1 /* $NetBSD: tcp_subr.c,v 1.298 2025/02/26 04:49:45 andvar 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, 2008 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 * 49 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 50 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 51 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 52 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 53 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 54 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 55 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 56 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 57 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 58 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 59 * POSSIBILITY OF SUCH DAMAGE. 60 */ 61 62 /* 63 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 64 * The Regents of the University of California. All rights reserved. 65 * 66 * Redistribution and use in source and binary forms, with or without 67 * modification, are permitted provided that the following conditions 68 * are met: 69 * 1. Redistributions of source code must retain the above copyright 70 * notice, this list of conditions and the following disclaimer. 71 * 2. Redistributions in binary form must reproduce the above copyright 72 * notice, this list of conditions and the following disclaimer in the 73 * documentation and/or other materials provided with the distribution. 74 * 3. Neither the name of the University nor the names of its contributors 75 * may be used to endorse or promote products derived from this software 76 * without specific prior written permission. 77 * 78 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 79 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 80 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 81 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 82 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 83 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 84 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 85 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 86 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 87 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 88 * SUCH DAMAGE. 89 * 90 * @(#)tcp_subr.c 8.2 (Berkeley) 5/24/95 91 */ 92 93 #include <sys/cdefs.h> 94 __KERNEL_RCSID(0, "$NetBSD: tcp_subr.c,v 1.298 2025/02/26 04:49:45 andvar Exp $"); 95 96 #ifdef _KERNEL_OPT 97 #include "opt_inet.h" 98 #include "opt_ipsec.h" 99 #include "opt_inet_csum.h" 100 #include "opt_mbuftrace.h" 101 #endif 102 103 #include <sys/param.h> 104 #include <sys/atomic.h> 105 #include <sys/proc.h> 106 #include <sys/systm.h> 107 #include <sys/mbuf.h> 108 #include <sys/once.h> 109 #include <sys/socket.h> 110 #include <sys/socketvar.h> 111 #include <sys/protosw.h> 112 #include <sys/errno.h> 113 #include <sys/kernel.h> 114 #include <sys/pool.h> 115 #include <sys/md5.h> 116 #include <sys/cprng.h> 117 118 #include <net/route.h> 119 #include <net/if.h> 120 121 #include <netinet/in.h> 122 #include <netinet/in_systm.h> 123 #include <netinet/ip.h> 124 #include <netinet/in_pcb.h> 125 #include <netinet/ip_var.h> 126 #include <netinet/ip_icmp.h> 127 128 #ifdef INET6 129 #include <netinet/ip6.h> 130 #include <netinet6/in6_pcb.h> 131 #include <netinet6/ip6_var.h> 132 #include <netinet6/in6_var.h> 133 #include <netinet6/ip6protosw.h> 134 #include <netinet/icmp6.h> 135 #include <netinet6/nd6.h> 136 #endif 137 138 #include <netinet/tcp.h> 139 #include <netinet/tcp_fsm.h> 140 #include <netinet/tcp_seq.h> 141 #include <netinet/tcp_timer.h> 142 #include <netinet/tcp_var.h> 143 #include <netinet/tcp_vtw.h> 144 #include <netinet/tcp_private.h> 145 #include <netinet/tcp_congctl.h> 146 #include <netinet/tcp_syncache.h> 147 148 #ifdef IPSEC 149 #include <netipsec/ipsec.h> 150 #ifdef INET6 151 #include <netipsec/ipsec6.h> 152 #endif 153 #include <netipsec/key.h> 154 #endif 155 156 157 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 158 u_int32_t tcp_now; /* slow ticks, for RFC 1323 timestamps */ 159 160 percpu_t *tcpstat_percpu; 161 162 /* patchable/settable parameters for tcp */ 163 int tcp_mssdflt = TCP_MSS; 164 int tcp_minmss = TCP_MINMSS; 165 int tcp_rttdflt = TCPTV_SRTTDFLT / PR_SLOWHZ; 166 int tcp_do_rfc1323 = 1; /* window scaling / timestamps (obsolete) */ 167 int tcp_do_rfc1948 = 0; /* ISS by cryptographic hash */ 168 int tcp_do_sack = 1; /* selective acknowledgement */ 169 int tcp_do_win_scale = 1; /* RFC1323 window scaling */ 170 int tcp_do_timestamps = 1; /* RFC1323 timestamps */ 171 int tcp_ack_on_push = 0; /* set to enable immediate ACK-on-PUSH */ 172 int tcp_do_ecn = 0; /* Explicit Congestion Notification */ 173 #ifndef TCP_INIT_WIN 174 #define TCP_INIT_WIN 4 /* initial slow start window */ 175 #endif 176 #ifndef TCP_INIT_WIN_LOCAL 177 #define TCP_INIT_WIN_LOCAL 4 /* initial slow start window for local nets */ 178 #endif 179 /* 180 * Up to 5 we scale linearly, to reach 3 * 1460; then (iw) * 1460. 181 * This is to simulate current behavior for iw == 4 182 */ 183 int tcp_init_win_max[] = { 184 1 * 1460, 185 1 * 1460, 186 2 * 1460, 187 2 * 1460, 188 3 * 1460, 189 5 * 1460, 190 6 * 1460, 191 7 * 1460, 192 8 * 1460, 193 9 * 1460, 194 10 * 1460 195 }; 196 int tcp_init_win = TCP_INIT_WIN; 197 int tcp_init_win_local = TCP_INIT_WIN_LOCAL; 198 int tcp_mss_ifmtu = 0; 199 int tcp_rst_ppslim = 100; /* 100pps */ 200 int tcp_ackdrop_ppslim = 100; /* 100pps */ 201 int tcp_do_loopback_cksum = 0; 202 int tcp_do_abc = 1; /* RFC3465 Appropriate byte counting. */ 203 int tcp_abc_aggressive = 1; /* 1: L=2*SMSS 0: L=1*SMSS */ 204 int tcp_sack_tp_maxholes = 32; 205 int tcp_sack_globalmaxholes = 1024; 206 int tcp_sack_globalholes = 0; 207 int tcp_ecn_maxretries = 1; 208 int tcp_msl_enable = 1; /* enable TIME_WAIT truncation */ 209 int tcp_msl_loop = PR_SLOWHZ; /* MSL for loopback */ 210 int tcp_msl_local = 5 * PR_SLOWHZ; /* MSL for 'local' */ 211 int tcp_msl_remote = TCPTV_MSL; /* MSL otherwise */ 212 int tcp_msl_remote_threshold = TCPTV_SRTTDFLT; /* RTT threshold */ 213 int tcp_rttlocal = 0; /* Use RTT to decide who's 'local' */ 214 215 int tcp4_vtw_enable = 0; /* 1 to enable */ 216 int tcp6_vtw_enable = 0; /* 1 to enable */ 217 int tcp_vtw_was_enabled = 0; 218 int tcp_vtw_entries = 1 << 4; /* 16 vestigial TIME_WAIT entries */ 219 220 /* tcb hash */ 221 #ifndef TCBHASHSIZE 222 #define TCBHASHSIZE 128 223 #endif 224 int tcbhashsize = TCBHASHSIZE; 225 226 int tcp_freeq(struct tcpcb *); 227 static int tcp_iss_secret_init(void); 228 229 static void tcp_mtudisc_callback(struct in_addr); 230 231 #ifdef INET6 232 static void tcp6_mtudisc(struct inpcb *, int); 233 #endif 234 235 static struct pool tcpcb_pool; 236 237 static int tcp_drainwanted; 238 239 #ifdef TCP_CSUM_COUNTERS 240 #include <sys/device.h> 241 242 struct evcnt tcp_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 243 NULL, "tcp", "hwcsum bad"); 244 struct evcnt tcp_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 245 NULL, "tcp", "hwcsum ok"); 246 struct evcnt tcp_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 247 NULL, "tcp", "hwcsum data"); 248 struct evcnt tcp_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 249 NULL, "tcp", "swcsum"); 250 251 EVCNT_ATTACH_STATIC(tcp_hwcsum_bad); 252 EVCNT_ATTACH_STATIC(tcp_hwcsum_ok); 253 EVCNT_ATTACH_STATIC(tcp_hwcsum_data); 254 EVCNT_ATTACH_STATIC(tcp_swcsum); 255 256 #if defined(INET6) 257 struct evcnt tcp6_hwcsum_bad = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 258 NULL, "tcp6", "hwcsum