984 } 2985 } 2986 2987 #ifdef TCP_SIGNATURE 2988 if (tp->t_flags & TF_SIGNATURE) { 2989 2990 sav = tcp_signature_getsav(m, th); 2991 2992 if (sav == NULL && tp->t_state == TCPS_LISTEN) 2993 return (-1); 2994 } 2995 2996 if ((sigp ? TF_SIGNATURE : 0) ^ (tp->t_flags & TF_SIGNATURE)) { 2997 if (sav == NULL) 2998 return (-1); 2999 #ifdef FAST_IPSEC 3000 KEY_FREESAV(&sav); 3001 #else 3002 key_freesav(sav); 3003 #endif 3004 return (-1); 3005 } 3006 3007 if (sigp) { 3008 char sig[TCP_SIGLEN]; 3009 3010 TCP_FIELDS_TO_NET(th); 3011 if (tcp_signature(m, th, toff, sav, sig) < 0) { 3012 TCP_FIELDS_TO_HOST(th); 3013 if (sav == NULL) 3014 return (-1); 3015 #ifdef FAST_IPSEC 3016 KEY_FREESAV(&sav); 3017 #else 3018 key_freesav(sav); 3019 #endif 3020 return (-1); 3021 } 3022 TCP_FIELDS_TO_HOST(th); 3023 3024 if (bcmp(sig, sigp, TCP_SIGLEN)) { 3025 tcpstat.tcps_badsig++; 3026 if (sav == NULL) 3027 return (-1); 3028 #ifdef FAST_IPSEC 3029 KEY_FREESAV(&sav); 3030 #else 3031 key_freesav(sav); 3032 #endif 3033 return (-1); 3034 } else 3035 tcpstat.tcps_goodsig++; 3036 3037 key_sa_recordxfer(sav, m); 3038 #ifdef FAST_IPSEC 3039 KEY_FREESAV(&sav); 3040 #else 3041 key_freesav(sav); 3042 #endif 3043 } 3044 #endif 3045 3046 return (0); 3047 } 3048 3049 /* 3050 * Pull out of band byte out of a segment so 3051 * it doesn't appear in the user's data queue. 3052 * It is still reflected in the segment length for 3053 * sequencing purposes. 3054 */ 3055 void 3056 tcp_pulloutofband(struct socket *so, struct tcphdr *th, 3057 struct mbuf *m, int off) 3058 { 3059 int cnt = off + th->th_urp - 1; 3060 3061 while (cnt >= 0) { 3062 if (m->m_len > cnt) { 3063 char *cp = mtod(m, caddr_t) + cnt; 3064 struct tcpcb *tp = sototcpcb(so); 3065 3066 tp->t_iobc = *cp; 3067 tp->t_oobflags |= TCPOOB_HAVEDATA; 3068 bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1)); 3069 m->m_len--; 3070 return; 3071 } 3072 cnt -= m->m_len; 3073 m = m->m_next; 3074 if (m == 0) 3075 break; 3076 } 3077 panic("tcp_pulloutofband"); 3078 } 3079 3080 /* 3081 * Collect new round-trip time estimate 3082 * and update averages and current timeout. 3083 */ 3084 void 3085 tcp_xmit_timer(struct tcpcb *tp, uint32_t rtt) 3086 { 3087 int32_t delta; 3088 3089 tcpstat.tcps_rttupdated++; 3090 if (tp->t_srtt != 0) { 3091 /* 3092 * srtt is stored as fixed point with 3 bits after the 3093 * binary point (i.e., scaled by 8). The following magic 3094 * is equivalent to the smoothing algorithm in rfc793 with 3095 * an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed 3096 * point). Adjust rtt to origin 0. 3097 */ 3098 delta = (rtt << 2) - (tp->t_srtt >> TCP_RTT_SHIFT); 3099 if ((tp->t_srtt += delta) <= 0) 3100 tp->t_srtt = 1 << 2; 3101 /* 3102 * We accumulate a smoothed rtt variance (actually, a 3103 * smoothed mean difference), then set the retransmit 3104 * timer to smoothed rtt + 4 times the smoothed variance. 3105 * rttvar is stored as fixed point with 2 bits after the 3106 * binary point (scaled by 4). The following is 3107 * equivalent to rfc793 smoothing with an alpha of .75 3108 * (rttvar = rttvar*3/4 + |delta| / 4). This replaces 3109 * rfc793's wired-in beta. 3110 */ 3111 if (delta < 0) 3112 delta = -delta; 3113 delta -= (tp->t_rttvar >> TCP_RTTVAR_SHIFT); 3114 if ((tp->t_rttvar += delta) <= 0) 3115 tp->t_rttvar = 1 << 2; 3116 } else { 3117 /* 3118 * No rtt measurement yet - use the unsmoothed rtt. 3119 * Set the variance to half the rtt (so our first 3120 * retransmit happens at 3*rtt). 3121 */ 3122 tp->t_srtt = rtt << (TCP_RTT_SHIFT + 2); 3123 tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT + 2 - 1); 3124 } 3125 tp->t_rtttime = 0; 3126 tp->t_rxtshift = 0; 3127 3128 /* 3129 * the retransmit should happen at rtt + 4 * rttvar. 3130 * Because of the way we do the smoothing, srtt and rttvar 3131 * will each average +1/2 tick of bias. When we compute 3132 * the retransmit timer, we want 1/2 tick of rounding and 3133 * 1 extra tick because of +-1/2 tick uncertainty in the 3134 * firing of the timer. The bias will give us exactly the 3135 * 1.5 tick we need. But, because the bias is 3136 * statistical, we have to test that we don't drop below 3137 * the minimum feasible timer (which is 2 ticks). 3138 */ 3139 TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), 3140 max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX); 3141 3142 /* 3143 * We received an ack for a packet that wasn't retransmitted; 3144 * it is probably safe to discard any error indications we've 3145 * received recently. This isn't quite right, but close enough 3146 * for now (a route might have failed after we sent a segment, 3147 * and the return path might not be symmetrical). 3148 */ 3149 tp->t_softerror = 0; 3150 } 3151 3152 3153 /* 3154 * TCP compressed state engine. Currently used to hold compressed 3155 * state for SYN_RECEIVED. 