1 /* $NetBSD: tcp_output.c,v 1.168 2009/03/18 16:00:22 cegger Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /* 33 * @(#)COPYRIGHT 1.1 (NRL) 17 January 1995 34 * 35 * NRL grants permission for redistribution and use in source and binary 36 * forms, with or without modification, of the software and documentation 37 * created at NRL provided that the following conditions are met: 38 * 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. All advertising materials mentioning features or use of this software 45 * must display the following acknowledgements: 46 * This product includes software developed by the University of 47 * California, Berkeley and its contributors. 48 * This product includes software developed at the Information 49 * Technology Division, US Naval Research Laboratory. 50 * 4. Neither the name of the NRL nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS 55 * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 56 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A 57 * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NRL OR 58 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 59 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 60 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 61 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 62 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 63 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 64 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 65 * 66 * The views and conclusions contained in the software and documentation 67 * are those of the authors and should not be interpreted as representing 68 * official policies, either expressed or implied, of the US Naval 69 * Research Laboratory (NRL). 70 */ 71 72 /*- 73 * Copyright (c) 1997, 1998, 2001, 2005, 2006 The NetBSD Foundation, Inc. 74 * All rights reserved. 75 * 76 * This code is derived from software contributed to The NetBSD Foundation 77 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 78 * Facility, NASA Ames Research Center. 79 * This code is derived from software contributed to The NetBSD Foundation 80 * by Charles M. Hannum. 81 * This code is derived from software contributed to The NetBSD Foundation 82 * by Rui Paulo. 83 * 84 * Redistribution and use in source and binary forms, with or without 85 * modification, are permitted provided that the following conditions 86 * are met: 87 * 1. Redistributions of source code must retain the above copyright 88 * notice, this list of conditions and the following disclaimer. 89 * 2. Redistributions in binary form must reproduce the above copyright 90 * notice, this list of conditions and the following disclaimer in the 91 * documentation and/or other materials provided with the distribution. 92 * 93 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 94 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 95 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 96 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 97 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 98 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 99 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 100 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 101 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 102 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 103 * POSSIBILITY OF SUCH DAMAGE. 104 */ 105 106 /* 107 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 108 * The Regents of the University of California. All rights reserved. 109 * 110 * Redistribution and use in source and binary forms, with or without 111 * modification, are permitted provided that the following conditions 112 * are met: 113 * 1. Redistributions of source code must retain the above copyright 114 * notice, this list of conditions and the following disclaimer. 115 * 2. Redistributions in binary form must reproduce the above copyright 116 * notice, this list of conditions and the following disclaimer in the 117 * documentation and/or other materials provided with the distribution. 118 * 3. Neither the name of the University nor the names of its contributors 119 * may be used to endorse or promote products derived from this software 120 * without specific prior written permission. 121 * 122 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 123 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 124 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 125 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 126 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 127 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 128 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 129 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 130 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 131 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 132 * SUCH DAMAGE. 133 * 134 * @(#)tcp_output.c 8.4 (Berkeley) 5/24/95 135 */ 136 137 #include <sys/cdefs.h> 138 __KERNEL_RCSID(0, "$NetBSD: tcp_output.c,v 1.168 2009/03/18 16:00:22 cegger Exp $"); 139 140 #include "opt_inet.h" 141 #include "opt_ipsec.h" 142 #include "opt_tcp_debug.h" 143 144 #include <sys/param.h> 145 #include <sys/systm.h> 146 #include <sys/malloc.h> 147 #include <sys/mbuf.h> 148 #include <sys/protosw.h> 149 #include <sys/socket.h> 150 #include <sys/socketvar.h> 151 #include <sys/errno.h> 152 #include <sys/domain.h> 153 #include <sys/kernel.h> 154 #ifdef TCP_SIGNATURE 155 #include <sys/md5.h> 156 #endif 157 158 #include <net/if.h> 159 #include <net/route.h> 160 161 #include <netinet/in.h> 162 #include <netinet/in_systm.h> 163 #include <netinet/ip.h> 164 #include <netinet/in_pcb.h> 165 #include <netinet/ip_var.h> 166 167 #ifdef INET6 168 #ifndef INET 169 #include <netinet/in.h> 170 #endif 171 #include <netinet/ip6.h> 172 #include <netinet6/in6_var.h> 173 #include <netinet6/ip6_var.h> 174 #include <netinet6/in6_pcb.h> 175 #include <netinet6/nd6.h> 176 #endif 177 178 #ifdef FAST_IPSEC 179 #include <netipsec/ipsec.h> 180 #include <netipsec/key.h> 181 #ifdef INET6 182 #include <netipsec/ipsec6.h> 183 #endif 184 #endif /* FAST_IPSEC*/ 185 #ifdef IPSEC 186 #include <netinet6/ipsec.h> 187 #endif 188 189 #include <netinet/tcp.h> 190 #define TCPOUTFLAGS 191 #include <netinet/tcp_fsm.h> 192 #include <netinet/tcp_seq.h> 193 #include <netinet/tcp_timer.h> 194 #include <netinet/tcp_var.h> 195 #include <netinet/tcp_private.h> 196 #include <netinet/tcp_congctl.h> 197 #include <netinet/tcpip.h> 198 #include <netinet/tcp_debug.h> 199 #include <netinet/in_offload.h> 200 #include <netinet6/in6_offload.h> 201 202 #ifdef IPSEC 203 #include <netkey/key.h> 204 #endif 205 206 #ifdef notyet 207 extern struct mbuf *m_copypack(); 208 #endif 209 210 /* 211 * Knob to enable Congestion Window Monitoring, and control 212 * the burst size it allows. Default burst is 4 packets, per 213 * the Internet draft. 