bad"); 259 struct evcnt tcp6_hwcsum_ok = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 260 NULL, "tcp6", "hwcsum ok"); 261 struct evcnt tcp6_hwcsum_data = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 262 NULL, "tcp6", "hwcsum data"); 263 struct evcnt tcp6_swcsum = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 264 NULL, "tcp6", "swcsum"); 265 266 EVCNT_ATTACH_STATIC(tcp6_hwcsum_bad); 267 EVCNT_ATTACH_STATIC(tcp6_hwcsum_ok); 268 EVCNT_ATTACH_STATIC(tcp6_hwcsum_data); 269 EVCNT_ATTACH_STATIC(tcp6_swcsum); 270 #endif /* defined(INET6) */ 271 #endif /* TCP_CSUM_COUNTERS */ 272 273 274 #ifdef TCP_OUTPUT_COUNTERS 275 #include <sys/device.h> 276 277 struct evcnt tcp_output_bigheader = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 278 NULL, "tcp", "output big header"); 279 struct evcnt tcp_output_predict_hit = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 280 NULL, "tcp", "output predict hit"); 281 struct evcnt tcp_output_predict_miss = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 282 NULL, "tcp", "output predict miss"); 283 struct evcnt tcp_output_copysmall = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 284 NULL, "tcp", "output copy small"); 285 struct evcnt tcp_output_copybig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 286 NULL, "tcp", "output copy big"); 287 struct evcnt tcp_output_refbig = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 288 NULL, "tcp", "output reference big"); 289 290 EVCNT_ATTACH_STATIC(tcp_output_bigheader); 291 EVCNT_ATTACH_STATIC(tcp_output_predict_hit); 292 EVCNT_ATTACH_STATIC(tcp_output_predict_miss); 293 EVCNT_ATTACH_STATIC(tcp_output_copysmall); 294 EVCNT_ATTACH_STATIC(tcp_output_copybig); 295 EVCNT_ATTACH_STATIC(tcp_output_refbig); 296 297 #endif /* TCP_OUTPUT_COUNTERS */ 298 299 #ifdef TCP_REASS_COUNTERS 300 #include <sys/device.h> 301 302 struct evcnt tcp_reass_ = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 303 NULL, "tcp_reass", "calls"); 304 struct evcnt tcp_reass_empty = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 305 &tcp_reass_, "tcp_reass", "insert into empty queue"); 306 struct evcnt tcp_reass_iteration[8] = { 307 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", ">7 iterations"), 308 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "1 iteration"), 309 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "2 iterations"), 310 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "3 iterations"), 311 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "4 iterations"), 312 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "5 iterations"), 313 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "6 iterations"), 314 EVCNT_INITIALIZER(EVCNT_TYPE_MISC, &tcp_reass_, "tcp_reass", "7 iterations"), 315 }; 316 struct evcnt tcp_reass_prependfirst = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 317 &tcp_reass_, "tcp_reass", "prepend to first"); 318 struct evcnt tcp_reass_prepend = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 319 &tcp_reass_, "tcp_reass", "prepend"); 320 struct evcnt tcp_reass_insert = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 321 &tcp_reass_, "tcp_reass", "insert"); 322 struct evcnt tcp_reass_inserttail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 323 &tcp_reass_, "tcp_reass", "insert at tail"); 324 struct evcnt tcp_reass_append = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 325 &tcp_reass_, "tcp_reass", "append"); 326 struct evcnt tcp_reass_appendtail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 327 &tcp_reass_, "tcp_reass", "append to tail fragment"); 328 struct evcnt tcp_reass_overlaptail = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 329 &tcp_reass_, "tcp_reass", "overlap at end"); 330 struct evcnt tcp_reass_overlapfront = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 331 &tcp_reass_, "tcp_reass", "overlap at start"); 332 struct evcnt tcp_reass_segdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 333 &tcp_reass_, "tcp_reass", "duplicate segment"); 334 struct evcnt tcp_reass_fragdup = EVCNT_INITIALIZER(EVCNT_TYPE_MISC, 335 &tcp_reass_, "tcp_reass", "duplicate fragment"); 336 337 EVCNT_ATTACH_STATIC(tcp_reass_); 338 EVCNT_ATTACH_STATIC(tcp_reass_empty); 339 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 0); 340 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 1); 341 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 2); 342 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 3); 343 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 4); 344 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 5); 345 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 6); 346 EVCNT_ATTACH_STATIC2(tcp_reass_iteration, 7); 347 EVCNT_ATTACH_STATIC(tcp_reass_prependfirst); 348 EVCNT_ATTACH_STATIC(tcp_reass_prepend); 349 EVCNT_ATTACH_STATIC(tcp_reass_insert); 350 EVCNT_ATTACH_STATIC(tcp_reass_inserttail); 351 EVCNT_ATTACH_STATIC(tcp_reass_append); 352 EVCNT_ATTACH_STATIC(tcp_reass_appendtail); 353 EVCNT_ATTACH_STATIC(tcp_reass_overlaptail); 354 EVCNT_ATTACH_STATIC(tcp_reass_overlapfront); 355 EVCNT_ATTACH_STATIC(tcp_reass_segdup); 356 EVCNT_ATTACH_STATIC(tcp_reass_fragdup); 357 358 #endif /* TCP_REASS_COUNTERS */ 359 360 #ifdef MBUFTRACE 361 struct mowner tcp_mowner = MOWNER_INIT("tcp", ""); 362 struct mowner tcp_rx_mowner = MOWNER_INIT("tcp", "rx"); 363 struct mowner tcp_tx_mowner = MOWNER_INIT("tcp", "tx"); 364 struct mowner tcp_sock_mowner = MOWNER_INIT("tcp", "sock"); 365 struct mowner tcp_sock_rx_mowner = MOWNER_INIT("tcp", "sock rx"); 366 struct mowner tcp_sock_tx_mowner = MOWNER_INIT("tcp", "sock tx"); 367 #endif 368 369 static int 370 do_tcpinit(void) 371 { 372 373 inpcb_init(&tcbtable, tcbhashsize, tcbhashsize); 374 pool_init(&tcpcb_pool, sizeof(struct tcpcb), 0, 0, 0, "tcpcbpl", 375 NULL, IPL_SOFTNET); 376 377 tcp_usrreq_init(); 378 379 /* Initialize timer state. */ 380 tcp_timer_init(); 381 382 /* Initialize the compressed state engine. */ 383 syn_cache_init(); 384 385 /* Initialize the congestion control algorithms. */ 386 tcp_congctl_init(); 387 388 /* Initialize the TCPCB template. */ 389 tcp_tcpcb_template(); 390 391 /* Initialize reassembly queue */ 392 tcpipqent_init(); 393 394 /* SACK */ 395 tcp_sack_init(); 396 397 MOWNER_ATTACH(&tcp_tx_mowner); 398 MOWNER_ATTACH(&tcp_rx_mowner); 399 MOWNER_ATTACH(&tcp_reass_mowner); 400 MOWNER_ATTACH(&tcp_sock_mowner); 401 MOWNER_ATTACH(&tcp_sock_tx_mowner); 402 MOWNER_ATTACH(&tcp_sock_rx_mowner); 403 MOWNER_ATTACH(&tcp_mowner); 404 405 tcpstat_percpu = percpu_alloc(sizeof(uint64_t) * TCP_NSTATS); 406 407 vtw_earlyinit(); 408 409 tcp_slowtimo_init(); 410 411 return 0; 412 } 413 414 void 415 tcp_init_common(unsigned basehlen) 416 { 417 static ONCE_DECL(dotcpinit); 418 unsigned hlen = basehlen + sizeof(struct tcphdr); 419 unsigned oldhlen; 420 421 if (max_linkhdr + hlen > MHLEN) 422 panic("tcp_init"); 423 while ((oldhlen = max_protohdr) < hlen) 424 atomic_cas_uint(&max_protohdr, oldhlen, hlen); 425 426 RUN_ONCE(&dotcpinit, do_tcpinit); 427 } 428 429 /* 430 * Tcp initialization 431 */ 432 void 433 tcp_init(void) 434 { 435 436 icmp_mtudisc_callback_register(tcp_mtudisc_callback); 437 438 tcp_init_common(sizeof(struct ip)); 439 } 440 441 /* 442 * Create template to be used to send tcp packets on a connection. 