tcp_subr.c revision 1.139
1 /* $NetBSD: tcp_subr.c,v 1.139 2003/03/01 04:40:28 thorpej Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /*- 33 * Copyright (c) 1997, 1998, 2000, 2001 The NetBSD Foundation, Inc. 34 * All rights reserved. 35 * 36 * This code is derived from software contributed to The NetBSD Foundation 37 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 38 * Facility, NASA Ames Research Center. 39 * 40 * Redistribution and use in source and binary forms, with or without 41 * modification, are permitted provided that the following conditions 42 * are met: 43 * 1. Redistributions of source code must retain the above copyright 44 * notice, this list of conditions and the following disclaimer. 45 * 2. Redistributions in binary form must reproduce the above copyright 46 * notice, this list of conditions and the following disclaimer in the 47 * documentation and/or other materials provided with the distribution. 48 * 3. All advertising materials mentioning features or use of this software 49 * must display the following acknowledgement: 50 * This product includes software developed by the NetBSD 51 * Foundation, Inc. and its contributors. 52 * 4. Neither the name of The NetBSD Foundation nor the names of its 53 * contributors may be used to endorse or promote products derived 54 * from this software without specific prior written permission. 55 * 56 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 57 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 58 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 59 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 60 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 61 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 62 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 63 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 64 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 65 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 66 * POSSIBILITY OF SUCH DAMAGE. 67 */ 68 69 /* 70 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 71 * The Regents of the University of California. All rights reserved. 72 * 73 * Redistribution and use in source and binary forms, with or without 74 * modification, are permitted provided that the following conditions 75 * are met: 76 * 1. Redistributions of source code must retain the above copyright 77 * notice, this list of conditions and the following disclaimer. 78 * 2. Redistributions in binary form must reproduce the above copyright 79 * notice, this list of conditions and the following disclaimer in the 80 * documentation and/or other materials provided with the distribution. 81 * 3. All advertising materials mentioning features or use of this software 82 * must display the following acknowledgement: 83 * This product includes software developed by the University of 84 * California, Berkeley and its contributors. 85 * 4. Neither the name of the University nor the names of its contributors 86 * may be used to endorse or promote products derived from this software 87 * without specific prior written permission. 88 * 89 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 90 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 91 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 92 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 93 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 94 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 95 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 96 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 97 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 98 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 99 * SUCH DAMAGE. 100 * 101 * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 102 */ 103 104 #include <sys/cdefs.h> 105 __KERNEL_RCSID(0, "$NetBSD: tcp_subr.c,v 1.139 2003/03/01 04:40:28 thorpej Exp $"); 106 107 #include "opt_inet.h" 108 #include "opt_ipsec.h" 109 #include "opt_tcp_compat_42.h" 110 #include "opt_inet_csum.h" 111 #include "rnd.h" 112 113 #include <sys/param.h> 114 #include <sys/proc.h> 115 #include <sys/systm.h> 116 #include <sys/malloc.h> 117 #include <sys/mbuf.h> 118 #include <sys/socket.h> 119 #include <sys/socketvar.h> 120 #include <sys/protosw.h> 121 #include <sys/errno.h> 122 #include <sys/kernel.h> 123 #include <sys/pool.h> 124 #if NRND > 0 125 #include <sys/md5.h> 126 #include <sys/rnd.h> 127 #endif 128 129 #include <net/route.h> 130 #include <net/if.h> 131 132 #include <netinet/in.h> 133 #include <netinet/in_systm.h> 134 #include <netinet/ip.h> 135 #include <netinet/in_pcb.h> 136 #include <netinet/ip_var.h> 137 #include <netinet/ip_icmp.h> 138 139 #ifdef INET6 140 #ifndef INET 141 #include <netinet/in.h> 142 #endif 143 #include <netinet/ip6.h> 144 #include <netinet6/in6_pcb.h> 145 #include <netinet6/ip6_var.h> 146 #include <netinet6/in6_var.h> 147 #include <netinet6/ip6protosw.h> 148 #include <netinet/icmp6.h> 149 #include <netinet6/nd6.h> 150 #endif 151 152 #include <netinet/tcp.h> 153 #include <netinet/tcp_fsm.h> 154 #include <netinet/tcp_seq.h> 155 #include <netinet/tcp_timer.h> 156 #include <netinet/tcp_var.h> 157 #include <netinet/tcpip.h> 158 159 #ifdef IPSEC 160 #include <netinet6/ipsec.h> 161 #endif /*IPSEC*/ 162 163 #ifdef INET6 164 struct in6pcb tcb6; 165 #endif 166 167 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 168 struct tcpstat tcpstat; /* tcp statistics */ 169 u_int32_t tcp_now; /* for RFC 1323 timestamps */ 170 171 /* patchable/settable parameters for tcp */ 172 int tcp_mssdflt = TCP_MSS; 173 int tcp_rttdflt = TCPTV_SRTTDFLT / PR_SLOWHZ; 174 int tcp_do_rfc1323 = 1; /* window scaling / timestamps (obsolete) */ 175 #if NRND > 0 176 int tcp_do_rfc1948 = 0; /* ISS by cryptographic hash */ 177 #endif 178 int tcp_do_sack = 1; /* selective acknowledgement */ 179 int tcp_do_win_scale = 1; /* RFC1323 window scaling */ 180 int tcp_do_timestamps = 1; /* RFC1323 timestamps */ 181 int tcp_do_newreno = 0; /* Use the New Reno algorithms */ 182 int tcp_ack_on_push = 0; /* set to enable immediate ACK-on-PUSH */ 183 int tcp_init_win = 1; /* initial slow start window */ 184 int tcp_init_win_local = 4; /* initial slow start window for local nets */ 185 int tcp_mss_ifmtu = 0; 186 #ifdef TCP_COMPAT_42 187 int tcp_compat_42 = 1; 188 #else 189 int tcp_compat_42 = 0; 190 #endif 191 int tcp_rst_ppslim = 100; /* 100pps */ 192 193 /* tcb hash */ 194 #ifndef TCBHASHSIZE 195 #define TCBHASHSIZE 128 196 #endif 197 int tcbhashsize = TCBHASHSIZE; 198 199 /* syn hash parameters */ 200 #define TCP_SYN_HASH_SIZE 293 201 #define TCP_SYN_BUCKET_SIZE 35 202 int tcp_syn_cache_size = TCP_SYN_HASH_SIZE; 203 int tcp_syn_cache_limit = TCP_SYN_HASH_SIZE*TCP_SYN_BUCKET_SIZE; 204 int tcp_syn_bucket_limit = 3*TCP_SYN_BUCKET_SIZE; 205 struct syn_cache_head tcp_syn_cache[TCP_SYN_HASH_SIZE]; 206 207 int tcp_freeq __P((struct tcpcb *)); 208 209 #ifdef INET 210 void tcp_mtudisc_callback __P((struct in_addr)); 211 #endif 212 #ifdef INET6 213 void tcp6_mtudisc_callback __P((struct in6_addr *)); 214 #endif 215 216 void tcp_mtudisc __P((struct inpcb *, int)); 217 #ifdef INET6 218 void tcp6_mtudisc __P((struct in6pcb *, int)); 219 #endif 220 221 struct pool tcpcb_pool; 222 223 #ifdef TCP_CSUM_COUNTERS 224 #include <sys/device.h> 225 226 struct evcnt tcp_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 227 NULL, "tcp", "hwcsum bad"); 228 struct evcnt tcp_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 229 NULL, "tcp", "hwcsum ok"); 230 struct evcnt tcp_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 231 NULL, "tcp", "hwcsum data"); 232 struct evcnt