tcp_subr.c revision 1.145
1 /* $NetBSD: tcp_subr.c,v 1.145 2003/08/07 16:33:18 agc 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. Neither the name of the University nor the names of its contributors 82 * may be used to endorse or promote products derived from this software 83 * without specific prior written permission. 84 * 85 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 86 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 87 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 88 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 89 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 90 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 91 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 92 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 93 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 94 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 95 * SUCH DAMAGE. 96 * 97 * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 98 */ 99 100 #include <sys/cdefs.h> 101 __KERNEL_RCSID(0, "$NetBSD: tcp_subr.c,v 1.145 2003/08/07 16:33:18 agc Exp $"); 102 103 #include "opt_inet.h" 104 #include "opt_ipsec.h" 105 #include "opt_tcp_compat_42.h" 106 #include "opt_inet_csum.h" 107 #include "opt_mbuftrace.h" 108 #include "rnd.h" 109 110 #include <sys/param.h> 111 #include <sys/proc.h> 112 #include <sys/systm.h> 113 #include <sys/malloc.h> 114 #include <sys/mbuf.h> 115 #include <sys/socket.h> 116 #include <sys/socketvar.h> 117 #include <sys/protosw.h> 118 #include <sys/errno.h> 119 #include <sys/kernel.h> 120 #include <sys/pool.h> 121 #if NRND > 0 122 #include <sys/md5.h> 123 #include <sys/rnd.h> 124 #endif 125 126 #include <net/route.h> 127 #include <net/if.h> 128 129 #include <netinet/in.h> 130 #include <netinet/in_systm.h> 131 #include <netinet/ip.h> 132 #include <netinet/in_pcb.h> 133 #include <netinet/ip_var.h> 134 #include <netinet/ip_icmp.h> 135 136 #ifdef INET6 137 #ifndef INET 138 #include <netinet/in.h> 139 #endif 140 #include <netinet/ip6.h> 141 #include <netinet6/in6_pcb.h> 142 #include <netinet6/ip6_var.h> 143 #include <netinet6/in6_var.h> 144 #include <netinet6/ip6protosw.h> 145 #include <netinet/icmp6.h> 146 #include <netinet6/nd6.h> 147 #endif 148 149 #include <netinet/tcp.h> 150 #include <netinet/tcp_fsm.h> 151 #include <netinet/tcp_seq.h> 152 #include <netinet/tcp_timer.h> 153 #include <netinet/tcp_var.h> 154 #include <netinet/tcpip.h> 155 156 #ifdef IPSEC 157 #include <netinet6/ipsec.h> 158 #endif /*IPSEC*/ 159 160 #ifdef INET6 161 struct in6pcb tcb6; 162 #endif 163 164 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 165 struct tcpstat tcpstat; /* tcp statistics */ 166 u_int32_t tcp_now; /* for RFC 1323 timestamps */ 167 168 /* patchable/settable parameters for tcp */ 169 int tcp_mssdflt = TCP_MSS; 170 int tcp_rttdflt = TCPTV_SRTTDFLT / PR_SLOWHZ; 171 int tcp_do_rfc1323 = 1; /* window scaling / timestamps (obsolete) */ 172 #if NRND > 0 173 int tcp_do_rfc1948 = 0; /* ISS by cryptographic hash */ 174 #endif 175 int tcp_do_sack = 1; /* selective acknowledgement */ 176 int tcp_do_win_scale = 1; /* RFC1323 window scaling */ 177 int tcp_do_timestamps = 1; /* RFC1323 timestamps */ 178 int tcp_do_newreno = 0; /* Use the New Reno algorithms */ 179 int tcp_ack_on_push = 0; /* set to enable immediate ACK-on-PUSH */ 180 #ifndef TCP_INIT_WIN 181 #define TCP_INIT_WIN 1 /* initial slow start window */ 182 #endif 183 #ifndef TCP_INIT_WIN_LOCAL 184 #define TCP_INIT_WIN_LOCAL 4 /* initial slow start window for local nets */ 185 #endif 186 int tcp_init_win = TCP_INIT_WIN; 187 int tcp_init_win_local = TCP_INIT_WIN_LOCAL; 188 int tcp_mss_ifmtu = 0; 189 #ifdef TCP_COMPAT_42 190 int tcp_compat_42 = 1; 191 #else 192 int tcp_compat_42 = 0; 193 #endif 194 int tcp_rst_ppslim = 100; /* 100pps */ 195 196 /* tcb hash */ 197 #ifndef TCBHASHSIZE 198 #define TCBHASHSIZE 128 199 #endif 200 int tcbhashsize = TCBHASHSIZE; 201 202 /* syn hash parameters */ 203 #define TCP_SYN_HASH_SIZE 293 204 #define TCP_SYN_BUCKET_SIZE 35 205 int tcp_syn_cache_size = TCP_SYN_HASH_SIZE; 206 int tcp_syn_cache_limit = TCP_SYN_HASH_SIZE*TCP_SYN_BUCKET_SIZE; 207 int tcp_syn_bucket_limit = 3*TCP_SYN_BUCKET_SIZE; 208 struct syn_cache_head tcp_syn_cache[TCP_SYN_HASH_SIZE]; 209 210 int tcp_freeq __P((struct tcpcb *)); 211 212 #ifdef INET 213 void tcp_mtudisc_callback __P((struct in_addr)); 214 #endif 215 #ifdef INET6 216 void tcp6_mtudisc_callback __P((struct in6_addr *)); 217 #endif 218 219 void tcp_mtudisc __P((struct inpcb *, int)); 220 #ifdef INET6 221 void tcp6_mtudisc __P((struct in6pcb *, int)); 222 #endif 223 224 struct pool tcpcb_pool; 225 226 #ifdef TCP_CSUM_COUNTERS 227 #include <sys/device.h> 228 229 struct evcnt tcp_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 230 NULL, "tcp", "hwcsum bad"); 231 struct evcnt tcp_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 232 NULL, "tcp", "hwcsum ok"); 233 struct evcnt tcp_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 234 NULL, "tcp", "hwcsum data"); 235 struct evcnt tcp_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 236 NULL, "tcp", "swcsum"); 237 #endif /* TCP_CSUM_COUNTERS */ 238 239 #ifdef TCP_OUTPUT_COUNTERS 240 #include <sys/device.h> 241 242 struct evcnt tcp_output_bigheader = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 243 NULL, "tcp", "output big header"); 244 struct evcnt tcp_output_copysmall = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 245 NULL, "tcp", "output copy small"); 246 struct evcnt tcp_output_copybig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 247 NULL, "tcp", "output copy big"); 248 struct evcnt tcp_output_refbig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 249 NULL, "tcp", "output reference big"); 250 #endif /* TCP_OUTPUT_COUNTERS */ 251 252 #ifdef TCP_REASS_COUNTERS 253 #include <sys/device.h> 254 255 struct evcnt tcp_reass_ = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 256 NULL, "tcp_reass", "calls"); 257 struct evcnt tcp_reass_empty = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 258 &tcp_reass_, "tcp_reass", "insert into empty queue"); 259 struct evcnt tcp_reass_iteration[8] = { 260 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", ">7 iterations"), 261 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "1 iteration"), 262 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "2 iterations"), 263 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "3 iterations"), 264 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "4 iterations"), 265 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "5 iterations"), 266 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "6 iterations"), 267 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "7 iterations"), 268 }; 269 struct evcnt tcp_reass_prependfirst = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 270 &tcp_reass_, "tcp_reass", "prepend to first"); 271 struct evcnt tcp_reass_prepend = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 272 &tcp_reass_, "tcp_reass", "prepend"); 273 struct evcnt tcp_reass_insert = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 274 &tcp_reass_, "tcp_reass", "insert"); 275 struct evcnt tcp_reass_inserttail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 276 &tcp_reass_, "tcp_reass", "insert at tail"); 277 struct evcnt tcp_reass_append = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 278 &tcp_reass_, "tcp_reass", "append"); 