443 * Call after host entry created, allocates an mbuf and fills 444 * in a skeletal tcp/ip header, minimizing the amount of work 445 * necessary when the connection is used. 446 */ 447 struct mbuf * 448 tcp_template(struct tcpcb *tp) 449 { 450 struct inpcb *inp = tp->t_inpcb; 451 struct tcphdr *n; 452 struct mbuf *m; 453 int hlen; 454 455 switch (tp->t_family) { 456 case AF_INET: 457 hlen = sizeof(struct ip); 458 if (inp->inp_af == AF_INET) 459 break; 460 #ifdef INET6 461 if (inp->inp_af == AF_INET6) { 462 /* mapped addr case */ 463 if (IN6_IS_ADDR_V4MAPPED(&in6p_laddr(inp)) 464 && IN6_IS_ADDR_V4MAPPED(&in6p_faddr(inp))) 465 break; 466 } 467 #endif 468 return NULL; /*EINVAL*/ 469 #ifdef INET6 470 case AF_INET6: 471 hlen = sizeof(struct ip6_hdr); 472 if (inp != NULL) { 473 /* more sainty check? */ 474 break; 475 } 476 return NULL; /*EINVAL*/ 477 #endif 478 default: 479 return NULL; /*EAFNOSUPPORT*/ 480 } 481 482 KASSERT(hlen + sizeof(struct tcphdr) <= MCLBYTES); 483 484 m = tp->t_template; 485 if (m && m->m_len == hlen + sizeof(struct tcphdr)) { 486 ; 487 } else { 488 m_freem(m); 489 m = tp->t_template = NULL; 490 MGETHDR(m, M_DONTWAIT, MT_HEADER); 491 if (m && hlen + sizeof(struct tcphdr) > MHLEN) { 492 MCLGET(m, M_DONTWAIT); 493 if ((m->m_flags & M_EXT) == 0) { 494 m_free(m); 495 m = NULL; 496 } 497 } 498 if (m == NULL) 499 return NULL; 500 MCLAIM(m, &tcp_mowner); 501 m->m_pkthdr.len = m->m_len = hlen + sizeof(struct tcphdr); 502 } 503 504 memset(mtod(m, void *), 0, m->m_len); 505 506 n = (struct tcphdr *)(mtod(m, char *) + hlen); 507 508 switch (tp->t_family) { 509 case AF_INET: 510 { 511 struct ipovly *ipov; 512 mtod(m, struct ip *)->ip_v = 4; 513 mtod(m, struct ip *)->ip_hl = hlen >> 2; 514 ipov = mtod(m, struct ipovly *); 515 ipov->ih_pr = IPPROTO_TCP; 516 ipov->ih_len = htons(sizeof(struct tcphdr)); 517 if (inp->inp_af == AF_INET) { 518 ipov->ih_src = in4p_laddr(inp); 519 ipov->ih_dst = in4p_faddr(inp); 520 } 521 #ifdef INET6 522 else if (inp->inp_af == AF_INET6) { 523 /* mapped addr case */ 524 bcopy(&in6p_laddr(inp).s6_addr32[3], &ipov->ih_src, 525 sizeof(ipov->ih_src)); 526 bcopy(&in6p_faddr(inp).s6_addr32[3], &ipov->ih_dst, 527 sizeof(ipov->ih_dst)); 528 } 529 #endif 530 531 /* 532 * Compute the pseudo-header portion of the checksum 533 * now. We incrementally add in the TCP option and 534 * payload lengths later, and then compute the TCP 535 * checksum right before the packet is sent off onto 536 * the wire. 537 */ 538 n->th_sum = in_cksum_phdr(ipov->ih_src.s_addr, 539 ipov->ih_dst.s_addr, 540 htons(sizeof(struct tcphdr) + IPPROTO_TCP)); 541 break; 542 } 543 #ifdef INET6 544 case AF_INET6: 545 { 546 struct ip6_hdr *ip6; 547 mtod(m, struct ip *)->ip_v = 6; 548 ip6 = mtod(m, struct ip6_hdr *); 549 ip6->ip6_nxt = IPPROTO_TCP; 550 ip6->ip6_plen = htons(sizeof(struct tcphdr)); 551 ip6->ip6_src = in6p_laddr(inp); 552 ip6->ip6_dst = in6p_faddr(inp); 553 ip6->ip6_flow = in6p_flowinfo(inp) & IPV6_FLOWINFO_MASK; 554 if (ip6_auto_flowlabel) { 555 ip6->ip6_flow &= ~IPV6_FLOWLABEL_MASK; 556 ip6->ip6_flow |= 557 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 558 } 559 ip6->ip6_vfc &= ~IPV6_VERSION_MASK; 560 ip6->ip6_vfc |= IPV6_VERSION; 561 562 /* 563 * Compute the pseudo-header portion of the checksum 564 * now. We incrementally add in the TCP option and 565 * payload lengths later, and then compute the TCP 566 * checksum right before the packet is sent off onto 567 * the wire. 568 */ 569 n->th_sum = in6_cksum_phdr(&in6p_laddr(inp), 570 &in6p_faddr(inp), htonl(sizeof(struct tcphdr)), 571 htonl(IPPROTO_TCP)); 572 break; 573 } 574 #endif 575 } 576 577 n->th_sport = inp->inp_lport; 578 n->th_dport = inp->inp_fport; 579 580 n->th_seq = 0; 581 n->th_ack = 0; 582 n->th_x2 = 0; 583 n->th_off = 5; 584 n->th_flags = 0; 585 n->th_win = 0; 586 n->th_urp = 0; 587 return m; 588 } 589 590 /* 591 * Send a single message to the TCP at address specified by 592 * the given TCP/IP header. If m == 0, then we make a copy 593 * of the tcpiphdr at ti and send directly to the addressed host. 594 * This is used to force keep alive messages out using the TCP 595 * template for a connection tp->t_template. If flags are given 596 * then we send a message back to the TCP which originated the 597 * segment ti, and discard the mbuf containing it and any other 598 * attached mbufs. 599 * 600 * In any case the ack and sequence number of the transmitted 601 * segment are as specified by the parameters. 602 */ 603 int 604 tcp_respond(struct tcpcb *tp, struct mbuf *mtemplate, struct mbuf *m, 605 struct tcphdr *th0, tcp_seq ack, tcp_seq seq, int flags) 606 { 607 struct route *ro; 608 int error, tlen, win = 0; 609 int hlen; 610 struct ip *ip; 611 #ifdef INET6 612 struct ip6_hdr *ip6; 613 #endif 614 int family; /* family on packet, not inpcb! */ 615 struct tcphdr *th; 616 617 if (tp != NULL && (flags & TH_RST) == 0) { 618 KASSERT(tp->t_inpcb != NULL); 619 620 win = sbspace(&tp->t_inpcb->inp_socket->so_rcv); 621 } 622 623 th = NULL; /* Quell uninitialized warning */ 624 ip = NULL; 625 #ifdef INET6 626 ip6 = NULL; 627 #endif 628 if (m == NULL) { 629 if (!mtemplate) 630 return EINVAL; 631 632 /* get family information from template */ 633 switch (mtod(mtemplate, struct ip *)->ip_v) { 634 case 4: 635 family = AF_INET; 636 hlen = sizeof(struct ip); 637 break; 638 #ifdef INET6 639 case 6: 640 family = AF_INET6; 641 hlen = sizeof(struct ip6_hdr); 642 break; 643 #endif 644 default: 645 return EAFNOSUPPORT; 646 } 647 648 MGETHDR(m, M_DONTWAIT, MT_HEADER); 649 if (m) { 650 MCLAIM(m, &tcp_tx_mowner); 651 MCLGET(m, M_DONTWAIT); 652 if ((m->m_flags & M_EXT) == 0) { 653 m_free(m); 654 m = NULL; 655 } 656 } 657 if (m == NULL) 658 return ENOBUFS; 659 660 tlen = 0; 661 662 m->m_data += max_linkhdr; 663 bcopy(mtod(mtemplate, void *), mtod(m, void *), 664 mtemplate->m_len); 665 switch (family) { 666 case AF_INET: 667 ip = mtod(m, struct ip *); 668 th = (struct tcphdr *)(ip + 1); 669 break; 670 #ifdef INET6 671 case AF_INET6: 672 ip6 = mtod(m, struct ip6_hdr *); 673 th = (struct tcphdr *)(ip6 + 1); 674 break; 675 #endif 676 } 677 flags = TH_ACK; 678 } else { 679 if ((m->m_flags & M_PKTHDR) == 0) { 680 m_freem(m); 681 return EINVAL; 682 } 683 KASSERT(th0 != NULL); 684 685 /* get family information from m */ 686 switch (mtod(m, struct ip *)->ip_v) { 687 case 4: 688 family = AF_INET; 689 hlen = sizeof(struct ip); 690 ip = mtod(m, struct ip *); 691 break; 692 #ifdef INET6 693 case 6: 694 family = AF_INET6; 695 hlen = sizeof(struct ip6_hdr); 696 ip6 = mtod(m, struct ip6_hdr *); 697 break; 698 #endif 699 default: 700 m_freem(m); 701 return EAFNOSUPPORT; 702 } 703 /* clear h/w csum flags inherited from rx packet */ 704 m->m_pkthdr.csum_flags = 0; 705 706 if ((flags & TH_SYN) == 0 || sizeof(*th0) > (th0->th_off << 2)) 707 tlen = sizeof(*th0); 708 else 709 tlen = th0->th_off << 2; 710 711 if (m->m_len > hlen + tlen && (m->m_flags & M_EXT) == 0 && 712 mtod(m, char *) + hlen == (char *)th0) { 713 m->m_len = hlen + tlen; 714 m_freem(m->m_next); 715 m->m_next = NULL; 716 } else { 717 struct mbuf *n; 718 719 KASSERT(max_linkhdr + hlen + tlen <= MCLBYTES); 720 721 MGETHDR(n, M_DONTWAIT, MT_HEADER); 722 if (n && max_linkhdr + hlen + tlen > MHLEN) { 723 MCLGET(n, M_DONTWAIT); 724 if ((n->m_flags & M_EXT) == 0) { 725 m_freem(n); 726 n = NULL; 727 } 728 } 729 if (!n) { 730 m_freem(m); 731 return ENOBUFS; 732 } 733 734 MCLAIM(n, &tcp_tx_mowner); 735 n->m_data += max_linkhdr; 736 n->m_len = hlen + tlen; 737 m_copyback(n, 0, hlen, mtod(m, void *)); 738 m_copyback(n, hlen, tlen, (void *)th0); 739 740 m_freem(m); 741 m = n; 742 n = NULL; 743 } 744 745 #define xchg(a,b,type) { type t; t=a; a=b; b=t; } 746 switch (family) { 747 case AF_INET: 748 ip = mtod(m, struct ip *); 749 th = (struct tcphdr *)(ip + 1); 750 ip->ip_p = IPPROTO_TCP; 751 xchg(ip->ip_dst, ip->ip_src, struct in_addr); 752 ip->ip_p = IPPROTO_TCP; 753 break; 754 #ifdef INET6 755 case AF_INET6: 756 ip6 = mtod(m, struct ip6_hdr *); 757 th = (struct tcphdr *)(ip6 + 1); 758 ip6->ip6_nxt = IPPROTO_TCP; 759 xchg(ip6->ip6_dst, ip6->ip6_src, struct in6_addr); 760 ip6->ip6_nxt = IPPROTO_TCP; 761 break; 762 #endif 763 } 764 xchg(th->th_dport, th->th_sport, u_int16_t); 765 #undef xchg 766 tlen = 0; /*be friendly with the following code*/ 767 } 768 th->th_seq = htonl(seq); 769 th->th_ack = htonl(ack); 770 th->th_x2 = 0; 771 if ((flags & TH_SYN) == 0) { 772 if (tp) 773 win >>= tp->rcv_scale; 774 if (win > TCP_MAXWIN) 775 win = TCP_MAXWIN; 776 th->th_win = htons((u_int16_t)win); 777 th->th_off = sizeof (struct tcphdr) >> 2; 778 tlen += sizeof(*th); 779 } else { 780 tlen += th->th_off << 2; 781 } 782 m->m_len = hlen + tlen; 783 m->m_pkthdr.len = hlen + tlen; 784 m_reset_rcvif(m); 785 th->th_flags = flags; 786 th->th_urp = 0; 787 788 switch (family) { 789 case AF_INET: 790 { 791 struct ipovly *ipov = (struct ipovly *)ip; 792 memset(ipov->ih_x1, 0, sizeof ipov->ih_x1); 793 ipov->ih_len = htons((u_int16_t)tlen); 794 795 th->th_sum = 0; 796 th->th_sum = in_cksum(m, hlen + tlen); 797 ip->ip_len = htons(hlen + tlen); 798 ip->ip_ttl = ip_defttl; 799 break; 800 } 801 #ifdef INET6 802 case AF_INET6: 803 { 804 th->th_sum = 0; 805 th->th_sum = in6_cksum(m, IPPROTO_TCP, sizeof(struct ip6_hdr), 806 tlen); 807 ip6->ip6_plen = htons(tlen); 808 if (tp && tp->t_inpcb->inp_af == AF_INET6) 809 ip6->ip6_hlim = in6pcb_selecthlim_rt(tp->t_inpcb); 810 else 811 ip6->ip6_hlim = ip6_defhlim; 812 ip6->ip6_flow &= ~IPV6_FLOWINFO_MASK; 813 if (ip6_auto_flowlabel) { 814 ip6->ip6_flow |= 815 (htonl(ip6_randomflowlabel()) & IPV6_FLOWLABEL_MASK); 816 } 817 break; 818 } 819 #endif 820 } 821 822 if (tp != NULL && tp->t_inpcb->inp_af == AF_INET) { 823 ro = &tp->t_inpcb->inp_route; 824 KASSERT(family == AF_INET); 825 KASSERT(in_hosteq(ip->ip_dst, in4p_faddr(tp->t_inpcb))); 826 } 827 #ifdef INET6 828 else if (tp != NULL && tp->t_inpcb->inp_af == AF_INET6) { 829 ro = (struct route *)&tp->t_inpcb->inp_route; 830 831 #ifdef DIAGNOSTIC 832 if (family == AF_INET) { 833 if (!IN6_IS_ADDR_V4MAPPED(&in6p_faddr(tp->t_inpcb))) 834 panic("tcp_respond: not mapped addr"); 835 if (memcmp(&ip->ip_dst, 836 &in6p_faddr(tp->t_inpcb).s6_addr32[3], 837 sizeof(ip->ip_dst)) != 0) { 838 panic("tcp_respond: ip_dst != in6p_faddr"); 839 } 840 } else if (family == AF_INET6) { 841 if (!IN6_ARE_ADDR_EQUAL(&ip6->ip6_dst, 842 &in6p_faddr(tp->t_inpcb))) 843 panic("tcp_respond: ip6_dst != in6p_faddr"); 844 } else 845 panic("tcp_respond: address family mismatch"); 846 #endif 847 } 848 #endif 849 else 850 ro = NULL; 851 852 switch (family) { 853 case AF_INET: 854 error = ip_output(m, NULL, ro, 855 (tp && tp->t_mtudisc ? IP_MTUDISC : 0), NULL, 856 tp ? tp->t_inpcb : NULL); 857 break; 858 #ifdef INET6 859 case AF_INET6: 860 error = ip6_output(m, NULL, ro, 0, NULL, 861 tp ? tp->t_inpcb : NULL, NULL); 862 break; 863 #endif 864 default: 865 error = EAFNOSUPPORT; 866 break; 867 } 868 869 return error; 870 } 871 872 /* 873 * Template TCPCB. Rather than zeroing a new TCPCB and initializing 874 * a bunch of members individually, we maintain this template for the 875 * static and mostly-static components of the TCPCB, and copy it into 876 * the new TCPCB instead. 877 */ 878 static struct tcpcb tcpcb_template = { 879 .t_srtt = TCPTV_SRTTBASE, 880 .t_rttmin = TCPTV_MIN, 881 882 .snd_cwnd = TCP_MAXWIN << TCP_MAX_WINSHIFT, 883 .snd_ssthresh = TCP_MAXWIN << TCP_MAX_WINSHIFT, 884 .snd_numholes = 0, 885 .snd_cubic_wmax = 0, 886 .snd_cubic_wmax_last = 0, 887 .snd_cubic_ctime = 0, 888 889 .t_partialacks = -1, 890 .t_bytes_acked = 0, 891 .t_sndrexmitpack = 0, 892 .t_rcvoopack = 0, 893 .t_sndzerowin = 0, 894 }; 895 896 /* 897 * Updates the TCPCB template whenever a parameter that would affect 898 * the template is changed. 899 */ 900 void 901 tcp_tcpcb_template(void) 902 { 903 struct tcpcb *tp = &tcpcb_template; 904 int flags; 905 906 tp->t_peermss = tcp_mssdflt; 907 tp->t_ourmss = tcp_mssdflt; 908 tp->t_segsz = tcp_mssdflt; 909 910 flags = 0; 911 if (tcp_do_rfc1323 && tcp_do_win_scale) 912 flags |= TF_REQ_SCALE; 913 if (tcp_do_rfc1323 && tcp_do_timestamps) 914 flags |= TF_REQ_TSTMP; 915 tp->t_flags = flags; 916 917 /* 918 * Init srtt to TCPTV_SRTTBASE (0), so we can tell that we have no 919 * rtt estimate. Set rttvar so that srtt + 2 * rttvar gives 920 * reasonable initial retransmit time. 921 */ 922 tp->t_rttvar = tcp_rttdflt * PR_SLOWHZ << (TCP_RTTVAR_SHIFT + 2 - 1); 923 TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp), 924 TCPTV_MIN, TCPTV_REXMTMAX); 925 926 /* Keep Alive */ 927 tp->t_keepinit = MIN(tcp_keepinit, TCP_TIMER_MAXTICKS); 928 tp->t_keepidle = MIN(tcp_keepidle, TCP_TIMER_MAXTICKS); 929 tp->t_keepintvl = MIN(tcp_keepintvl, TCP_TIMER_MAXTICKS); 930 tp->t_keepcnt = MAX(1, MIN(tcp_keepcnt, TCP_TIMER_MAXTICKS)); 931 tp->t_maxidle = tp->t_keepcnt * MIN(tp->t_keepintvl, 932 TCP_TIMER_MAXTICKS/tp->t_keepcnt); 933 934 /* MSL */ 935 tp->t_msl = TCPTV_MSL; 936 } 937 938 /* 939 * Create a new TCP control block, making an 940 * empty reassembly queue and hooking it to the argument 941 * protocol control block. 942 */ 943 struct tcpcb * 944 tcp_newtcpcb(int family, struct inpcb *inp) 945 { 946 struct tcpcb *tp; 947 int i; 948 949 /* XXX Consider using a pool_cache for speed. */ 950 tp = pool_get(&tcpcb_pool, PR_NOWAIT); /* splsoftnet via tcp_usrreq */ 951 if (tp == NULL) 952 return NULL; 953 memcpy(tp, &tcpcb_template, sizeof(*tp)); 954 TAILQ_INIT(&tp->segq); 955 TAILQ_INIT(&tp->timeq); 956 tp->t_family = family; /* may be overridden later on */ 957 TAILQ_INIT(&tp->snd_holes); 958 LIST_INIT(&tp->t_sc); /* XXX can template this */ 959 960 /* Don't sweat this loop; hopefully the compiler will unroll it. */ 961 for (i = 0; i < TCPT_NTIMERS; i++) { 962 callout_init(&tp->t_timer[i], CALLOUT_MPSAFE); 963 TCP_TIMER_INIT(tp, i); 964 } 965 callout_init(&tp->t_delack_ch, CALLOUT_MPSAFE); 966 967 switch (family) { 968 case AF_INET: 969 in4p_ip(inp).ip_ttl = ip_defttl; 970 inp->inp_ppcb = (void *)tp; 971 972 tp->t_inpcb = inp; 973 tp->t_mtudisc = ip_mtudisc; 974 break; 975 #ifdef INET6 976 case AF_INET6: 977 in6p_ip6(inp).ip6_hlim = in6pcb_selecthlim_rt(inp); 978 inp->inp_ppcb = (void *)tp; 979 980 tp->t_inpcb = inp; 981 /* for IPv6, always try to run path MTU discovery */ 982 tp->t_mtudisc = 1; 983 break; 984 #endif /* INET6 */ 985 default: 986 for (i = 0; i < TCPT_NTIMERS; i++) 987 callout_destroy(&tp->t_timer[i]); 988 callout_destroy(&tp->t_delack_ch); 989 pool_put(&tcpcb_pool, tp); /* splsoftnet via tcp_usrreq */ 990 return NULL; 991 } 992 993 /* 994 * Initialize our timebase. When we send timestamps, we take 995 * the delta from tcp_now -- this means each connection always 996 * gets a timebase of 1, which makes it, among other things, 997 * more difficult to determine how long a system has been up, 998 * and thus how many TCP sequence increments have occurred. 999 * 1000 * We start with 1, because 0 doesn't work with linux, which 1001 * considers timestamp 0 in a SYN packet as a bug and disables 1002 * timestamps. 