3156 */ 3157 3158 u_long syn_cache_count; 3159 u_int32_t syn_hash1, syn_hash2; 3160 3161 #define SYN_HASH(sa, sp, dp) \ 3162 ((((sa)->s_addr^syn_hash1)*(((((u_int32_t)(dp))<<16) + \ 3163 ((u_int32_t)(sp)))^syn_hash2))) 3164 #ifndef INET6 3165 #define SYN_HASHALL(hash, src, dst) \ 3166 do { \ 3167 hash = SYN_HASH(&((const struct sockaddr_in *)(src))->sin_addr, \ 3168 ((const struct sockaddr_in *)(src))->sin_port, \ 3169 ((const struct sockaddr_in *)(dst))->sin_port); \ 3170 } while (/*CONSTCOND*/ 0) 3171 #else 3172 #define SYN_HASH6(sa, sp, dp) \ 3173 ((((sa)->s6_addr32[0] ^ (sa)->s6_addr32[3] ^ syn_hash1) * \ 3174 (((((u_int32_t)(dp))<<16) + ((u_int32_t)(sp)))^syn_hash2)) \ 3175 & 0x7fffffff) 3176 3177 #define SYN_HASHALL(hash, src, dst) \ 3178 do { \ 3179 switch ((src)->sa_family) { \ 3180 case AF_INET: \ 3181 hash = SYN_HASH(&((const struct sockaddr_in *)(src))->sin_addr, \ 3182 ((const struct sockaddr_in *)(src))->sin_port, \ 3183 ((const struct sockaddr_in *)(dst))->sin_port); \ 3184 break; \ 3185 case AF_INET6: \ 3186 hash = SYN_HASH6(&((const struct sockaddr_in6 *)(src))->sin6_addr, \ 3187 ((const struct sockaddr_in6 *)(src))->sin6_port, \ 3188 ((const struct sockaddr_in6 *)(dst))->sin6_port); \ 3189 break; \ 3190 default: \ 3191 hash = 0; \ 3192 } \ 3193 } while (/*CONSTCOND*/0) 3194 #endif /* INET6 */ 3195 3196 #define SYN_CACHE_RM(sc) \ 3197 do { \ 3198 TAILQ_REMOVE(&tcp_syn_cache[(sc)->sc_bucketidx].sch_bucket, \ 3199 (sc), sc_bucketq); \ 3200 (sc)->sc_tp = NULL; \ 3201 LIST_REMOVE((sc), sc_tpq); \ 3202 tcp_syn_cache[(sc)->sc_bucketidx].sch_length--; \ 3203 callout_stop(&(sc)->sc_timer); \ 3204 syn_cache_count--; \ 3205 } while (/*CONSTCOND*/0) 3206 3207 #define SYN_CACHE_PUT(sc) \ 3208 do { \ 3209 if ((sc)->sc_ipopts) \ 3210 (void) m_free((sc)->sc_ipopts); \ 3211 if ((sc)->sc_route4.ro_rt != NULL) \ 3212 rtflush(&(sc)->sc_route4); \ 3213 if (callout_invoking(&(sc)->sc_timer)) \ 3214 (sc)->sc_flags |= SCF_DEAD; \ 3215 else \ 3216 pool_put(&syn_cache_pool, (sc)); \ 3217 } while (/*CONSTCOND*/0) 3218 3219 POOL_INIT(syn_cache_pool, sizeof(struct syn_cache), 0, 0, 0, "synpl", NULL); 3220 3221 /* 3222 * We don't estimate RTT with SYNs, so each packet starts with the default 3223 * RTT and each timer step has a fixed timeout value. 3224 */ 3225 #define SYN_CACHE_TIMER_ARM(sc) \ 3226 do { \ 3227 TCPT_RANGESET((sc)->sc_rxtcur, \ 3228 TCPTV_SRTTDFLT * tcp_backoff[(sc)->sc_rxtshift], TCPTV_MIN, \ 3229 TCPTV_REXMTMAX); \ 3230 callout_reset(&(sc)->sc_timer, \ 3231 (sc)->sc_rxtcur * (hz / PR_SLOWHZ), syn_cache_timer, (sc)); \ 3232 } while (/*CONSTCOND*/0) 3233 3234 #define SYN_CACHE_TIMESTAMP(sc) (tcp_now - (sc)->sc_timebase) 3235 3236 void 3237 syn_cache_init(void) 3238 { 3239 int i; 3240 3241 /* Initialize the hash buckets. */ 3242 for (i = 0; i < tcp_syn_cache_size; i++) 3243 TAILQ_INIT(&tcp_syn_cache[i].sch_bucket); 3244 } 3245 3246 void 3247 syn_cache_insert(struct syn_cache *sc, struct tcpcb *tp) 3248 { 3249 struct syn_cache_head *scp; 3250 struct syn_cache *sc2; 3251 int s; 3252 3253 /* 3254 * If there are no entries in the hash table, reinitialize 3255 * the hash secrets. 3256 */ 3257 if (syn_cache_count == 0) { 3258 syn_hash1 = arc4random(); 3259 syn_hash2 = arc4random(); 3260 } 3261 3262 SYN_HASHALL(sc->sc_hash, &sc->sc_src.sa, &sc->sc_dst.sa); 3263 sc->sc_bucketidx = sc->sc_hash % tcp_syn_cache_size; 3264 scp = &tcp_syn_cache[sc->sc_bucketidx]; 3265 3266 /* 3267 * Make sure that we don't overflow the per-bucket 3268 * limit or the total cache size limit. 3269 */ 3270 s = splsoftnet(); 3271 if (scp->sch_length >= tcp_syn_bucket_limit) { 3272 tcpstat.tcps_sc_bucketoverflow++; 3273 /* 3274 * The bucket is full. Toss the oldest element in the 3275 * bucket. This will be the first entry in the bucket. 3276 */ 3277 sc2 = TAILQ_FIRST(&scp->sch_bucket); 3278 #ifdef DIAGNOSTIC 3279 /* 3280 * This should never happen; we should always find an 3281 * entry in our bucket. 3282 */ 3283 if (sc2 == NULL) 3284 panic("syn_cache_insert: bucketoverflow: impossible"); 3285 #endif 3286 SYN_CACHE_RM(sc2); 3287 SYN_CACHE_PUT(sc2); /* calls pool_put but see spl above */ 3288 } else if (syn_cache_count >= tcp_syn_cache_limit) { 3289 struct syn_cache_head *scp2, *sce; 3290 3291 tcpstat.tcps_sc_overflowed++; 3292 /* 3293 * The cache is full. Toss the oldest entry in the 3294 * first non-empty bucket we can find. 3295 * 3296 * XXX We would really like to toss the oldest 3297 * entry in the cache, but we hope that this 3298 * condition doesn't happen very often. 3299 */ 3300 scp2 = scp; 3301 if (TAILQ_EMPTY(&scp2->sch_bucket)) { 3302 sce = &tcp_syn_cache[tcp_syn_cache_size]; 3303 for (++scp2; scp2 != scp; scp2++) { 3304 if (scp2 >= sce) 3305 scp2 = &tcp_syn_cache[0]; 3306 if (! TAILQ_EMPTY(&scp2->sch_bucket)) 3307 break; 3308 } 3309 #ifdef DIAGNOSTIC 3310 /* 3311 * This should never happen; we should always find a 3312 * non-empty bucket. 