279 struct evcnt tcp_reass_appendtail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 280 &tcp_reass_, "tcp_reass", "append to tail fragment"); 281 struct evcnt tcp_reass_overlaptail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 282 &tcp_reass_, "tcp_reass", "overlap at end"); 283 struct evcnt tcp_reass_overlapfront = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 284 &tcp_reass_, "tcp_reass", "overlap at start"); 285 struct evcnt tcp_reass_segdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 286 &tcp_reass_, "tcp_reass", "duplicate segment"); 287 struct evcnt tcp_reass_fragdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 288 &tcp_reass_, "tcp_reass", "duplicate fragment"); 289 290 #endif /* TCP_REASS_COUNTERS */ 291 292 #ifdef MBUFTRACE 293 struct mowner tcp_mowner = { "tcp" }; 294 struct mowner tcp_rx_mowner = { "tcp", "rx" }; 295 struct mowner tcp_tx_mowner = { "tcp", "tx" }; 296 #endif 297 298 /* 299 * Tcp initialization 300 */ 301 void 302 tcp_init() 303 { 304 int hlen; 305 306 pool_init(&tcpcb_pool, sizeof(struct tcpcb), 0, 0, 0, "tcpcbpl", 307 NULL); 308 in_pcbinit(&tcbtable, tcbhashsize, tcbhashsize); 309 #ifdef INET6 310 tcb6.in6p_next = tcb6.in6p_prev = &tcb6; 311 #endif 312 313 hlen = sizeof(struct ip) + sizeof(struct tcphdr); 314 #ifdef INET6 315 if (sizeof(struct ip) < sizeof(struct ip6_hdr)) 316 hlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 317 #endif 318 if (max_protohdr < hlen) 319 max_protohdr = hlen; 320 if (max_linkhdr + hlen > MHLEN) 321 panic("tcp_init"); 322 323 #ifdef INET 324 icmp_mtudisc_callback_register(tcp_mtudisc_callback); 325 #endif 326 #ifdef INET6 327 icmp6_mtudisc_callback_register(tcp6_mtudisc_callback); 328 #endif 329 330 /* Initialize timer state. */ 331 tcp_timer_init(); 332 333 /* Initialize the compressed state engine. */ 334 syn_cache_init(); 335 336 #ifdef TCP_CSUM_COUNTERS 337 evcnt_attach_static(&tcp_hwcsum_bad); 338 evcnt_attach_static(&tcp_hwcsum_ok); 339 evcnt_attach_static(&tcp_hwcsum_data); 340 evcnt_attach_static(&tcp_swcsum); 341 #endif /* TCP_CSUM_COUNTERS */ 342 343 #ifdef TCP_OUTPUT_COUNTERS 344 evcnt_attach_static(&tcp_output_bigheader); 345 evcnt_attach_static(&tcp_output_copysmall); 346 evcnt_attach_static(&tcp_output_copybig); 347 evcnt_attach_static(&tcp_output_refbig); 348 #endif /* TCP_OUTPUT_COUNTERS */ 349 350 #ifdef TCP_REASS_COUNTERS 351 evcnt_attach_static(&tcp_reass_); 352 evcnt_attach_static(&tcp_reass_empty); 353 evcnt_attach_static(&tcp_reass_iteration[0]); 354 evcnt_attach_static(&tcp_reass_iteration[1]); 355 evcnt_attach_static(&tcp_reass_iteration[2]); 356 evcnt_attach_static(&tcp_reass_iteration[3]); 357 evcnt_attach_static(&tcp_reass_iteration[4]); 358 evcnt_attach_static(&tcp_reass_iteration[5]); 359 evcnt_attach_static(&tcp_reass_iteration[6]); 360 evcnt_attach_static(&tcp_reass_iteration[7]); 361 evcnt_attach_static(&tcp_reass_prependfirst); 362 evcnt_attach_static(&tcp_reass_prepend); 363 evcnt_attach_static(&tcp_reass_insert); 364 evcnt_attach_static(&tcp_reass_inserttail); 365 evcnt_attach_static(&tcp_reass_append); 366 evcnt_attach_static(&tcp_reass_appendtail); 367 evcnt_attach_static(&tcp_reass_overlaptail); 368 evcnt_attach_static(&tcp_reass_overlapfront); 369 evcnt_attach_static(&tcp_reass_segdup); 370 evcnt_attach_static(&tcp_reass_fragdup); 371 #endif /* TCP_REASS_COUNTERS */ 372 373 MOWNER_ATTACH(&tcp_tx_mowner); 374 MOWNER_ATTACH(&tcp_rx_mowner); 375 MOWNER_ATTACH(&tcp_mowner); 376 } 377 378 /* 379 * Create template to be used to send tcp packets on a connection. 380 * Call after host entry created, allocates an mbuf and fills 381 * in a skeletal tcp/ip header, minimizing the amount of work 382 * necessary when the connection is used. 383 */ 384 struct mbuf * 385 tcp_template(tp) 386 struct tcpcb *tp; 387 { 388 struct inpcb *inp = tp->t_inpcb; 389 #ifdef INET6 390 struct in6pcb *in6p = tp->t_in6pcb; 391 #endif 392 struct tcphdr *n; 393 struct mbuf *m; 394 int hlen; 395 396 switch (tp->t_family) { 397 case AF_INET: 398 hlen = sizeof(struct ip); 399 if (inp) 400 break; 401 #ifdef INET6 402 if (in6p) { 403 /* mapped addr case */ 404 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr) 405 && IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) 406 break; 407 } 408 #endif 409 return NULL; /*EINVAL*/ 410 #ifdef INET6 411 case AF_INET6: 412 hlen = sizeof(struct ip6_hdr); 413 if (in6p) { 414 /* more sainty check? */ 415 break; 416 } 417 return NULL; /*EINVAL*/ 418 #endif 419 default: 420 hlen = 0; /*pacify gcc*/ 421 return NULL; /*EAFNOSUPPORT*/ 422 } 423 #ifdef DIAGNOSTIC 424 if (hlen + sizeof(struct tcphdr) > MCLBYTES) 425 panic("mclbytes too small for t_template"); 426 #endif 427 m = tp->t_template; 428 if (m && m->m_len == hlen + sizeof(struct tcphdr)) 429 ; 430 else { 431 if (m) 432 m_freem(m); 433 m = tp->t_template = NULL; 434 MGETHDR(m, M_DONTWAIT, MT_HEADER); 435 if (m && hlen + sizeof(struct tcphdr) > MHLEN) { 436 MCLGET(m, M_DONTWAIT); 437 if ((m->m_flags & M_EXT) == 0) { 438 m_free(m); 439 m = NULL; 440 } 441 } 442 if (m == NULL) 443 return NULL; 444 MCLAIM(m, &tcp_mowner); 445 m->m_pkthdr.len = m->m_len = hlen + sizeof(struct tcphdr); 446 } 447 448 bzero(mtod(m, caddr_t), m->m_len); 449 450 n = (struct tcphdr *)(mtod(m, caddr_t) + hlen); 451 452 switch (tp->t_family) { 453 case AF_INET: 454 { 455 struct ipovly *ipov; 456 mtod(m, struct ip *)->ip_v = 4; 457 ipov = mtod(m, struct ipovly *); 458 ipov->ih_pr = IPPROTO_TCP; 459 ipov->ih_len = htons(sizeof(struct tcphdr)); 460 if (inp) { 461 ipov->ih_src = inp->inp_laddr; 462 ipov->ih_dst = inp->inp_faddr; 463 } 464 #ifdef INET6 465 else if (in6p) { 466 /* mapped addr case */ 467 bcopy(&in6p->in6p_laddr.s6_addr32[3], &ipov->ih_src, 468 sizeof(ipov->ih_src)); 469 bcopy(&in6p->in6p_faddr.s6_addr32[3], &ipov->ih_dst, 470 sizeof(ipov->ih_dst)); 471 } 472 #endif 473 /* 474 * Compute the pseudo-header portion of the checksum 475 * now. We incrementally add in the TCP option and 476 * payload lengths later, and then compute the TCP 477 * checksum right before the packet is sent off onto 478 * the wire. 479 */ 480 n->th_sum = in_cksum_phdr(ipov->ih_src.s_addr, 481 ipov->ih_dst.s_addr, 482 htons(sizeof(struct tcphdr) + IPPROTO_TCP)); 483 break; 484 } 485 #ifdef INET6 486 case AF_INET6: 487 { 488 struct ip6_hdr *ip6; 489 mtod(m, struct ip *)->ip_v = 6; 490 ip6 = mtod(m, struct ip6_hdr *); 491 ip6->ip6_nxt = IPPROTO_TCP; 492 ip6->ip6_plen = htons(sizeof(struct tcphdr)); 493 ip6->ip6_src = in6p->in6p_laddr; 494 ip6->ip6_dst = in6p->in6p_faddr; 495 ip6->ip6_flow = in6p->in6p_flowinfo & IPV6_FLOWINFO_MASK; 496 if (ip6_auto_flowlabel) { 497 ip6->ip6_flow &= ~IPV6_FLOWLABEL_MASK; 498 ip6->ip6_flow |= 499 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 500 } 501 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 502 ip6->ip6_vfc |= IPV6_VERSION; 503 504 /* 505 * Compute the pseudo-header portion of the checksum 506 * now. We incrementally add in the TCP option and 507 * payload lengths later, and then compute the TCP 508 * checksum right before the packet is sent off onto 509 * the wire. 510 */ 511 n->th_sum = in6_cksum_phdr(&in6p->in6p_laddr, 512 &in6p->in6p_faddr, htonl(sizeof(struct tcphdr)), 513 htonl(IPPROTO_TCP)); 514 break; 515 } 516 #endif 517 } 518 if (inp) { 519 n->th_sport = inp->inp_lport; 520 n->th_dport = inp->inp_fport; 521 } 522 #ifdef INET6 523 else if (in6p) { 524 n->th_sport = in6p->in6p_lport; 525 n->th_dport = in6p->in6p_fport; 526 } 527 #endif 528 n->th_seq = 0; 529 n->th_ack = 0; 530 n->th_x2 = 0; 531 n->th_off = 5; 532 n->th_flags = 0; 533 n->th_win = 0; 534 n->th_urp = 0; 535 return (m); 536 } 537 538 /* 539 * Send a single message to the TCP at address specified by 540 * the given TCP/IP header. If m == 0, then we make a copy 541 * of the tcpiphdr at ti and send directly to the addressed host. 542 * This is used to force keep alive messages out using the TCP 543 * template for a connection tp->t_template. If flags are given 544 * then we send a message back to the TCP which originated the 545 * segment ti, and discard the mbuf containing it and any other 546 * attached mbufs. 