1003 */ 1004 tp->ts_timebase = tcp_now - 1; 1005 1006 tcp_congctl_select(tp, tcp_congctl_global_name); 1007 1008 return tp; 1009 } 1010 1011 /* 1012 * Drop a TCP connection, reporting 1013 * the specified error. If connection is synchronized, 1014 * then send a RST to peer. 1015 */ 1016 struct tcpcb * 1017 tcp_drop(struct tcpcb *tp, int errno) 1018 { 1019 struct socket *so; 1020 1021 KASSERT(tp->t_inpcb != NULL); 1022 1023 so = tp->t_inpcb->inp_socket; 1024 if (so == NULL) 1025 return NULL; 1026 1027 if (TCPS_HAVERCVDSYN(tp->t_state)) { 1028 tp->t_state = TCPS_CLOSED; 1029 (void) tcp_output(tp); 1030 TCP_STATINC(TCP_STAT_DROPS); 1031 } else 1032 TCP_STATINC(TCP_STAT_CONNDROPS); 1033 if (errno == ETIMEDOUT && tp->t_softerror) 1034 errno = tp->t_softerror; 1035 so->so_error = errno; 1036 return (tcp_close(tp)); 1037 } 1038 1039 /* 1040 * Close a TCP control block: 1041 * discard all space held by the tcp 1042 * discard internet protocol block 1043 * wake up any sleepers 1044 */ 1045 struct tcpcb * 1046 tcp_close(struct tcpcb *tp) 1047 { 1048 struct inpcb *inp; 1049 struct socket *so; 1050 #ifdef RTV_RTT 1051 struct rtentry *rt = NULL; 1052 #endif 1053 struct route *ro; 1054 int j; 1055 1056 inp = tp->t_inpcb; 1057 so = inp->inp_socket; 1058 ro = &inp->inp_route; 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 = rtcache_validate(ro)) != NULL && 1075 !in_nullhost(satocsin(rt_getkey(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 threshold 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 rtcache_unref(rt, ro); 1128 #endif /* RTV_RTT */ 1129 /* free the reassembly queue, if any */ 1130 TCP_REASS_LOCK(tp); 1131 (void) tcp_freeq(tp); 1132 TCP_REASS_UNLOCK(tp); 1133 1134 /* free the SACK holes list. */ 1135 tcp_free_sackholes(tp); 1136 tcp_congctl_release(tp); 1137 syn_cache_cleanup(tp); 1138 1139 if (tp->t_template) { 1140 m_free(tp->t_template); 1141 tp->t_template = NULL; 1142 } 1143 1144 /* 1145 * Detaching the pcb will unlock the socket/tcpcb, and stopping 1146 * the timers can also drop the lock. We need to prevent access 1147 * to the tcpcb as it's half torn down. Flag the pcb as dead 1148 * (prevents access by timers) and only then detach it. 1149 */ 1150 tp->t_flags |= TF_DEAD; 1151 inp->inp_ppcb = NULL; 1152 soisdisconnected(so); 1153 inpcb_destroy(inp); 1154 /* 1155 * pcb is no longer visble elsewhere, so we can safely release 1156 * the lock in callout_halt() if needed. 1157 */ 1158 TCP_STATINC(TCP_STAT_CLOSED); 1159 for (j = 0; j < TCPT_NTIMERS; j++) { 1160 callout_halt(&tp->t_timer[j], softnet_lock); 1161 callout_destroy(&tp->t_timer[j]); 1162 } 1163 callout_halt(&tp->t_delack_ch, softnet_lock); 1164 callout_destroy(&tp->t_delack_ch); 1165 pool_put(&tcpcb_pool, tp); 1166 1167 return NULL; 1168 } 1169 1170 int 1171 tcp_freeq(struct tcpcb *tp) 1172 { 1173 struct ipqent *qe; 1174 int rv = 0; 1175 1176 TCP_REASS_LOCK_CHECK(tp); 1177 1178 while ((qe = TAILQ_FIRST(&tp->segq)) != NULL) { 1179 TAILQ_REMOVE(&tp->segq, qe, ipqe_q); 1180 TAILQ_REMOVE(&tp->timeq, qe, ipqe_timeq); 1181 m_freem(qe->ipqe_m); 1182 tcpipqent_free(qe); 1183 rv = 1; 1184 } 1185 tp->t_segqlen = 0; 1186 KASSERT(TAILQ_EMPTY(&tp->timeq)); 1187 return (rv); 1188 } 1189 1190 void 1191 tcp_fasttimo(void) 1192 { 1193 if (tcp_drainwanted) { 1194 tcp_drain(); 1195 tcp_drainwanted = 0; 1196 } 1197 } 1198 1199 void 1200 tcp_drainstub(void) 1201 { 1202 tcp_drainwanted = 1; 1203 } 1204 1205 /* 1206 * Protocol drain routine. Called when memory is in short supply. 1207 * Called from pr_fasttimo thus a callout context. 1208 */ 1209 void 1210 tcp_drain(void) 1211 { 1212 struct inpcb *inp; 1213 struct tcpcb *tp; 1214 1215 mutex_enter(softnet_lock); 1216 KERNEL_LOCK(1, NULL); 1217 1218 /* 1219 * Free the sequence queue of all TCP connections. 1220 */ 1221 TAILQ_FOREACH(inp, &tcbtable.inpt_queue, inp_queue) { 1222 tp = intotcpcb(inp); 1223 if (tp != NULL) { 1224 /* 1225 * If the tcpcb is already busy, 1226 * just bail out now. 1227 */ 1228 if (tcp_reass_lock_try(tp) == 0) 1229 continue; 1230 if (tcp_freeq(tp)) 1231 TCP_STATINC(TCP_STAT_CONNSDRAINED); 1232 TCP_REASS_UNLOCK(tp); 1233 } 1234 } 1235 1236 KERNEL_UNLOCK_ONE(NULL); 1237 mutex_exit(softnet_lock); 1238 } 1239 1240 /* 1241 * Notify a tcp user of an asynchronous error; 1242 * store error as soft error, but wake up user 1243 * (for now, won't do anything until can select for soft error). 1244 */ 1245 void 1246 tcp_notify(struct inpcb *inp, int error) 1247 { 1248 struct tcpcb *tp = (struct tcpcb *)inp->inp_ppcb; 1249 struct socket *so = inp->inp_socket; 1250 1251 /* 1252 * Ignore some errors if we are hooked up. 1253 * If connection hasn't completed, has retransmitted several times, 1254 * and receives a second error, give up now. This is better 1255 * than waiting a long time to establish a connection that 1256 * can never complete. 1257 */ 1258 if (tp->t_state == TCPS_ESTABLISHED && 1259 (error == EHOSTUNREACH || error == ENETUNREACH || 1260 error == EHOSTDOWN)) { 1261 return; 1262 } else if (TCPS_HAVEESTABLISHED(tp->t_state) == 0 && 1263 tp->t_rxtshift > 3 && tp->t_softerror) 1264 so->so_error = error; 1265 else 1266 tp->t_softerror = error; 1267 cv_broadcast(&so->so_cv); 1268 sorwakeup(so); 1269 sowwakeup(so); 1270 } 1271 1272 #ifdef INET6 1273 void * 1274 tcp6_ctlinput(int cmd, const struct sockaddr *sa, void *d) 1275 { 1276 struct tcphdr th; 1277 void (*notify)(struct inpcb *, int) = tcp_notify; 1278 int nmatch; 1279 struct ip6_hdr *ip6; 1280 const struct sockaddr_in6 *sa6_src = NULL; 1281 const struct sockaddr_in6 *sa6 = (const struct sockaddr_in6 *)sa; 1282 struct mbuf *m; 1283 int off; 1284 1285 if (sa->sa_family != AF_INET6 || 1286 sa->sa_len != sizeof(struct sockaddr_in6)) 1287 return NULL; 1288 if ((unsigned)cmd >= PRC_NCMDS) 1289 return NULL; 1290 else if (cmd == PRC_QUENCH) { 1291 /* 1292 * Don't honor ICMP Source Quench messages meant for 1293 * TCP connections. 1294 */ 1295 return NULL; 1296 } else if (PRC_IS_REDIRECT(cmd)) 1297 notify = in6pcb_rtchange, d = NULL; 1298 else if (cmd == PRC_MSGSIZE) 1299 ; /* special code is present, see below */ 1300 else if (cmd == PRC_HOSTDEAD) 1301 d = NULL; 1302 else if (inet6ctlerrmap[cmd] == 0) 1303 return NULL; 1304 1305 /* if the parameter is from icmp6, decode it. */ 1306 if (d != NULL) { 1307 struct ip6ctlparam *ip6cp = (struct ip6ctlparam *)d; 1308 m = ip6cp->ip6c_m; 1309 ip6 = ip6cp->ip6c_ip6; 1310 off = ip6cp->ip6c_off; 1311 sa6_src = ip6cp->ip6c_src; 1312 } else { 1313 m = NULL; 1314 ip6 = NULL; 1315 sa6_src = &sa6_any; 1316 off = 0; 1317 } 1318 1319 if (ip6) { 1320 /* check if we can safely examine src and dst ports */ 1321 if (m->m_pkthdr.len < off + sizeof(th)) { 1322 if (cmd == PRC_MSGSIZE) 1323 icmp6_mtudisc_update((struct ip6ctlparam *)d, 0); 1324 return NULL; 1325 } 1326 1327 memset(&th, 0, sizeof(th)); 1328 m_copydata(m, off, sizeof(th), (void *)&th); 1329 1330 if (cmd == PRC_MSGSIZE) { 1331 int valid = 0; 1332 1333 /* 1334 * Check to see if we have a valid TCP connection 1335 * corresponding to the address in the ICMPv6 message 1336 * payload. 1337 */ 1338 if (in6pcb_lookup(&tcbtable, &sa6->sin6_addr, 1339 th.th_dport, 1340 (const struct in6_addr *)&sa6_src->sin6_addr, 1341 th.th_sport, 0, 0)) 1342 valid++; 1343 1344 /* 1345 * Depending on the value of "valid" and routing table 1346 * size (mtudisc_{hi,lo}wat), we will: 1347 * - recalculate the new MTU and create the 1348 * corresponding routing entry, or 1349 * - ignore the MTU change notification. 