3313 */ 3314 if (scp2 == scp) 3315 panic("syn_cache_insert: cacheoverflow: " 3316 "impossible"); 3317 #endif 3318 } 3319 sc2 = TAILQ_FIRST(&scp2->sch_bucket); 3320 SYN_CACHE_RM(sc2); 3321 SYN_CACHE_PUT(sc2); /* calls pool_put but see spl above */ 3322 } 3323 3324 /* 3325 * Initialize the entry's timer. 3326 */ 3327 sc->sc_rxttot = 0; 3328 sc->sc_rxtshift = 0; 3329 SYN_CACHE_TIMER_ARM(sc); 3330 3331 /* Link it from tcpcb entry */ 3332 LIST_INSERT_HEAD(&tp->t_sc, sc, sc_tpq); 3333 3334 /* Put it into the bucket. */ 3335 TAILQ_INSERT_TAIL(&scp->sch_bucket, sc, sc_bucketq); 3336 scp->sch_length++; 3337 syn_cache_count++; 3338 3339 tcpstat.tcps_sc_added++; 3340 splx(s); 3341 } 3342 3343 /* 3344 * Walk the timer queues, looking for SYN,ACKs that need to be retransmitted. 3345 * If we have retransmitted an entry the maximum number of times, expire 3346 * that entry. 3347 */ 3348 void 3349 syn_cache_timer(void *arg) 3350 { 3351 struct syn_cache *sc = arg; 3352 int s; 3353 3354 s = splsoftnet(); 3355 callout_ack(&sc->sc_timer); 3356 3357 if (__predict_false(sc->sc_flags & SCF_DEAD)) { 3358 tcpstat.tcps_sc_delayed_free++; 3359 pool_put(&syn_cache_pool, sc); 3360 splx(s); 3361 return; 3362 } 3363 3364 if (__predict_false(sc->sc_rxtshift == TCP_MAXRXTSHIFT)) { 3365 /* Drop it -- too many retransmissions. */ 3366 goto dropit; 3367 } 3368 3369 /* 3370 * Compute the total amount of time this entry has 3371 * been on a queue. If this entry has been on longer 3372 * than the keep alive timer would allow, expire it. 3373 */ 3374 sc->sc_rxttot += sc->sc_rxtcur; 3375 if (sc->sc_rxttot >= TCPTV_KEEP_INIT) 3376 goto dropit; 3377 3378 tcpstat.tcps_sc_retransmitted++; 3379 (void) syn_cache_respond(sc, NULL); 3380 3381 /* Advance the timer back-off. */ 3382 sc->sc_rxtshift++; 3383 SYN_CACHE_TIMER_ARM(sc); 3384 3385 splx(s); 3386 return; 3387 3388 dropit: 3389 tcpstat.tcps_sc_timed_out++; 3390 SYN_CACHE_RM(sc); 3391 SYN_CACHE_PUT(sc); /* calls pool_put but see spl above */ 3392 splx(s); 3393 } 3394 3395 /* 3396 * Remove syn cache created by the specified tcb entry, 3397 * because this does not make sense to keep them 3398 * (if there's no tcb entry, syn cache entry will never be used) 3399 */ 3400 void 3401 syn_cache_cleanup(struct tcpcb *tp) 3402 { 3403 struct syn_cache *sc, *nsc; 3404 int s; 3405 3406 s = splsoftnet(); 3407 3408 for (sc = LIST_FIRST(&tp->t_sc); sc != NULL; sc = nsc) { 3409 nsc = LIST_NEXT(sc, sc_tpq); 3410 3411 #ifdef DIAGNOSTIC 3412 if (sc->sc_tp != tp) 3413 panic("invalid sc_tp in syn_cache_cleanup"); 3414 #endif 3415 SYN_CACHE_RM(sc); 3416 SYN_CACHE_PUT(sc); /* calls pool_put but see spl above */ 3417 } 3418 /* just for safety */ 3419 LIST_INIT(&tp->t_sc); 3420 3421 splx(s); 3422 } 3423 3424 /* 3425 * Find an entry in the syn cache. 3426 */ 3427 struct syn_cache * 3428 syn_cache_lookup(const struct sockaddr *src, const struct sockaddr *dst, 3429 struct syn_cache_head **headp) 3430 { 3431 struct syn_cache *sc; 3432 struct syn_cache_head *scp; 3433 u_int32_t hash; 3434 int s; 3435 3436 SYN_HASHALL(hash, src, dst); 3437 3438 scp = &tcp_syn_cache[hash % tcp_syn_cache_size]; 3439 *headp = scp; 3440 s = splsoftnet(); 3441 for (sc = TAILQ_FIRST(&scp->sch_bucket); sc != NULL; 3442 sc = TAILQ_NEXT(sc, sc_bucketq)) { 3443 if (sc->sc_hash != hash) 3444 continue; 3445 if (!bcmp(&sc->sc_src, src, src->sa_len) && 3446 !bcmp(&sc->sc_dst, dst, dst->sa_len)) { 3447 splx(s); 3448 return (sc); 3449 } 3450 } 3451 splx(s); 3452 return (NULL); 3453 } 3454 3455 /* 3456 * This function gets called when we receive an ACK for a 3457 * socket in the LISTEN state. We look up the connection 3458 * in the syn cache, and if its there, we pull it out of 3459 * the cache and turn it into a full-blown connection in 3460 * the SYN-RECEIVED state. 3461 * 3462 * The return values may not be immediately obvious, and their effects 3463 * can be subtle, so here they are: 3464 * 3465 * NULL SYN was not found in cache; caller should drop the 3466 * packet and send an RST. 3467 * 3468 * -1 We were unable to create the new connection, and are 3469 * aborting it. An ACK,RST is being sent to the peer 3470 * (unless we got screwey sequence numbners; see below), 3471 * because the 3-way handshake has been completed. Caller 3472 * should not free the mbuf, since we may be using it. If 3473 * we are not, we will free it. 3474 * 3475 * Otherwise, the return value is a pointer to the new socket 3476 * associated with the connection. 