tcp_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 233 NULL, "tcp", "swcsum"); 234 #endif /* TCP_CSUM_COUNTERS */ 235 236 #ifdef TCP_OUTPUT_COUNTERS 237 #include <sys/device.h> 238 239 struct evcnt tcp_output_bigheader = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 240 NULL, "tcp", "output big header"); 241 struct evcnt tcp_output_copysmall = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 242 NULL, "tcp", "output copy small"); 243 struct evcnt tcp_output_copybig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 244 NULL, "tcp", "output copy big"); 245 struct evcnt tcp_output_refbig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 246 NULL, "tcp", "output reference big"); 247 #endif /* TCP_OUTPUT_COUNTERS */ 248 249 #ifdef TCP_REASS_COUNTERS 250 #include <sys/device.h> 251 252 struct evcnt tcp_reass_ = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 253 NULL, "tcp_reass", "calls"); 254 struct evcnt tcp_reass_empty = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 255 &tcp_reass_, "tcp_reass", "insert into empty queue"); 256 struct evcnt tcp_reass_iteration[8] = { 257 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", ">7 iterations"), 258 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "1 iteration"), 259 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "2 iterations"), 260 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "3 iterations"), 261 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "4 iterations"), 262 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "5 iterations"), 263 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "6 iterations"), 264 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "7 iterations"), 265 }; 266 struct evcnt tcp_reass_prependfirst = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 267 &tcp_reass_, "tcp_reass", "prepend to first"); 268 struct evcnt tcp_reass_prepend = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 269 &tcp_reass_, "tcp_reass", "prepend"); 270 struct evcnt tcp_reass_insert = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 271 &tcp_reass_, "tcp_reass", "insert"); 272 struct evcnt tcp_reass_inserttail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 273 &tcp_reass_, "tcp_reass", "insert at tail"); 274 struct evcnt tcp_reass_append = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 275 &tcp_reass_, "tcp_reass", "append"); 276 struct evcnt tcp_reass_appendtail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 277 &tcp_reass_, "tcp_reass", "append to tail fragment"); 278 struct evcnt tcp_reass_overlaptail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 279 &tcp_reass_, "tcp_reass", "overlap at end"); 280 struct evcnt tcp_reass_overlapfront = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 281 &tcp_reass_, "tcp_reass", "overlap at start"); 282 struct evcnt tcp_reass_segdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 283 &tcp_reass_, "tcp_reass", "duplicate segment"); 284 struct evcnt tcp_reass_fragdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 285 &tcp_reass_, "tcp_reass", "duplicate fragment"); 286 287 #endif /* TCP_REASS_COUNTERS */ 288 289 #ifdef MBUFTRACE 290 struct mowner tcp_mowner = { "tcp" }; 291 struct mowner tcp_rx_mowner = { "tcp", "rx" }; 292 struct mowner tcp_tx_mowner = { "tcp", "tx" }; 293 #endif 294 295 /* 296 * Tcp initialization 297 */ 298 void 299 tcp_init() 300 { 301 int hlen; 302 303 pool_init(&tcpcb_pool, sizeof(struct tcpcb), 0, 0, 0, "tcpcbpl", 304 NULL); 305 in_pcbinit(&tcbtable, tcbhashsize, tcbhashsize); 306 #ifdef INET6 307 tcb6.in6p_next = tcb6.in6p_prev = &tcb6; 308 #endif 309 310 hlen = sizeof(struct ip) + sizeof(struct tcphdr); 311 #ifdef INET6 312 if (sizeof(struct ip) < sizeof(struct ip6_hdr)) 313 hlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 314 #endif 315 if (max_protohdr < hlen) 316 max_protohdr = hlen; 317 if (max_linkhdr + hlen > MHLEN) 318 panic("tcp_init"); 319 320 #ifdef INET 321 icmp_mtudisc_callback_register(tcp_mtudisc_callback); 322 #endif 323 #ifdef INET6 324 icmp6_mtudisc_callback_register(tcp6_mtudisc_callback); 325 #endif 326 327 /* Initialize timer state. */ 328 tcp_timer_init(); 329 330 /* Initialize the compressed state engine. */ 331 syn_cache_init(); 332 333 #ifdef TCP_CSUM_COUNTERS 334 evcnt_attach_static(&tcp_hwcsum_bad); 335 evcnt_attach_static(&tcp_hwcsum_ok); 336 evcnt_attach_static(&tcp_hwcsum_data); 337 evcnt_attach_static(&tcp_swcsum); 338 #endif /* TCP_CSUM_COUNTERS */ 339 340 #ifdef TCP_OUTPUT_COUNTERS 341 evcnt_attach_static(&tcp_output_bigheader); 342 evcnt_attach_static(&tcp_output_copysmall); 343 evcnt_attach_static(&tcp_output_copybig); 344 evcnt_attach_static(&tcp_output_refbig); 345 #endif /* TCP_OUTPUT_COUNTERS */ 346 347 #ifdef TCP_REASS_COUNTERS 348 evcnt_attach_static(&tcp_reass_); 349 evcnt_attach_static(&tcp_reass_empty); 350 evcnt_attach_static(&tcp_reass_iteration[0]); 351 evcnt_attach_static(&tcp_reass_iteration[1]); 352 evcnt_attach_static(&tcp_reass_iteration[2]); 353 evcnt_attach_static(&tcp_reass_iteration[3]); 354 evcnt_attach_static(&tcp_reass_iteration[4]); 355 evcnt_attach_static(&tcp_reass_iteration[5]); 356 evcnt_attach_static(&tcp_reass_iteration[6]); 357 evcnt_attach_static(&tcp_reass_iteration[7]); 358 evcnt_attach_static(&tcp_reass_prependfirst); 359 evcnt_attach_static(&tcp_reass_prepend); 360 evcnt_attach_static(&tcp_reass_insert); 361 evcnt_attach_static(&tcp_reass_inserttail); 362 evcnt_attach_static(&tcp_reass_append); 363 evcnt_attach_static(&tcp_reass_appendtail); 364 evcnt_attach_static(&tcp_reass_overlaptail); 365 evcnt_attach_static(&tcp_reass_overlapfront); 366 evcnt_attach_static(&tcp_reass_segdup); 367 evcnt_attach_static(&tcp_reass_fragdup); 368 #endif /* TCP_REASS_COUNTERS */ 369 370 MOWNER_ATTACH(&tcp_tx_mowner); 371 MOWNER_ATTACH(&tcp_rx_mowner); 372 MOWNER_ATTACH(&tcp_mowner); 373 } 374 375 /* 376 * Create template to be used to send tcp packets on a connection. 377 * Call after host entry created, allocates an mbuf and fills 378 * in a skeletal tcp/ip header, minimizing the amount of work 379 * necessary when the connection is used. 380 */ 381 struct mbuf * 382 tcp_template(tp) 383 struct tcpcb *tp; 384 { 385 struct inpcb *inp = tp->t_inpcb; 386 #ifdef INET6 387 struct in6pcb *in6p = tp->t_in6pcb; 388 #endif 389 struct tcphdr *n; 390 struct mbuf *m; 391 int hlen; 392 393 switch (tp->t_family) { 394 case AF_INET: 395 hlen = sizeof(struct ip); 396 if (inp) 397 break; 398 #ifdef INET6 399 if (in6p) { 400 /* mapped addr case */ 401 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 402 && IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 403 break; 404 } 405 #endif 406 return NULL; /*EINVAL*/ 407 #ifdef INET6 408 case AF_INET6: 409 hlen = sizeof(struct ip6_hdr); 410 if (in6p) { 411 /* more sainty check? */ 412 break; 413 } 414 return NULL; /*EINVAL*/ 415 #endif 416 default: 417 hlen = 0; /*pacify gcc*/ 418 return NULL; /*EAFNOSUPPORT*/ 419 } 420 #ifdef DIAGNOSTIC 421 if (hlen + sizeof(struct tcphdr) > MCLBYTES) 422 panic("mclbytes too small for t_template"); 423 #endif 424 m = tp->t_template; 425 if (m && m->m_len == hlen + sizeof(struct tcphdr)) 426 ; 427 else { 428 if (m) 429 m_freem(m); 430 m = tp->t_template = NULL; 431 MGETHDR(m, M_DONTWAIT, MT_HEADER); 432 if (m && hlen + sizeof(struct tcphdr) > MHLEN) { 433 MCLGET(m, M_DONTWAIT); 434 if ((m->m_flags & M_EXT) == 0) { 435 m_free(m); 436 m = NULL; 437 } 438 } 439 if (m == NULL) 440 return NULL; 441 MCLAIM(m, &tcp_mowner); 442 m->m_pkthdr.len = m->m_len = hlen + sizeof(struct tcphdr); 443 } 444 445 bzero(mtod(m, caddr_t), m->m_len); 446 447 n = (struct tcphdr *)(mtod(m, caddr_t) + hlen); 448 449 switch (tp->t_family) { 450 case AF_INET: 451 { 452 struct ipovly *ipov; 453 mtod(m, struct ip *)->ip_v = 4; 454 ipov = mtod(m, struct ipovly *); 455 ipov->ih_pr = IPPROTO_TCP; 456 ipov->ih_len = htons(sizeof(struct tcphdr)); 457 if (inp) { 458 ipov->ih_src = inp->inp_laddr; 459 ipov->ih_dst = inp->inp_faddr; 460 } 461 #ifdef INET6 462 else if (in6p) { 463 /* mapped addr case */ 464 bcopy(&in6p->in6p_laddr.s6_addr32[3], &ipov->ih_src, 465 sizeof(ipov->ih_src)); 466 bcopy(&in6p->in6p_faddr.s6_addr32[3], &ipov->ih_dst, 467 sizeof(ipov->ih_dst)); 468 } 469 #endif 470 /* 471 * Compute the pseudo-header portion of the checksum 472 * now. We incrementally add in the TCP option and 473 * payload lengths later, and then compute the TCP 474 * checksum right before the packet is sent off onto 475 * the wire. 476 */ 477 n->th_sum = in_cksum_phdr(ipov->ih_src.s_addr, 478 ipov->ih_dst.s_addr, 479 htons(sizeof(struct tcphdr) + IPPROTO_TCP)); 480 break; 481 } 482 #ifdef INET6 483 case AF_INET6: 484 { 485 struct ip6_hdr *ip6; 486 mtod(m, struct ip *)->ip_v = 6; 487 ip6 = mtod(m, struct ip6_hdr *); 488 ip6->ip6_nxt = IPPROTO_TCP; 489 ip6->ip6_plen = htons(sizeof(struct tcphdr)); 490 ip6->ip6_src = in6p->in6p_laddr; 491 ip6->ip6_dst = in6p->in6p_faddr; 492 ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK; 493 if (ip6_auto_flowlabel) { 494 ip6->ip6_flow &= ~IPV6_FLOWLABEL_MASK; 495 ip6->ip6_flow |= 496 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 497 } 498 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 499 ip6->ip6_vfc |= IPV6_VERSION; 500 501 /* 502 * Compute the pseudo-header portion of the checksum 503 * now. We incrementally add in the TCP option and 504 * payload lengths later, and then compute the TCP 505 * checksum right before the packet is sent off onto 506 * the wire. 507 */ 508 n->th_sum = in6_cksum_phdr(&in6p->in6p_laddr, 509 &in6p->in6p_faddr, htonl(sizeof(struct tcphdr)), 510 htonl(IPPROTO_TCP)); 511 break; 512 } 513 #endif 514 } 515 if (inp) { 516 n->th_sport = inp->inp_lport; 517 n->th_dport = inp->inp_fport; 518 } 519 #ifdef INET6 520 else if (in6p) { 521 n->th_sport = in6p->in6p_lport; 522 n->th_dport = in6p->in6p_fport; 523 } 524 #endif 525 n->th_seq = 0; 526 n->th_ack = 0; 527 n->th_x2 = 0; 528 n->th_off = 5; 529 n->th_flags = 0; 530 n->th_win = 0; 531 n->th_urp = 0; 532 return (m); 533 } 534 535 /* 536 * Send a single message to the TCP at address specified by 537 * the given TCP/IP header. If m == 0, then we make a copy 538 * of the tcpiphdr at ti and send directly to the addressed host. 539 * This is used to force keep alive messages out using the TCP 540 * template for a connection tp->t_template. If flags are given 541 * then we send a message back to the TCP which originated the 542 * segment ti, and discard the mbuf containing it and any other 543 * attached mbufs. 544 * 545 * In any case the ack and sequence number of the transmitted 546 * segment are as specified by the parameters. 547 */ 548 int 549 tcp_respond(tp, template, m, th0, ack, seq, flags) 550 struct tcpcb *tp; 551 struct mbuf *template; 552 struct mbuf *m; 553 struct tcphdr *th0; 554 tcp_seq ack, seq; 555 int flags; 556 { 557 struct route *ro; 558 int error, tlen, win = 0; 559 int hlen; 560 struct ip *ip; 561 #ifdef INET6 562 struct ip6_hdr *ip6; 563 #endif 564 int family; /* family on packet, not inpcb/in6pcb! */ 565 struct tcphdr *th; 566 567 if (tp != NULL && (flags & TH_RST) == 0) { 568 #ifdef DIAGNOSTIC 569 if (tp->t_inpcb && tp->t_in6pcb) 570 panic("tcp_respond: both t_inpcb and t_in6pcb are set"); 571 #endif 572 #ifdef INET 573 if (tp->t_inpcb) 574 win = sbspace(&tp->t_inpcb->inp_socket->so_rcv); 575 #endif 576 #ifdef INET6 577 if (tp->t_in6pcb) 578 win = sbspace(&tp->t_in6pcb->in6p_socket->so_rcv); 579 #endif 580 } 581 582 th = NULL; /* Quell uninitialized warning */ 583 ip = NULL; 584 #ifdef INET6 585 ip6 = NULL; 586 #endif 587 if (m == 0) { 588 if (!template) 589 return EINVAL; 590 591 /* get family information from template */ 592 switch (mtod(template, struct ip *)->ip_v) { 593 case 4: 594 family = AF_INET; 595 hlen = sizeof(struct ip); 596 break; 597 #ifdef INET6 598 case 6: 599 family = AF_INET6; 600 hlen = sizeof(struct ip6_hdr); 601 break; 602 #endif 603 default: 604 return EAFNOSUPPORT; 605 } 606 607 MGETHDR(m, M_DONTWAIT, MT_HEADER); 608 if (m) { 609 MCLAIM(m, &tcp_tx_mowner); 610 MCLGET(m, M_DONTWAIT); 611 if ((m->m_flags & M_EXT) == 0) { 612 m_free(m); 613 m = NULL; 614 } 615 } 616 if (m == NULL) 617 return (ENOBUFS); 618 619 if (tcp_compat_42) 620 tlen = 1; 621 else 622 tlen = 0; 623 624 m->m_data += max_linkhdr; 625 bcopy(mtod(template, caddr_t), mtod(m, caddr_t), 626 template->m_len); 627 switch (family) { 628 case AF_INET: 629 ip = mtod(m, struct ip *); 630 th = (struct tcphdr *)(ip + 1); 631 break; 632 #ifdef INET6 633 case AF_INET6: 634 ip6 = mtod(m, struct ip6_hdr *); 635 th = (struct tcphdr *)(ip6 + 1); 636 break; 637 #endif 638 #if 0 639 default: 640 /* noone will visit here */ 641 m_freem(m); 642 return EAFNOSUPPORT; 643 #endif 644 } 645 flags = TH_ACK; 646 } else { 647 648 if ((m->m_flags & M_PKTHDR) == 0) { 649 #if 0 650 printf("non PKTHDR to tcp_respond\n"); 651 #endif 652 m_freem(m); 653 return EINVAL; 654 } 655 #ifdef DIAGNOSTIC 656 if (!th0) 657 panic("th0 == NULL in tcp_respond"); 658 #endif 659 660 /* get family information from m */ 661 switch (mtod(m, struct ip *)->ip_v) { 662 case 4: 663 family = AF_INET; 664 hlen = sizeof(struct ip); 665 ip = mtod(m, struct ip *); 666 break; 667 #ifdef INET6 668 case 6: 669 family = AF_INET6; 670 hlen = sizeof(struct ip6_hdr); 671 ip6 = mtod(m, struct ip6_hdr *); 672 break; 673 #endif 674 default: 675 m_freem(m); 676 return EAFNOSUPPORT; 677 } 678 if ((flags & TH_SYN) == 0 || sizeof(*th0) > (th0->th_off << 2)) 679 tlen = sizeof(*th0); 680 else 681 tlen = th0->th_off << 2; 682 683 if (m->m_len > hlen + tlen && (m->m_flags & M_EXT) == 0 && 684 mtod(m, caddr_t) + hlen == (caddr_t)th0) { 685 m->m_len = hlen + tlen; 686 m_freem(m->m_next); 687 m->m_next = NULL; 688 } else { 689 struct mbuf *n; 690 691 #ifdef DIAGNOSTIC 692 if (max_linkhdr + hlen + tlen > MCLBYTES) { 693 m_freem(m); 694 return EMSGSIZE; 695 } 696 #endif 697 MGETHDR(n, M_DONTWAIT, MT_HEADER); 698 if (n && max_linkhdr + hlen + tlen > MHLEN) { 699 MCLGET(n, M_DONTWAIT); 700 if ((n->m_flags & M_EXT) == 0) { 701 m_freem(n); 702 n = NULL; 703 } 704 } 705 if (!n) { 706 m_freem(m); 707 return ENOBUFS; 708 } 709 710 MCLAIM(n, &tcp_tx_mowner); 711 n->m_data += max_linkhdr; 712 n->m_len = hlen + tlen; 713 m_copyback(n, 0, hlen, mtod(m, caddr_t)); 714 m_copyback(n, hlen, tlen, (caddr_t)th0); 715 716 m_freem(m); 717 m = n; 718 n = NULL; 719 } 720 721 #define xchg(a,b,type) { type t; t=a; a=b; b=t; } 722 switch (family) { 723 case AF_INET: 724 ip = mtod(m, struct ip *); 725 th = (struct tcphdr *)(ip + 1); 726 ip->ip_p = IPPROTO_TCP; 727 xchg(ip->ip_dst, ip->ip_src, struct in_addr); 728 ip->ip_p = IPPROTO_TCP; 729 break; 730 #ifdef INET6 731 case AF_INET6: 732 ip6 = mtod(m, struct ip6_hdr *); 733 th = (struct tcphdr *)(ip6 + 1); 734 ip6->ip6_nxt = IPPROTO_TCP; 735 xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); 736 ip6->ip6_nxt = IPPROTO_TCP; 737 break; 738 #endif 739 #if 0 740 default: 741 /* noone will visit here */ 742 m_freem(m); 743 return EAFNOSUPPORT; 744 #endif 745 } 746 xchg(th->th_dport, th->th_sport, u_int16_t); 747 #undef xchg 748 tlen = 0; /*be friendly with the following code*/ 749 } 750 th->th_seq = htonl(seq); 751 th->th_ack = htonl(ack); 752 th->th_x2 = 0; 753 if ((flags & TH_SYN) == 0) { 754 if (tp) 