1350 */ 1351 icmp6_mtudisc_update((struct ip6ctlparam *)d, valid); 1352 1353 /* 1354 * no need to call in6pcb_notify, it should have been 1355 * called via callback if necessary 1356 */ 1357 return NULL; 1358 } 1359 1360 nmatch = in6pcb_notify(&tcbtable, sa, th.th_dport, 1361 (const struct sockaddr *)sa6_src, th.th_sport, cmd, NULL, notify); 1362 if (nmatch == 0 && syn_cache_count && 1363 (inet6ctlerrmap[cmd] == EHOSTUNREACH || 1364 inet6ctlerrmap[cmd] == ENETUNREACH || 1365 inet6ctlerrmap[cmd] == EHOSTDOWN)) 1366 syn_cache_unreach((const struct sockaddr *)sa6_src, 1367 sa, &th); 1368 } else { 1369 (void) in6pcb_notify(&tcbtable, sa, 0, 1370 (const struct sockaddr *)sa6_src, 0, cmd, NULL, notify); 1371 } 1372 1373 return NULL; 1374 } 1375 #endif 1376 1377 /* assumes that ip header and tcp header are contiguous on mbuf */ 1378 void * 1379 tcp_ctlinput(int cmd, const struct sockaddr *sa, void *v) 1380 { 1381 struct ip *ip = v; 1382 struct tcphdr *th; 1383 struct icmp *icp; 1384 extern const int inetctlerrmap[]; 1385 void (*notify)(struct inpcb *, int) = tcp_notify; 1386 int errno; 1387 int nmatch; 1388 struct tcpcb *tp; 1389 u_int mtu; 1390 tcp_seq seq; 1391 struct inpcb *inp; 1392 #ifdef INET6 1393 struct in6_addr src6, dst6; 1394 #endif 1395 1396 if (sa->sa_family != AF_INET || 1397 sa->sa_len != sizeof(struct sockaddr_in)) 1398 return NULL; 1399 if ((unsigned)cmd >= PRC_NCMDS) 1400 return NULL; 1401 errno = inetctlerrmap[cmd]; 1402 if (cmd == PRC_QUENCH) 1403 /* 1404 * Don't honor ICMP Source Quench messages meant for 1405 * TCP connections. 1406 */ 1407 return NULL; 1408 else if (PRC_IS_REDIRECT(cmd)) 1409 notify = inpcb_rtchange, ip = 0; 1410 else if (cmd == PRC_MSGSIZE && ip && ip->ip_v == 4) { 1411 /* 1412 * Check to see if we have a valid TCP connection 1413 * corresponding to the address in the ICMP message 1414 * payload. 1415 * 1416 * Boundary check is made in icmp_input(), with ICMP_ADVLENMIN. 1417 */ 1418 th = (struct tcphdr *)((char *)ip + (ip->ip_hl << 2)); 1419 #ifdef INET6 1420 in6_in_2_v4mapin6(&ip->ip_src, &src6); 1421 in6_in_2_v4mapin6(&ip->ip_dst, &dst6); 1422 #endif 1423 if ((inp = inpcb_lookup(&tcbtable, ip->ip_dst, 1424 th->th_dport, ip->ip_src, th->th_sport, 0)) != NULL) 1425 ; 1426 #ifdef INET6 1427 else if ((inp = in6pcb_lookup(&tcbtable, &dst6, 1428 th->th_dport, &src6, th->th_sport, 0, 0)) != NULL) 1429 ; 1430 #endif 1431 else 1432 return NULL; 1433 1434 /* 1435 * Now that we've validated that we are actually communicating 1436 * with the host indicated in the ICMP message, locate the 1437 * ICMP header, recalculate the new MTU, and create the 1438 * corresponding routing entry. 1439 */ 1440 icp = (struct icmp *)((char *)ip - 1441 offsetof(struct icmp, icmp_ip)); 1442 tp = intotcpcb(inp); 1443 if (tp == NULL) 1444 return NULL; 1445 seq = ntohl(th->th_seq); 1446 if (SEQ_LT(seq, tp->snd_una) || SEQ_GT(seq, tp->snd_max)) 1447 return NULL; 1448 /* 1449 * If the ICMP message advertises a Next-Hop MTU 1450 * equal or larger than the maximum packet size we have 1451 * ever sent, drop the message. 1452 */ 1453 mtu = (u_int)ntohs(icp->icmp_nextmtu); 1454 if (mtu >= tp->t_pmtud_mtu_sent) 1455 return NULL; 1456 if (mtu >= tcp_hdrsz(tp) + tp->t_pmtud_mss_acked) { 1457 /* 1458 * Calculate new MTU, and create corresponding 1459 * route (traditional PMTUD). 1460 */ 1461 tp->t_flags &= ~TF_PMTUD_PEND; 1462 icmp_mtudisc(icp, ip->ip_dst); 1463 } else { 1464 /* 1465 * Record the information got in the ICMP 1466 * message; act on it later. 1467 * If we had already recorded an ICMP message, 1468 * replace the old one only if the new message 1469 * refers to an older TCP segment 1470 */ 1471 if (tp->t_flags & TF_PMTUD_PEND) { 1472 if (SEQ_LT(tp->t_pmtud_th_seq, seq)) 1473 return NULL; 1474 } else 1475 tp->t_flags |= TF_PMTUD_PEND; 1476 tp->t_pmtud_th_seq = seq; 1477 tp->t_pmtud_nextmtu = icp->icmp_nextmtu; 1478 tp->t_pmtud_ip_len = icp->icmp_ip.ip_len; 1479 tp->t_pmtud_ip_hl = icp->icmp_ip.ip_hl; 1480 } 1481 return NULL; 1482 } else if (cmd == PRC_HOSTDEAD) 1483 ip = 0; 1484 else if (errno == 0) 1485 return NULL; 1486 if (ip && ip->ip_v == 4 && sa->sa_family == AF_INET) { 1487 th = (struct tcphdr *)((char *)ip + (ip->ip_hl << 2)); 1488 nmatch = inpcb_notify(&tcbtable, satocsin(sa)->sin_addr, 1489 th->th_dport, ip->ip_src, th->th_sport, errno, notify); 1490 if (nmatch == 0 && syn_cache_count && 1491 (inetctlerrmap[cmd] == EHOSTUNREACH || 1492 inetctlerrmap[cmd] == ENETUNREACH || 1493 inetctlerrmap[cmd] == EHOSTDOWN)) { 1494 struct sockaddr_in sin; 1495 memset(&sin, 0, sizeof(sin)); 1496 sin.sin_len = sizeof(sin); 1497 sin.sin_family = AF_INET; 1498 sin.sin_port = th->th_sport; 1499 sin.sin_addr = ip->ip_src; 1500 syn_cache_unreach((struct sockaddr *)&sin, sa, th); 1501 } 1502 1503 /* XXX mapped address case */ 1504 } else 1505 inpcb_notifyall(&tcbtable, satocsin(sa)->sin_addr, errno, 1506 notify); 1507 return NULL; 1508 } 1509 1510 /* 1511 * When a source quench is received, we are being notified of congestion. 1512 * Close the congestion window down to the Loss Window (one segment). 1513 * We will gradually open it again as we proceed. 1514 */ 1515 void 1516 tcp_quench(struct inpcb *inp) 1517 { 1518 struct tcpcb *tp = intotcpcb(inp); 1519 1520 if (tp) { 1521 tp->snd_cwnd = tp->t_segsz; 1522 tp->t_bytes_acked = 0; 1523 } 1524 } 1525 1526 /* 1527 * Path MTU Discovery handlers. 1528 */ 1529 void 1530 tcp_mtudisc_callback(struct in_addr faddr) 1531 { 1532 #ifdef INET6 1533 struct in6_addr in6; 1534 #endif 1535 1536 inpcb_notifyall(&tcbtable, faddr, EMSGSIZE, tcp_mtudisc); 1537 #ifdef INET6 1538 in6_in_2_v4mapin6(&faddr, &in6); 1539 tcp6_mtudisc_callback(&in6); 1540 #endif 1541 } 1542 1543 /* 1544 * On receipt of path MTU corrections, flush old route and replace it 1545 * with the new one. Retransmit all unacknowledged packets, to ensure 1546 * that all packets will be received. 1547 */ 1548 void 1549 tcp_mtudisc(struct inpcb *inp, int errno) 1550 { 1551 struct tcpcb *tp = intotcpcb(inp); 1552 struct rtentry *rt; 1553 1554 if (tp == NULL) 1555 return; 1556 1557 rt = inpcb_rtentry(inp); 1558 if (rt != NULL) { 1559 /* 1560 * If this was not a host route, remove and realloc. 1561 */ 1562 if ((rt->rt_flags & RTF_HOST) == 0) { 1563 inpcb_rtentry_unref(rt, inp); 1564 inpcb_rtchange(inp, errno); 1565 if ((rt = inpcb_rtentry(inp)) == NULL) 1566 return; 1567 } 1568 1569 /* 1570 * Slow start out of the error condition. We 1571 * use the MTU because we know it's smaller 1572 * than the previously transmitted segment. 1573 * 1574 * Note: This is more conservative than the 1575 * suggestion in draft-floyd-incr-init-win-03. 1576 */ 1577 if (rt->rt_rmx.rmx_mtu != 0) 1578 tp->snd_cwnd = 1579 TCP_INITIAL_WINDOW(tcp_init_win, 1580 rt->rt_rmx.rmx_mtu); 1581 inpcb_rtentry_unref(rt, inp); 1582 } 1583 1584 /* 1585 * Resend unacknowledged packets. 1586 */ 1587 tp->snd_nxt = tp->sack_newdata = tp->snd_una; 1588 tcp_output(tp); 1589 } 1590 1591 #ifdef INET6 1592 /* 1593 * Path MTU Discovery handlers. 1594 */ 1595 void 1596 tcp6_mtudisc_callback(struct in6_addr *faddr) 1597 { 1598 struct sockaddr_in6 sin6; 1599 1600 memset(&sin6, 0, sizeof(sin6)); 1601 sin6.sin6_family = AF_INET6; 1602 sin6.sin6_len = sizeof(struct sockaddr_in6); 1603 sin6.sin6_addr = *faddr; 1604 (void) in6pcb_notify(&tcbtable, (struct sockaddr *)&sin6, 0, 1605 (const struct sockaddr *)&sa6_any, 0, PRC_MSGSIZE, NULL, tcp6_mtudisc); 1606 } 1607 1608 void 1609 tcp6_mtudisc(struct inpcb *inp, int errno) 1610 { 1611 struct tcpcb *tp = intotcpcb(inp); 1612 struct rtentry *rt; 1613 1614 if (tp == NULL) 1615 return; 1616 1617 rt = in6pcb_rtentry(inp); 1618 if (rt != NULL) { 1619 /* 1620 * If this was not a host route, remove and realloc. 