547 * 548 * In any case the ack and sequence number of the transmitted 549 * segment are as specified by the parameters. 550 */ 551 int 552 tcp_respond(tp, template, m, th0, ack, seq, flags) 553 struct tcpcb *tp; 554 struct mbuf *template; 555 struct mbuf *m; 556 struct tcphdr *th0; 557 tcp_seq ack, seq; 558 int flags; 559 { 560 struct route *ro; 561 int error, tlen, win = 0; 562 int hlen; 563 struct ip *ip; 564 #ifdef INET6 565 struct ip6_hdr *ip6; 566 #endif 567 int family; /* family on packet, not inpcb/in6pcb! */ 568 struct tcphdr *th; 569 570 if (tp != NULL && (flags & TH_RST) == 0) { 571 #ifdef DIAGNOSTIC 572 if (tp->t_inpcb && tp->t_in6pcb) 573 panic("tcp_respond: both t_inpcb and t_in6pcb are set"); 574 #endif 575 #ifdef INET 576 if (tp->t_inpcb) 577 win = sbspace(&tp->t_inpcb->inp_socket->so_rcv); 578 #endif 579 #ifdef INET6 580 if (tp->t_in6pcb) 581 win = sbspace(&tp->t_in6pcb->in6p_socket->so_rcv); 582 #endif 583 } 584 585 th = NULL; /* Quell uninitialized warning */ 586 ip = NULL; 587 #ifdef INET6 588 ip6 = NULL; 589 #endif 590 if (m == 0) { 591 if (!template) 592 return EINVAL; 593 594 /* get family information from template */ 595 switch (mtod(template, struct ip *)->ip_v) { 596 case 4: 597 family = AF_INET; 598 hlen = sizeof(struct ip); 599 break; 600 #ifdef INET6 601 case 6: 602 family = AF_INET6; 603 hlen = sizeof(struct ip6_hdr); 604 break; 605 #endif 606 default: 607 return EAFNOSUPPORT; 608 } 609 610 MGETHDR(m, M_DONTWAIT, MT_HEADER); 611 if (m) { 612 MCLAIM(m, &tcp_tx_mowner); 613 MCLGET(m, M_DONTWAIT); 614 if ((m->m_flags & M_EXT) == 0) { 615 m_free(m); 616 m = NULL; 617 } 618 } 619 if (m == NULL) 620 return (ENOBUFS); 621 622 if (tcp_compat_42) 623 tlen = 1; 624 else 625 tlen = 0; 626 627 m->m_data += max_linkhdr; 628 bcopy(mtod(template, caddr_t), mtod(m, caddr_t), 629 template->m_len); 630 switch (family) { 631 case AF_INET: 632 ip = mtod(m, struct ip *); 633 th = (struct tcphdr *)(ip + 1); 634 break; 635 #ifdef INET6 636 case AF_INET6: 637 ip6 = mtod(m, struct ip6_hdr *); 638 th = (struct tcphdr *)(ip6 + 1); 639 break; 640 #endif 641 #if 0 642 default: 643 /* noone will visit here */ 644 m_freem(m); 645 return EAFNOSUPPORT; 646 #endif 647 } 648 flags = TH_ACK; 649 } else { 650 651 if ((m->m_flags & M_PKTHDR) == 0) { 652 #if 0 653 printf("non PKTHDR to tcp_respond\n"); 654 #endif 655 m_freem(m); 656 return EINVAL; 657 } 658 #ifdef DIAGNOSTIC 659 if (!th0) 660 panic("th0 == NULL in tcp_respond"); 661 #endif 662 663 /* get family information from m */ 664 switch (mtod(m, struct ip *)->ip_v) { 665 case 4: 666 family = AF_INET; 667 hlen = sizeof(struct ip); 668 ip = mtod(m, struct ip *); 669 break; 670 #ifdef INET6 671 case 6: 672 family = AF_INET6; 673 hlen = sizeof(struct ip6_hdr); 674 ip6 = mtod(m, struct ip6_hdr *); 675 break; 676 #endif 677 default: 678 m_freem(m); 679 return EAFNOSUPPORT; 680 } 681 if ((flags & TH_SYN) == 0 || sizeof(*th0) > (th0->th_off << 2)) 682 tlen = sizeof(*th0); 683 else 684 tlen = th0->th_off << 2; 685 686 if (m->m_len > hlen + tlen && (m->m_flags & M_EXT) == 0 && 687 mtod(m, caddr_t) + hlen == (caddr_t)th0) { 688 m->m_len = hlen + tlen; 689 m_freem(m->m_next); 690 m->m_next = NULL; 691 } else { 692 struct mbuf *n; 693 694 #ifdef DIAGNOSTIC 695 if (max_linkhdr + hlen + tlen > MCLBYTES) { 696 m_freem(m); 697 return EMSGSIZE; 698 } 699 #endif 700 MGETHDR(n, M_DONTWAIT, MT_HEADER); 701 if (n && max_linkhdr + hlen + tlen > MHLEN) { 702 MCLGET(n, M_DONTWAIT); 703 if ((n->m_flags & M_EXT) == 0) { 704 m_freem(n); 705 n = NULL; 706 } 707 } 708 if (!n) { 709 m_freem(m); 710 return ENOBUFS; 711 } 712 713 MCLAIM(n, &tcp_tx_mowner); 714 n->m_data += max_linkhdr; 715 n->m_len = hlen + tlen; 716 m_copyback(n, 0, hlen, mtod(m, caddr_t)); 717 m_copyback(n, hlen, tlen, (caddr_t)th0); 718 719 m_freem(m); 720 m = n; 721 n = NULL; 722 } 723 724 #define xchg(a,b,type) { type t; t=a; a=b; b=t; } 725 switch (family) { 726 case AF_INET: 727 ip = mtod(m, struct ip *); 728 th = (struct tcphdr *)(ip + 1); 729 ip->ip_p = IPPROTO_TCP; 730 xchg(ip->ip_dst, ip->ip_src, struct in_addr); 731 ip->ip_p = IPPROTO_TCP; 732 break; 733 #ifdef INET6 734 case AF_INET6: 735 ip6 = mtod(m, struct ip6_hdr *); 736 th = (struct tcphdr *)(ip6 + 1); 737 ip6->ip6_nxt = IPPROTO_TCP; 738 xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); 739 ip6->ip6_nxt = IPPROTO_TCP; 740 break; 741 #endif 742 #if 0 743 default: 744 /* noone will visit here */ 745 m_freem(m); 746 return EAFNOSUPPORT; 747 #endif 748 } 749 xchg(th->th_dport, th->th_sport, u_int16_t); 750 #undef xchg 751 tlen = 0; /*be friendly with the following code*/ 752 } 753 th->th_seq = htonl(seq); 754 th->th_ack = htonl(ack); 755 th->th_x2 = 0; 756 if ((flags & TH_SYN) == 0) { 757 if (tp) 758 win >>= tp->rcv_scale; 759 if (win > TCP_MAXWIN) 760 win = TCP_MAXWIN; 761 th->th_win = htons((u_int16_t)win); 762 th->th_off = sizeof (struct tcphdr) >> 2; 763 tlen += sizeof(*th); 764 } else 765 tlen += th->th_off << 2; 766 m->m_len = hlen + tlen; 767 m->m_pkthdr.len = hlen + tlen; 768 m->m_pkthdr.rcvif = (struct ifnet *) 0; 769 th->th_flags = flags; 770 th->th_urp = 0; 771 772 switch (family) { 773 #ifdef INET 774 case AF_INET: 775 { 776 struct ipovly *ipov = (struct ipovly *)ip; 777 bzero(ipov->ih_x1, sizeof ipov->ih_x1); 778 ipov->ih_len = htons((u_int16_t)tlen); 779 780 th->th_sum = 0; 781 th->th_sum = in_cksum(m, hlen + tlen); 782 ip->ip_len = htons(hlen + tlen); 783 ip->ip_ttl = ip_defttl; 784 break; 785 } 786 #endif 787 #ifdef INET6 788 case AF_INET6: 789 { 790 th->th_sum = 0; 791 th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), 792 tlen); 793 ip6->ip6_plen = ntohs(tlen); 794 if (tp && tp->t_in6pcb) { 795 struct ifnet *oifp; 796 ro = (struct route *)&tp->t_in6pcb->in6p_route; 797 oifp = ro->ro_rt ? ro->ro_rt->rt_ifp : NULL; 798 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, oifp); 799 } else 800 ip6->ip6_hlim = ip6_defhlim; 801 ip6->ip6_flow &= ~IPV6_FLOWINFO_MASK; 802 if (ip6_auto_flowlabel) { 803 ip6->ip6_flow |= 804 (htonl(ip6_flow_seq++) & IPV6_FLOWLABEL_MASK); 805 } 806 break; 807 } 808 #endif 809 } 810 811 #ifdef IPSEC 812 (void)ipsec_setsocket(m, NULL); 813 #endif /*IPSEC*/ 814 815 if (tp != NULL && tp->t_inpcb != NULL) { 816 ro = &tp->t_inpcb->inp_route; 817 #ifdef IPSEC 818 if (ipsec_setsocket(m, tp->t_inpcb->inp_socket) != 0) { 819 m_freem(m); 820 return ENOBUFS; 821 } 822 #endif 823 #ifdef DIAGNOSTIC 824 if (family != AF_INET) 825 panic("tcp_respond: address family mismatch"); 826 if (!in_hosteq(ip->ip_dst, tp->t_inpcb->inp_faddr)) { 827 panic("tcp_respond: ip_dst %x != inp_faddr %x", 828 ntohl(ip->ip_dst.s_addr), 829 ntohl(tp->t_inpcb->inp_faddr.s_addr)); 830 } 831 #endif 832 } 833 #ifdef INET6 834 else