755 win >>= tp->rcv_scale; 756 if (win > TCP_MAXWIN) 757 win = TCP_MAXWIN; 758 th->th_win = htons((u_int16_t)win); 759 th->th_off = sizeof (struct tcphdr) >> 2; 760 tlen += sizeof(*th); 761 } else 762 tlen += th->th_off << 2; 763 m->m_len = hlen + tlen; 764 m->m_pkthdr.len = hlen + tlen; 765 m->m_pkthdr.rcvif = (struct ifnet *) 0; 766 th->th_flags = flags; 767 th->th_urp = 0; 768 769 switch (family) { 770 #ifdef INET 771 case AF_INET: 772 { 773 struct ipovly *ipov = (struct ipovly *)ip; 774 bzero(ipov->ih_x1, sizeof ipov->ih_x1); 775 ipov->ih_len = htons((u_int16_t)tlen); 776 777 th->th_sum = 0; 778 th->th_sum = in_cksum(m, hlen + tlen); 779 ip->ip_len = htons(hlen + tlen); 780 ip->ip_ttl = ip_defttl; 781 break; 782 } 783 #endif 784 #ifdef INET6 785 case AF_INET6: 786 { 787 th->th_sum = 0; 788 th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), 789 tlen); 790 ip6->ip6_plen = ntohs(tlen); 791 if (tp && tp->t_in6pcb) { 792 struct ifnet *oifp; 793 ro = (struct route *)&tp->t_in6pcb->in6p_route; 794 oifp = ro->ro_rt ? ro->ro_rt->rt_ifp : NULL; 795 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, oifp); 796 } else 797 ip6->ip6_hlim = ip6_defhlim; 798 ip6->ip6_flow &= ~IPV6_FLOWINFO_MASK; 799 if (ip6_auto_flowlabel) { 800 ip6->ip6_flow |= 801 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 802 } 803 break; 804 } 805 #endif 806 } 807 808 #ifdef IPSEC 809 (void)ipsec_setsocket(m, NULL); 810 #endif /*IPSEC*/ 811 812 if (tp != NULL && tp->t_inpcb != NULL) { 813 ro = &tp->t_inpcb->inp_route; 814 #ifdef IPSEC 815 if (ipsec_setsocket(m, tp->t_inpcb->inp_socket) != 0) { 816 m_freem(m); 817 return ENOBUFS; 818 } 819 #endif 820 #ifdef DIAGNOSTIC 821 if (family != AF_INET) 822 panic("tcp_respond: address family mismatch"); 823 if (!in_hosteq(ip->ip_dst, tp->t_inpcb->inp_faddr)) { 824 panic("tcp_respond: ip_dst %x != inp_faddr %x", 825 ntohl(ip->ip_dst.s_addr), 826 ntohl(tp->t_inpcb->inp_faddr.s_addr)); 827 } 828 #endif 829 } 830 #ifdef INET6 831 else if (tp != NULL && tp->t_in6pcb != NULL) { 832 ro = (struct route *)&tp->t_in6pcb->in6p_route; 833 #ifdef IPSEC 834 if (ipsec_setsocket(m, tp->t_in6pcb->in6p_socket) != 0) { 835 m_freem(m); 836 return ENOBUFS; 837 } 838 #endif 839 #ifdef DIAGNOSTIC 840 if (family == AF_INET) { 841 if (!IN6_IS_ADDR_V4MAPPED(&tp->t_in6pcb->in6p_faddr)) 842 panic("tcp_respond: not mapped addr"); 843 if (bcmp(&ip->ip_dst, 844 &tp->t_in6pcb->in6p_faddr.s6_addr32[3], 845 sizeof(ip->ip_dst)) != 0) { 846 panic("tcp_respond: ip_dst != in6p_faddr"); 847 } 848 } else if (family == AF_INET6) { 849 if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 850 &tp->t_in6pcb->in6p_faddr)) 851 panic("tcp_respond: ip6_dst != in6p_faddr"); 852 } else 853 panic("tcp_respond: address family mismatch"); 854 #endif 855 } 856 #endif 857 else 858 ro = NULL; 859 860 switch (family) { 861 #ifdef INET 862 case AF_INET: 863 error = ip_output(m, NULL, ro, 864 (tp && tp->t_mtudisc ? IP_MTUDISC : 0), 865 NULL); 866 break; 867 #endif 868 #ifdef INET6 869 case AF_INET6: 870 error = ip6_output(m, NULL, (struct route_in6 *)ro, 0, NULL, 871 NULL); 872 break; 873 #endif 874 default: 875 error = EAFNOSUPPORT; 876 break; 877 } 878 879 return (error); 880 } 881 882 /* 883 * Create a new TCP control block, making an 884 * empty reassembly queue and hooking it to the argument 885 * protocol control block. 886 */ 887 struct tcpcb * 888 tcp_newtcpcb(family, aux) 889 int family; /* selects inpcb, or in6pcb */ 890 void *aux; 891 { 892 struct tcpcb *tp; 893 int i; 894 895 switch (family) { 896 case PF_INET: 897 break; 898 #ifdef INET6 899 case PF_INET6: 900 break; 901 #endif 902 default: 903 return NULL; 904 } 905 906 tp = pool_get(&tcpcb_pool, PR_NOWAIT); 907 if (tp == NULL) 908 return (NULL); 909 bzero((caddr_t)tp, sizeof(struct tcpcb)); 910 TAILQ_INIT(&tp->segq); 911 TAILQ_INIT(&tp->timeq); 912 tp->t_family = family; /* may be overridden later on */ 913 tp->t_peermss = tcp_mssdflt; 914 tp->t_ourmss = tcp_mssdflt; 915 tp->t_segsz = tcp_mssdflt; 916 LIST_INIT(&tp->t_sc); 917 918 callout_init(&tp->t_delack_ch); 919 for (i = 0; i < TCPT_NTIMERS; i++) 920 TCP_TIMER_INIT(tp, i); 921 922 tp->t_flags = 0; 923 if (tcp_do_rfc1323 && tcp_do_win_scale) 924 tp->t_flags |= TF_REQ_SCALE; 925 if (tcp_do_rfc1323 && tcp_do_timestamps) 926 tp->t_flags |= TF_REQ_TSTMP; 927 if (tcp_do_sack == 2) 928 tp->t_flags |= TF_WILL_SACK; 929 else if (tcp_do_sack == 1) 930 tp->t_flags |= TF_WILL_SACK|TF_IGNR_RXSACK; 931 tp->t_flags |= TF_CANT_TXSACK; 932 switch (family) { 933 case PF_INET: 934 tp->t_inpcb = (struct inpcb *)aux; 935 tp->t_mtudisc = ip_mtudisc; 936 break; 937 #ifdef INET6 938 case PF_INET6: 939 tp->t_in6pcb = (struct in6pcb *)aux; 940 /* for IPv6, always try to run path MTU discovery */ 941 tp->t_mtudisc = 1; 942 break; 943 #endif 944 } 945 /* 946 * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no 947 * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives 948 * reasonable initial retransmit time. 949 */ 950 tp->t_srtt = TCPTV_SRTTBASE; 951 tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << (TCP_RTTVAR_SHIFT + 2 - 1); 952 tp->t_rttmin = TCPTV_MIN; 953 TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), 954 TCPTV_MIN, TCPTV_REXMTMAX); 955 tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; 956 tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; 957 if (family == AF_INET) { 958 struct inpcb *inp = (struct inpcb *)aux; 959 inp->inp_ip.ip_ttl = ip_defttl; 960 inp->inp_ppcb = (caddr_t)tp; 961 } 962 #ifdef INET6 963 else if (family == AF_INET6) { 964 struct in6pcb *in6p = (struct in6pcb *)aux; 965 in6p->in6p_ip6.ip6_hlim = in6_selecthlim(in6p, 966 in6p->in6p_route.ro_rt ? in6p->in6p_route.ro_rt->rt_ifp 967 : NULL); 968 in6p->in6p_ppcb = (caddr_t)tp; 969 } 970 #endif 971 972 /* 973 * Initialize our timebase. When we send timestamps, we take 974 * the delta from tcp_now -- this means each connection always 975 * gets a timebase of 0, which makes it, among other things, 976 * more difficult to determine how long a system has been up, 977 * and thus how many TCP sequence increments have occurred. 978 */ 979 tp->ts_timebase = tcp_now; 980 981 return (tp); 982 } 983 984 /* 985 * Drop a TCP connection, reporting 986 * the specified error. If connection is synchronized, 987 * then send a RST to peer. 988 */ 989 struct tcpcb * 990 tcp_drop(tp, errno) 991 struct tcpcb *tp; 992 int errno; 993 { 994 struct socket *so = NULL; 995 996 #ifdef DIAGNOSTIC 997 if (tp->t_inpcb && tp->t_in6pcb) 998 panic("tcp_drop: both t_inpcb and t_in6pcb are set"); 999 #endif 1000 #ifdef INET 1001 if (tp->t_inpcb) 1002 so = tp->t_inpcb->inp_socket; 1003 #endif 1004 #ifdef INET6 1005 if (tp->t_in6pcb) 1006 so = tp->t_in6pcb->in6p_socket; 1007 #endif 1008 if (!so) 1009 return NULL; 1010 1011 if (TCPS_HAVERCVDSYN(tp->t_state)) { 1012 tp->t_state = TCPS_CLOSED; 1013 (void) tcp_output(tp); 1014 tcpstat.tcps_drops++; 1015 } else 1016 tcpstat.tcps_conndrops++; 1017 if (errno == ETIMEDOUT && tp->t_softerror) 1018 errno = tp->t_softerror; 1019 so->so_error = errno; 1020 return (tcp_close(tp)); 1021 } 1022 1023 /* 1024 * Close a TCP control block: 1025 * discard all space held by the tcp 1026 * discard internet protocol block 1027 * wake up any sleepers 1028 */ 1029 struct tcpcb * 1030 tcp_close(tp) 1031 struct tcpcb *tp; 1032 { 1033 struct inpcb *inp; 1034 #ifdef INET6 1035 struct in6pcb *in6p; 1036 #endif 1037 struct socket *so; 1038 #ifdef RTV_RTT 1039 struct rtentry *rt; 1040 #endif 1041 struct route *ro; 1042 1043 inp = tp->t_inpcb; 1044 #ifdef INET6 1045 in6p = tp->t_in6pcb; 1046 #endif 1047 so = NULL; 1048 ro = NULL; 1049 if (inp) { 1050 so = inp->inp_socket; 1051 ro = &inp->inp_route; 1052 } 1053 #ifdef INET6 1054 else if (in6p) { 1055 so = in6p->in6p_socket; 1056 ro = (struct route *)&in6p->in6p_route; 1057 } 1058 #endif 1059 1060 #ifdef RTV_RTT 1061 /* 1062 * If we sent enough data to get some meaningful characteristics, 1063 * save them in the routing entry. 