214 */ 215 int tcp_cwm = 0; 216 int tcp_cwm_burstsize = 4; 217 218 int tcp_do_autosndbuf = 0; 219 int tcp_autosndbuf_inc = 8 * 1024; 220 int tcp_autosndbuf_max = 256 * 1024; 221 222 #ifdef TCP_OUTPUT_COUNTERS 223 #include <sys/device.h> 224 225 extern struct evcnt tcp_output_bigheader; 226 extern struct evcnt tcp_output_predict_hit; 227 extern struct evcnt tcp_output_predict_miss; 228 extern struct evcnt tcp_output_copysmall; 229 extern struct evcnt tcp_output_copybig; 230 extern struct evcnt tcp_output_refbig; 231 232 #define TCP_OUTPUT_COUNTER_INCR(ev) (ev)->ev_count++ 233 #else 234 235 #define TCP_OUTPUT_COUNTER_INCR(ev) /* nothing */ 236 237 #endif /* TCP_OUTPUT_COUNTERS */ 238 239 static 240 #ifndef GPROF 241 inline 242 #endif 243 int 244 tcp_segsize(struct tcpcb *tp, int *txsegsizep, int *rxsegsizep, 245 bool *alwaysfragp) 246 { 247 #ifdef INET 248 struct inpcb *inp = tp->t_inpcb; 249 #endif 250 #ifdef INET6 251 struct in6pcb *in6p = tp->t_in6pcb; 252 #endif 253 struct socket *so = NULL; 254 struct rtentry *rt; 255 struct ifnet *ifp; 256 int size; 257 int hdrlen; 258 int optlen; 259 260 *alwaysfragp = false; 261 262 #ifdef DIAGNOSTIC 263 if (tp->t_inpcb && tp->t_in6pcb) 264 panic("tcp_segsize: both t_inpcb and t_in6pcb are set"); 265 #endif 266 switch (tp->t_family) { 267 #ifdef INET 268 case AF_INET: 269 hdrlen = sizeof(struct ip) + sizeof(struct tcphdr); 270 break; 271 #endif 272 #ifdef INET6 273 case AF_INET6: 274 hdrlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 275 break; 276 #endif 277 default: 278 size = tcp_mssdflt; 279 goto out; 280 } 281 282 rt = NULL; 283 #ifdef INET 284 if (inp) { 285 rt = in_pcbrtentry(inp); 286 so = inp->inp_socket; 287 } 288 #endif 289 #ifdef INET6 290 if (in6p) { 291 rt = in6_pcbrtentry(in6p); 292 so = in6p->in6p_socket; 293 } 294 #endif 295 if (rt == NULL) { 296 size = tcp_mssdflt; 297 goto out; 298 } 299 300 ifp = rt->rt_ifp; 301 302 size = tcp_mssdflt; 303 if (tp->t_mtudisc && rt->rt_rmx.rmx_mtu != 0) { 304 #ifdef INET6 305 if (in6p && rt->rt_rmx.rmx_mtu < IPV6_MMTU) { 306 /* 307 * RFC2460 section 5, last paragraph: if path MTU is 308 * smaller than 1280, use 1280 as packet size and 309 * attach fragment header. 310 */ 311 size = IPV6_MMTU - hdrlen - sizeof(struct ip6_frag); 312 *alwaysfragp = true; 313 } else 314 size = rt->rt_rmx.rmx_mtu - hdrlen; 315 #else 316 size = rt->rt_rmx.rmx_mtu - hdrlen; 317 #endif 318 } else if (ifp->if_flags & IFF_LOOPBACK) 319 size = ifp->if_mtu - hdrlen; 320 #ifdef INET 321 else if (inp && tp->t_mtudisc) 322 size = ifp->if_mtu - hdrlen; 323 else if (inp && in_localaddr(inp->inp_faddr)) 324 size = ifp->if_mtu - hdrlen; 325 #endif 326 #ifdef INET6 327 else if (in6p) { 328 #ifdef INET 329 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 330 /* mapped addr case */ 331 struct in_addr d; 332 bcopy(&in6p->in6p_faddr.s6_addr32[3], &d, sizeof(d)); 333 if (tp->t_mtudisc || in_localaddr(d)) 334 size = ifp->if_mtu - hdrlen; 335 } else 336 #endif 337 { 338 /* 339 * for IPv6, path MTU discovery is always turned on, 340 * or the node must use packet size <= 1280. 341 */ 342 size = tp->t_mtudisc ? IN6_LINKMTU(ifp) : IPV6_MMTU; 343 size -= hdrlen; 344 } 345 } 346 #endif 347 out: 348 /* 349 * Now we must make room for whatever extra TCP/IP options are in 350 * the packet. 351 */ 352 optlen = tcp_optlen(tp); 353 354 /* 355 * XXX tp->t_ourmss should have the right size, but without this code 356 * fragmentation will occur... need more investigation 357 */ 358 #ifdef INET 359 if (inp) { 360 #if defined(IPSEC) || defined(FAST_IPSEC) 361 if (! IPSEC_PCB_SKIP_IPSEC(inp->inp_sp, IPSEC_DIR_OUTBOUND)) 362 optlen += ipsec4_hdrsiz_tcp(tp); 363 #endif 364 optlen += ip_optlen(inp); 365 } 366 #endif 367 #ifdef INET6 368 #ifdef INET 369 if (in6p && tp->t_family == AF_INET) { 370 #if defined(IPSEC) || defined(FAST_IPSEC) 371 if (! IPSEC_PCB_SKIP_IPSEC(in6p->in6p_sp, IPSEC_DIR_OUTBOUND)) 372 optlen += ipsec4_hdrsiz_tcp(tp); 373 #endif 374 /* XXX size -= ip_optlen(in6p); */ 375 } else 376 #endif 377 if (in6p && tp->t_family == AF_INET6) { 378 #if defined(IPSEC) || defined(FAST_IPSEC) 379 if (! IPSEC_PCB_SKIP_IPSEC(in6p->in6p_sp, IPSEC_DIR_OUTBOUND)) 380 optlen += ipsec6_hdrsiz_tcp(tp); 381 #endif 382 optlen += ip6_optlen(in6p); 383 } 384 #endif 385 size -= optlen; 386 387 /* there may not be any room for data if mtu is too small */ 388 if (size < 0) 389 return (EMSGSIZE); 390 391 /* 392 * *rxsegsizep holds *estimated* inbound segment size (estimation 393 * assumes that path MTU is the same for both ways). this is only 394 * for silly window avoidance, do not use the value for other purposes. 395 * 396 * ipseclen is subtracted from both sides, this may not be right. 397 * I'm not quite sure about this (could someone comment). 398 */ 399 *txsegsizep = min(tp->t_peermss - optlen, size); 400 /* 401 * Never send more than half a buffer full. This insures that we can 402 * always keep 2 packets on the wire, no matter what SO_SNDBUF is, and 403 * therefore acks will never be delayed unless we run out of data to 404 * transmit. 405 */ 406 if (so) 407 *txsegsizep = min(so->so_snd.sb_hiwat >> 1, *txsegsizep); 408 *rxsegsizep = min(tp->t_ourmss - optlen, size); 409 410 if (*txsegsizep != tp->t_segsz) { 411 /* 412 * If the new segment size is larger, we don't want to 413 * mess up the congestion window, but if it is smaller 414 * we'll have to reduce the congestion window to ensure 415 * that we don't get into trouble with initial windows 416 * and the rest. In any case, if the segment size 417 * has changed, chances are the path has, too, and 418 * our congestion window will be different. 