3477 */ 3478 struct socket * 3479 syn_cache_get(struct sockaddr *src, struct sockaddr *dst, 3480 struct tcphdr *th, unsigned int hlen, unsigned int tlen, 3481 struct socket *so, struct mbuf *m) 3482 { 3483 struct syn_cache *sc; 3484 struct syn_cache_head *scp; 3485 struct inpcb *inp = NULL; 3486 #ifdef INET6 3487 struct in6pcb *in6p = NULL; 3488 #endif 3489 struct tcpcb *tp = 0; 3490 struct mbuf *am; 3491 int s; 3492 struct socket *oso; 3493 3494 s = splsoftnet(); 3495 if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) { 3496 splx(s); 3497 return (NULL); 3498 } 3499 3500 /* 3501 * Verify the sequence and ack numbers. Try getting the correct 3502 * response again. 3503 */ 3504 if ((th->th_ack != sc->sc_iss + 1) || 3505 SEQ_LEQ(th->th_seq, sc->sc_irs) || 3506 SEQ_GT(th->th_seq, sc->sc_irs + 1 + sc->sc_win)) { 3507 (void) syn_cache_respond(sc, m); 3508 splx(s); 3509 return ((struct socket *)(-1)); 3510 } 3511 3512 /* Remove this cache entry */ 3513 SYN_CACHE_RM(sc); 3514 splx(s); 3515 3516 /* 3517 * Ok, create the full blown connection, and set things up 3518 * as they would have been set up if we had created the 3519 * connection when the SYN arrived. If we can't create 3520 * the connection, abort it. 3521 */ 3522 /* 3523 * inp still has the OLD in_pcb stuff, set the 3524 * v6-related flags on the new guy, too. This is 3525 * done particularly for the case where an AF_INET6 3526 * socket is bound only to a port, and a v4 connection 3527 * comes in on that port. 3528 * we also copy the flowinfo from the original pcb 3529 * to the new one. 3530 */ 3531 oso = so; 3532 so = sonewconn(so, SS_ISCONNECTED); 3533 if (so == NULL) 3534 goto resetandabort; 3535 3536 switch (so->so_proto->pr_domain->dom_family) { 3537 #ifdef INET 3538 case AF_INET: 3539 inp = sotoinpcb(so); 3540 break; 3541 #endif 3542 #ifdef INET6 3543 case AF_INET6: 3544 in6p = sotoin6pcb(so); 3545 break; 3546 #endif 3547 } 3548 switch (src->sa_family) { 3549 #ifdef INET 3550 case AF_INET: 3551 if (inp) { 3552 inp->inp_laddr = ((struct sockaddr_in *)dst)->sin_addr; 3553 inp->inp_lport = ((struct sockaddr_in *)dst)->sin_port; 3554 inp->inp_options = ip_srcroute(); 3555 in_pcbstate(inp, INP_BOUND); 3556 if (inp->inp_options == NULL) { 3557 inp->inp_options = sc->sc_ipopts; 3558 sc->sc_ipopts = NULL; 3559 } 3560 } 3561 #ifdef INET6 3562 else if (in6p) { 3563 /* IPv4 packet to AF_INET6 socket */ 3564 bzero(&in6p->in6p_laddr, sizeof(in6p->in6p_laddr)); 3565 in6p->in6p_laddr.s6_addr16[5] = htons(0xffff); 3566 bcopy(&((struct sockaddr_in *)dst)->sin_addr, 3567 &in6p->in6p_laddr.s6_addr32[3], 3568 sizeof(((struct sockaddr_in *)dst)->sin_addr)); 3569 in6p->in6p_lport = ((struct sockaddr_in *)dst)->sin_port; 3570 in6totcpcb(in6p)->t_family = AF_INET; 3571 if (sotoin6pcb(oso)->in6p_flags & IN6P_IPV6_V6ONLY) 3572 in6p->in6p_flags |= IN6P_IPV6_V6ONLY; 3573 else 3574 in6p->in6p_flags &= ~IN6P_IPV6_V6ONLY; 3575 in6_pcbstate(in6p, IN6P_BOUND); 3576 } 3577 #endif 3578 break; 3579 #endif 3580 #ifdef INET6 3581 case AF_INET6: 3582 if (in6p) { 3583 in6p->in6p_laddr = ((struct sockaddr_in6 *)dst)->sin6_addr; 3584 in6p->in6p_lport = ((struct sockaddr_in6 *)dst)->sin6_port; 3585 in6_pcbstate(in6p, IN6P_BOUND); 3586 } 3587 break; 3588 #endif 3589 } 3590 #ifdef INET6 3591 if (in6p && in6totcpcb(in6p)->t_family == AF_INET6 && sotoinpcb(oso)) { 3592 struct in6pcb *oin6p = sotoin6pcb(oso); 3593 /* inherit socket options from the listening socket */ 3594 in6p->in6p_flags |= (oin6p->in6p_flags & IN6P_CONTROLOPTS); 3595 if (in6p->in6p_flags & IN6P_CONTROLOPTS) { 3596 m_freem(in6p->in6p_options); 3597 in6p->in6p_options = 0; 3598 } 3599 ip6_savecontrol(in6p, &in6p->in6p_options, 3600 mtod(m, struct ip6_hdr *), m); 3601 } 3602 #endif 3603 3604 #if defined(IPSEC) || defined(FAST_IPSEC) 3605 /* 3606 * we make a copy of policy, instead of sharing the policy, 3607 * for better behavior in terms of SA lookup and dead SA removal. 3608 */ 3609 if (inp) { 3610 /* copy old policy into new socket's */ 3611 if (ipsec_copy_pcbpolicy(sotoinpcb(oso)->inp_sp, inp->inp_sp)) 3612 printf("tcp_input: could not copy policy\n"); 3613 } 3614 #ifdef INET6 3615 else if (in6p) { 3616 /* copy old policy into new socket's */ 3617 if (ipsec_copy_pcbpolicy(sotoin6pcb(oso)->in6p_sp, 3618 in6p->in6p_sp)) 3619 printf("tcp_input: could not copy policy\n"); 3620 } 3621 #endif 3622 #endif 3623 3624 /* 3625 * Give the new socket our cached route reference. 3626 */ 3627 if (inp) { 3628 inp->inp_route = sc->sc_route4; /* struct assignment */ 3629 /* XXX the following is gross */ 3630 if (inp->inp_route.ro_rt != NULL) { 3631 inp->inp_route.ro_rt->rt_refcnt++; 3632 rtcache(&inp->inp_route); 3633 rtflush(&sc->sc_route4); 3634 } 3635 } 3636 #ifdef INET6 3637 else { 3638 in6p->in6p_route = sc->sc_route6; 3639 /* XXX the following is gross */ 3640 if (in6p->in6p_route.ro_rt != NULL) { 3641 in6p->in6p_route.ro_rt->rt_refcnt++; 3642 rtcache((struct route *)&in6p->in6p_route); 3643 rtflush((struct route *)&sc->sc_route6); 3644 } 3645 } 3646 #endif 3647 3648 am = m_get(M_DONTWAIT, MT_SONAME); /* XXX */ 3649 if (am == NULL) 3650 goto resetandabort; 3651 MCLAIM(am, &tcp_mowner); 3652 am->m_len = src->sa_len; 