'Enough' is arbitrarily 1064 * defined as the sendpipesize (default 4K) * 16. This would 1065 * give us 16 rtt samples assuming we only get one sample per 1066 * window (the usual case on a long haul net). 16 samples is 1067 * enough for the srtt filter to converge to within 5% of the correct 1068 * value; fewer samples and we could save a very bogus rtt. 1069 * 1070 * Don't update the default route's characteristics and don't 1071 * update anything that the user "locked". 1072 */ 1073 if (SEQ_LT(tp->iss + so->so_snd.sb_hiwat * 16, tp->snd_max) && 1074 ro && (rt = ro->ro_rt) && 1075 !in_nullhost(satosin(rt_key(rt))->sin_addr)) { 1076 u_long i = 0; 1077 1078 if ((rt->rt_rmx.rmx_locks & RTV_RTT) == 0) { 1079 i = tp->t_srtt * 1080 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1081 if (rt->rt_rmx.rmx_rtt && i) 1082 /* 1083 * filter this update to half the old & half 1084 * the new values, converting scale. 1085 * See route.h and tcp_var.h for a 1086 * description of the scaling constants. 1087 */ 1088 rt->rt_rmx.rmx_rtt = 1089 (rt->rt_rmx.rmx_rtt + i) / 2; 1090 else 1091 rt->rt_rmx.rmx_rtt = i; 1092 } 1093 if ((rt->rt_rmx.rmx_locks & RTV_RTTVAR) == 0) { 1094 i = tp->t_rttvar * 1095 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTTVAR_SHIFT + 2)); 1096 if (rt->rt_rmx.rmx_rttvar && i) 1097 rt->rt_rmx.rmx_rttvar = 1098 (rt->rt_rmx.rmx_rttvar + i) / 2; 1099 else 1100 rt->rt_rmx.rmx_rttvar = i; 1101 } 1102 /* 1103 * update the pipelimit (ssthresh) if it has been updated 1104 * already or if a pipesize was specified & the threshhold 1105 * got below half the pipesize. I.e., wait for bad news 1106 * before we start updating, then update on both good 1107 * and bad news. 1108 */ 1109 if (((rt->rt_rmx.rmx_locks & RTV_SSTHRESH) == 0 && 1110 (i = tp->snd_ssthresh) && rt->rt_rmx.rmx_ssthresh) || 1111 i < (rt->rt_rmx.rmx_sendpipe / 2)) { 1112 /* 1113 * convert the limit from user data bytes to 1114 * packets then to packet data bytes. 1115 */ 1116 i = (i + tp->t_segsz / 2) / tp->t_segsz; 1117 if (i < 2) 1118 i = 2; 1119 i *= (u_long)(tp->t_segsz + sizeof (struct tcpiphdr)); 1120 if (rt->rt_rmx.rmx_ssthresh) 1121 rt->rt_rmx.rmx_ssthresh = 1122 (rt->rt_rmx.rmx_ssthresh + i) / 2; 1123 else 1124 rt->rt_rmx.rmx_ssthresh = i; 1125 } 1126 } 1127 #endif /* RTV_RTT */ 1128 /* free the reassembly queue, if any */ 1129 TCP_REASS_LOCK(tp); 1130 (void) tcp_freeq(tp); 1131 TCP_REASS_UNLOCK(tp); 1132 1133 tcp_canceltimers(tp); 1134 TCP_CLEAR_DELACK(tp); 1135 syn_cache_cleanup(tp); 1136 1137 if (tp->t_template) { 1138 m_free(tp->t_template); 1139 tp->t_template = NULL; 1140 } 1141 pool_put(&tcpcb_pool, tp); 1142 if (inp) { 1143 inp->inp_ppcb = 0; 1144 soisdisconnected(so); 1145 in_pcbdetach(inp); 1146 } 1147 #ifdef INET6 1148 else if (in6p) { 1149 in6p->in6p_ppcb = 0; 1150 soisdisconnected(so); 1151 in6_pcbdetach(in6p); 1152 } 1153 #endif 1154 tcpstat.tcps_closed++; 1155 return ((struct tcpcb *)0); 1156 } 1157 1158 int 1159 tcp_freeq(tp) 1160 struct tcpcb *tp; 1161 { 1162 struct ipqent *qe; 1163 int rv = 0; 1164 #ifdef TCPREASS_DEBUG 1165 int i = 0; 1166 #endif 1167 1168 TCP_REASS_LOCK_CHECK(tp); 1169 1170 while ((qe = TAILQ_FIRST(&tp->segq)) != NULL) { 1171 #ifdef TCPREASS_DEBUG 1172 printf("tcp_freeq[%p,%d]: %u:%u(%u) 0x%02x\n", 1173 tp, i++, qe->ipqe_seq, qe->ipqe_seq + qe->ipqe_len, 1174 qe->ipqe_len, qe->ipqe_flags & (TH_SYN|TH_FIN|TH_RST)); 1175 #endif 1176 TAILQ_REMOVE(&tp->segq, qe, ipqe_q); 1177 TAILQ_REMOVE(&tp->timeq, qe, ipqe_timeq); 1178 m_freem(qe->ipqe_m); 1179 pool_put(&ipqent_pool, qe); 1180 rv = 1; 1181 } 1182 return (rv); 1183 } 1184 1185 /* 1186 * Protocol drain routine. Called when memory is in short supply. 1187 */ 1188 void 1189 tcp_drain() 1190 { 1191 struct inpcb *inp; 1192 struct tcpcb *tp; 1193 1194 /* 1195 * Free the sequence queue of all TCP connections. 1196 */ 1197 inp = CIRCLEQ_FIRST(&tcbtable.inpt_queue); 1198 if (inp) /* XXX */ 1199 CIRCLEQ_FOREACH(inp, &tcbtable.inpt_queue, inp_queue) { 1200 if ((tp = intotcpcb(inp)) != NULL) { 1201 /* 1202 * We may be called from a device's interrupt 1203 * context. If the tcpcb is already busy, 1204 * just bail out now. 1205 */ 1206 if (tcp_reass_lock_try(tp) == 0) 1207 continue; 1208 if (tcp_freeq(tp)) 1209 tcpstat.tcps_connsdrained++; 1210 TCP_REASS_UNLOCK(tp); 1211 } 1212 } 1213 } 1214 1215 #ifdef INET6 1216 void 1217 tcp6_drain() 1218 { 1219 struct in6pcb *in6p; 1220 struct tcpcb *tp; 1221 struct in6pcb *head = &tcb6; 1222 1223 /* 1224 * Free the sequence queue of all TCP connections. 1225 */ 1226 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 1227 if ((tp = in6totcpcb(in6p)) != NULL) { 1228 /* 1229 * We may be called from a device's interrupt 1230 * context. If the tcpcb is already busy, 1231 * just bail out now. 1232 */ 1233 if (tcp_reass_lock_try(tp) == 0) 1234 continue; 1235 if (tcp_freeq(tp)) 1236 tcpstat.tcps_connsdrained++; 1237 TCP_REASS_UNLOCK(tp); 1238 } 1239 } 1240 } 1241 #endif 1242 1243 /* 1244 * Notify a tcp user of an asynchronous error; 1245 * store error as soft error, but wake up user 1246 * (for now, won't do anything until can select for soft error). 1247 */ 1248 void 1249 tcp_notify(inp, error) 1250 struct inpcb *inp; 1251 int error; 1252 { 1253 struct tcpcb *tp = (struct tcpcb *)inp->inp_ppcb; 1254 struct socket *so = inp->inp_socket; 1255 1256 /* 1257 * Ignore some errors if we are hooked up. 1258 * If connection hasn't completed, has retransmitted several times, 1259 * and receives a second error, give up now. This is better 1260 * than waiting a long time to establish a connection that 1261 * can never complete. 1262 */ 1263 if (tp->t_state == TCPS_ESTABLISHED && 1264 (error == EHOSTUNREACH || error == ENETUNREACH || 1265 error == EHOSTDOWN)) { 1266 return; 1267 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1268 tp->t_rxtshift > 3 && tp->t_softerror) 1269 so->so_error = error; 1270 else 1271 tp->t_softerror = error; 1272 wakeup((caddr_t) &so->so_timeo); 1273 sorwakeup(so); 1274 sowwakeup(so); 1275 } 1276 1277 #ifdef INET6 1278 void 1279 tcp6_notify(in6p, error) 1280 struct in6pcb *in6p; 1281 int error; 1282 { 1283 struct tcpcb *tp = (struct tcpcb *)in6p->in6p_ppcb; 1284 struct socket *so = in6p->in6p_socket; 1285 1286 /* 1287 * Ignore some errors if we are hooked up. 1288 * If connection hasn't completed, has retransmitted several times, 1289 * and receives a second error, give up now. This is better 1290 * than waiting a long time to establish a connection that 1291 * can never complete. 