419 */ 420 if (*txsegsizep < tp->t_segsz) { 421 tp->snd_cwnd = max((tp->snd_cwnd / tp->t_segsz) 422 * *txsegsizep, *txsegsizep); 423 tp->snd_ssthresh = max((tp->snd_ssthresh / tp->t_segsz) 424 * *txsegsizep, *txsegsizep); 425 } 426 tp->t_segsz = *txsegsizep; 427 } 428 429 return (0); 430 } 431 432 static 433 #ifndef GPROF 434 inline 435 #endif 436 int 437 tcp_build_datapkt(struct tcpcb *tp, struct socket *so, int off, 438 long len, int hdrlen, struct mbuf **mp) 439 { 440 struct mbuf *m, *m0; 441 uint64_t *tcps; 442 443 tcps = TCP_STAT_GETREF(); 444 if (tp->t_force && len == 1) 445 tcps[TCP_STAT_SNDPROBE]++; 446 else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) { 447 tcps[TCP_STAT_SNDREXMITPACK]++; 448 tcps[TCP_STAT_SNDREXMITBYTE] += len; 449 } else { 450 tcps[TCP_STAT_SNDPACK]++; 451 tcps[TCP_STAT_SNDBYTE] += len; 452 } 453 TCP_STAT_PUTREF(); 454 #ifdef notyet 455 if ((m = m_copypack(so->so_snd.sb_mb, off, 456 (int)len, max_linkhdr + hdrlen)) == 0) 457 return (ENOBUFS); 458 /* 459 * m_copypack left space for our hdr; use it. 460 */ 461 m->m_len += hdrlen; 462 m->m_data -= hdrlen; 463 #else 464 MGETHDR(m, M_DONTWAIT, MT_HEADER); 465 if (__predict_false(m == NULL)) 466 return (ENOBUFS); 467 MCLAIM(m, &tcp_tx_mowner); 468 469 /* 470 * XXX Because other code assumes headers will fit in 471 * XXX one header mbuf. 472 * 473 * (This code should almost *never* be run.) 474 */ 475 if (__predict_false((max_linkhdr + hdrlen) > MHLEN)) { 476 TCP_OUTPUT_COUNTER_INCR(&tcp_output_bigheader); 477 MCLGET(m, M_DONTWAIT); 478 if ((m->m_flags & M_EXT) == 0) { 479 m_freem(m); 480 return (ENOBUFS); 481 } 482 } 483 484 m->m_data += max_linkhdr; 485 m->m_len = hdrlen; 486 487 /* 488 * To avoid traversing the whole sb_mb chain for correct 489 * data to send, remember last sent mbuf, its offset and 490 * the sent size. When called the next time, see if the 491 * data to send is directly following the previous transfer. 492 * This is important for large TCP windows. 493 */ 494 if (off == 0 || tp->t_lastm == NULL || 495 (tp->t_lastoff + tp->t_lastlen) != off) { 496 TCP_OUTPUT_COUNTER_INCR(&tcp_output_predict_miss); 497 /* 498 * Either a new packet or a retransmit. 499 * Start from the beginning. 500 */ 501 tp->t_lastm = so->so_snd.sb_mb; 502 tp->t_inoff = off; 503 } else { 504 TCP_OUTPUT_COUNTER_INCR(&tcp_output_predict_hit); 505 tp->t_inoff += tp->t_lastlen; 506 } 507 508 /* Traverse forward to next packet */ 509 while (tp->t_inoff > 0) { 510 if (tp->t_lastm == NULL) 511 panic("tp->t_lastm == NULL"); 512 if (tp->t_inoff < tp->t_lastm->m_len) 513 break; 514 tp->t_inoff -= tp->t_lastm->m_len; 515 tp->t_lastm = tp->t_lastm->m_next; 516 } 517 518 tp->t_lastoff = off; 519 tp->t_lastlen = len; 520 m0 = tp->t_lastm; 521 off = tp->t_inoff; 522 523 if (len <= M_TRAILINGSPACE(m)) { 524 m_copydata(m0, off, (int) len, mtod(m, char *) + hdrlen); 525 m->m_len += len; 526 TCP_OUTPUT_COUNTER_INCR(&tcp_output_copysmall); 527 } else { 528 m->m_next = m_copym(m0, off, (int) len, M_DONTWAIT); 529 if (m->m_next == NULL) { 530 m_freem(m); 531 return (ENOBUFS); 532 } 533 #ifdef TCP_OUTPUT_COUNTERS 534 if (m->m_next->m_flags & M_EXT) 535 TCP_OUTPUT_COUNTER_INCR(&tcp_output_refbig); 536 else 537 TCP_OUTPUT_COUNTER_INCR(&tcp_output_copybig); 538 #endif /* TCP_OUTPUT_COUNTERS */ 539 } 540 #endif 541 542 *mp = m; 543 return (0); 544 } 545 546 /* 547 * Tcp output routine: figure out what should be sent and send it. 548 */ 549 int 550 tcp_output(struct tcpcb *tp) 551 { 552 struct rtentry *rt; 553 struct socket *so; 554 struct route *ro; 555 long len, win; 556 int off, flags, error; 557 struct mbuf *m; 558 struct ip *ip; 559 #ifdef INET6 560 struct ip6_hdr *ip6; 561 #endif 562 struct tcphdr *th; 563 u_char opt[MAX_TCPOPTLEN]; 564 unsigned optlen, hdrlen, packetlen; 565 unsigned int sack_numblks; 566 int idle, sendalot, txsegsize, rxsegsize; 567 int txsegsize_nosack; 568 int maxburst = TCP_MAXBURST; 569 int af; /* address family on the wire */ 570 int iphdrlen; 571 int has_tso4, has_tso6; 572 int has_tso, use_tso; 573 bool alwaysfrag; 574 int sack_rxmit; 575 int sack_bytes_rxmt; 576 struct sackhole *p; 577 #ifdef TCP_SIGNATURE 578 int sigoff = 0; 579 #endif 580 uint64_t *tcps; 581 582 #ifdef DIAGNOSTIC 583 if (tp->t_inpcb && tp->t_in6pcb) 584 panic("tcp_output: both t_inpcb and t_in6pcb are set"); 585 #endif 586 so = NULL; 587 ro = NULL; 588 if (tp->t_inpcb) { 589 so = tp->t_inpcb->inp_socket; 590 ro = &tp->t_inpcb->inp_route; 591 } 592 #ifdef INET6 593 else if (tp->t_in6pcb) { 594 so = tp->t_in6pcb->in6p_socket; 595 ro = &tp->t_in6pcb->in6p_route; 596 } 597 #endif 598 599 switch (af = tp->t_family) { 600 #ifdef INET 601 case AF_INET: 602 if (tp->t_inpcb) 603 break; 604 #ifdef INET6 605 /* mapped addr case */ 606 if (tp->t_in6pcb) 607 break; 608 #endif 609 return (EINVAL); 610 #endif 611 #ifdef INET6 612 case AF_INET6: 613 if (tp->t_in6pcb) 614 break; 615 return (EINVAL); 616 #endif 617 default: 618 return (EAFNOSUPPORT); 619 } 620 621 if (tcp_segsize(tp, &txsegsize, &rxsegsize, &alwaysfrag)) 622 return (EMSGSIZE); 623 624 idle = (tp->snd_max == tp->snd_una); 625 626 /* 627 * Determine if we can use TCP segmentation offload: 628 * - If we're using IPv4 629 * - If there is not an IPsec policy that prevents it 630 * - If the interface can do it 631 */ 632 has_tso4 = has_tso6 = false; 633 #if defined(INET) 634 has_tso4 = tp->t_inpcb != NULL && 635 #if defined(IPSEC) || defined(FAST_IPSEC) 636 IPSEC_PCB_SKIP_IPSEC(tp->t_inpcb->inp_sp, 637 IPSEC_DIR_OUTBOUND) && 638 #endif 639 (rt = rtcache_validate(&tp->t_inpcb->inp_route)) != NULL && 640 (rt->rt_ifp->if_capenable & IFCAP_TSOv4) != 0; 641 #endif /* defined(INET) */ 642 #if defined(INET6) 643 has_tso6 = tp->t_in6pcb != NULL && 644 #if defined(IPSEC) || defined(FAST_IPSEC) 645 IPSEC_PCB_SKIP_IPSEC(tp->t_in6pcb->in6p_sp, 646 IPSEC_DIR_OUTBOUND) && 647 #endif 648 (rt = rtcache_validate(&tp->t_in6pcb->in6p_route)) != NULL && 649 (rt->rt_ifp->if_capenable & IFCAP_TSOv6) != 0; 650 #endif /* defined(INET6) */ 651 has_tso = (has_tso4 || has_tso6) && !alwaysfrag; 652 653 /* 654 * Restart Window computation. From draft-floyd-incr-init-win-03: 655 * 656 * Optionally, a TCP MAY set the restart window to the 657 * minimum of the value used for the initial window and 658 * the current value of cwnd (in other words, using a 659 * larger value for the restart window should never increase 660 * the size of cwnd). 