1621 */ 1622 if ((rt->rt_flags & RTF_HOST) == 0) { 1623 in6pcb_rtentry_unref(rt, inp); 1624 in6pcb_rtchange(inp, errno); 1625 rt = in6pcb_rtentry(inp); 1626 if (rt == NULL) 1627 return; 1628 } 1629 1630 /* 1631 * Slow start out of the error condition. We 1632 * use the MTU because we know it's smaller 1633 * than the previously transmitted segment. 1634 * 1635 * Note: This is more conservative than the 1636 * suggestion in draft-floyd-incr-init-win-03. 1637 */ 1638 if (rt->rt_rmx.rmx_mtu != 0) { 1639 tp->snd_cwnd = TCP_INITIAL_WINDOW(tcp_init_win, 1640 rt->rt_rmx.rmx_mtu); 1641 } 1642 in6pcb_rtentry_unref(rt, inp); 1643 } 1644 1645 /* 1646 * Resend unacknowledged packets. 1647 */ 1648 tp->snd_nxt = tp->sack_newdata = tp->snd_una; 1649 tcp_output(tp); 1650 } 1651 #endif /* INET6 */ 1652 1653 /* 1654 * Compute the MSS to advertise to the peer. Called only during 1655 * the 3-way handshake. If we are the server (peer initiated 1656 * connection), we are called with a pointer to the interface 1657 * on which the SYN packet arrived. If we are the client (we 1658 * initiated connection), we are called with a pointer to the 1659 * interface out which this connection should go. 1660 * 1661 * NOTE: Do not subtract IP option/extension header size nor IPsec 1662 * header size from MSS advertisement. MSS option must hold the maximum 1663 * segment size we can accept, so it must always be: 1664 * max(if mtu) - ip header - tcp header 1665 */ 1666 u_long 1667 tcp_mss_to_advertise(const struct ifnet *ifp, int af) 1668 { 1669 extern u_long in_maxmtu; 1670 u_long mss = 0; 1671 u_long hdrsiz; 1672 1673 /* 1674 * In order to avoid defeating path MTU discovery on the peer, 1675 * we advertise the max MTU of all attached networks as our MSS, 1676 * per RFC 1191, section 3.1. 1677 * 1678 * We provide the option to advertise just the MTU of 1679 * the interface on which we hope this connection will 1680 * be receiving. If we are responding to a SYN, we 1681 * will have a pretty good idea about this, but when 1682 * initiating a connection there is a bit more doubt. 1683 * 1684 * We also need to ensure that loopback has a large enough 1685 * MSS, as the loopback MTU is never included in in_maxmtu. 1686 */ 1687 1688 if (ifp != NULL) 1689 switch (af) { 1690 #ifdef INET6 1691 case AF_INET6: /* FALLTHROUGH */ 1692 #endif 1693 case AF_INET: 1694 mss = ifp->if_mtu; 1695 break; 1696 } 1697 1698 if (tcp_mss_ifmtu == 0) 1699 switch (af) { 1700 #ifdef INET6 1701 case AF_INET6: /* FALLTHROUGH */ 1702 #endif 1703 case AF_INET: 1704 mss = uimax(in_maxmtu, mss); 1705 break; 1706 } 1707 1708 switch (af) { 1709 case AF_INET: 1710 hdrsiz = sizeof(struct ip); 1711 break; 1712 #ifdef INET6 1713 case AF_INET6: 1714 hdrsiz = sizeof(struct ip6_hdr); 1715 break; 1716 #endif 1717 default: 1718 hdrsiz = 0; 1719 break; 1720 } 1721 hdrsiz += sizeof(struct tcphdr); 1722 if (mss > hdrsiz) 1723 mss -= hdrsiz; 1724 1725 mss = uimax(tcp_mssdflt, mss); 1726 return (mss); 1727 } 1728 1729 /* 1730 * Set connection variables based on the peer's advertised MSS. 1731 * We are passed the TCPCB for the actual connection. If we 1732 * are the server, we are called by the compressed state engine 1733 * when the 3-way handshake is complete. If we are the client, 1734 * we are called when we receive the SYN,ACK from the server. 1735 * 1736 * NOTE: Our advertised MSS value must be initialized in the TCPCB 1737 * before this routine is called! 1738 */ 1739 void 1740 tcp_mss_from_peer(struct tcpcb *tp, int offer) 1741 { 1742 struct socket *so; 1743 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1744 struct rtentry *rt; 1745 #endif 1746 u_long bufsize; 1747 int mss; 1748 1749 KASSERT(tp->t_inpcb != NULL); 1750 1751 so = NULL; 1752 rt = NULL; 1753 1754 so = tp->t_inpcb->inp_socket; 1755 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1756 rt = inpcb_rtentry(tp->t_inpcb); 1757 #endif 1758 1759 /* 1760 * As per RFC1122, use the default MSS value, unless they 1761 * sent us an offer. Do not accept offers less than 256 bytes. 1762 */ 1763 mss = tcp_mssdflt; 1764 if (offer) 1765 mss = offer; 1766 mss = uimax(mss, 256); /* sanity */ 1767 tp->t_peermss = mss; 1768 mss -= tcp_optlen(tp); 1769 if (tp->t_inpcb->inp_af == AF_INET) 1770 mss -= ip_optlen(tp->t_inpcb); 1771 #ifdef INET6 1772 if (tp->t_inpcb->inp_af == AF_INET6) 1773 mss -= ip6_optlen(tp->t_inpcb); 1774 #endif 1775 /* 1776 * XXX XXX What if mss goes negative or zero? This can happen if a 1777 * socket has large IPv6 options. We crash below. 1778 */ 1779 1780 /* 1781 * If there's a pipesize, change the socket buffer to that size. 1782 * Make the socket buffer an integral number of MSS units. If 1783 * the MSS is larger than the socket buffer, artificially decrease 1784 * the MSS. 1785 */ 1786 #ifdef RTV_SPIPE 1787 if (rt != NULL && rt->rt_rmx.rmx_sendpipe != 0) 1788 bufsize = rt->rt_rmx.rmx_sendpipe; 1789 else 1790 #endif 1791 { 1792 KASSERT(so != NULL); 1793 bufsize = so->so_snd.sb_hiwat; 1794 } 1795 if (bufsize < mss) 1796 mss = bufsize; 1797 else { 1798 bufsize = roundup(bufsize, mss); 1799 if (bufsize > sb_max) 1800 bufsize = sb_max; 1801 (void) sbreserve(&so->so_snd, bufsize, so); 1802 } 1803 tp->t_segsz = mss; 1804 1805 #ifdef RTV_SSTHRESH 1806 if (rt != NULL && rt->rt_rmx.rmx_ssthresh) { 1807 /* 1808 * There's some sort of gateway or interface buffer 1809 * limit on the path. Use this to set the slow 1810 * start threshold, but set the threshold to no less 1811 * than 2 * MSS. 1812 */ 1813 tp->snd_ssthresh = uimax(2 * mss, rt->rt_rmx.rmx_ssthresh); 1814 } 1815 #endif 1816 #if defined(RTV_SPIPE) || defined(RTV_SSTHRESH) 1817 inpcb_rtentry_unref(rt, tp->t_inpcb); 1818 #endif 1819 } 1820 1821 /* 1822 * Processing necessary when a TCP connection is established. 1823 */ 1824 void 1825 tcp_established(struct tcpcb *tp) 1826 { 1827 struct socket *so; 1828 #ifdef RTV_RPIPE 1829 struct rtentry *rt; 1830 #endif 1831 u_long bufsize; 1832 1833 KASSERT(tp->t_inpcb != NULL); 1834 1835 so = NULL; 1836 rt = NULL; 1837 1838 /* This is a while() to reduce the dreadful stairstepping below */ 1839 while (tp->t_inpcb->inp_af == AF_INET) { 1840 so = tp->t_inpcb->inp_socket; 1841 #if defined(RTV_RPIPE) 1842 rt = inpcb_rtentry(tp->t_inpcb); 1843 #endif 1844 if (__predict_true(tcp_msl_enable)) { 1845 if (in4p_laddr(tp->t_inpcb).s_addr == INADDR_LOOPBACK) { 1846 tp->t_msl = tcp_msl_loop ? tcp_msl_loop : (TCPTV_MSL >> 2); 1847 break; 1848 } 1849 1850 if (__predict_false(tcp_rttlocal)) { 1851 /* This may be adjusted by tcp_input */ 1852 tp->t_msl = tcp_msl_local ? tcp_msl_local : (TCPTV_MSL >> 1); 1853 break; 1854 } 1855 if (in_localaddr(in4p_faddr(tp->t_inpcb))) { 1856 tp->t_msl = tcp_msl_local ? tcp_msl_local : (TCPTV_MSL >> 1); 1857 break; 1858 } 1859 } 1860 tp->t_msl = tcp_msl_remote ? tcp_msl_remote : TCPTV_MSL; 1861 break; 1862 } 1863 1864 /* Clamp to a reasonable range. */ 1865 tp->t_msl = MIN(tp->t_msl, TCP_MAXMSL); 1866 1867 #ifdef INET6 1868 while (tp->t_inpcb->inp_af == AF_INET6) { 1869 so = tp->t_inpcb->inp_socket; 1870 #if defined(RTV_RPIPE) 1871 rt = in6pcb_rtentry(tp->t_inpcb); 1872 #endif 1873 if (__predict_true(tcp_msl_enable)) { 1874 extern const struct in6_addr in6addr_loopback; 1875 1876 if (IN6_ARE_ADDR_EQUAL(&in6p_laddr(tp->t_inpcb), 1877 &in6addr_loopback)) { 1878 tp->t_msl = tcp_msl_loop ? tcp_msl_loop : (TCPTV_MSL >> 2); 1879 break; 1880 } 1881 1882 if (__predict_false(tcp_rttlocal)) { 1883 /* This may be adjusted by tcp_input */ 1884 tp->t_msl = tcp_msl_local ? tcp_msl_local : (TCPTV_MSL >> 1); 1885 break; 1886 } 1887 if (in6_localaddr(&in6p_faddr(tp->t_inpcb))) { 1888 tp->t_msl = tcp_msl_local ? tcp_msl_local : (TCPTV_MSL >> 1); 1889 break; 1890 } 1891 } 1892 tp->t_msl = tcp_msl_remote ? tcp_msl_remote : TCPTV_MSL; 1893 break; 1894 } 1895 1896 /* Clamp to a reasonable range. */ 1897 tp->t_msl = MIN(tp->t_msl, TCP_MAXMSL); 1898 #endif 1899 1900 tp->t_state = TCPS_ESTABLISHED; 1901 TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle); 1902 1903 #ifdef RTV_RPIPE 1904 if (rt != NULL && rt->rt_rmx.rmx_recvpipe != 0) 1905 bufsize = rt->rt_rmx.rmx_recvpipe; 1906 else 1907 #endif 1908 { 1909 KASSERT(so != NULL); 1910 bufsize = so->so_rcv.sb_hiwat; 1911 } 1912 if (bufsize > tp->t_ourmss) { 1913 bufsize = roundup(bufsize, tp->t_ourmss); 1914 if (bufsize > sb_max) 1915 bufsize = sb_max; 1916 (void) sbreserve(&so->so_rcv, bufsize, so); 1917 } 1918 #ifdef RTV_RPIPE 1919 inpcb_rtentry_unref(rt, tp->t_inpcb); 1920 #endif 1921 } 1922 1923 /* 1924 * Check if there's an initial rtt or rttvar. Convert from the 1925 * route-table units to scaled multiples of the slow timeout timer. 