3653 bcopy(src, mtod(am, caddr_t), src->sa_len); 3654 if (inp) { 3655 if (in_pcbconnect(inp, am, NULL)) { 3656 (void) m_free(am); 3657 goto resetandabort; 3658 } 3659 } 3660 #ifdef INET6 3661 else if (in6p) { 3662 if (src->sa_family == AF_INET) { 3663 /* IPv4 packet to AF_INET6 socket */ 3664 struct sockaddr_in6 *sin6; 3665 sin6 = mtod(am, struct sockaddr_in6 *); 3666 am->m_len = sizeof(*sin6); 3667 bzero(sin6, sizeof(*sin6)); 3668 sin6->sin6_family = AF_INET6; 3669 sin6->sin6_len = sizeof(*sin6); 3670 sin6->sin6_port = ((struct sockaddr_in *)src)->sin_port; 3671 sin6->sin6_addr.s6_addr16[5] = htons(0xffff); 3672 bcopy(&((struct sockaddr_in *)src)->sin_addr, 3673 &sin6->sin6_addr.s6_addr32[3], 3674 sizeof(sin6->sin6_addr.s6_addr32[3])); 3675 } 3676 if (in6_pcbconnect(in6p, am, NULL)) { 3677 (void) m_free(am); 3678 goto resetandabort; 3679 } 3680 } 3681 #endif 3682 else { 3683 (void) m_free(am); 3684 goto resetandabort; 3685 } 3686 (void) m_free(am); 3687 3688 if (inp) 3689 tp = intotcpcb(inp); 3690 #ifdef INET6 3691 else if (in6p) 3692 tp = in6totcpcb(in6p); 3693 #endif 3694 else 3695 tp = NULL; 3696 tp->t_flags = sototcpcb(oso)->t_flags & TF_NODELAY; 3697 if (sc->sc_request_r_scale != 15) { 3698 tp->requested_s_scale = sc->sc_requested_s_scale; 3699 tp->request_r_scale = sc->sc_request_r_scale; 3700 tp->snd_scale = sc->sc_requested_s_scale; 3701 tp->rcv_scale = sc->sc_request_r_scale; 3702 tp->t_flags |= TF_REQ_SCALE|TF_RCVD_SCALE; 3703 } 3704 if (sc->sc_flags & SCF_TIMESTAMP) 3705 tp->t_flags |= TF_REQ_TSTMP|TF_RCVD_TSTMP; 3706 tp->ts_timebase = sc->sc_timebase; 3707 3708 tp->t_template = tcp_template(tp); 3709 if (tp->t_template == 0) { 3710 tp = tcp_drop(tp, ENOBUFS); /* destroys socket */ 3711 so = NULL; 3712 m_freem(m); 3713 goto abort; 3714 } 3715 3716 tp->iss = sc->sc_iss; 3717 tp->irs = sc->sc_irs; 3718 tcp_sendseqinit(tp); 3719 tcp_rcvseqinit(tp); 3720 tp->t_state = TCPS_SYN_RECEIVED; 3721 TCP_TIMER_ARM(tp, TCPT_KEEP, TCPTV_KEEP_INIT); 3722 tcpstat.tcps_accepts++; 3723 3724 if ((sc->sc_flags & SCF_SACK_PERMIT) && tcp_do_sack) 3725 tp->t_flags |= TF_WILL_SACK; 3726 3727 if ((sc->sc_flags & SCF_ECN_PERMIT) && tcp_do_ecn) 3728 tp->t_flags |= TF_ECN_PERMIT; 3729 3730 #ifdef TCP_SIGNATURE 3731 if (sc->sc_flags & SCF_SIGNATURE) 3732 tp->t_flags |= TF_SIGNATURE; 3733 #endif 3734 3735 /* Initialize tp->t_ourmss before we deal with the peer's! */ 3736 tp->t_ourmss = sc->sc_ourmaxseg; 3737 tcp_mss_from_peer(tp, sc->sc_peermaxseg); 3738 3739 /* 3740 * Initialize the initial congestion window. If we 3741 * had to retransmit the SYN,ACK, we must initialize cwnd 3742 * to 1 segment (i.e. the Loss Window). 3743 */ 3744 if (sc->sc_rxtshift) 3745 tp->snd_cwnd = tp->t_peermss; 3746 else { 3747 int ss = tcp_init_win; 3748 #ifdef INET 3749 if (inp != NULL && in_localaddr(inp->inp_faddr)) 3750 ss = tcp_init_win_local; 3751 #endif 3752 #ifdef INET6 3753 if (in6p != NULL && in6_localaddr(&in6p->in6p_faddr)) 3754 ss = tcp_init_win_local; 3755 #endif 3756 tp->snd_cwnd = TCP_INITIAL_WINDOW(ss, tp->t_peermss); 3757 } 3758 3759 tcp_rmx_rtt(tp); 3760 tp->snd_wl1 = sc->sc_irs; 3761 tp->rcv_up = sc->sc_irs + 1; 3762 3763 /* 3764 * This is what whould have happened in tcp_output() when 3765 * the SYN,ACK was sent. 3766 */ 3767 tp->snd_up = tp->snd_una; 3768 tp->snd_max = tp->snd_nxt = tp->iss+1; 3769 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 3770 if (sc->sc_win > 0 && SEQ_GT(tp->rcv_nxt + sc->sc_win, tp->rcv_adv)) 3771 tp->rcv_adv = tp->rcv_nxt + sc->sc_win; 3772 tp->last_ack_sent = tp->rcv_nxt; 3773 tp->t_partialacks = -1; 3774 tp->t_dupacks = 0; 3775 3776 tcpstat.tcps_sc_completed++; 3777 s = splsoftnet(); 3778 SYN_CACHE_PUT(sc); 3779 splx(s); 3780 return (so); 3781 3782 resetandabort: 3783 (void)tcp_respond(NULL, m, m, th, (tcp_seq)0, th->th_ack, TH_RST); 3784 abort: 3785 if (so != NULL) 3786 (void) soabort(so); 3787 s = splsoftnet(); 3788 SYN_CACHE_PUT(sc); 3789 splx(s); 3790 tcpstat.tcps_sc_aborted++; 3791 return ((struct socket *)(-1)); 3792 } 3793 3794 /* 3795 * This function is called when we get a RST for a 3796 * non-existent connection, so that we can see if the 3797 * connection is in the syn cache. If it is, zap it. 3798 */ 3799 3800 void 3801 syn_cache_reset(struct sockaddr *src, struct sockaddr *dst, struct tcphdr *th) 3802 { 3803 struct syn_cache *sc; 3804 struct syn_cache_head *scp; 3805 int s = splsoftnet(); 3806 3807 if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) { 3808 splx(s); 3809 return; 3810 } 3811 if (SEQ_LT(th->th_seq, sc->sc_irs) || 3812 SEQ_GT(th->th_seq, sc->sc_irs+1)) { 3813 splx(s); 3814 return; 3815 } 3816 SYN_CACHE_RM(sc); 3817 tcpstat.tcps_sc_reset++; 3818 SYN_CACHE_PUT(sc); /* calls pool_put but see spl above */ 3819 splx(s); 3820 } 3821 3822 void 3823 syn_cache_unreach(const struct sockaddr *src, const struct sockaddr *dst, 3824 struct tcphdr *th) 3825 { 3826 struct syn_cache *sc; 3827 struct syn_cache_head *scp; 3828 int s; 3829 3830 s = splsoftnet(); 3831 if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) { 3832 splx(s); 3833 return; 3834 } 3835 /* If the sequence number != sc_iss, then it's a bogus ICMP msg */ 3836 if (ntohl (th->th_seq) != sc->sc_iss) { 3837 splx(s); 3838 return; 3839 } 3840 3841 /* 3842 * If we've retransmitted 3 times and this is our second error, 3843 * we remove the entry. Otherwise, we allow it to continue on. 3844 * This prevents us from incorrectly nuking an entry during a 3845 * spurious network outage. 