if (tp != NULL && tp->t_in6pcb != NULL) { 835 ro = (struct route *)&tp->t_in6pcb->in6p_route; 836 #ifdef IPSEC 837 if (ipsec_setsocket(m, tp->t_in6pcb->in6p_socket) != 0) { 838 m_freem(m); 839 return ENOBUFS; 840 } 841 #endif 842 #ifdef DIAGNOSTIC 843 if (family == AF_INET) { 844 if (!IN6_IS_ADDR_V4MAPPED(&tp->t_in6pcb->in6p_faddr)) 845 panic("tcp_respond: not mapped addr"); 846 if (bcmp(&ip->ip_dst, 847 &tp->t_in6pcb->in6p_faddr.s6_addr32[3], 848 sizeof(ip->ip_dst)) != 0) { 849 panic("tcp_respond: ip_dst != in6p_faddr"); 850 } 851 } else if (family == AF_INET6) { 852 if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 853 &tp->t_in6pcb->in6p_faddr)) 854 panic("tcp_respond: ip6_dst != in6p_faddr"); 855 } else 856 panic("tcp_respond: address family mismatch"); 857 #endif 858 } 859 #endif 860 else 861 ro = NULL; 862 863 switch (family) { 864 #ifdef INET 865 case AF_INET: 866 error = ip_output(m, NULL, ro, 867 (tp && tp->t_mtudisc ? IP_MTUDISC : 0), 868 NULL); 869 break; 870 #endif 871 #ifdef INET6 872 case AF_INET6: 873 error = ip6_output(m, NULL, (struct route_in6 *)ro, 0, NULL, 874 NULL); 875 break; 876 #endif 877 default: 878 error = EAFNOSUPPORT; 879 break; 880 } 881 882 return (error); 883 } 884 885 /* 886 * Create a new TCP control block, making an 887 * empty reassembly queue and hooking it to the argument 888 * protocol control block. 889 */ 890 struct tcpcb * 891 tcp_newtcpcb(family, aux) 892 int family; /* selects inpcb, or in6pcb */ 893 void *aux; 894 { 895 struct tcpcb *tp; 896 int i; 897 898 switch (family) { 899 case PF_INET: 900 break; 901 #ifdef INET6 902 case PF_INET6: 903 break; 904 #endif 905 default: 906 return NULL; 907 } 908 909 tp = pool_get(&tcpcb_pool, PR_NOWAIT); 910 if (tp == NULL) 911 return (NULL); 912 bzero((caddr_t)tp, sizeof(struct tcpcb)); 913 TAILQ_INIT(&tp->segq); 914 TAILQ_INIT(&tp->timeq); 915 tp->t_family = family; /* may be overridden later on */ 916 tp->t_peermss = tcp_mssdflt; 917 tp->t_ourmss = tcp_mssdflt; 918 tp->t_segsz = tcp_mssdflt; 919 LIST_INIT(&tp->t_sc); 920 921 tp->t_lastm = NULL; 922 tp->t_lastoff = 0; 923 924 callout_init(&tp->t_delack_ch); 925 for (i = 0; i < TCPT_NTIMERS; i++) 926 TCP_TIMER_INIT(tp, i); 927 928 tp->t_flags = 0; 929 if (tcp_do_rfc1323 && tcp_do_win_scale) 930 tp->t_flags |= TF_REQ_SCALE; 931 if (tcp_do_rfc1323 && tcp_do_timestamps) 932 tp->t_flags |= TF_REQ_TSTMP; 933 if (tcp_do_sack == 2) 934 tp->t_flags |= TF_WILL_SACK; 935 else if (tcp_do_sack == 1) 936 tp->t_flags |= TF_WILL_SACK|TF_IGNR_RXSACK; 937 tp->t_flags |= TF_CANT_TXSACK; 938 switch (family) { 939 case PF_INET: 940 tp->t_inpcb = (struct inpcb *)aux; 941 tp->t_mtudisc = ip_mtudisc; 942 break; 943 #ifdef INET6 944 case PF_INET6: 945 tp->t_in6pcb = (struct in6pcb *)aux; 946 /* for IPv6, always try to run path MTU discovery */ 947 tp->t_mtudisc = 1; 948 break; 949 #endif 950 } 951 /* 952 * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no 953 * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives 954 * reasonable initial retransmit time. 955 */ 956 tp->t_srtt = TCPTV_SRTTBASE; 957 tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << (TCP_RTTVAR_SHIFT + 2 - 1); 958 tp->t_rttmin = TCPTV_MIN; 959 TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), 960 TCPTV_MIN, TCPTV_REXMTMAX); 961 tp->snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT; 962 tp->snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT; 963 if (family == AF_INET) { 964 struct inpcb *inp = (struct inpcb *)aux; 965 inp->inp_ip.ip_ttl = ip_defttl; 966 inp->inp_ppcb = (caddr_t)tp; 967 } 968 #ifdef INET6 969 else if (family == AF_INET6) { 970 struct in6pcb *in6p = (struct in6pcb *)aux; 971 in6p->in6p_ip6.ip6_hlim = in6_selecthlim(in6p, 972 in6p->in6p_route.ro_rt ? in6p->in6p_route.ro_rt->rt_ifp 973 : NULL); 974 in6p->in6p_ppcb = (caddr_t)tp; 975 } 976 #endif 977 978 /* 979 * Initialize our timebase. When we send timestamps, we take 980 * the delta from tcp_now -- this means each connection always 981 * gets a timebase of 0, which makes it, among other things, 982 * more difficult to determine how long a system has been up, 983 * and thus how many TCP sequence increments have occurred. 984 */ 985 tp->ts_timebase = tcp_now; 986 987 return (tp); 988 } 989 990 /* 991 * Drop a TCP connection, reporting 992 * the specified error. If connection is synchronized, 993 * then send a RST to peer. 994 */ 995 struct tcpcb * 996 tcp_drop(tp, errno) 997 struct tcpcb *tp; 998 int errno; 999 { 1000 struct socket *so = NULL; 1001 1002 #ifdef DIAGNOSTIC 1003 if (tp->t_inpcb && tp->t_in6pcb) 1004 panic("tcp_drop: both t_inpcb and t_in6pcb are set"); 1005 #endif 1006 #ifdef INET 1007 if (tp->t_inpcb) 1008 so = tp->t_inpcb->inp_socket; 1009 #endif 1010 #ifdef INET6 1011 if (tp->t_in6pcb) 1012 so = tp->t_in6pcb->in6p_socket; 1013 #endif 1014 if (!so) 1015 return NULL; 1016 1017 if (TCPS_HAVERCVDSYN(tp->t_state)) { 1018 tp->t_state = TCPS_CLOSED; 1019 (void) tcp_output(tp); 1020 tcpstat.tcps_drops++; 1021 } else 1022 tcpstat.tcps_conndrops++; 1023 if (errno == ETIMEDOUT && tp->t_softerror) 1024 errno = tp->t_softerror; 1025 so->so_error = errno; 1026 return (tcp_close(tp)); 1027 } 1028 1029 /* 1030 * Return whether this tcpcb is marked as dead, indicating 1031 * to the calling timer function that no further action should 1032 * be taken, as we are about to release this tcpcb. The release 1033 * of the storage will be done if this is the last timer running. 1034 * 1035 * This is typically called from the callout handler function before 1036 * callout_ack() is done, therefore we need to test the number of 1037 * running timer functions against 1 below, not 0. 1038 */ 1039 int 1040 tcp_isdead(tp) 1041 struct tcpcb *tp; 1042 { 1043 int dead = (tp->t_flags & TF_DEAD); 1044 1045 if (__predict_false(dead)) { 1046 if (tcp_timers_invoking(tp) > 1) 1047 /* not quite there yet -- count separately? */ 1048 return dead; 1049 tcpstat.tcps_delayed_free++; 1050 pool_put(&tcpcb_pool, tp); 1051 } 1052 return dead; 1053 } 1054 1055 /* 1056 * Close a TCP control block: 1057 * discard all space held by the tcp 1058 * discard internet protocol block 1059 * wake up any sleepers 1060 */ 1061 struct tcpcb * 1062 tcp_close(tp) 1063 struct tcpcb *tp; 1064 { 1065 struct inpcb *inp; 1066 #ifdef INET6 1067 struct in6pcb *in6p; 1068 #endif 1069 struct socket *so; 1070 #ifdef RTV_RTT 1071 struct rtentry *rt; 1072 #endif 1073 struct route *ro; 1074 1075 inp = tp->t_inpcb; 1076 #ifdef INET6 1077 in6p = tp->t_in6pcb; 1078 #endif 1079 so = NULL; 1080 ro = NULL; 1081 if (inp) { 1082 so = inp->inp_socket; 1083 ro = &inp->inp_route; 1084 } 1085 #ifdef INET6 1086 else if (in6p) { 1087 so = in6p->in6p_socket; 1088 ro = (struct route *)&in6p->in6p_route; 1089 } 1090 #endif 1091 1092 #ifdef RTV_RTT 1093 /* 1094 * If we sent enough data to get some meaningful characteristics, 1095 * save them in the routing entry. 'Enough' is arbitrarily 1096 * defined as the sendpipesize (default 4K) * 16. This would 1097 * give us 16 rtt samples assuming we only get one sample per 1098 * window (the usual case on a long haul net). 16 samples is 1099 * enough for the srtt filter to converge to within 5% of the correct 1100 * value; fewer samples and we could save a very bogus rtt. 1101 * 1102 * Don't update the default route's characteristics and don't 1103 * update anything that the user "locked". 