1926 * Called only during the 3-way handshake. 1927 */ 1928 void 1929 tcp_rmx_rtt(struct tcpcb *tp) 1930 { 1931 #ifdef RTV_RTT 1932 struct rtentry *rt = NULL; 1933 int rtt; 1934 1935 KASSERT(tp->t_inpcb != NULL); 1936 1937 rt = inpcb_rtentry(tp->t_inpcb); 1938 if (rt == NULL) 1939 return; 1940 1941 if (tp->t_srtt == 0 && (rtt = rt->rt_rmx.rmx_rtt)) { 1942 /* 1943 * XXX The lock bit for MTU indicates that the value 1944 * is also a minimum value; this is subject to time. 1945 */ 1946 if (rt->rt_rmx.rmx_locks & RTV_RTT) 1947 TCPT_RANGESET(tp->t_rttmin, 1948 rtt / (RTM_RTTUNIT / PR_SLOWHZ), 1949 TCPTV_MIN, TCPTV_REXMTMAX); 1950 tp->t_srtt = rtt / 1951 ((RTM_RTTUNIT / PR_SLOWHZ) >> (TCP_RTT_SHIFT + 2)); 1952 if (rt->rt_rmx.rmx_rttvar) { 1953 tp->t_rttvar = rt->rt_rmx.rmx_rttvar / 1954 ((RTM_RTTUNIT / PR_SLOWHZ) >> 1955 (TCP_RTTVAR_SHIFT + 2)); 1956 } else { 1957 /* Default variation is +- 1 rtt */ 1958 tp->t_rttvar = 1959 tp->t_srtt >> (TCP_RTT_SHIFT - TCP_RTTVAR_SHIFT); 1960 } 1961 TCPT_RANGESET(tp->t_rxtcur, 1962 ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2), 1963 tp->t_rttmin, TCPTV_REXMTMAX); 1964 } 1965 inpcb_rtentry_unref(rt, tp->t_inpcb); 1966 #endif 1967 } 1968 1969 tcp_seq tcp_iss_seq = 0; /* tcp initial seq # */ 1970 1971 /* 1972 * Get a new sequence value given a tcp control block 1973 */ 1974 tcp_seq 1975 tcp_new_iss(struct tcpcb *tp) 1976 { 1977 1978 if (tp->t_inpcb->inp_af == AF_INET) { 1979 return tcp_new_iss1(&in4p_laddr(tp->t_inpcb), 1980 &in4p_faddr(tp->t_inpcb), tp->t_inpcb->inp_lport, 1981 tp->t_inpcb->inp_fport, sizeof(in4p_laddr(tp->t_inpcb))); 1982 } 1983 #ifdef INET6 1984 if (tp->t_inpcb->inp_af == AF_INET6) { 1985 return tcp_new_iss1(&in6p_laddr(tp->t_inpcb), 1986 &in6p_faddr(tp->t_inpcb), tp->t_inpcb->inp_lport, 1987 tp->t_inpcb->inp_fport, sizeof(in6p_laddr(tp->t_inpcb))); 1988 } 1989 #endif 1990 1991 panic("tcp_new_iss: unreachable"); 1992 } 1993 1994 static u_int8_t tcp_iss_secret[16]; /* 128 bits; should be plenty */ 1995 1996 /* 1997 * Initialize RFC 1948 ISS Secret 1998 */ 1999 static int 2000 tcp_iss_secret_init(void) 2001 { 2002 cprng_strong(kern_cprng, 2003 tcp_iss_secret, sizeof(tcp_iss_secret), 0); 2004 2005 return 0; 2006 } 2007 2008 /* 2009 * This routine actually generates a new TCP initial sequence number. 2010 */ 2011 tcp_seq 2012 tcp_new_iss1(void *laddr, void *faddr, u_int16_t lport, u_int16_t fport, 2013 size_t addrsz) 2014 { 2015 tcp_seq tcp_iss; 2016 2017 if (tcp_do_rfc1948) { 2018 MD5_CTX ctx; 2019 u_int8_t hash[16]; /* XXX MD5 knowledge */ 2020 static ONCE_DECL(tcp_iss_secret_control); 2021 2022 /* 2023 * If we haven't been here before, initialize our cryptographic 2024 * hash secret. 2025 */ 2026 RUN_ONCE(&tcp_iss_secret_control, tcp_iss_secret_init); 2027 2028 /* 2029 * Compute the base value of the ISS. It is a hash 2030 * of (saddr, sport, daddr, dport, secret). 2031 */ 2032 MD5Init(&ctx); 2033 2034 MD5Update(&ctx, (u_char *) laddr, addrsz); 2035 MD5Update(&ctx, (u_char *) &lport, sizeof(lport)); 2036 2037 MD5Update(&ctx, (u_char *) faddr, addrsz); 2038 MD5Update(&ctx, (u_char *) &fport, sizeof(fport)); 2039 2040 MD5Update(&ctx, tcp_iss_secret, sizeof(tcp_iss_secret)); 2041 2042 MD5Final(hash, &ctx); 2043 2044 memcpy(&tcp_iss, hash, sizeof(tcp_iss)); 2045 2046 #ifdef TCPISS_DEBUG 2047 printf("ISS hash 0x%08x, ", tcp_iss); 2048 #endif 2049 } else { 2050 /* 2051 * Randomize. 2052 */ 2053 tcp_iss = cprng_fast32() & TCP_ISS_RANDOM_MASK; 2054 #ifdef TCPISS_DEBUG 2055 printf("ISS random 0x%08x, ", tcp_iss); 2056 #endif 2057 } 2058 2059 /* 2060 * Add the offset in to the computed value. 2061 */ 2062 tcp_iss += tcp_iss_seq; 2063 #ifdef TCPISS_DEBUG 2064 printf("ISS %08x\n", tcp_iss); 2065 #endif 2066 return tcp_iss; 2067 } 2068 2069 #if defined(IPSEC) 2070 /* compute ESP/AH header size for TCP, including outer IP header. */ 2071 size_t 2072 ipsec4_hdrsiz_tcp(struct tcpcb *tp) 2073 { 2074 struct inpcb *inp; 2075 size_t hdrsiz; 2076 2077 /* XXX mapped addr case (tp->t_inpcb) */ 2078 if (!tp || !tp->t_template || !(inp = tp->t_inpcb)) 2079 return 0; 2080 switch (tp->t_family) { 2081 case AF_INET: 2082 /* XXX: should use correct direction. */ 2083 hdrsiz = ipsec_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, inp); 2084 break; 2085 default: 2086 hdrsiz = 0; 2087 break; 2088 } 2089 2090 return hdrsiz; 2091 } 2092 2093 #ifdef INET6 2094 size_t 2095 ipsec6_hdrsiz_tcp(struct tcpcb *tp) 2096 { 2097 struct inpcb *inp; 2098 size_t hdrsiz; 2099 2100 if (!tp || !tp->t_template || !(inp = tp->t_inpcb)) 2101 return 0; 2102 switch (tp->t_family) { 2103 case AF_INET6: 2104 /* XXX: should use correct direction. */ 2105 hdrsiz = ipsec_hdrsiz(tp->t_template, IPSEC_DIR_OUTBOUND, inp); 2106 break; 2107 case AF_INET: 2108 /* mapped address case - tricky */ 2109 default: 2110 hdrsiz = 0; 2111 break; 2112 } 2113 2114 return hdrsiz; 2115 } 2116 #endif 2117 #endif /*IPSEC*/ 2118 2119 /* 2120 * Determine the length of the TCP options for this connection. 2121 * 2122 * XXX: What do we do for SACK, when we add that? Just reserve 2123 * all of the space? Otherwise we can't exactly be incrementing 2124 * cwnd by an amount that varies depending on the amount we last 2125 * had to SACK! 2126 */ 2127 2128 u_int 2129 tcp_optlen(struct tcpcb *tp) 2130 { 2131 u_int optlen; 2132 2133 optlen = 0; 2134 if ((tp->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP|TF_NOOPT)) == 2135 (TF_REQ_TSTMP | TF_RCVD_TSTMP)) 2136 optlen += TCPOLEN_TSTAMP_APPA; 2137 2138 #ifdef TCP_SIGNATURE 2139 if (tp->t_flags & TF_SIGNATURE) 2140 optlen += TCPOLEN_SIGLEN; 2141 #endif 2142 2143 return optlen; 2144 } 2145 2146 u_int 2147 tcp_hdrsz(struct tcpcb *tp) 2148 { 2149 u_int hlen; 2150 2151 switch (tp->t_family) { 2152 #ifdef INET6 2153 case AF_INET6: 2154 hlen = sizeof(struct ip6_hdr); 2155 break; 2156 #endif 2157 case AF_INET: 2158 hlen = sizeof(struct ip); 2159 break; 2160 default: 2161 hlen = 0; 2162 break; 2163 } 2164 hlen += sizeof(struct tcphdr); 2165 2166 if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && 2167 (tp->t_flags & TF_RCVD_TSTMP) == TF_RCVD_TSTMP) 2168 hlen += TCPOLEN_TSTAMP_APPA; 2169 #ifdef TCP_SIGNATURE 2170 if (tp->t_flags & TF_SIGNATURE) 2171 hlen += TCPOLEN_SIGLEN; 2172 #endif 2173 return hlen; 2174 } 2175 2176 void 2177 tcp_statinc(u_int stat) 2178 { 2179 2180 KASSERT(stat < TCP_NSTATS); 2181 TCP_STATINC(stat); 2182 } 2183 2184 void 2185 tcp_statadd(u_int stat, uint64_t val) 2186 { 2187 2188 KASSERT(stat < TCP_NSTATS); 2189 TCP_STATADD(stat, val); 2190 } 2191
Indexes created Sun Sep 21 20:09:37 GMT 2025