3846 * 3847 * See tcp_notify(). 3848 */ 3849 if ((sc->sc_flags & SCF_UNREACH) == 0 || sc->sc_rxtshift < 3) { 3850 sc->sc_flags |= SCF_UNREACH; 3851 splx(s); 3852 return; 3853 } 3854 3855 SYN_CACHE_RM(sc); 3856 tcpstat.tcps_sc_unreach++; 3857 SYN_CACHE_PUT(sc); /* calls pool_put but see spl above */ 3858 splx(s); 3859 } 3860 3861 /* 3862 * Given a LISTEN socket and an inbound SYN request, add 3863 * this to the syn cache, and send back a segment: 3864 * <SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK> 3865 * to the source. 3866 * 3867 * IMPORTANT NOTE: We do _NOT_ ACK data that might accompany the SYN. 3868 * Doing so would require that we hold onto the data and deliver it 3869 * to the application. However, if we are the target of a SYN-flood 3870 * DoS attack, an attacker could send data which would eventually 3871 * consume all available buffer space if it were ACKed. By not ACKing 3872 * the data, we avoid this DoS scenario. 3873 */ 3874 3875 int 3876 syn_cache_add(struct sockaddr *src, struct sockaddr *dst, struct tcphdr *th, 3877 unsigned int hlen, struct socket *so, struct mbuf *m, u_char *optp, 3878 int optlen, struct tcp_opt_info *oi) 3879 { 3880 struct tcpcb tb, *tp; 3881 long win; 3882 struct syn_cache *sc; 3883 struct syn_cache_head *scp; 3884 struct mbuf *ipopts; 3885 struct tcp_opt_info opti; 3886 int s; 3887 3888 tp = sototcpcb(so); 3889 3890 bzero(&opti, sizeof(opti)); 3891 3892 /* 3893 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN 3894 * 3895 * Note this check is performed in tcp_input() very early on. 3896 */ 3897 3898 /* 3899 * Initialize some local state. 3900 */ 3901 win = sbspace(&so->so_rcv); 3902 if (win > TCP_MAXWIN) 3903 win = TCP_MAXWIN; 3904 3905 switch (src->sa_family) { 3906 #ifdef INET 3907 case AF_INET: 3908 /* 3909 * Remember the IP options, if any. 3910 */ 3911 ipopts = ip_srcroute(); 3912 break; 3913 #endif 3914 default: 3915 ipopts = NULL; 3916 } 3917 3918 #ifdef TCP_SIGNATURE 3919 if (optp || (tp->t_flags & TF_SIGNATURE)) 3920 #else 3921 if (optp) 3922 #endif 3923 { 3924 tb.t_flags = tcp_do_rfc1323 ? (TF_REQ_SCALE|TF_REQ_TSTMP) : 0; 3925 #ifdef TCP_SIGNATURE 3926 tb.t_flags |= (tp->t_flags & TF_SIGNATURE); 3927 #endif 3928 tb.t_state = TCPS_LISTEN; 3929 if (tcp_dooptions(&tb, optp, optlen, th, m, m->m_pkthdr.len - 3930 sizeof(struct tcphdr) - optlen - hlen, oi) < 0) 3931 return (0); 3932 } else 3933 tb.t_flags = 0; 3934 3935 /* 3936 * See if we already have an entry for this connection. 3937 * If we do, resend the SYN,ACK. We do not count this 3938 * as a retransmission (XXX though maybe we should). 3939 */ 3940 if ((sc = syn_cache_lookup(src, dst, &scp)) != NULL) { 3941 tcpstat.tcps_sc_dupesyn++; 3942 if (ipopts) { 3943 /* 3944 * If we were remembering a previous source route, 3945 * forget it and use the new one we've been given. 3946 */ 3947 if (sc->sc_ipopts) 3948 (void) m_free(sc->sc_ipopts); 3949 sc->sc_ipopts = ipopts; 3950 } 3951 sc->sc_timestamp = tb.ts_recent; 3952 if (syn_cache_respond(sc, m) == 0) { 3953 tcpstat.tcps_sndacks++; 3954 tcpstat.tcps_sndtotal++; 3955 } 3956 return (1); 3957 } 3958 3959 s = splsoftnet(); 3960 sc = pool_get(&syn_cache_pool, PR_NOWAIT); 3961 splx(s); 3962 if (sc == NULL) { 3963 if (ipopts) 3964 (void) m_free(ipopts); 3965 return (0); 3966 } 3967 3968 /* 3969 * Fill in the cache, and put the necessary IP and TCP 3970 * options into the reply. 3971 */ 3972 bzero(sc, sizeof(struct syn_cache)); 3973 callout_init(&sc->sc_timer); 3974 bcopy(src, &sc->sc_src, src->sa_len); 3975 bcopy(dst, &sc->sc_dst, dst->sa_len); 3976 sc->sc_flags = 0; 3977 sc->sc_ipopts = ipopts; 3978 sc->sc_irs = th->th_seq; 3979 switch (src->sa_family) { 3980 #ifdef INET 3981 case AF_INET: 3982 { 3983 struct sockaddr_in *srcin = (void *) src; 3984 struct sockaddr_in *dstin = (void *) dst; 3985 3986 sc->sc_iss = tcp_new_iss1(&dstin->sin_addr, 3987 &srcin->sin_addr, dstin->sin_port, 3988 srcin->sin_port, sizeof(dstin->sin_addr), 0); 3989 break; 3990 } 3991 #endif /* INET */ 3992 #ifdef INET6 3993 case AF_INET6: 3994 { 3995 struct sockaddr_in6 *srcin6 = (void *) src; 3996 struct sockaddr_in6 *dstin6 = (void *) dst; 3997 3998 sc->sc_iss = tcp_new_iss1(&dstin6->sin6_addr, 3999 &srcin6->sin6_addr, dstin6->sin6_port, 4000 srcin6->sin6_port, sizeof(dstin6->sin6_addr), 0); 4001 break; 4002 } 4003 #endif /* INET6 */ 4004 } 4005 sc->sc_peermaxseg = oi->maxseg; 4006 sc->sc_ourmaxseg = tcp_mss_to_advertise(m->m_flags & M_PKTHDR ? 