1104 */ 1105 if (SEQ_LT(tp->iss + so->so_snd.sb_hiwat * 16, tp->snd_max) && 1106 ro && (rt = ro->ro_rt) && 1107 !in_nullhost(satosin(rt_key(rt))->sin_addr)) { 1108 u_long i = 0; 1109 1110 if ((rt->rt_rmx.rmx_locks & RTV_RTT) == 0) { 1111 i = tp->t_srtt * 1112 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1113 if (rt->rt_rmx.rmx_rtt && i) 1114 /* 1115 * filter this update to half the old & half 1116 * the new values, converting scale. 1117 * See route.h and tcp_var.h for a 1118 * description of the scaling constants. 1119 */ 1120 rt->rt_rmx.rmx_rtt = 1121 (rt->rt_rmx.rmx_rtt + i) / 2; 1122 else 1123 rt->rt_rmx.rmx_rtt = i; 1124 } 1125 if ((rt->rt_rmx.rmx_locks & RTV_RTTVAR) == 0) { 1126 i = tp->t_rttvar * 1127 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTTVAR_SHIFT + 2)); 1128 if (rt->rt_rmx.rmx_rttvar && i) 1129 rt->rt_rmx.rmx_rttvar = 1130 (rt->rt_rmx.rmx_rttvar + i) / 2; 1131 else 1132 rt->rt_rmx.rmx_rttvar = i; 1133 } 1134 /* 1135 * update the pipelimit (ssthresh) if it has been updated 1136 * already or if a pipesize was specified & the threshhold 1137 * got below half the pipesize. I.e., wait for bad news 1138 * before we start updating, then update on both good 1139 * and bad news. 1140 */ 1141 if (((rt->rt_rmx.rmx_locks & RTV_SSTHRESH) == 0 && 1142 (i = tp->snd_ssthresh) && rt->rt_rmx.rmx_ssthresh) || 1143 i < (rt->rt_rmx.rmx_sendpipe / 2)) { 1144 /* 1145 * convert the limit from user data bytes to 1146 * packets then to packet data bytes. 1147 */ 1148 i = (i + tp->t_segsz / 2) / tp->t_segsz; 1149 if (i < 2) 1150 i = 2; 1151 i *= (u_long)(tp->t_segsz + sizeof (struct tcpiphdr)); 1152 if (rt->rt_rmx.rmx_ssthresh) 1153 rt->rt_rmx.rmx_ssthresh = 1154 (rt->rt_rmx.rmx_ssthresh + i) / 2; 1155 else 1156 rt->rt_rmx.rmx_ssthresh = i; 1157 } 1158 } 1159 #endif /* RTV_RTT */ 1160 /* free the reassembly queue, if any */ 1161 TCP_REASS_LOCK(tp); 1162 (void) tcp_freeq(tp); 1163 TCP_REASS_UNLOCK(tp); 1164 1165 tcp_canceltimers(tp); 1166 TCP_CLEAR_DELACK(tp); 1167 syn_cache_cleanup(tp); 1168 1169 if (tp->t_template) { 1170 m_free(tp->t_template); 1171 tp->t_template = NULL; 1172 } 1173 if (tcp_timers_invoking(tp)) 1174 tp->t_flags |= TF_DEAD; 1175 else 1176 pool_put(&tcpcb_pool, tp); 1177 1178 if (inp) { 1179 inp->inp_ppcb = 0; 1180 soisdisconnected(so); 1181 in_pcbdetach(inp); 1182 } 1183 #ifdef INET6 1184 else if (in6p) { 1185 in6p->in6p_ppcb = 0; 1186 soisdisconnected(so); 1187 in6_pcbdetach(in6p); 1188 } 1189 #endif 1190 tcpstat.tcps_closed++; 1191 return ((struct tcpcb *)0); 1192 } 1193 1194 int 1195 tcp_freeq(tp) 1196 struct tcpcb *tp; 1197 { 1198 struct ipqent *qe; 1199 int rv = 0; 1200 #ifdef TCPREASS_DEBUG 1201 int i = 0; 1202 #endif 1203 1204 TCP_REASS_LOCK_CHECK(tp); 1205 1206 while ((qe = TAILQ_FIRST(&tp->segq)) != NULL) { 1207 #ifdef TCPREASS_DEBUG 1208 printf("tcp_freeq[%p,%d]: %u:%u(%u) 0x%02x\n", 1209 tp, i++, qe->ipqe_seq, qe->ipqe_seq + qe->ipqe_len, 1210 qe->ipqe_len, qe->ipqe_flags & (TH_SYN|TH_FIN|TH_RST)); 1211 #endif 1212 TAILQ_REMOVE(&tp->segq, qe, ipqe_q); 1213 TAILQ_REMOVE(&tp->timeq, qe, ipqe_timeq); 1214 m_freem(qe->ipqe_m); 1215 pool_put(&ipqent_pool, qe); 1216 rv = 1; 1217 } 1218 return (rv); 1219 } 1220 1221 /* 1222 * Protocol drain routine. Called when memory is in short supply. 1223 */ 1224 void 1225 tcp_drain() 1226 { 1227 struct inpcb *inp; 1228 struct tcpcb *tp; 1229 1230 /* 1231 * Free the sequence queue of all TCP connections. 1232 */ 1233 inp = CIRCLEQ_FIRST(&tcbtable.inpt_queue); 1234 if (inp) /* XXX */ 1235 CIRCLEQ_FOREACH(inp, &tcbtable.inpt_queue, inp_queue) { 1236 if ((tp = intotcpcb(inp)) != NULL) { 1237 /* 1238 * We may be called from a device's interrupt 1239 * context. If the tcpcb is already busy, 1240 * just bail out now. 1241 */ 1242 if (tcp_reass_lock_try(tp) == 0) 1243 continue; 1244 if (tcp_freeq(tp)) 1245 tcpstat.tcps_connsdrained++; 1246 TCP_REASS_UNLOCK(tp); 1247 } 1248 } 1249 } 1250 1251 #ifdef INET6 1252 void 1253 tcp6_drain() 1254 { 1255 struct in6pcb *in6p; 1256 struct tcpcb *tp; 1257 struct in6pcb *head = &tcb6; 1258 1259 /* 1260 * Free the sequence queue of all TCP connections. 1261 */ 1262 for (in6p = head->in6p_next; in6p != head; in6p = in6p->in6p_next) { 1263 if ((tp = in6totcpcb(in6p)) != NULL) { 1264 /* 1265 * We may be called from a device's interrupt 1266 * context. If the tcpcb is already busy, 1267 * just bail out now. 1268 */ 1269 if (tcp_reass_lock_try(tp) == 0) 1270 continue; 1271 if (tcp_freeq(tp)) 1272 tcpstat.tcps_connsdrained++; 1273 TCP_REASS_UNLOCK(tp); 1274 } 1275 } 1276 } 1277 #endif 1278 1279 /* 1280 * Notify a tcp user of an asynchronous error; 1281 * store error as soft error, but wake up user 1282 * (for now, won't do anything until can select for soft error). 1283 */ 1284 void 1285 tcp_notify(inp, error) 1286 struct inpcb *inp; 1287 int error; 1288 { 1289 struct tcpcb *tp = (struct tcpcb *)inp->inp_ppcb; 1290 struct socket *so = inp->inp_socket; 1291 1292 /* 1293 * Ignore some errors if we are hooked up. 1294 * If connection hasn't completed, has retransmitted several times, 1295 * and receives a second error, give up now. This is better 1296 * than waiting a long time to establish a connection that 1297 * can never complete. 1298 */ 1299 if (tp->t_state == TCPS_ESTABLISHED && 1300 (error == EHOSTUNREACH || error == ENETUNREACH || 1301 error == EHOSTDOWN)) { 1302 return; 1303 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1304 tp->t_rxtshift > 3 && tp->t_softerror) 1305 so->so_error = error; 1306 else 1307 tp->t_softerror = error; 1308 wakeup((caddr_t) &so->so_timeo); 1309 sorwakeup(so); 1310 sowwakeup(so); 1311 } 1312 1313 #ifdef INET6 1314 void 1315 tcp6_notify(in6p, error) 1316 struct in6pcb *in6p; 1317 int error; 1318 { 1319 struct tcpcb *tp = (struct tcpcb *)in6p->in6p_ppcb; 1320 struct socket *so = in6p->in6p_socket; 1321 1322 /* 1323 * Ignore some errors if we are hooked up. 1324 * If connection hasn't completed, has retransmitted several times, 1325 * and receives a second error, give up now. This is better 1326 * than waiting a long time to establish a connection that 1327 * can never complete. 1328 */ 1329 if (tp->t_state == TCPS_ESTABLISHED && 1330 (error == EHOSTUNREACH || error == ENETUNREACH || 1331 error == EHOSTDOWN)) { 1332 return; 1333 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1334 tp->t_rxtshift > 3 && tp->t_softerror) 1335 so->so_error = error; 1336 else 1337 tp->t_softerror = error; 1338 wakeup((caddr_t) &so->so_timeo); 1339 sorwakeup(so); 1340 sowwakeup(so); 1341 } 1342 #endif 1343 1344 #ifdef INET6 1345 void 1346 tcp6_ctlinput(cmd, sa, d) 1347 int cmd; 1348 struct sockaddr *sa; 1349 void *d; 1350 { 1351 struct tcphdr th; 1352 void (*notify) __P((struct in6pcb *, int)) = tcp6_notify; 1353 int nmatch; 1354 struct ip6_hdr *ip6; 1355 const struct sockaddr_in6 *sa6_src = NULL; 1356 struct sockaddr_in6 *sa6 = (struct sockaddr_in6 *)sa; 1357 struct mbuf *m; 1358 int off; 1359 1360 if (sa->sa_family != AF_INET6 || 1361 sa->sa_len != sizeof(struct sockaddr_in6)) 1362 return; 1363 if ((unsigned)cmd >= PRC_NCMDS) 1364 return; 1365 else if (cmd == PRC_QUENCH) { 1366 /* XXX there's no PRC_QUENCH in IPv6 */ 1367 notify = tcp6_quench; 1368 } else if (PRC_IS_REDIRECT(cmd)) 1369 notify = in6_rtchange, d = NULL; 1370 else if (cmd == PRC_MSGSIZE) 1371 ; /* special code is present, see below */ 1372 else if (cmd == PRC_HOSTDEAD) 1373 d = NULL; 1374 else if (inet6ctlerrmap[cmd] == 0) 1375 return; 1376 1377 /* if the parameter is from icmp6, decode it. */ 1378 if (d != NULL) { 1379 struct ip6ctlparam *ip6cp = (struct ip6ctlparam *)d; 1380 m = ip6cp->ip6c_m; 1381 ip6 = ip6cp->ip6c_ip6; 1382 off = ip6cp->ip6c_off; 1383 sa6_src = ip6cp->ip6c_src; 1384 } else { 1385 m = NULL; 1386 ip6 = NULL; 1387 sa6_src = &sa6_any; 1388 } 1389 1390 if (ip6) { 1391 /* 1392 * XXX: We assume that when ip6 is non NULL, 1393 * M and OFF are valid. 