661 */ 662 if (tcp_cwm) { 663 /* 664 * Hughes/Touch/Heidemann Congestion Window Monitoring. 665 * Count the number of packets currently pending 666 * acknowledgement, and limit our congestion window 667 * to a pre-determined allowed burst size plus that count. 668 * This prevents bursting once all pending packets have 669 * been acknowledged (i.e. transmission is idle). 670 * 671 * XXX Link this to Initial Window? 672 */ 673 tp->snd_cwnd = min(tp->snd_cwnd, 674 (tcp_cwm_burstsize * txsegsize) + 675 (tp->snd_nxt - tp->snd_una)); 676 } else { 677 if (idle && (tcp_now - tp->t_rcvtime) >= tp->t_rxtcur) { 678 /* 679 * We have been idle for "a while" and no acks are 680 * expected to clock out any data we send -- 681 * slow start to get ack "clock" running again. 682 */ 683 int ss = tcp_init_win; 684 #ifdef INET 685 if (tp->t_inpcb && 686 in_localaddr(tp->t_inpcb->inp_faddr)) 687 ss = tcp_init_win_local; 688 #endif 689 #ifdef INET6 690 if (tp->t_in6pcb && 691 in6_localaddr(&tp->t_in6pcb->in6p_faddr)) 692 ss = tcp_init_win_local; 693 #endif 694 tp->snd_cwnd = min(tp->snd_cwnd, 695 TCP_INITIAL_WINDOW(ss, txsegsize)); 696 } 697 } 698 699 txsegsize_nosack = txsegsize; 700 again: 701 use_tso = has_tso; 702 if ((tp->t_flags & (TF_ECN_SND_CWR|TF_ECN_SND_ECE)) != 0) { 703 /* don't duplicate CWR/ECE. */ 704 use_tso = 0; 705 } 706 TCP_REASS_LOCK(tp); 707 sack_numblks = tcp_sack_numblks(tp); 708 if (sack_numblks) { 709 int sackoptlen; 710 711 sackoptlen = TCP_SACK_OPTLEN(sack_numblks); 712 if (sackoptlen > txsegsize_nosack) { 713 sack_numblks = 0; /* give up SACK */ 714 txsegsize = txsegsize_nosack; 715 } else { 716 if ((tp->rcv_sack_flags & TCPSACK_HAVED) != 0) { 717 /* don't duplicate D-SACK. */ 718 use_tso = 0; 719 } 720 txsegsize = txsegsize_nosack - sackoptlen; 721 } 722 } else { 723 txsegsize = txsegsize_nosack; 724 } 725 726 /* 727 * Determine length of data that should be transmitted, and 728 * flags that should be used. If there is some data or critical 729 * controls (SYN, RST) to send, then transmit; otherwise, 730 * investigate further. 731 * 732 * Readjust SACK information to avoid resending duplicate data. 733 */ 734 if (TCP_SACK_ENABLED(tp) && SEQ_LT(tp->snd_nxt, tp->snd_max)) 735 tcp_sack_adjust(tp); 736 sendalot = 0; 737 off = tp->snd_nxt - tp->snd_una; 738 win = min(tp->snd_wnd, tp->snd_cwnd); 739 740 flags = tcp_outflags[tp->t_state]; 741 742 /* 743 * Send any SACK-generated retransmissions. If we're explicitly trying 744 * to send out new data (when sendalot is 1), bypass this function. 745 * If we retransmit in fast recovery mode, decrement snd_cwnd, since 746 * we're replacing a (future) new transmission with a retransmission 747 * now, and we previously incremented snd_cwnd in tcp_input(). 748 */ 749 /* 750 * Still in sack recovery , reset rxmit flag to zero. 751 */ 752 sack_rxmit = 0; 753 sack_bytes_rxmt = 0; 754 len = 0; 755 p = NULL; 756 do { 757 long cwin; 758 if (!TCP_SACK_ENABLED(tp)) 759 break; 760 if (tp->t_partialacks < 0) 761 break; 762 p = tcp_sack_output(tp, &sack_bytes_rxmt); 763 if (p == NULL) 764 break; 765 766 cwin = min(tp->snd_wnd, tp->snd_cwnd) - sack_bytes_rxmt; 767 if (cwin < 0) 768 cwin = 0; 769 /* Do not retransmit SACK segments beyond snd_recover */ 770 if (SEQ_GT(p->end, tp->snd_recover)) { 771 /* 772 * (At least) part of sack hole extends beyond 773 * snd_recover. Check to see if we can rexmit data 774 * for this hole. 775 */ 776 if (SEQ_GEQ(p->rxmit, tp->snd_recover)) { 777 /* 778 * Can't rexmit any more data for this hole. 779 * That data will be rexmitted in the next 780 * sack recovery episode, when snd_recover 781 * moves past p->rxmit. 782 */ 783 p = NULL; 784 break; 785 } 786 /* Can rexmit part of the current hole */ 787 len = ((long)ulmin(cwin, tp->snd_recover - p->rxmit)); 788 } else 789 len = ((long)ulmin(cwin, p->end - p->rxmit)); 790 off = p->rxmit - tp->snd_una; 791 if (off + len > so->so_snd.sb_cc) { 792 /* 1 for TH_FIN */ 793 KASSERT(off + len == so->so_snd.sb_cc + 1); 794 KASSERT(p->rxmit + len == tp->snd_max); 795 len = so->so_snd.sb_cc - off; 796 } 797 if (len > 0) { 798 sack_rxmit = 1; 799 sendalot = 1; 800 } 801 } while (/*CONSTCOND*/0); 802 803 /* 804 * If in persist timeout with window of 0, send 1 byte. 805 * Otherwise, if window is small but nonzero 806 * and timer expired, we will send what we can 807 * and go to transmit state. 808 */ 809 if (tp->t_force) { 810 if (win == 0) { 811 /* 812 * If we still have some data to send, then 813 * clear the FIN bit. Usually this would 814 * happen below when it realizes that we 815 * aren't sending all the data. However, 816 * if we have exactly 1 byte of unset data, 817 * then it won't clear the FIN bit below, 818 * and if we are in persist state, we wind 819 * up sending the packet without recording 820 * that we sent the FIN bit. 821 * 822 * We can't just blindly clear the FIN bit, 823 * because if we don't have any more data 824 * to send then the probe will be the FIN 825 * itself. 826 */ 827 if (off < so->so_snd.sb_cc) 828 flags &= ~TH_FIN; 829 win = 1; 830 } else { 831 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 832 tp->t_rxtshift = 0; 833 } 834 } 835 836 if (sack_rxmit == 0) { 837 if (TCP_SACK_ENABLED(tp) && tp->t_partialacks >= 0) { 838 long cwin; 839 840 /* 841 * We are inside of a SACK recovery episode and are 842 * sending new data, having retransmitted all the 843 * data possible in the scoreboard. 844 */ 845 if (tp->snd_wnd < so->so_snd.sb_cc) { 846 len = tp->snd_wnd - off; 847 flags &= ~TH_FIN; 848 } else { 849 len = so->so_snd.sb_cc - off; 850 } 851 852 /* 853 * From FreeBSD: 854 * Don't remove this (len > 0) check ! 