4007 m->m_pkthdr.rcvif : NULL, 4008 sc->sc_src.sa.sa_family); 4009 sc->sc_win = win; 4010 sc->sc_timebase = tcp_now; /* see tcp_newtcpcb() */ 4011 sc->sc_timestamp = tb.ts_recent; 4012 if ((tb.t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) == 4013 (TF_REQ_TSTMP|TF_RCVD_TSTMP)) 4014 sc->sc_flags |= SCF_TIMESTAMP; 4015 if ((tb.t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == 4016 (TF_RCVD_SCALE|TF_REQ_SCALE)) { 4017 sc->sc_requested_s_scale = tb.requested_s_scale; 4018 sc->sc_request_r_scale = 0; 4019 while (sc->sc_request_r_scale < TCP_MAX_WINSHIFT && 4020 TCP_MAXWIN << sc->sc_request_r_scale < 4021 so->so_rcv.sb_hiwat) 4022 sc->sc_request_r_scale++; 4023 } else { 4024 sc->sc_requested_s_scale = 15; 4025 sc->sc_request_r_scale = 15; 4026 } 4027 if ((tb.t_flags & TF_SACK_PERMIT) && tcp_do_sack) 4028 sc->sc_flags |= SCF_SACK_PERMIT; 4029 4030 /* 4031 * ECN setup packet recieved. 4032 */ 4033 if ((th->th_flags & (TH_ECE|TH_CWR)) && tcp_do_ecn) 4034 sc->sc_flags |= SCF_ECN_PERMIT; 4035 4036 #ifdef TCP_SIGNATURE 4037 if (tb.t_flags & TF_SIGNATURE) 4038 sc->sc_flags |= SCF_SIGNATURE; 4039 #endif 4040 sc->sc_tp = tp; 4041 if (syn_cache_respond(sc, m) == 0) { 4042 syn_cache_insert(sc, tp); 4043 tcpstat.tcps_sndacks++; 4044 tcpstat.tcps_sndtotal++; 4045 } else { 4046 s = splsoftnet(); 4047 SYN_CACHE_PUT(sc); 4048 splx(s); 4049 tcpstat.tcps_sc_dropped++; 4050 } 4051 return (1); 4052 } 4053 4054 int 4055 syn_cache_respond(struct syn_cache *sc, struct mbuf *m) 4056 { 4057 struct route *ro; 4058 u_int8_t *optp; 4059 int optlen, error; 4060 u_int16_t tlen; 4061 struct ip *ip = NULL; 4062 #ifdef INET6 4063 struct ip6_hdr *ip6 = NULL; 4064 #endif 4065 struct tcpcb *tp = NULL; 4066 struct tcphdr *th; 4067 u_int hlen; 4068 struct socket *so; 4069 4070 switch (sc->sc_src.sa.sa_family) { 4071 case AF_INET: 4072 hlen = sizeof(struct ip); 4073 ro = &sc->sc_route4; 4074 break; 4075 #ifdef INET6 4076 case AF_INET6: 4077 hlen = sizeof(struct ip6_hdr); 4078 ro = (struct route *)&sc->sc_route6; 4079 break; 4080 #endif 4081 default: 4082 if (m) 4083 m_freem(m); 4084 return (EAFNOSUPPORT); 4085 } 4086 4087 /* Compute the size of the TCP options. */ 4088 optlen = 4 + (sc->sc_request_r_scale != 15 ? 4 : 0) + 4089 ((sc->sc_flags & SCF_SACK_PERMIT) ? (TCPOLEN_SACK_PERMITTED + 2) : 0) + 4090 #ifdef TCP_SIGNATURE 4091 ((sc->sc_flags & SCF_SIGNATURE) ? (TCPOLEN_SIGNATURE + 2) : 0) + 4092 #endif 4093 ((sc->sc_flags & SCF_TIMESTAMP) ? TCPOLEN_TSTAMP_APPA : 0); 4094 4095 tlen = hlen + sizeof(struct tcphdr) + optlen; 4096 4097 /* 4098 * Create the IP+TCP header from scratch. 4099 */ 4100 if (m) 4101 m_freem(m); 4102 #ifdef DIAGNOSTIC 4103 if (max_linkhdr + tlen > MCLBYTES) 4104 return (ENOBUFS); 4105 #endif 4106 MGETHDR(m, M_DONTWAIT, MT_DATA); 4107 if (m && tlen > MHLEN) { 4108 MCLGET(m, M_DONTWAIT); 4109 if ((m->m_flags & M_EXT) == 0) { 4110 m_freem(m); 4111 m = NULL; 4112 } 4113 } 4114 if (m == NULL) 4115 return (ENOBUFS); 4116 MCLAIM(m, &tcp_tx_mowner); 4117 4118 /* Fixup the mbuf. */ 4119 m->m_data += max_linkhdr; 4120 m->m_len = m->m_pkthdr.len = tlen; 4121 if (sc->sc_tp) { 4122 tp = sc->sc_tp; 4123 if (tp->t_inpcb) 4124 so = tp->t_inpcb->inp_socket; 4125 #ifdef INET6 4126 else if (tp->t_in6pcb) 4127 so = tp->t_in6pcb->in6p_socket; 4128 #endif 4129 else 4130 so = NULL; 4131 } else 4132 so = NULL; 4133 m->m_pkthdr.rcvif = NULL; 4134 memset(mtod(m, u_char *), 0, tlen); 4135 4136 switch (sc->sc_src.sa.sa_family) { 4137 case AF_INET: 4138 ip = mtod(m, struct ip *); 4139 ip->ip_v = 4; 4140 ip->ip_dst = sc->sc_src.sin.sin_addr; 4141 ip->ip_src = sc->sc_dst.sin.sin_addr; 4142 ip->ip_p = IPPROTO_TCP; 4143 th = (struct tcphdr *)(ip + 1); 4144 th->th_dport = sc->sc_src.sin.sin_port; 4145 th->th_sport = sc->sc_dst.sin.sin_port; 4146 break; 4147 #ifdef INET6 4148 case AF_INET6: 4149 ip6 = mtod(m, struct ip6_hdr *); 4150 ip6->ip6_vfc = IPV6_VERSION; 4151 ip6->ip6_dst = sc->sc_src.sin6.sin6_addr; 4152 ip6->ip6_src = sc->sc_dst.sin6.sin6_addr; 4153 ip6->ip6_nxt = IPPROTO_TCP; 4154 /* ip6_plen will be updated in ip6_output() */ 4155 th = (struct tcphdr *)(ip6 + 1); 4156 th->th_dport = sc->sc_src.sin6.sin6_port; 4157 th->th_sport = sc->sc_dst.sin6.sin6_port; 4158 break; 4159 #endif 4160 default: 4161 th = NULL; 4162 } 4163 4164 th->th_seq = htonl(sc->sc_iss); 4165 th->th_ack = htonl(sc->sc_irs + 1); 4166 th->th_off = (sizeof(struct tcphdr) + optlen) >> 2; 4167 th->th_flags = TH_SYN|TH_ACK; 4168 th->th_win = htons(sc->sc_win); 4169 /* th_sum already 0 */ 4170 /* th_urp already 0 */ 4171 4172 /* Tack on the TCP options. */ 4173 optp = (u_int8_t *)(th + 1); 