1394 */ 1395 1396 /* check if we can safely examine src and dst ports */ 1397 if (m->m_pkthdr.len < off + sizeof(th)) { 1398 if (cmd == PRC_MSGSIZE) 1399 icmp6_mtudisc_update((struct ip6ctlparam *)d, 0); 1400 return; 1401 } 1402 1403 bzero(&th, sizeof(th)); 1404 m_copydata(m, off, sizeof(th), (caddr_t)&th); 1405 1406 if (cmd == PRC_MSGSIZE) { 1407 int valid = 0; 1408 1409 /* 1410 * Check to see if we have a valid TCP connection 1411 * corresponding to the address in the ICMPv6 message 1412 * payload. 1413 */ 1414 if (in6_pcblookup_connect(&tcb6, &sa6->sin6_addr, 1415 th.th_dport, (struct in6_addr *)&sa6_src->sin6_addr, 1416 th.th_sport, 0)) 1417 valid++; 1418 1419 /* 1420 * Depending on the value of "valid" and routing table 1421 * size (mtudisc_{hi,lo}wat), we will: 1422 * - recalcurate the new MTU and create the 1423 * corresponding routing entry, or 1424 * - ignore the MTU change notification. 1425 */ 1426 icmp6_mtudisc_update((struct ip6ctlparam *)d, valid); 1427 1428 /* 1429 * no need to call in6_pcbnotify, it should have been 1430 * called via callback if necessary 1431 */ 1432 return; 1433 } 1434 1435 nmatch = in6_pcbnotify(&tcb6, sa, th.th_dport, 1436 (struct sockaddr *)sa6_src, th.th_sport, cmd, NULL, notify); 1437 if (nmatch == 0 && syn_cache_count && 1438 (inet6ctlerrmap[cmd] == EHOSTUNREACH || 1439 inet6ctlerrmap[cmd] == ENETUNREACH || 1440 inet6ctlerrmap[cmd] == EHOSTDOWN)) 1441 syn_cache_unreach((struct sockaddr *)sa6_src, 1442 sa, &th); 1443 } else { 1444 (void) in6_pcbnotify(&tcb6, sa, 0, (struct sockaddr *)sa6_src, 1445 0, cmd, NULL, notify); 1446 } 1447 } 1448 #endif 1449 1450 #ifdef INET 1451 /* assumes that ip header and tcp header are contiguous on mbuf */ 1452 void * 1453 tcp_ctlinput(cmd, sa, v) 1454 int cmd; 1455 struct sockaddr *sa; 1456 void *v; 1457 { 1458 struct ip *ip = v; 1459 struct tcphdr *th; 1460 struct icmp *icp; 1461 extern const int inetctlerrmap[]; 1462 void (*notify) __P((struct inpcb *, int)) = tcp_notify; 1463 int errno; 1464 int nmatch; 1465 #ifdef INET6 1466 struct in6_addr src6, dst6; 1467 #endif 1468 1469 if (sa->sa_family != AF_INET || 1470 sa->sa_len != sizeof(struct sockaddr_in)) 1471 return NULL; 1472 if ((unsigned)cmd >= PRC_NCMDS) 1473 return NULL; 1474 errno = inetctlerrmap[cmd]; 1475 if (cmd == PRC_QUENCH) 1476 notify = tcp_quench; 1477 else if (PRC_IS_REDIRECT(cmd)) 1478 notify = in_rtchange, ip = 0; 1479 else if (cmd == PRC_MSGSIZE && ip && ip->ip_v == 4) { 1480 /* 1481 * Check to see if we have a valid TCP connection 1482 * corresponding to the address in the ICMP message 1483 * payload. 1484 * 1485 * Boundary check is made in icmp_input(), with ICMP_ADVLENMIN. 1486 */ 1487 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1488 #ifdef INET6 1489 memset(&src6, 0, sizeof(src6)); 1490 memset(&dst6, 0, sizeof(dst6)); 1491 src6.s6_addr16[5] = dst6.s6_addr16[5] = 0xffff; 1492 memcpy(&src6.s6_addr32[3], &ip->ip_src, sizeof(struct in_addr)); 1493 memcpy(&dst6.s6_addr32[3], &ip->ip_dst, sizeof(struct in_addr)); 1494 #endif 1495 if (in_pcblookup_connect(&tcbtable, ip->ip_dst, th->th_dport, 1496 ip->ip_src, th->th_sport) != NULL) 1497 ; 1498 #ifdef INET6 1499 else if (in6_pcblookup_connect(&tcb6, &dst6, 1500 th->th_dport, &src6, th->th_sport, 0) != NULL) 1501 ; 1502 #endif 1503 else 1504 return NULL; 1505 1506 /* 1507 * Now that we've validated that we are actually communicating 1508 * with the host indicated in the ICMP message, locate the 1509 * ICMP header, recalculate the new MTU, and create the 1510 * corresponding routing entry. 1511 */ 1512 icp = (struct icmp *)((caddr_t)ip - 1513 offsetof(struct icmp, icmp_ip)); 1514 icmp_mtudisc(icp, ip->ip_dst); 1515 1516 return NULL; 1517 } else if (cmd == PRC_HOSTDEAD) 1518 ip = 0; 1519 else if (errno == 0) 1520 return NULL; 1521 if (ip && ip->ip_v == 4 && sa->sa_family == AF_INET) { 1522 th = (struct tcphdr *)((caddr_t)ip + (ip->ip_hl << 2)); 1523 nmatch = in_pcbnotify(&tcbtable, satosin(sa)->sin_addr, 1524 th->th_dport, ip->ip_src, th->th_sport, errno, notify); 1525 if (nmatch == 0 && syn_cache_count && 1526 (inetctlerrmap[cmd] == EHOSTUNREACH || 1527 inetctlerrmap[cmd] == ENETUNREACH || 1528 inetctlerrmap[cmd] == EHOSTDOWN)) { 1529 struct sockaddr_in sin; 1530 bzero(&sin, sizeof(sin)); 1531 sin.sin_len = sizeof(sin); 1532 sin.sin_family = AF_INET; 1533 sin.sin_port = th->th_sport; 1534 sin.sin_addr = ip->ip_src; 1535 syn_cache_unreach((struct sockaddr *)&sin, sa, th); 1536 } 1537 1538 /* XXX mapped address case */ 1539 } else 1540 in_pcbnotifyall(&tcbtable, satosin(sa)->sin_addr, errno, 1541 notify); 1542 return NULL; 1543 } 1544 1545 /* 1546 * When a source quence is received, we are being notifed of congestion. 1547 * Close the congestion window down to the Loss Window (one segment). 1548 * We will gradually open it again as we proceed. 1549 */ 1550 void 1551 tcp_quench(inp, errno) 1552 struct inpcb *inp; 1553 int errno; 1554 { 1555 struct tcpcb *tp = intotcpcb(inp); 1556 1557 if (tp) 1558 tp->snd_cwnd = tp->t_segsz; 1559 } 1560 #endif 1561 1562 #ifdef INET6 1563 void 1564 tcp6_quench(in6p, errno) 1565 struct in6pcb *in6p; 1566 int errno; 1567 { 1568 struct tcpcb *tp = in6totcpcb(in6p); 1569 1570 if (tp) 1571 tp->snd_cwnd = tp->t_segsz; 1572 } 1573 #endif 1574 1575 #ifdef INET 1576 /* 1577 * Path MTU Discovery handlers. 1578 */ 1579 void 1580 tcp_mtudisc_callback(faddr) 1581 struct in_addr faddr; 1582 { 1583 #ifdef INET6 1584 struct in6_addr in6; 1585 #endif 1586 1587 in_pcbnotifyall(&tcbtable, faddr, EMSGSIZE, tcp_mtudisc); 1588 #ifdef INET6 1589 memset(&in6, 0, sizeof(in6)); 1590 in6.s6_addr16[5] = 0xffff; 1591 memcpy(&in6.s6_addr32[3], &faddr, sizeof(struct in_addr)); 1592 tcp6_mtudisc_callback(&in6); 1593 #endif 1594 } 1595 1596 /* 1597 * On receipt of path MTU corrections, flush old route and replace it 1598 * with the new one. Retransmit all unacknowledged packets, to ensure 1599 * that all packets will be received. 1600 */ 1601 void 1602 tcp_mtudisc(inp, errno) 1603 struct inpcb *inp; 1604 int errno; 1605 { 1606 struct tcpcb *tp = intotcpcb(inp); 1607 struct rtentry *rt = in_pcbrtentry(inp); 1608 1609 if (tp != 0) { 1610 if (rt != 0) { 1611 /* 1612 * If this was not a host route, remove and realloc. 1613 */ 1614 if ((rt->rt_flags & RTF_HOST) == 0) { 1615 in_rtchange(inp, errno); 1616 if ((rt = in_pcbrtentry(inp)) == 0) 1617 return; 1618 } 1619 1620 /* 1621 * Slow start out of the error condition. We 1622 * use the MTU because we know it's smaller 1623 * than the previously transmitted segment. 1624 * 1625 * Note: This is more conservative than the 1626 * suggestion in draft-floyd-incr-init-win-03. 1627 */ 1628 if (rt->rt_rmx.rmx_mtu != 0) 1629 tp->snd_cwnd = 1630 TCP_INITIAL_WINDOW(tcp_init_win, 1631 rt->rt_rmx.rmx_mtu); 1632 } 1633 1634 /* 1635 * Resend unacknowledged packets. 1636 */ 1637 tp->snd_nxt = tp->snd_una; 1638 tcp_output(tp); 1639 } 1640 } 1641 #endif 1642 1643 #ifdef INET6 1644 /* 1645 * Path MTU Discovery handlers. 