855 * We explicitly check for len > 0 here (although it 856 * isn't really necessary), to work around a gcc 857 * optimization issue - to force gcc to compute 858 * len above. Without this check, the computation 859 * of len is bungled by the optimizer. 860 */ 861 if (len > 0) { 862 cwin = tp->snd_cwnd - 863 (tp->snd_nxt - tp->sack_newdata) - 864 sack_bytes_rxmt; 865 if (cwin < 0) 866 cwin = 0; 867 if (cwin < len) { 868 len = cwin; 869 flags &= ~TH_FIN; 870 } 871 } 872 } else if (win < so->so_snd.sb_cc) { 873 len = win - off; 874 flags &= ~TH_FIN; 875 } else { 876 len = so->so_snd.sb_cc - off; 877 } 878 } 879 880 if (len < 0) { 881 /* 882 * If FIN has been sent but not acked, 883 * but we haven't been called to retransmit, 884 * len will be -1. Otherwise, window shrank 885 * after we sent into it. If window shrank to 0, 886 * cancel pending retransmit, pull snd_nxt back 887 * to (closed) window, and set the persist timer 888 * if it isn't already going. If the window didn't 889 * close completely, just wait for an ACK. 890 * 891 * If we have a pending FIN, either it has already been 892 * transmitted or it is outside the window, so drop it. 893 * If the FIN has been transmitted, but this is not a 894 * retransmission, then len must be -1. Therefore we also 895 * prevent here the sending of `gratuitous FINs'. This 896 * eliminates the need to check for that case below (e.g. 897 * to back up snd_nxt before the FIN so that the sequence 898 * number is correct). 899 */ 900 len = 0; 901 flags &= ~TH_FIN; 902 if (win == 0) { 903 TCP_TIMER_DISARM(tp, TCPT_REXMT); 904 tp->t_rxtshift = 0; 905 tp->snd_nxt = tp->snd_una; 906 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) 907 tcp_setpersist(tp); 908 } 909 } 910 911 /* 912 * Automatic sizing enables the performance of large buffers 913 * and most of the efficiency of small ones by only allocating 914 * space when it is needed. 915 * 916 * The criteria to step up the send buffer one notch are: 917 * 1. receive window of remote host is larger than send buffer 918 * (with a fudge factor of 5/4th); 919 * 2. send buffer is filled to 7/8th with data (so we actually 920 * have data to make use of it); 921 * 3. send buffer fill has not hit maximal automatic size; 922 * 4. our send window (slow start and cogestion controlled) is 923 * larger than sent but unacknowledged data in send buffer. 924 * 925 * The remote host receive window scaling factor may limit the 926 * growing of the send buffer before it reaches its allowed 927 * maximum. 928 * 929 * It scales directly with slow start or congestion window 930 * and does at most one step per received ACK. This fast 931 * scaling has the drawback of growing the send buffer beyond 932 * what is strictly necessary to make full use of a given 933 * delay*bandwith product. However testing has shown this not 934 * to be much of an problem. At worst we are trading wasting 935 * of available bandwith (the non-use of it) for wasting some 936 * socket buffer memory. 937 * 938 * TODO: Shrink send buffer during idle periods together 939 * with congestion window. Requires another timer. 940 */ 941 if (tcp_do_autosndbuf && so->so_snd.sb_flags & SB_AUTOSIZE) { 942 if ((tp->snd_wnd / 4 * 5) >= so->so_snd.sb_hiwat && 943 so->so_snd.sb_cc >= (so->so_snd.sb_hiwat / 8 * 7) && 944 so->so_snd.sb_cc < tcp_autosndbuf_max && 945 win >= (so->so_snd.sb_cc - (tp->snd_nxt - tp->snd_una))) { 946 if (!sbreserve(&so->so_snd, 947 min(so->so_snd.sb_hiwat + tcp_autosndbuf_inc, 948 tcp_autosndbuf_max), so)) 949 so->so_snd.sb_flags &= ~SB_AUTOSIZE; 950 } 951 } 952 953 if (len > txsegsize) { 954 if (use_tso) { 955 /* 956 * Truncate TSO transfers to IP_MAXPACKET, and make 957 * sure that we send equal size transfers down the 958 * stack (rather than big-small-big-small-...). 959 */ 960 #ifdef INET6 961 #if IPV6_MAXPACKET != IP_MAXPACKET 962 #error IPV6_MAXPACKET != IP_MAXPACKET 963 #endif 964 #endif 965 len = (min(len, IP_MAXPACKET) / txsegsize) * txsegsize; 966 if (len <= txsegsize) { 967 use_tso = 0; 968 } 969 } else 970 len = txsegsize; 971 flags &= ~TH_FIN; 972 sendalot = 1; 973 } else 974 use_tso = 0; 975 if (sack_rxmit) { 976 if (SEQ_LT(p->rxmit + len, tp->snd_una + so->so_snd.sb_cc)) 977 flags &= ~TH_FIN; 978 } 979 980 win = sbspace(&so->so_rcv); 981 982 /* 983 * Sender silly window avoidance. If connection is idle 984 * and can send all data, a maximum segment, 985 * at least a maximum default-size segment do it, 986 * or are forced, do it; otherwise don't bother. 987 * If peer's buffer is tiny, then send 988 * when window is at least half open. 989 * If retransmitting (possibly after persist timer forced us 990 * to send into a small window), then must resend. 991 */ 992 if (len) { 993 if (len >= txsegsize) 994 goto send; 995 if ((so->so_state & SS_MORETOCOME) == 0 && 996 ((idle || tp->t_flags & TF_NODELAY) && 997 len + off >= so->so_snd.sb_cc)) 998 goto send; 999 if (tp->t_force) 1000 goto send; 1001 if (len >= tp->max_sndwnd / 2) 1002 goto send; 1003 if (SEQ_LT(tp->snd_nxt, tp->snd_max)) 1004 goto send; 1005 if (sack_rxmit) 1006 goto send; 1007 } 1008 1009 /* 1010 * Compare available window to amount of window known to peer 1011 * (as advertised window less next expected input). If the 1012 * difference is at least twice the size of the largest segment 1013 * we expect to receive (i.e. two segments) or at least 50% of 1014 * the maximum possible window, then want to send a window update 1015 * to peer. 1016 */ 1017 if (win > 0) { 1018 /* 1019 * "adv" is the amount we can increase the window, 1020 * taking into account that we are limited by 1021 * TCP_MAXWIN << tp->rcv_scale. 