1292 */ 1293 if (tp->t_state == TCPS_ESTABLISHED && 1294 (error == EHOSTUNREACH || error == ENETUNREACH || 1295 error == EHOSTDOWN)) { 1296 return; 1297 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1298 tp->t_rxtshift > 3 && tp->t_softerror) 1299 so->so_error = error; 1300 else 1301 tp->t_softerror = error; 1302 wakeup((caddr_t) &so->so_timeo); 1303 sorwakeup(so); 1304 sowwakeup(so); 1305 } 1306 #endif 1307 1308 #ifdef INET6 1309 void 1310 tcp6_ctlinput(cmd, sa, d) 1311 int cmd; 1312 struct sockaddr *sa; 1313 void *d; 1314 { 1315 struct tcphdr th; 1316 void (*notify) __P((struct in6pcb *, int)) = tcp6_notify; 1317 int nmatch; 1318 struct ip6_hdr *ip6; 1319 const struct sockaddr_in6 *sa6_src = NULL; 1320 struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa; 1321 struct mbuf *m; 1322 int off; 1323 1324 if (sa->sa_family != AF_INET6 || 1325 sa->sa_len != sizeof(struct sockaddr_in6)) 1326 return; 1327 if ((unsigned)cmd >= PRC_NCMDS) 1328 return; 1329 else if (cmd == PRC_QUENCH) { 1330 /* XXX there's no PRC_QUENCH in IPv6 */ 1331 notify = tcp6_quench; 1332 } else if (PRC_IS_REDIRECT(cmd)) 1333 notify = in6_rtchange, d = NULL; 1334 else if (cmd == PRC_MSGSIZE) 1335 ; /* special code is present, see below */ 1336 else if (cmd == PRC_HOSTDEAD) 1337 d = NULL; 1338 else if (inet6ctlerrmap[cmd] == 0) 1339 return; 1340 1341 /* if the parameter is from icmp6, decode it. */ 1342 if (d != NULL) { 1343 struct ip6ctlparam *ip6cp = (struct ip6ctlparam *)d; 1344 m = ip6cp->ip6c_m; 1345 ip6 = ip6cp->ip6c_ip6; 1346 off = ip6cp->ip6c_off; 1347 sa6_src = ip6cp->ip6c_src; 1348 } else { 1349 m = NULL; 1350 ip6 = NULL; 1351 sa6_src = &sa6_any; 1352 } 1353 1354 if (ip6) { 1355 /* 1356 * XXX: We assume that when ip6 is non NULL, 1357 * M and OFF are valid. 1358 */ 1359 1360 /* check if we can safely examine src and dst ports */ 1361 if (m->m_pkthdr.len < off + sizeof(th)) { 1362 if (cmd == PRC_MSGSIZE) 1363 icmp6_mtudisc_update((struct ip6ctlparam *)d, 0); 1364 return; 1365 } 1366 1367 bzero(&th, sizeof(th)); 1368 m_copydata(m, off, sizeof(th), (caddr_t)&th); 1369 1370 if (cmd == PRC_MSGSIZE) { 1371 int valid = 0; 1372 1373 /* 1374 * Check to see if we have a valid TCP connection 1375 * corresponding to the address in the ICMPv6 message 1376 * payload. 1377 */ 1378 if (in6_pcblookup_connect(&tcb6, &sa6->sin6_addr, 1379 th.th_dport, (struct in6_addr *)&sa6_src->sin6_addr, 1380 th.th_sport, 0)) 1381 valid++; 1382 1383 /* 1384 * Depending on the value of "valid" and routing table 1385 * size (mtudisc_{hi,lo}wat), we will: 1386 * - recalcurate the new MTU and create the 1387 * corresponding routing entry, or 1388 * - ignore the MTU change notification. 1389 */ 1390 icmp6_mtudisc_update((struct ip6ctlparam *)d, valid); 1391 1392 /* 1393 * no need to call in6_pcbnotify, it should have been 1394 * called via callback if necessary 1395 */ 1396 return; 1397 } 1398 1399 nmatch = in6_pcbnotify(&tcb6, sa, th.th_dport, 1400 (struct sockaddr *)sa6_src, th.th_sport, cmd, NULL, notify); 1401 if (nmatch == 0 && syn_cache_count && 1402 (inet6ctlerrmap[cmd] == EHOSTUNREACH || 1403 inet6ctlerrmap[cmd] == ENETUNREACH || 1404 inet6ctlerrmap[cmd] == EHOSTDOWN)) 1405 syn_cache_unreach((struct sockaddr *)sa6_src, 1406 sa, &th); 1407 } else { 1408 (void) in6_pcbnotify(&tcb6, sa, 0, (struct sockaddr *)sa6_src, 1409 0, cmd, NULL, notify); 1410 } 1411 } 1412 #endif 1413 1414 #ifdef INET 1415 /* assumes that ip header and tcp header are contiguous on mbuf */ 1416 void * 1417 tcp_ctlinput(cmd, sa, v) 1418 int cmd; 1419 struct sockaddr *sa; 1420 void *v; 1421 { 1422 struct ip *ip = v; 1423 struct tcphdr *th; 1424 struct icmp *icp; 1425 extern const int inetctlerrmap[]; 1426 void (*notify) __P((struct inpcb *, int)) = tcp_notify; 1427 int errno; 1428 int nmatch; 1429 #ifdef INET6 1430 struct in6_addr src6, dst6; 1431 #endif 1432 1433 if (sa->sa_family != AF_INET || 1434 sa->sa_len != sizeof(struct sockaddr_in)) 1435 return NULL; 1436 if ((unsigned)cmd >= PRC_NCMDS) 1437 return NULL; 1438 errno = inetctlerrmap[cmd]; 1439 if (cmd == PRC_QUENCH) 1440 notify = tcp_quench; 1441 else if (PRC_IS_REDIRECT(cmd)) 1442 notify = in_rtchange, ip = 0; 1443 else if (cmd == PRC_MSGSIZE && ip && ip->ip_v == 4) { 1444 /* 1445 * Check to see if we have a valid TCP connection 1446 * corresponding to the address in the ICMP message 1447 * payload. 1448 * 1449 * Boundary check is made in icmp_input(), with ICMP_ADVLENMIN. 1450 */ 1451 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1452 #ifdef INET6 1453 memset(&src6, 0, sizeof(src6)); 1454 memset(&dst6, 0, sizeof(dst6)); 1455 src6.s6_addr16[5] = dst6.s6_addr16[5] = 0xffff; 1456 memcpy(&src6.s6_addr32[3], &ip->ip_src, sizeof(struct in_addr)); 1457 memcpy(&dst6.s6_addr32[3], &ip->ip_dst, sizeof(struct in_addr)); 1458 #endif 1459 if (in_pcblookup_connect(&tcbtable, ip->ip_dst, th->th_dport, 1460 ip->ip_src, th->th_sport) != NULL) 1461 ; 1462 #ifdef INET6 1463 else if (in6_pcblookup_connect(&tcb6, &dst6, 1464 th->th_dport, &src6, th->th_sport, 0) != NULL) 1465 ; 1466 #endif 1467 else 1468 return NULL; 1469 1470 /* 1471 * Now that we've validated that we are actually communicating 1472 * with the host indicated in the ICMP message, locate the 1473 * ICMP header, recalculate the new MTU, and create the 1474 * corresponding routing entry. 1475 */ 1476 icp = (struct icmp *)((caddr_t)ip - 1477 offsetof(struct icmp, icmp_ip)); 1478 icmp_mtudisc(icp, ip->ip_dst); 1479 1480 return NULL; 1481 } else if (cmd == PRC_HOSTDEAD) 1482 ip = 0; 1483 else if (errno == 0) 1484 return NULL; 1485 if (ip && ip->ip_v == 4 && sa->sa_family == AF_INET) { 1486 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1487 nmatch = in_pcbnotify(&tcbtable, satosin(sa)->sin_addr, 1488 th->th_dport, ip->ip_src, th->th_sport, errno, notify); 1489 if (nmatch == 0 && syn_cache_count && 1490 (inetctlerrmap[cmd] == EHOSTUNREACH || 1491 inetctlerrmap[cmd] == ENETUNREACH || 1492 inetctlerrmap[cmd] == EHOSTDOWN)) { 1493 struct sockaddr_in sin; 1494 bzero(&sin, sizeof(sin)); 1495 sin.sin_len = sizeof(sin); 1496 sin.sin_family = AF_INET; 1497 sin.sin_port = th->th_sport; 1498 sin.sin_addr = ip->ip_src; 1499 syn_cache_unreach((struct sockaddr *)&sin, sa, th); 1500 } 1501 1502 /* XXX mapped address case */ 1503 } else 1504 in_pcbnotifyall(&tcbtable, satosin(sa)->sin_addr, errno, 1505 notify); 1506 return NULL; 1507 } 1508 1509 /* 1510 * When a source quence is received, we are being notifed of congestion. 1511 * Close the congestion window down to the Loss Window (one segment). 1512 * We will gradually open it again as we proceed. 1513 */ 1514 void 1515 tcp_quench(inp, errno) 1516 struct inpcb *inp; 1517 int errno; 1518 { 1519 struct tcpcb *tp = intotcpcb(inp); 1520 1521 if (tp) 1522 tp->snd_cwnd = tp->t_segsz; 1523 } 1524 #endif 1525 1526 #ifdef INET6 1527 void 1528 tcp6_quench(in6p, errno) 1529 struct in6pcb *in6p; 1530 int errno; 1531 { 1532 struct tcpcb *tp = in6totcpcb(in6p); 1533 1534 if (tp) 1535 tp->snd_cwnd = tp->t_segsz; 1536 } 1537 #endif 1538 1539 #ifdef INET 1540 /* 1541 * Path MTU Discovery handlers. 1542 */ 1543 void 1544 tcp_mtudisc_callback(faddr) 1545 struct in_addr faddr; 1546 { 1547 #ifdef INET6 1548 struct in6_addr in6; 1549 #endif 1550 1551 in_pcbnotifyall(&tcbtable, faddr, EMSGSIZE, tcp_mtudisc); 1552 #ifdef INET6 1553 memset(&in6, 0, sizeof(in6)); 1554 in6.s6_addr16[5] = 0xffff; 1555 memcpy(&in6.s6_addr32[3], &faddr, sizeof(struct in_addr)); 1556 tcp6_mtudisc_callback(&in6); 1557 #endif 1558 } 1559 1560 /* 1561 * On receipt of path MTU corrections, flush old route and replace it 1562 * with the new one. Retransmit all unacknowledged packets, to ensure 1563 * that all packets will be received. 1564 */ 1565 void 1566 tcp_mtudisc(inp, errno) 1567 struct inpcb *inp; 1568 int errno; 1569 { 1570 struct tcpcb *tp = intotcpcb(inp); 1571 struct rtentry *rt = in_pcbrtentry(inp); 1572 1573 if (tp != 0) { 1574 if (rt != 0) { 1575 /* 1576 * If this was not a host route, remove and realloc. 1577 */ 1578 if ((rt->rt_flags & RTF_HOST) == 0) { 1579 in_rtchange(inp, errno); 1580 if ((rt = in_pcbrtentry(inp)) == 0) 1581 return; 1582 } 1583 1584 /* 1585 * Slow start out of the error condition. We 1586 * use the MTU because we know it's smaller 1587 * than the previously transmitted segment. 1588 * 1589 * Note: This is more conservative than the 1590 * suggestion in draft-floyd-incr-init-win-03. 