1646 */ 1647 void 1648 tcp6_mtudisc_callback(faddr) 1649 struct in6_addr *faddr; 1650 { 1651 struct sockaddr_in6 sin6; 1652 1653 bzero(&sin6, sizeof(sin6)); 1654 sin6.sin6_family = AF_INET6; 1655 sin6.sin6_len = sizeof(struct sockaddr_in6); 1656 sin6.sin6_addr = *faddr; 1657 (void) in6_pcbnotify(&tcb6, (struct sockaddr *)&sin6, 0, 1658 (struct sockaddr *)&sa6_any, 0, PRC_MSGSIZE, NULL, tcp6_mtudisc); 1659 } 1660 1661 void 1662 tcp6_mtudisc(in6p, errno) 1663 struct in6pcb *in6p; 1664 int errno; 1665 { 1666 struct tcpcb *tp = in6totcpcb(in6p); 1667 struct rtentry *rt = in6_pcbrtentry(in6p); 1668 1669 if (tp != 0) { 1670 if (rt != 0) { 1671 /* 1672 * If this was not a host route, remove and realloc. 1673 */ 1674 if ((rt->rt_flags & RTF_HOST) == 0) { 1675 in6_rtchange(in6p, errno); 1676 if ((rt = in6_pcbrtentry(in6p)) == 0) 1677 return; 1678 } 1679 1680 /* 1681 * Slow start out of the error condition. We 1682 * use the MTU because we know it's smaller 1683 * than the previously transmitted segment. 1684 * 1685 * Note: This is more conservative than the 1686 * suggestion in draft-floyd-incr-init-win-03. 1687 */ 1688 if (rt->rt_rmx.rmx_mtu != 0) 1689 tp->snd_cwnd = 1690 TCP_INITIAL_WINDOW(tcp_init_win, 1691 rt->rt_rmx.rmx_mtu); 1692 } 1693 1694 /* 1695 * Resend unacknowledged packets. 1696 */ 1697 tp->snd_nxt = tp->snd_una; 1698 tcp_output(tp); 1699 } 1700 } 1701 #endif /* INET6 */ 1702 1703 /* 1704 * Compute the MSS to advertise to the peer. Called only during 1705 * the 3-way handshake. If we are the server (peer initiated 1706 * connection), we are called with a pointer to the interface 1707 * on which the SYN packet arrived. If we are the client (we 1708 * initiated connection), we are called with a pointer to the 1709 * interface out which this connection should go. 1710 * 1711 * NOTE: Do not subtract IP option/extension header size nor IPsec 1712 * header size from MSS advertisement. MSS option must hold the maximum 1713 * segment size we can accept, so it must always be: 1714 * max(if mtu) - ip header - tcp header 1715 */ 1716 u_long 1717 tcp_mss_to_advertise(ifp, af) 1718 const struct ifnet *ifp; 1719 int af; 1720 { 1721 extern u_long in_maxmtu; 1722 u_long mss = 0; 1723 u_long hdrsiz; 1724 1725 /* 1726 * In order to avoid defeating path MTU discovery on the peer, 1727 * we advertise the max MTU of all attached networks as our MSS, 1728 * per RFC 1191, section 3.1. 1729 * 1730 * We provide the option to advertise just the MTU of 1731 * the interface on which we hope this connection will 1732 * be receiving. If we are responding to a SYN, we 1733 * will have a pretty good idea about this, but when 1734 * initiating a connection there is a bit more doubt. 1735 * 1736 * We also need to ensure that loopback has a large enough 1737 * MSS, as the loopback MTU is never included in in_maxmtu. 1738 */ 1739 1740 if (ifp != NULL) 1741 switch (af) { 1742 case AF_INET: 1743 mss = ifp->if_mtu; 1744 break; 1745 #ifdef INET6 1746 case AF_INET6: 1747 mss = IN6_LINKMTU(ifp); 1748 break; 1749 #endif 1750 } 1751 1752 if (tcp_mss_ifmtu == 0) 1753 switch (af) { 1754 case AF_INET: 1755 mss = max(in_maxmtu, mss); 1756 break; 1757 #ifdef INET6 1758 case AF_INET6: 1759 mss = max(in6_maxmtu, mss); 1760 break; 1761 #endif 1762 } 1763 1764 switch (af) { 1765 case AF_INET: 1766 hdrsiz = sizeof(struct ip); 1767 break; 1768 #ifdef INET6 1769 case AF_INET6: 1770 hdrsiz = sizeof(struct ip6_hdr); 1771 break; 1772 #endif 1773 default: 1774 hdrsiz = 0; 1775 break; 1776 } 1777 hdrsiz += sizeof(struct tcphdr); 1778 if (mss > hdrsiz) 1779 mss -= hdrsiz; 1780 1781 mss = max(tcp_mssdflt, mss); 1782 return (mss); 1783 } 1784 1785 /* 1786 * Set connection variables based on the peer's advertised MSS. 1787 * We are passed the TCPCB for the actual connection. If we 1788 * are the server, we are called by the compressed state engine 1789 * when the 3-way handshake is complete. If we are the client, 1790 * we are called when we receive the SYN,ACK from the server. 1791 * 1792 * NOTE: Our advertised MSS value must be initialized in the TCPCB 1793 * before this routine is called! 1794 */ 1795 void 1796 tcp_mss_from_peer(tp, offer) 1797 struct tcpcb *tp; 1798 int offer; 1799 { 1800 struct socket *so; 1801 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1802 struct rtentry *rt; 1803 #endif 1804 u_long bufsize; 1805 int mss; 1806 1807 #ifdef DIAGNOSTIC 1808 if (tp->t_inpcb && tp->t_in6pcb) 1809 panic("tcp_mss_from_peer: both t_inpcb and t_in6pcb are set"); 1810 #endif 1811 so = NULL; 1812 rt = NULL; 1813 #ifdef INET 1814 if (tp->t_inpcb) { 1815 so = tp->t_inpcb->inp_socket; 1816 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1817 rt = in_pcbrtentry(tp->t_inpcb); 1818 #endif 1819 } 1820 #endif 1821 #ifdef INET6 1822 if (tp->t_in6pcb) { 1823 so = tp->t_in6pcb->in6p_socket; 1824 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1825 rt = in6_pcbrtentry(tp->t_in6pcb); 1826 #endif 1827 } 1828 #endif 1829 1830 /* 1831 * As per RFC1122, use the default MSS value, unless they 1832 * sent us an offer. Do not accept offers less than 32 bytes. 1833 */ 1834 mss = tcp_mssdflt; 1835 if (offer) 1836 mss = offer; 1837 mss = max(mss, 32); /* sanity */ 1838 tp->t_peermss = mss; 1839 mss -= tcp_optlen(tp); 1840 #ifdef INET 1841 if (tp->t_inpcb) 1842 mss -= ip_optlen(tp->t_inpcb); 1843 #endif 1844 #ifdef INET6 1845 if (tp->t_in6pcb) 1846 mss -= ip6_optlen(tp->t_in6pcb); 1847 #endif 1848 1849 /* 1850 * If there's a pipesize, change the socket buffer to that size. 1851 * Make the socket buffer an integral number of MSS units. If 1852 * the MSS is larger than the socket buffer, artificially decrease 1853 * the MSS. 1854 */ 1855 #ifdef RTV_SPIPE 1856 if (rt != NULL && rt->rt_rmx.rmx_sendpipe != 0) 1857 bufsize = rt->rt_rmx.rmx_sendpipe; 1858 else 1859 #endif 1860 bufsize = so->so_snd.sb_hiwat; 1861 if (bufsize < mss) 1862 mss = bufsize; 1863 else { 1864 bufsize = roundup(bufsize, mss); 1865 if (bufsize > sb_max) 1866 bufsize = sb_max; 1867 (void) sbreserve(&so->so_snd, bufsize); 1868 } 1869 tp->t_segsz = mss; 1870 1871 #ifdef RTV_SSTHRESH 1872 if (rt != NULL && rt->rt_rmx.rmx_ssthresh) { 1873 /* 1874 * There's some sort of gateway or interface buffer 1875 * limit on the path. Use this to set the slow 1876 * start threshold, but set the threshold to no less 1877 * than 2 * MSS. 1878 */ 1879 tp->snd_ssthresh = max(2 * mss, rt->rt_rmx.rmx_ssthresh); 1880 } 1881 #endif 1882 } 1883 1884 /* 1885 * Processing necessary when a TCP connection is established. 1886 */ 1887 void 1888 tcp_established(tp) 1889 struct tcpcb *tp; 1890 { 1891 struct socket *so; 1892 #ifdef RTV_RPIPE 1893 struct rtentry *rt; 1894 #endif 1895 u_long bufsize; 1896 1897 #ifdef DIAGNOSTIC 1898 if (tp->t_inpcb && tp->t_in6pcb) 1899 panic("tcp_established: both t_inpcb and t_in6pcb are set"); 1900 #endif 1901 so = NULL; 1902 rt = NULL; 1903 #ifdef INET 1904 if (tp->t_inpcb) { 1905 so = tp->t_inpcb->inp_socket; 1906 #if defined(RTV_RPIPE) 1907 rt = in_pcbrtentry(tp->t_inpcb); 1908 #endif 1909 } 1910 #endif 1911 #ifdef INET6 1912 if (tp->t_in6pcb) { 1913 so = tp->t_in6pcb->in6p_socket; 1914 #if defined(RTV_RPIPE) 1915 rt = in6_pcbrtentry(tp->t_in6pcb); 1916 #endif 1917 } 1918 #endif 1919 1920 tp->t_state = TCPS_ESTABLISHED; 1921 TCP_TIMER_ARM(tp, TCPT_KEEP, tcp_keepidle); 1922 1923 #ifdef RTV_RPIPE 1924 if (rt != NULL && rt->rt_rmx.rmx_recvpipe != 0) 1925 bufsize = rt->rt_rmx.rmx_recvpipe; 1926 else 1927 #endif 1928 bufsize = so->so_rcv.sb_hiwat; 1929 if (bufsize > tp->t_ourmss) { 1930 bufsize = roundup(bufsize, tp->t_ourmss); 1931 if (bufsize > sb_max) 1932 bufsize = sb_max; 1933 (void) sbreserve(&so->so_rcv, bufsize); 1934 } 1935 } 1936 1937 /* 1938 * Check if there's an initial rtt or rttvar. Convert from the 1939 * route-table units to scaled multiples of the slow timeout timer. 