4174 *optp++ = TCPOPT_MAXSEG; 4175 *optp++ = 4; 4176 *optp++ = (sc->sc_ourmaxseg >> 8) & 0xff; 4177 *optp++ = sc->sc_ourmaxseg & 0xff; 4178 4179 if (sc->sc_request_r_scale != 15) { 4180 *((u_int32_t *)optp) = htonl(TCPOPT_NOP << 24 | 4181 TCPOPT_WINDOW << 16 | TCPOLEN_WINDOW << 8 | 4182 sc->sc_request_r_scale); 4183 optp += 4; 4184 } 4185 4186 if (sc->sc_flags & SCF_TIMESTAMP) { 4187 u_int32_t *lp = (u_int32_t *)(optp); 4188 /* Form timestamp option as shown in appendix A of RFC 1323. */ 4189 *lp++ = htonl(TCPOPT_TSTAMP_HDR); 4190 *lp++ = htonl(SYN_CACHE_TIMESTAMP(sc)); 4191 *lp = htonl(sc->sc_timestamp); 4192 optp += TCPOLEN_TSTAMP_APPA; 4193 } 4194 4195 if (sc->sc_flags & SCF_SACK_PERMIT) { 4196 u_int8_t *p = optp; 4197 4198 /* Let the peer know that we will SACK. */ 4199 p[0] = TCPOPT_SACK_PERMITTED; 4200 p[1] = 2; 4201 p[2] = TCPOPT_NOP; 4202 p[3] = TCPOPT_NOP; 4203 optp += 4; 4204 } 4205 4206 /* 4207 * Send ECN SYN-ACK setup packet. 4208 * Routes can be asymetric, so, even if we receive a packet 4209 * with ECE and CWR set, we must not assume no one will block 4210 * the ECE packet we are about to send. 4211 */ 4212 if ((sc->sc_flags & SCF_ECN_PERMIT) && tp && 4213 SEQ_GEQ(tp->snd_nxt, tp->snd_max)) { 4214 th->th_flags |= TH_ECE; 4215 tcpstat.tcps_ecn_shs++; 4216 4217 /* 4218 * draft-ietf-tcpm-ecnsyn-00.txt 4219 * 4220 * "[...] a TCP node MAY respond to an ECN-setup 4221 * SYN packet by setting ECT in the responding 4222 * ECN-setup SYN/ACK packet, indicating to routers 4223 * that the SYN/ACK packet is ECN-Capable. 4224 * This allows a congested router along the path 4225 * to mark the packet instead of dropping the 4226 * packet as an indication of congestion." 4227 * 4228 * "[...] There can be a great benefit in setting 4229 * an ECN-capable codepoint in SYN/ACK packets [...] 4230 * Congestion is most likely to occur in 4231 * the server-to-client direction. As a result, 4232 * setting an ECN-capable codepoint in SYN/ACK 4233 * packets can reduce the occurence of three-second 4234 * retransmit timeouts resulting from the drop 4235 * of SYN/ACK packets." 4236 * 4237 * Page 4 and 6, January 2006. 4238 */ 4239 4240 switch (sc->sc_src.sa.sa_family) { 4241 #ifdef INET 4242 case AF_INET: 4243 ip->ip_tos |= IPTOS_ECN_ECT0; 4244 break; 4245 #endif 4246 #ifdef INET6 4247 case AF_INET6: 4248 ip6->ip6_flow |= htonl(IPTOS_ECN_ECT0 << 20); 4249 break; 4250 #endif 4251 } 4252 tcpstat.tcps_ecn_ect++; 4253 } 4254 4255 #ifdef TCP_SIGNATURE 4256 if (sc->sc_flags & SCF_SIGNATURE) { 4257 struct secasvar *sav; 4258 u_int8_t *sigp; 4259 4260 sav = tcp_signature_getsav(m, th); 4261 4262 if (sav == NULL) { 4263 if (m) 4264 m_freem(m); 4265 return (EPERM); 4266 } 4267 4268 *optp++ = TCPOPT_SIGNATURE; 4269 *optp++ = TCPOLEN_SIGNATURE; 4270 sigp = optp; 4271 bzero(optp, TCP_SIGLEN); 4272 optp += TCP_SIGLEN; 4273 *optp++ = TCPOPT_NOP; 4274 *optp++ = TCPOPT_EOL; 4275 4276 (void)tcp_signature(m, th, hlen, sav, sigp); 4277 4278 key_sa_recordxfer(sav, m); 4279 #ifdef FAST_IPSEC 4280 KEY_FREESAV(&sav); 4281 #else 4282 key_freesav(sav); 4283 #endif 4284 } 4285 #endif 4286 4287 /* Compute the packet's checksum. */ 4288 switch (sc->sc_src.sa.sa_family) { 4289 case AF_INET: 4290 ip->ip_len = htons(tlen - hlen); 4291 th->th_sum = 0; 4292 th->th_sum = in4_cksum(m, IPPROTO_TCP, hlen, tlen - hlen); 4293 break; 4294 #ifdef INET6 4295 case AF_INET6: 4296 ip6->ip6_plen = htons(tlen - hlen); 4297 th->th_sum = 0; 4298 th->th_sum = in6_cksum(m, IPPROTO_TCP, hlen, tlen - hlen); 4299 break; 4300 #endif 4301 } 4302 4303 /* 4304 * Fill in some straggling IP bits. Note the stack expects 4305 * ip_len to be in host order, for convenience. 4306 */ 4307 switch (sc->sc_src.sa.sa_family) { 4308 #ifdef INET 4309 case AF_INET: 4310 ip->ip_len = htons(tlen); 4311 ip->ip_ttl = ip_defttl; 4312 /* XXX tos? */ 4313 break; 4314 #endif 4315 #ifdef INET6 4316 case AF_INET6: 4317 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 4318 ip6->ip6_vfc |= IPV6_VERSION; 4319 ip6->ip6_plen = htons(tlen - hlen); 4320 /* ip6_hlim will be initialized afterwards */ 4321 /* XXX flowlabel? */ 4322 break; 4323 #endif 4324 } 4325 4326 /* XXX use IPsec policy on listening socket, on SYN ACK */ 4327 tp = sc->sc_tp; 4328 4329 switch (sc->sc_src.sa.sa_family) { 4330 #ifdef INET 4331 case AF_INET: 4332 error = ip_output(m, sc->sc_ipopts, ro, 4333 (ip_mtudisc ? IP_MTUDISC : 0), 4334 (struct ip_moptions *)NULL, so); 4335 break; 4336 #endif 4337 #ifdef INET6 4338 case AF_INET6: 4339 ip6->ip6_hlim = in6_selecthlim(NULL, 4340 ro->ro_rt ? ro->ro_rt->rt_ifp : NULL); 4341 4342 error = ip6_output(m, NULL /*XXX*/, (struct route_in6 *)ro, 0, 4343 (struct ip6_moptions *)0, so, NULL); 4344 break; 4345 #endif 4346 default: 4347 error = EAFNOSUPPORT; 4348 break; 4349 } 4350 return (error); 4351 } 4352
Indexes created Sun Sep 21 20:09:37 GMT 2025