1591 */ 1592 if (rt->rt_rmx.rmx_mtu != 0) 1593 tp->snd_cwnd = 1594 TCP_INITIAL_WINDOW(tcp_init_win, 1595 rt->rt_rmx.rmx_mtu); 1596 } 1597 1598 /* 1599 * Resend unacknowledged packets. 1600 */ 1601 tp->snd_nxt = tp->snd_una; 1602 tcp_output(tp); 1603 } 1604 } 1605 #endif 1606 1607 #ifdef INET6 1608 /* 1609 * Path MTU Discovery handlers. 1610 */ 1611 void 1612 tcp6_mtudisc_callback(faddr) 1613 struct in6_addr *faddr; 1614 { 1615 struct sockaddr_in6 sin6; 1616 1617 bzero(&sin6, sizeof(sin6)); 1618 sin6.sin6_family = AF_INET6; 1619 sin6.sin6_len = sizeof(struct sockaddr_in6); 1620 sin6.sin6_addr = *faddr; 1621 (void) in6_pcbnotify(&tcb6, (struct sockaddr *)&sin6, 0, 1622 (struct sockaddr *)&sa6_any, 0, PRC_MSGSIZE, NULL, tcp6_mtudisc); 1623 } 1624 1625 void 1626 tcp6_mtudisc(in6p, errno) 1627 struct in6pcb *in6p; 1628 int errno; 1629 { 1630 struct tcpcb *tp = in6totcpcb(in6p); 1631 struct rtentry *rt = in6_pcbrtentry(in6p); 1632 1633 if (tp != 0) { 1634 if (rt != 0) { 1635 /* 1636 * If this was not a host route, remove and realloc. 1637 */ 1638 if ((rt->rt_flags & RTF_HOST) == 0) { 1639 in6_rtchange(in6p, errno); 1640 if ((rt = in6_pcbrtentry(in6p)) == 0) 1641 return; 1642 } 1643 1644 /* 1645 * Slow start out of the error condition. We 1646 * use the MTU because we know it's smaller 1647 * than the previously transmitted segment. 1648 * 1649 * Note: This is more conservative than the 1650 * suggestion in draft-floyd-incr-init-win-03. 1651 */ 1652 if (rt->rt_rmx.rmx_mtu != 0) 1653 tp->snd_cwnd = 1654 TCP_INITIAL_WINDOW(tcp_init_win, 1655 rt->rt_rmx.rmx_mtu); 1656 } 1657 1658 /* 1659 * Resend unacknowledged packets. 1660 */ 1661 tp->snd_nxt = tp->snd_una; 1662 tcp_output(tp); 1663 } 1664 } 1665 #endif /* INET6 */ 1666 1667 /* 1668 * Compute the MSS to advertise to the peer. Called only during 1669 * the 3-way handshake. If we are the server (peer initiated 1670 * connection), we are called with a pointer to the interface 1671 * on which the SYN packet arrived. If we are the client (we 1672 * initiated connection), we are called with a pointer to the 1673 * interface out which this connection should go. 1674 * 1675 * NOTE: Do not subtract IP option/extension header size nor IPsec 1676 * header size from MSS advertisement. MSS option must hold the maximum 1677 * segment size we can accept, so it must always be: 1678 * max(if mtu) - ip header - tcp header 1679 */ 1680 u_long 1681 tcp_mss_to_advertise(ifp, af) 1682 const struct ifnet *ifp; 1683 int af; 1684 { 1685 extern u_long in_maxmtu; 1686 u_long mss = 0; 1687 u_long hdrsiz; 1688 1689 /* 1690 * In order to avoid defeating path MTU discovery on the peer, 1691 * we advertise the max MTU of all attached networks as our MSS, 1692 * per RFC 1191, section 3.1. 1693 * 1694 * We provide the option to advertise just the MTU of 1695 * the interface on which we hope this connection will 1696 * be receiving. If we are responding to a SYN, we 1697 * will have a pretty good idea about this, but when 1698 * initiating a connection there is a bit more doubt. 1699 * 1700 * We also need to ensure that loopback has a large enough 1701 * MSS, as the loopback MTU is never included in in_maxmtu. 1702 */ 1703 1704 if (ifp != NULL) 1705 switch (af) { 1706 case AF_INET: 1707 mss = ifp->if_mtu; 1708 break; 1709 #ifdef INET6 1710 case AF_INET6: 1711 mss = IN6_LINKMTU(ifp); 1712 break; 1713 #endif 1714 } 1715 1716 if (tcp_mss_ifmtu == 0) 1717 switch (af) { 1718 case AF_INET: 1719 mss = max(in_maxmtu, mss); 1720 break; 1721 #ifdef INET6 1722 case AF_INET6: 1723 mss = max(in6_maxmtu, mss); 1724 break; 1725 #endif 1726 } 1727 1728 switch (af) { 1729 case AF_INET: 1730 hdrsiz = sizeof(struct ip); 1731 break; 1732 #ifdef INET6 1733 case AF_INET6: 1734 hdrsiz = sizeof(struct ip6_hdr); 1735 break; 1736 #endif 1737 default: 1738 hdrsiz = 0; 1739 break; 1740 } 1741 hdrsiz += sizeof(struct tcphdr); 1742 if (mss > hdrsiz) 1743 mss -= hdrsiz; 1744 1745 mss = max(tcp_mssdflt, mss); 1746 return (mss); 1747 } 1748 1749 /* 1750 * Set connection variables based on the peer's advertised MSS. 1751 * We are passed the TCPCB for the actual connection. If we 1752 * are the server, we are called by the compressed state engine 1753 * when the 3-way handshake is complete. If we are the client, 1754 * we are called when we receive the SYN,ACK from the server. 1755 * 1756 * NOTE: Our advertised MSS value must be initialized in the TCPCB 1757 * before this routine is called! 1758 */ 1759 void 1760 tcp_mss_from_peer(tp, offer) 1761 struct tcpcb *tp; 1762 int offer; 1763 { 1764 struct socket *so; 1765 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1766 struct rtentry *rt; 1767 #endif 1768 u_long bufsize; 1769 int mss; 1770 1771 #ifdef DIAGNOSTIC 1772 if (tp->t_inpcb && tp->t_in6pcb) 1773 panic("tcp_mss_from_peer: both t_inpcb and t_in6pcb are set"); 1774 #endif 1775 so = NULL; 1776 rt = NULL; 1777 #ifdef INET 1778 if (tp->t_inpcb) { 1779 so = tp->t_inpcb->inp_socket; 1780 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1781 rt = in_pcbrtentry(tp->t_inpcb); 1782 #endif 1783 } 1784 #endif 1785 #ifdef INET6 1786 if (tp->t_in6pcb) { 1787 so = tp->t_in6pcb->in6p_socket; 1788 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1789 rt = in6_pcbrtentry(tp->t_in6pcb); 1790 #endif 1791 } 1792 #endif 1793 1794 /* 1795 * As per RFC1122, use the default MSS value, unless they 1796 * sent us an offer. Do not accept offers less than 32 bytes. 1797 */ 1798 mss = tcp_mssdflt; 1799 if (offer) 1800 mss = offer; 1801 mss = max(mss, 32); /* sanity */ 1802 tp->t_peermss = mss; 1803 mss -= tcp_optlen(tp); 1804 #ifdef INET 1805 if (tp->t_inpcb) 1806 mss -= ip_optlen(tp->t_inpcb); 1807 #endif 1808 #ifdef INET6 1809 if (tp->t_in6pcb) 1810 mss -= ip6_optlen(tp->t_in6pcb); 1811 #endif 1812 1813 /* 1814 * If there's a pipesize, change the socket buffer to that size. 1815 * Make the socket buffer an integral number of MSS units. If 1816 * the MSS is larger than the socket buffer, artificially decrease 1817 * the MSS. 1818 */ 1819 #ifdef RTV_SPIPE 1820 if (rt != NULL && rt->rt_rmx.rmx_sendpipe != 0) 1821 bufsize = rt->rt_rmx.rmx_sendpipe; 1822 else 1823 #endif 1824 bufsize = so->so_snd.sb_hiwat; 1825 if (bufsize < mss) 1826 mss = bufsize; 1827 else { 1828 bufsize = roundup(bufsize, mss); 1829 if (bufsize > sb_max) 1830 bufsize = sb_max; 1831 (void) sbreserve(&so->so_snd, bufsize); 1832 } 1833 tp->t_segsz = mss; 1834 1835 #ifdef RTV_SSTHRESH 1836 if (rt != NULL && rt->rt_rmx.rmx_ssthresh) { 1837 /* 1838 * There's some sort of gateway or interface buffer 1839 * limit on the path. Use this to set the slow 1840 * start threshold, but set the threshold to no less 1841 * than 2 * MSS. 1842 */ 1843 tp->snd_ssthresh = max(2 * mss, rt->rt_rmx.rmx_ssthresh); 1844 } 1845 #endif 1846 } 1847 1848 /* 1849 * Processing necessary when a TCP connection is established. 1850 */ 1851 void 1852 tcp_established(tp) 1853 struct tcpcb *tp; 1854 { 1855 struct socket *so; 1856 #ifdef RTV_RPIPE 1857 struct rtentry *rt; 1858 #endif 1859 u_long bufsize; 1860 1861 #ifdef DIAGNOSTIC 1862 if (tp->t_inpcb && tp->t_in6pcb) 1863 panic("tcp_established: both t_inpcb and t_in6pcb are set"); 1864 #endif 1865 so = NULL; 1866 rt = NULL; 1867 #ifdef INET 1868 if (tp->t_inpcb) { 1869 so = tp->t_inpcb->inp_socket; 1870 #if defined(RTV_RPIPE) 1871 rt = in_pcbrtentry(tp->t_inpcb); 1872 #endif 1873 } 1874 #endif 1875 #ifdef INET6 1876 if (tp->t_in6pcb) { 1877 so = tp->t_in6pcb->in6p_socket; 1878 #if defined(RTV_RPIPE) 1879 rt = in6_pcbrtentry(tp->t_in6pcb); 1880 #endif 1881 } 1882 #endif 1883 1884 tp->t_state = TCPS_ESTABLISHED; 1885 TCP_TIMER_ARM(tp, TCPT_KEEP, tcp_keepidle); 1886 1887 #ifdef RTV_RPIPE 1888 if (rt != NULL && rt->rt_rmx.rmx_recvpipe != 0) 1889 bufsize = rt->rt_rmx.rmx_recvpipe; 1890 else 1891 #endif 1892 bufsize = so->so_rcv.sb_hiwat; 1893 if (bufsize > tp->t_ourmss) { 1894 bufsize = roundup(bufsize, tp->t_ourmss); 1895 if (bufsize > sb_max) 1896 bufsize = sb_max; 1897 (void) sbreserve(&so->so_rcv, bufsize); 1898 } 1899 } 1900 1901 /* 1902 * Check if there's an initial rtt or rttvar. Convert from the 1903 * route-table units to scaled multiples of the slow timeout timer. 