1940 * Called only during the 3-way handshake. 1941 */ 1942 void 1943 tcp_rmx_rtt(tp) 1944 struct tcpcb *tp; 1945 { 1946 #ifdef RTV_RTT 1947 struct rtentry *rt = NULL; 1948 int rtt; 1949 1950 #ifdef DIAGNOSTIC 1951 if (tp->t_inpcb && tp->t_in6pcb) 1952 panic("tcp_rmx_rtt: both t_inpcb and t_in6pcb are set"); 1953 #endif 1954 #ifdef INET 1955 if (tp->t_inpcb) 1956 rt = in_pcbrtentry(tp->t_inpcb); 1957 #endif 1958 #ifdef INET6 1959 if (tp->t_in6pcb) 1960 rt = in6_pcbrtentry(tp->t_in6pcb); 1961 #endif 1962 if (rt == NULL) 1963 return; 1964 1965 if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt)) { 1966 /* 1967 * XXX The lock bit for MTU indicates that the value 1968 * is also a minimum value; this is subject to time. 1969 */ 1970 if (rt->rt_rmx.rmx_locks & RTV_RTT) 1971 TCPT_RANGESET(tp->t_rttmin, 1972 rtt / (RTM_RTTUNIT / PR_SLOWHZ), 1973 TCPTV_MIN, TCPTV_REXMTMAX); 1974 tp->t_srtt = rtt / 1975 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1976 if (rt->rt_rmx.rmx_rttvar) { 1977 tp->t_rttvar = rt->rt_rmx.rmx_rttvar / 1978 ((RTM_RTTUNIT / PR_SLOWHZ) >> 1979 (TCP_RTTVAR_SHIFT + 2)); 1980 } else { 1981 /* Default variation is +- 1 rtt */ 1982 tp->t_rttvar = 1983 tp->t_srtt >> (TCP_RTT_SHIFT - TCP_RTTVAR_SHIFT); 1984 } 1985 TCPT_RANGESET(tp->t_rxtcur, 1986 ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2), 1987 tp->t_rttmin, TCPTV_REXMTMAX); 1988 } 1989 #endif 1990 } 1991 1992 tcp_seq tcp_iss_seq = 0; /* tcp initial seq # */ 1993 #if NRND > 0 1994 u_int8_t tcp_iss_secret[16]; /* 128 bits; should be plenty */ 1995 #endif 1996 1997 /* 1998 * Get a new sequence value given a tcp control block 1999 */ 2000 tcp_seq 2001 tcp_new_iss(struct tcpcb *tp, tcp_seq addin) 2002 { 2003 2004 #ifdef INET 2005 if (tp->t_inpcb != NULL) { 2006 return (tcp_new_iss1(&tp->t_inpcb->inp_laddr, 2007 &tp->t_inpcb->inp_faddr, tp->t_inpcb->inp_lport, 2008 tp->t_inpcb->inp_fport, sizeof(tp->t_inpcb->inp_laddr), 2009 addin)); 2010 } 2011 #endif 2012 #ifdef INET6 2013 if (tp->t_in6pcb != NULL) { 2014 return (tcp_new_iss1(&tp->t_in6pcb->in6p_laddr, 2015 &tp->t_in6pcb->in6p_faddr, tp->t_in6pcb->in6p_lport, 2016 tp->t_in6pcb->in6p_fport, sizeof(tp->t_in6pcb->in6p_laddr), 2017 addin)); 2018 } 2019 #endif 2020 /* Not possible. */ 2021 panic("tcp_new_iss"); 2022 } 2023 2024 /* 2025 * This routine actually generates a new TCP initial sequence number. 2026 */ 2027 tcp_seq 2028 tcp_new_iss1(void *laddr, void *faddr, u_int16_t lport, u_int16_t fport, 2029 size_t addrsz, tcp_seq addin) 2030 { 2031 tcp_seq tcp_iss; 2032 2033 #if NRND > 0 2034 static int beenhere; 2035 2036 /* 2037 * If we haven't been here before, initialize our cryptographic 2038 * hash secret. 2039 */ 2040 if (beenhere == 0) { 2041 rnd_extract_data(tcp_iss_secret, sizeof(tcp_iss_secret), 2042 RND_EXTRACT_ANY); 2043 beenhere = 1; 2044 } 2045 2046 if (tcp_do_rfc1948) { 2047 MD5_CTX ctx; 2048 u_int8_t hash[16]; /* XXX MD5 knowledge */ 2049 2050 /* 2051 * Compute the base value of the ISS. It is a hash 2052 * of (saddr, sport, daddr, dport, secret). 2053 */ 2054 MD5Init(&ctx); 2055 2056 MD5Update(&ctx, (u_char *) laddr, addrsz); 2057 MD5Update(&ctx, (u_char *) &lport, sizeof(lport)); 2058 2059 MD5Update(&ctx, (u_char *) faddr, addrsz); 2060 MD5Update(&ctx, (u_char *) &fport, sizeof(fport)); 2061 2062 MD5Update(&ctx, tcp_iss_secret, sizeof(tcp_iss_secret)); 2063 2064 MD5Final(hash, &ctx); 2065 2066 memcpy(&tcp_iss, hash, sizeof(tcp_iss)); 2067 2068 /* 2069 * Now increment our "timer", and add it in to 2070 * the computed value. 2071 * 2072 * XXX Use `addin'? 2073 * XXX TCP_ISSINCR too large to use? 2074 */ 2075 tcp_iss_seq += TCP_ISSINCR; 2076 #ifdef TCPISS_DEBUG 2077 printf("ISS hash 0x%08x, ", tcp_iss); 2078 #endif 2079 tcp_iss += tcp_iss_seq + addin; 2080 #ifdef TCPISS_DEBUG 2081 printf("new ISS 0x%08x\n", tcp_iss); 2082 #endif 2083 } else 2084 #endif /* NRND > 0 */ 2085 { 2086 /* 2087 * Randomize. 2088 */ 2089 #if NRND > 0 2090 rnd_extract_data(&tcp_iss, sizeof(tcp_iss), RND_EXTRACT_ANY); 2091 #else 2092 tcp_iss = arc4random(); 2093 #endif 2094 2095 /* 2096 * If we were asked to add some amount to a known value, 2097 * we will take a random value obtained above, mask off 2098 * the upper bits, and add in the known value. We also 2099 * add in a constant to ensure that we are at least a 2100 * certain distance from the original value. 2101 * 2102 * This is used when an old connection is in timed wait 2103 * and we have a new one coming in, for instance. 2104 */ 2105 if (addin != 0) { 2106 #ifdef TCPISS_DEBUG 2107 printf("Random %08x, ", tcp_iss); 2108 #endif 2109 tcp_iss &= TCP_ISS_RANDOM_MASK; 2110 tcp_iss += addin + TCP_ISSINCR; 2111 #ifdef TCPISS_DEBUG 2112 printf("Old ISS %08x, ISS %08x\n", addin, tcp_iss); 2113 #endif 2114 } else { 2115 tcp_iss &= TCP_ISS_RANDOM_MASK; 2116 tcp_iss += tcp_iss_seq; 2117 tcp_iss_seq += TCP_ISSINCR; 2118 #ifdef TCPISS_DEBUG 2119 printf("ISS %08x\n", tcp_iss); 2120 #endif 2121 } 2122 } 2123 2124 if (tcp_compat_42) { 2125 /* 2126 * Limit it to the positive range for really old TCP 2127 * implementations. 2128 * Just AND off the top bit instead of checking if 2129 * is set first - saves a branch 50% of the time. 2130 */ 2131 tcp_iss &= 0x7fffffff; /* XXX */ 2132 } 2133 2134 return (tcp_iss); 2135 } 2136 2137 #ifdef IPSEC 2138 /* compute ESP/AH header size for TCP, including outer IP header. */ 2139 size_t 2140 ipsec4_hdrsiz_tcp(tp) 2141 struct tcpcb *tp; 2142 { 2143 struct inpcb *inp; 2144 size_t hdrsiz; 2145 2146 /* XXX mapped addr case (tp->t_in6pcb) */ 2147 if (!tp || !tp->t_template || !(inp = tp->t_inpcb)) 2148 return 0; 2149 switch (tp->t_family) { 2150 case AF_INET: 2151 /* XXX: should use currect direction. */ 2152 hdrsiz = ipsec4_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, inp); 2153 break; 2154 default: 2155 hdrsiz = 0; 2156 break; 2157 } 2158 2159 return hdrsiz; 2160 } 2161 2162 #ifdef INET6 2163 size_t 2164 ipsec6_hdrsiz_tcp(tp) 2165 struct tcpcb *tp; 2166 { 2167 struct in6pcb *in6p; 2168 size_t hdrsiz; 2169 2170 if (!tp || !tp->t_template || !(in6p = tp->t_in6pcb)) 2171 return 0; 2172 switch (tp->t_family) { 2173 case AF_INET6: 2174 /* XXX: should use currect direction. */ 2175 hdrsiz = ipsec6_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, in6p); 2176 break; 2177 case AF_INET: 2178 /* mapped address case - tricky */ 2179 default: 2180 hdrsiz = 0; 2181 break; 2182 } 2183 2184 return hdrsiz; 2185 } 2186 #endif 2187 #endif /*IPSEC*/ 2188 2189 /* 2190 * Determine the length of the TCP options for this connection. 2191 * 2192 * XXX: What do we do for SACK, when we add that? Just reserve 2193 * all of the space? Otherwise we can't exactly be incrementing 2194 * cwnd by an amount that varies depending on the amount we last 2195 * had to SACK! 2196 */ 2197 2198 u_int 2199 tcp_optlen(tp) 2200 struct tcpcb *tp; 2201 { 2202 if ((tp->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP|TF_NOOPT)) == 2203 (TF_REQ_TSTMP | TF_RCVD_TSTMP)) 2204 return TCPOLEN_TSTAMP_APPA; 2205 else 2206 return 0; 2207 } 2208
Indexes created Tue Sep 23 07:09:52 GMT 2025