1022 */ 1023 long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) - 1024 (tp->rcv_adv - tp->rcv_nxt); 1025 1026 if (adv >= (long) (2 * rxsegsize)) 1027 goto send; 1028 if (2 * adv >= (long) so->so_rcv.sb_hiwat) 1029 goto send; 1030 } 1031 1032 /* 1033 * Send if we owe peer an ACK. 1034 */ 1035 if (tp->t_flags & TF_ACKNOW) 1036 goto send; 1037 if (flags & (TH_SYN|TH_FIN|TH_RST)) 1038 goto send; 1039 if (SEQ_GT(tp->snd_up, tp->snd_una)) 1040 goto send; 1041 /* 1042 * In SACK, it is possible for tcp_output to fail to send a segment 1043 * after the retransmission timer has been turned off. Make sure 1044 * that the retransmission timer is set. 1045 */ 1046 if (TCP_SACK_ENABLED(tp) && SEQ_GT(tp->snd_max, tp->snd_una) && 1047 !TCP_TIMER_ISARMED(tp, TCPT_REXMT) && 1048 !TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) { 1049 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 1050 goto just_return; 1051 } 1052 1053 /* 1054 * TCP window updates are not reliable, rather a polling protocol 1055 * using ``persist'' packets is used to insure receipt of window 1056 * updates. The three ``states'' for the output side are: 1057 * idle not doing retransmits or persists 1058 * persisting to move a small or zero window 1059 * (re)transmitting and thereby not persisting 1060 * 1061 * tp->t_timer[TCPT_PERSIST] 1062 * is set when we are in persist state. 1063 * tp->t_force 1064 * is set when we are called to send a persist packet. 1065 * tp->t_timer[TCPT_REXMT] 1066 * is set when we are retransmitting 1067 * The output side is idle when both timers are zero. 1068 * 1069 * If send window is too small, there is data to transmit, and no 1070 * retransmit or persist is pending, then go to persist state. 1071 * If nothing happens soon, send when timer expires: 1072 * if window is nonzero, transmit what we can, 1073 * otherwise force out a byte. 1074 */ 1075 if (so->so_snd.sb_cc && TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 1076 TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 1077 tp->t_rxtshift = 0; 1078 tcp_setpersist(tp); 1079 } 1080 1081 /* 1082 * No reason to send a segment, just return. 1083 */ 1084 just_return: 1085 TCP_REASS_UNLOCK(tp); 1086 return (0); 1087 1088 send: 1089 /* 1090 * Before ESTABLISHED, force sending of initial options 1091 * unless TCP set not to do any options. 1092 * NOTE: we assume that the IP/TCP header plus TCP options 1093 * always fit in a single mbuf, leaving room for a maximum 1094 * link header, i.e. 1095 * max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MCLBYTES 1096 */ 1097 optlen = 0; 1098 switch (af) { 1099 #ifdef INET 1100 case AF_INET: 1101 iphdrlen = sizeof(struct ip) + sizeof(struct tcphdr); 1102 break; 1103 #endif 1104 #ifdef INET6 1105 case AF_INET6: 1106 iphdrlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 1107 break; 1108 #endif 1109 default: /*pacify gcc*/ 1110 iphdrlen = 0; 1111 break; 1112 } 1113 hdrlen = iphdrlen; 1114 if (flags & TH_SYN) { 1115 struct rtentry *synrt; 1116 1117 synrt = NULL; 1118 #ifdef INET 1119 if (tp->t_inpcb) 1120 synrt = in_pcbrtentry(tp->t_inpcb); 1121 #endif 1122 #ifdef INET6 1123 if (tp->t_in6pcb) 1124 synrt = in6_pcbrtentry(tp->t_in6pcb); 1125 #endif 1126 1127 tp->snd_nxt = tp->iss; 1128 tp->t_ourmss = tcp_mss_to_advertise(synrt != NULL ? 1129 synrt->rt_ifp : NULL, af); 1130 if ((tp->t_flags & TF_NOOPT) == 0) { 1131 opt[0] = TCPOPT_MAXSEG; 1132 opt[1] = 4; 1133 opt[2] = (tp->t_ourmss >> 8) & 0xff; 1134 opt[3] = tp->t_ourmss & 0xff; 1135 optlen = 4; 1136 1137 if ((tp->t_flags & TF_REQ_SCALE) && 1138 ((flags & TH_ACK) == 0 || 1139 (tp->t_flags & TF_RCVD_SCALE))) { 1140 *((u_int32_t *) (opt + optlen)) = htonl( 1141 TCPOPT_NOP << 24 | 1142 TCPOPT_WINDOW << 16 | 1143 TCPOLEN_WINDOW << 8 | 1144 tp->request_r_scale); 1145 optlen += 4; 1146 } 1147 if (tcp_do_sack) { 1148 u_int8_t *cp = (u_int8_t *)(opt + optlen); 1149 1150 cp[0] = TCPOPT_SACK_PERMITTED; 1151 cp[1] = 2; 1152 cp[2] = TCPOPT_NOP; 1153 cp[3] = TCPOPT_NOP; 1154 optlen += 4; 1155 } 1156 } 1157 } 1158 1159 /* 1160 * Send a timestamp and echo-reply if this is a SYN and our side 1161 * wants to use timestamps (TF_REQ_TSTMP is set) or both our side 1162 * and our peer have sent timestamps in our SYN's. 1163 */ 1164 if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && 1165 (flags & TH_RST) == 0 && 1166 ((flags & (TH_SYN|TH_ACK)) == TH_SYN || 1167 (tp->t_flags & TF_RCVD_TSTMP))) { 1168 u_int32_t *lp = (u_int32_t *)(opt + optlen); 1169 1170 /* Form timestamp option as shown in appendix A of RFC 1323. */ 1171 *lp++ = htonl(TCPOPT_TSTAMP_HDR); 1172 *lp++ = htonl(TCP_TIMESTAMP(tp)); 1173 *lp = htonl(tp->ts_recent); 1174 optlen += TCPOLEN_TSTAMP_APPA; 1175 1176 /* Set receive buffer autosizing timestamp. */ 1177 if (tp->rfbuf_ts == 0 && (so->so_rcv.sb_flags & SB_AUTOSIZE)) 1178 tp->rfbuf_ts = TCP_TIMESTAMP(tp); 1179 } 1180 1181 /* 1182 * Tack on the SACK block if it is necessary. 1183 */ 1184 if (sack_numblks) { 1185 int sack_len; 1186 u_char *bp = (u_char *)(opt + optlen); 1187 u_int32_t *lp = (u_int32_t *)(bp + 4); 1188 struct ipqent *tiqe; 1189 1190 sack_len = sack_numblks * 8 + 2; 1191 bp[0] = TCPOPT_NOP; 1192 bp[1] = TCPOPT_NOP; 1193 bp[2] = TCPOPT_SACK; 1194 bp[3] = sack_len; 1195 if ((tp->rcv_sack_flags & TCPSACK_HAVED) != 0) { 1196 sack_numblks--; 1197 *lp++ = htonl(tp->rcv_dsack_block.left); 1198 *lp++ = htonl(tp->rcv_dsack_block.right); 1199 tp->rcv_sack_flags &= ~TCPSACK_HAVED; 1200 } 1201 for (tiqe = TAILQ_FIRST(&tp->timeq); 1202 sack_numblks > 0; tiqe = TAILQ_NEXT(tiqe, ipqe_timeq)) { 1203 KASSERT(tiqe != NULL); 1204 sack_numblks--; 1205 *lp++ = htonl(tiqe->ipqe_seq); 1206 *lp++ = htonl(tiqe->ipqe_seq + tiqe->ipqe_len + 1207 ((tiqe->ipqe_flags & TH_FIN) != 0 ? 1 : 0)); 1208 } 1209 optlen += sack_len + 2; 1210 } 1211 TCP_REASS_UNLOCK(tp); 1212 1213 #ifdef TCP_SIGNATURE 1214 if (tp->t_flags & TF_SIGNATURE) { 1215 u_char *bp; 1216 /* 1217 * Initialize TCP-MD5 option (RFC2385) 1218 */ 1219 bp = (u_char *)opt + optlen; 1220 *bp++ = TCPOPT_SIGNATURE; 1221 *bp++ = TCPOLEN_SIGNATURE; 1222 sigoff = optlen + 2; 1223 memset(bp, 0, TCP_SIGLEN); 1224 bp += TCP_SIGLEN; 1225 optlen += TCPOLEN_SIGNATURE; 1226 /* 1227 * Terminate options list and maintain 32-bit alignment. 1228 */ 1229 *bp++ = TCPOPT_NOP; 1230 *bp++ = TCPOPT_EOL; 1231 optlen += 2; 1232 } 1233 #endif /* TCP_SIGNATURE */ 1234 1235 hdrlen += optlen; 1236 1237 #ifdef DIAGNOSTIC 1238 if (!use_tso && len > txsegsize) 1239 panic("tcp data to be sent is larger than segment"); 1240 else if (use_tso && len > IP_MAXPACKET) 1241 panic("tcp data to be sent is larger than max TSO size"); 1242 if (max_linkhdr + hdrlen > MCLBYTES) 1243 panic("tcphdr too big"); 1244 #endif 1245 1246 /* 1247 * Grab a header mbuf, attaching a copy of data to 1248 * be transmitted, and initialize the header from 1249 * the template for sends on this connection. 