1904 * Called only during the 3-way handshake. 1905 */ 1906 void 1907 tcp_rmx_rtt(tp) 1908 struct tcpcb *tp; 1909 { 1910 #ifdef RTV_RTT 1911 struct rtentry *rt = NULL; 1912 int rtt; 1913 1914 #ifdef DIAGNOSTIC 1915 if (tp->t_inpcb && tp->t_in6pcb) 1916 panic("tcp_rmx_rtt: both t_inpcb and t_in6pcb are set"); 1917 #endif 1918 #ifdef INET 1919 if (tp->t_inpcb) 1920 rt = in_pcbrtentry(tp->t_inpcb); 1921 #endif 1922 #ifdef INET6 1923 if (tp->t_in6pcb) 1924 rt = in6_pcbrtentry(tp->t_in6pcb); 1925 #endif 1926 if (rt == NULL) 1927 return; 1928 1929 if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt)) { 1930 /* 1931 * XXX The lock bit for MTU indicates that the value 1932 * is also a minimum value; this is subject to time. 1933 */ 1934 if (rt->rt_rmx.rmx_locks & RTV_RTT) 1935 TCPT_RANGESET(tp->t_rttmin, 1936 rtt / (RTM_RTTUNIT / PR_SLOWHZ), 1937 TCPTV_MIN, TCPTV_REXMTMAX); 1938 tp->t_srtt = rtt / 1939 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1940 if (rt->rt_rmx.rmx_rttvar) { 1941 tp->t_rttvar = rt->rt_rmx.rmx_rttvar / 1942 ((RTM_RTTUNIT / PR_SLOWHZ) >> 1943 (TCP_RTTVAR_SHIFT + 2)); 1944 } else { 1945 /* Default variation is +- 1 rtt */ 1946 tp->t_rttvar = 1947 tp->t_srtt >> (TCP_RTT_SHIFT - TCP_RTTVAR_SHIFT); 1948 } 1949 TCPT_RANGESET(tp->t_rxtcur, 1950 ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2), 1951 tp->t_rttmin, TCPTV_REXMTMAX); 1952 } 1953 #endif 1954 } 1955 1956 tcp_seq tcp_iss_seq = 0; /* tcp initial seq # */ 1957 #if NRND > 0 1958 u_int8_t tcp_iss_secret[16]; /* 128 bits; should be plenty */ 1959 #endif 1960 1961 /* 1962 * Get a new sequence value given a tcp control block 1963 */ 1964 tcp_seq 1965 tcp_new_iss(struct tcpcb *tp, tcp_seq addin) 1966 { 1967 1968 #ifdef INET 1969 if (tp->t_inpcb != NULL) { 1970 return (tcp_new_iss1(&tp->t_inpcb->inp_laddr, 1971 &tp->t_inpcb->inp_faddr, tp->t_inpcb->inp_lport, 1972 tp->t_inpcb->inp_fport, sizeof(tp->t_inpcb->inp_laddr), 1973 addin)); 1974 } 1975 #endif 1976 #ifdef INET6 1977 if (tp->t_in6pcb != NULL) { 1978 return (tcp_new_iss1(&tp->t_in6pcb->in6p_laddr, 1979 &tp->t_in6pcb->in6p_faddr, tp->t_in6pcb->in6p_lport, 1980 tp->t_in6pcb->in6p_fport, sizeof(tp->t_in6pcb->in6p_laddr), 1981 addin)); 1982 } 1983 #endif 1984 /* Not possible. */ 1985 panic("tcp_new_iss"); 1986 } 1987 1988 /* 1989 * This routine actually generates a new TCP initial sequence number. 1990 */ 1991 tcp_seq 1992 tcp_new_iss1(void *laddr, void *faddr, u_int16_t lport, u_int16_t fport, 1993 size_t addrsz, tcp_seq addin) 1994 { 1995 tcp_seq tcp_iss; 1996 1997 #if NRND > 0 1998 static int beenhere; 1999 2000 /* 2001 * If we haven't been here before, initialize our cryptographic 2002 * hash secret. 2003 */ 2004 if (beenhere == 0) { 2005 rnd_extract_data(tcp_iss_secret, sizeof(tcp_iss_secret), 2006 RND_EXTRACT_ANY); 2007 beenhere = 1; 2008 } 2009 2010 if (tcp_do_rfc1948) { 2011 MD5_CTX ctx; 2012 u_int8_t hash[16]; /* XXX MD5 knowledge */ 2013 2014 /* 2015 * Compute the base value of the ISS. It is a hash 2016 * of (saddr, sport, daddr, dport, secret). 2017 */ 2018 MD5Init(&ctx); 2019 2020 MD5Update(&ctx, (u_char *) laddr, addrsz); 2021 MD5Update(&ctx, (u_char *) &lport, sizeof(lport)); 2022 2023 MD5Update(&ctx, (u_char *) faddr, addrsz); 2024 MD5Update(&ctx, (u_char *) &fport, sizeof(fport)); 2025 2026 MD5Update(&ctx, tcp_iss_secret, sizeof(tcp_iss_secret)); 2027 2028 MD5Final(hash, &ctx); 2029 2030 memcpy(&tcp_iss, hash, sizeof(tcp_iss)); 2031 2032 /* 2033 * Now increment our "timer", and add it in to 2034 * the computed value. 2035 * 2036 * XXX Use `addin'? 2037 * XXX TCP_ISSINCR too large to use? 2038 */ 2039 tcp_iss_seq += TCP_ISSINCR; 2040 #ifdef TCPISS_DEBUG 2041 printf("ISS hash 0x%08x, ", tcp_iss); 2042 #endif 2043 tcp_iss += tcp_iss_seq + addin; 2044 #ifdef TCPISS_DEBUG 2045 printf("new ISS 0x%08x\n", tcp_iss); 2046 #endif 2047 } else 2048 #endif /* NRND > 0 */ 2049 { 2050 /* 2051 * Randomize. 2052 */ 2053 #if NRND > 0 2054 rnd_extract_data(&tcp_iss, sizeof(tcp_iss), RND_EXTRACT_ANY); 2055 #else 2056 tcp_iss = arc4random(); 2057 #endif 2058 2059 /* 2060 * If we were asked to add some amount to a known value, 2061 * we will take a random value obtained above, mask off 2062 * the upper bits, and add in the known value. We also 2063 * add in a constant to ensure that we are at least a 2064 * certain distance from the original value. 2065 * 2066 * This is used when an old connection is in timed wait 2067 * and we have a new one coming in, for instance. 2068 */ 2069 if (addin != 0) { 2070 #ifdef TCPISS_DEBUG 2071 printf("Random %08x, ", tcp_iss); 2072 #endif 2073 tcp_iss &= TCP_ISS_RANDOM_MASK; 2074 tcp_iss += addin + TCP_ISSINCR; 2075 #ifdef TCPISS_DEBUG 2076 printf("Old ISS %08x, ISS %08x\n", addin, tcp_iss); 2077 #endif 2078 } else { 2079 tcp_iss &= TCP_ISS_RANDOM_MASK; 2080 tcp_iss += tcp_iss_seq; 2081 tcp_iss_seq += TCP_ISSINCR; 2082 #ifdef TCPISS_DEBUG 2083 printf("ISS %08x\n", tcp_iss); 2084 #endif 2085 } 2086 } 2087 2088 if (tcp_compat_42) { 2089 /* 2090 * Limit it to the positive range for really old TCP 2091 * implementations. 2092 * Just AND off the top bit instead of checking if 2093 * is set first - saves a branch 50% of the time. 2094 */ 2095 tcp_iss &= 0x7fffffff; /* XXX */ 2096 } 2097 2098 return (tcp_iss); 2099 } 2100 2101 #ifdef IPSEC 2102 /* compute ESP/AH header size for TCP, including outer IP header. */ 2103 size_t 2104 ipsec4_hdrsiz_tcp(tp) 2105 struct tcpcb *tp; 2106 { 2107 struct inpcb *inp; 2108 size_t hdrsiz; 2109 2110 /* XXX mapped addr case (tp->t_in6pcb) */ 2111 if (!tp || !tp->t_template || !(inp = tp->t_inpcb)) 2112 return 0; 2113 switch (tp->t_family) { 2114 case AF_INET: 2115 /* XXX: should use currect direction. */ 2116 hdrsiz = ipsec4_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, inp); 2117 break; 2118 default: 2119 hdrsiz = 0; 2120 break; 2121 } 2122 2123 return hdrsiz; 2124 } 2125 2126 #ifdef INET6 2127 size_t 2128 ipsec6_hdrsiz_tcp(tp) 2129 struct tcpcb *tp; 2130 { 2131 struct in6pcb *in6p; 2132 size_t hdrsiz; 2133 2134 if (!tp || !tp->t_template || !(in6p = tp->t_in6pcb)) 2135 return 0; 2136 switch (tp->t_family) { 2137 case AF_INET6: 2138 /* XXX: should use currect direction. */ 2139 hdrsiz = ipsec6_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, in6p); 2140 break; 2141 case AF_INET: 2142 /* mapped address case - tricky */ 2143 default: 2144 hdrsiz = 0; 2145 break; 2146 } 2147 2148 return hdrsiz; 2149 } 2150 #endif 2151 #endif /*IPSEC*/ 2152 2153 /* 2154 * Determine the length of the TCP options for this connection. 2155 * 2156 * XXX: What do we do for SACK, when we add that? Just reserve 2157 * all of the space? Otherwise we can't exactly be incrementing 2158 * cwnd by an amount that varies depending on the amount we last 2159 * had to SACK! 2160 */ 2161 2162 u_int 2163 tcp_optlen(tp) 2164 struct tcpcb *tp; 2165 { 2166 if ((tp->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP|TF_NOOPT)) == 2167 (TF_REQ_TSTMP | TF_RCVD_TSTMP)) 2168 return TCPOLEN_TSTAMP_APPA; 2169 else 2170 return 0; 2171 } 2172
Indexes created Mon Sep 22 21:09:42 GMT 2025