1250 */ 1251 if (len) { 1252 error = tcp_build_datapkt(tp, so, off, len, hdrlen, &m); 1253 if (error) 1254 goto out; 1255 /* 1256 * If we're sending everything we've got, set PUSH. 1257 * (This will keep happy those implementations which only 1258 * give data to the user when a buffer fills or 1259 * a PUSH comes in.) 1260 */ 1261 if (off + len == so->so_snd.sb_cc) 1262 flags |= TH_PUSH; 1263 } else { 1264 tcps = TCP_STAT_GETREF(); 1265 if (tp->t_flags & TF_ACKNOW) 1266 tcps[TCP_STAT_SNDACKS]++; 1267 else if (flags & (TH_SYN|TH_FIN|TH_RST)) 1268 tcps[TCP_STAT_SNDCTRL]++; 1269 else if (SEQ_GT(tp->snd_up, tp->snd_una)) 1270 tcps[TCP_STAT_SNDURG]++; 1271 else 1272 tcps[TCP_STAT_SNDWINUP]++; 1273 TCP_STAT_PUTREF(); 1274 1275 MGETHDR(m, M_DONTWAIT, MT_HEADER); 1276 if (m != NULL && max_linkhdr + hdrlen > MHLEN) { 1277 MCLGET(m, M_DONTWAIT); 1278 if ((m->m_flags & M_EXT) == 0) { 1279 m_freem(m); 1280 m = NULL; 1281 } 1282 } 1283 if (m == NULL) { 1284 error = ENOBUFS; 1285 goto out; 1286 } 1287 MCLAIM(m, &tcp_tx_mowner); 1288 m->m_data += max_linkhdr; 1289 m->m_len = hdrlen; 1290 } 1291 m->m_pkthdr.rcvif = (struct ifnet *)0; 1292 switch (af) { 1293 #ifdef INET 1294 case AF_INET: 1295 ip = mtod(m, struct ip *); 1296 #ifdef INET6 1297 ip6 = NULL; 1298 #endif 1299 th = (struct tcphdr *)(ip + 1); 1300 break; 1301 #endif 1302 #ifdef INET6 1303 case AF_INET6: 1304 ip = NULL; 1305 ip6 = mtod(m, struct ip6_hdr *); 1306 th = (struct tcphdr *)(ip6 + 1); 1307 break; 1308 #endif 1309 default: /*pacify gcc*/ 1310 ip = NULL; 1311 #ifdef INET6 1312 ip6 = NULL; 1313 #endif 1314 th = NULL; 1315 break; 1316 } 1317 if (tp->t_template == 0) 1318 panic("tcp_output"); 1319 if (tp->t_template->m_len < iphdrlen) 1320 panic("tcp_output"); 1321 bcopy(mtod(tp->t_template, void *), mtod(m, void *), iphdrlen); 1322 1323 /* 1324 * If we are starting a connection, send ECN setup 1325 * SYN packet. If we are on a retransmit, we may 1326 * resend those bits a number of times as per 1327 * RFC 3168. 1328 */ 1329 if (tp->t_state == TCPS_SYN_SENT && tcp_do_ecn) { 1330 if (tp->t_flags & TF_SYN_REXMT) { 1331 if (tp->t_ecn_retries--) 1332 flags |= TH_ECE|TH_CWR; 1333 } else { 1334 flags |= TH_ECE|TH_CWR; 1335 tp->t_ecn_retries = tcp_ecn_maxretries; 1336 } 1337 } 1338 1339 if (TCP_ECN_ALLOWED(tp)) { 1340 /* 1341 * If the peer has ECN, mark data packets 1342 * ECN capable. Ignore pure ack packets, retransmissions 1343 * and window probes. 1344 */ 1345 if (len > 0 && SEQ_GEQ(tp->snd_nxt, tp->snd_max) && 1346 !(tp->t_force && len == 1)) { 1347 switch (af) { 1348 #ifdef INET 1349 case AF_INET: 1350 tp->t_inpcb->inp_ip.ip_tos |= IPTOS_ECN_ECT0; 1351 break; 1352 #endif 1353 #ifdef INET6 1354 case AF_INET6: 1355 ip6->ip6_flow |= htonl(IPTOS_ECN_ECT0 << 20); 1356 break; 1357 #endif 1358 } 1359 TCP_STATINC(TCP_STAT_ECN_ECT); 1360 } 1361 1362 /* 1363 * Reply with proper ECN notifications. 1364 */ 1365 if (tp->t_flags & TF_ECN_SND_CWR) { 1366 flags |= TH_CWR; 1367 tp->t_flags &= ~TF_ECN_SND_CWR; 1368 } 1369 if (tp->t_flags & TF_ECN_SND_ECE) { 1370 flags |= TH_ECE; 1371 } 1372 } 1373 1374 1375 /* 1376 * If we are doing retransmissions, then snd_nxt will 1377 * not reflect the first unsent octet. For ACK only 1378 * packets, we do not want the sequence number of the 1379 * retransmitted packet, we want the sequence number 1380 * of the next unsent octet. So, if there is no data 1381 * (and no SYN or FIN), use snd_max instead of snd_nxt 1382 * when filling in ti_seq. But if we are in persist 1383 * state, snd_max might reflect one byte beyond the 1384 * right edge of the window, so use snd_nxt in that 1385 * case, since we know we aren't doing a retransmission. 1386 * (retransmit and persist are mutually exclusive...) 1387 */ 1388 if (TCP_SACK_ENABLED(tp) && sack_rxmit) { 1389 th->th_seq = htonl(p->rxmit); 1390 p->rxmit += len; 1391 } else { 1392 if (len || (flags & (TH_SYN|TH_FIN)) || 1393 TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) 1394 th->th_seq = htonl(tp->snd_nxt); 1395 else 1396 th->th_seq = htonl(tp->snd_max); 1397 } 1398 th->th_ack = htonl(tp->rcv_nxt); 1399 if (optlen) { 1400 bcopy((void *)opt, (void *)(th + 1), optlen); 1401 th->th_off = (sizeof (struct tcphdr) + optlen) >> 2; 1402 } 1403 th->th_flags = flags; 1404 /* 1405 * Calculate receive window. Don't shrink window, 1406 * but avoid silly window syndrome. 1407 */ 1408 if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)rxsegsize) 1409 win = 0; 1410 if (win > (long)TCP_MAXWIN << tp->rcv_scale) 1411 win = (long)TCP_MAXWIN << tp->rcv_scale; 1412 if (win < (long)(int32_t)(tp->rcv_adv - tp->rcv_nxt)) 1413 win = (long)(int32_t)(tp->rcv_adv - tp->rcv_nxt); 1414 th->th_win = htons((u_int16_t) (win>>tp->rcv_scale)); 1415 if (SEQ_GT(tp->snd_up, tp->snd_nxt)) { 1416 u_int32_t urp = tp->snd_up - tp->snd_nxt; 1417 if (urp > IP_MAXPACKET) 1418 urp = IP_MAXPACKET; 1419 th->th_urp = htons((u_int16_t)urp); 1420 th->th_flags |= TH_URG; 1421 } else 1422 /* 1423 * If no urgent pointer to send, then we pull 1424 * the urgent pointer to the left edge of the send window 1425 * so that it doesn't drift into the send window on sequence 1426 * number wraparound. 1427 */ 1428 tp->snd_up = tp->snd_una; /* drag it along */ 1429 1430 #ifdef TCP_SIGNATURE 1431 if (sigoff && (tp->t_flags & TF_SIGNATURE)) { 1432 struct secasvar *sav; 1433 u_int8_t *sigp; 1434 1435 sav = tcp_signature_getsav(m, th); 1436 1437 if (sav == NULL) { 1438 if (m) 1439 m_freem(m); 1440 return (EPERM); 1441 } 1442 1443 m->m_pkthdr.len = hdrlen + len; 1444 sigp = (char *)th + sizeof(*th) + sigoff; 1445 tcp_signature(m, th, (char *)th - mtod(m, char *), sav, sigp); 1446 1447 key_sa_recordxfer(sav, m); 1448 #ifdef FAST_IPSEC 1449 KEY_FREESAV(&sav); 1450 #else 1451 key_freesav(sav); 1452 #endif 1453 } 1454 #endif 1455 1456 /* 1457 * Set ourselves up to be checksummed just before the packet 1458 * hits the wire. 1459 */ 1460 switch (af) { 1461 #ifdef INET 1462 case AF_INET: 1463 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 1464 if (use_tso) { 1465 m->m_pkthdr.segsz = txsegsize; 1466 m->m_pkthdr.csum_flags = M_CSUM_TSOv4; 1467 } else { 1468 m->m_pkthdr.csum_flags = M_CSUM_TCPv4; 1469 if (len + optlen) { 1470 /* Fixup the pseudo-header checksum. */ 1471 /* XXXJRT Not IP Jumbogram safe. */ 1472 th->th_sum = in_cksum_addword(th->th_sum, 1473 htons((u_int16_t) (len + optlen))); 1474 } 1475 } 1476 break; 1477 #endif 1478 #ifdef INET6 1479 case AF_INET6: 1480 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 1481 if (use_tso) { 1482 m->m_pkthdr.segsz = txsegsize; 1483 m->m_pkthdr.csum_flags = M_CSUM_TSOv6; 1484 } else { 1485 m->m_pkthdr.csum_flags = M_CSUM_TCPv6; 1486 if (len + optlen) { 1487 /* Fixup the pseudo-header checksum. */ 1488 /* XXXJRT: Not IPv6 Jumbogram safe. */ 1489 th->th_sum = in_cksum_addword(th->th_sum, 1490 htons((u_int16_t) (len + optlen))); 1491 } 1492 } 1493 break; 1494 #endif 1495 } 1496 1497 /* 1498 * In transmit state, time the transmission and arrange for 1499 * the retransmit. In persist state, just set snd_max. 1500 */ 1501 if (tp->t_force == 0 || TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 1502 tcp_seq startseq = tp->snd_nxt; 1503 1504 /* 1505 * Advance snd_nxt over sequence space of this segment. 1506 * There are no states in which we send both a SYN and a FIN, 1507 * so we collapse the tests for these flags. 1508 */ 1509 if (flags & (TH_SYN|TH_FIN)) 1510 tp->snd_nxt++; 1511 if (sack_rxmit) 1512 goto timer; 1513 tp->snd_nxt += len; 1514 if (SEQ_GT(tp->snd_nxt, tp->snd_max)) { 1515 tp->snd_max = tp->snd_nxt; 1516 /* 1517 * Time this transmission if not a retransmission and 1518 * not currently timing anything. 1519 */ 1520 if (tp->t_rtttime == 0) { 1521 tp->t_rtttime = tcp_now; 1522 tp->t_rtseq = startseq; 1523 TCP_STATINC(TCP_STAT_SEGSTIMED); 1524 } 1525 } 1526 1527 /* 1528 * Set retransmit timer if not currently set, 1529 * and not doing an ack or a keep-alive probe. 1530 * Initial value for retransmit timer is smoothed 1531 * round-trip time + 2 * round-trip time variance. 1532 * Initialize shift counter which is used for backoff 1533 * of retransmit time. 1534 */ 1535 timer: 1536 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 1537 ((sack_rxmit && tp->snd_nxt != tp->snd_max) || 1538 tp->snd_nxt != tp->snd_una)) { 1539 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) { 1540 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 1541 tp->t_rxtshift = 0; 1542 } 1543 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 1544 } 1545 } else 1546 if (SEQ_GT(tp->snd_nxt + len, tp->snd_max)) 1547 tp->snd_max = tp->snd_nxt + len; 1548 1549 #ifdef TCP_DEBUG 1550 /* 1551 * Trace. 1552 */ 1553 if (so->so_options & SO_DEBUG) 1554 tcp_trace(TA_OUTPUT, tp->t_state, tp, m, 0); 1555 #endif 1556 1557 /* 1558 * Fill in IP length and desired time to live and 1559 * send to IP level. There should be a better way 1560 * to handle ttl and tos; we could keep them in 1561 * the template, but need a way to checksum without them. 1562 */ 1563 m->m_pkthdr.len = hdrlen + len; 1564 1565 switch (af) { 1566 #ifdef INET 1567 case AF_INET: 1568 ip->ip_len = htons(m->m_pkthdr.len); 1569 packetlen = m->m_pkthdr.len; 1570 if (tp->t_inpcb) { 1571 ip->ip_ttl = tp->t_inpcb->inp_ip.ip_ttl; 1572 ip->ip_tos = tp->t_inpcb->inp_ip.ip_tos; 1573 } 1574 #ifdef INET6 1575 else if (tp->t_in6pcb) { 1576 ip->ip_ttl = in6_selecthlim(tp->t_in6pcb, NULL); /*XXX*/ 1577 ip->ip_tos = 0; /*XXX*/ 1578 } 1579 #endif 1580 break; 1581 #endif 1582 #ifdef INET6 1583 case AF_INET6: 1584 packetlen = m->m_pkthdr.len; 1585 ip6->ip6_nxt = IPPROTO_TCP; 1586 if (tp->t_in6pcb) { 1587 /* 1588 * we separately set hoplimit for every segment, since 1589 * the user might want to change the value via 1590 * setsockopt. Also, desired default hop limit might 1591 * be changed via Neighbor Discovery. 1592 */ 1593 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, 1594 (rt = rtcache_validate(ro)) != NULL ? rt->rt_ifp 1595 : NULL); 1596 } 1597 /* ip6->ip6_flow = ??? */ 1598 /* ip6_plen will be filled in ip6_output(). */ 1599 break; 1600 #endif 1601 default: /*pacify gcc*/ 1602 packetlen = 0; 1603 break; 1604 } 1605 1606 switch (af) { 1607 #ifdef INET 1608 case AF_INET: 1609 { 1610 struct mbuf *opts; 1611 1612 if (tp->t_inpcb) 1613 opts = tp->t_inpcb->inp_options; 1614 else 1615 opts = NULL; 1616 error = ip_output(m, opts, ro, 1617 (tp->t_mtudisc ? IP_MTUDISC : 0) | 1618 (so->so_options & SO_DONTROUTE), 1619 (struct ip_moptions *)0, so); 1620 break; 1621 } 1622 #endif 1623 #ifdef INET6 1624 case AF_INET6: 1625 { 1626 struct ip6_pktopts *opts; 1627 1628 if (tp->t_in6pcb) 1629 opts = tp->t_in6pcb->in6p_outputopts; 1630 else 1631 opts = NULL; 1632 error = ip6_output(m, opts, ro, so->so_options & SO_DONTROUTE, 1633 NULL, so, NULL); 1634 break; 1635 } 1636 #endif 1637 default: 1638 error = EAFNOSUPPORT; 1639 break; 1640 } 1641 if (error) { 1642 out: 1643 if (error == ENOBUFS) { 1644 TCP_STATINC(TCP_STAT_SELFQUENCH); 1645 #ifdef INET 1646 if (tp->t_inpcb) 1647 tcp_quench(tp->t_inpcb, 0); 1648 #endif 1649 #ifdef INET6 1650 if (tp->t_in6pcb) 1651 tcp6_quench(tp->t_in6pcb, 0); 1652 #endif 1653 error = 0; 1654 } else if ((error == EHOSTUNREACH || error == ENETDOWN) && 1655 TCPS_HAVERCVDSYN(tp->t_state)) { 1656 tp->t_softerror = error; 1657 error = 0; 1658 } 1659 1660 /* Back out the seqence number advance. */ 1661 if (sack_rxmit) 1662 p->rxmit -= len; 1663 1664 /* Restart the delayed ACK timer, if necessary. */ 1665 if (tp->t_flags & TF_DELACK) 1666 TCP_RESTART_DELACK(tp); 1667 1668 return (error); 1669 } 1670 1671 if (packetlen > tp->t_pmtud_mtu_sent) 1672 tp->t_pmtud_mtu_sent = packetlen; 1673 1674 tcps = TCP_STAT_GETREF(); 1675 tcps[TCP_STAT_SNDTOTAL]++; 1676 if (tp->t_flags & TF_DELACK) 1677 tcps[TCP_STAT_DELACK]++; 1678 TCP_STAT_PUTREF(); 1679 1680 /* 1681 * Data sent (as far as we can tell). 1682 * If this advertises a larger window than any other segment, 1683 * then remember the size of the advertised window. 1684 * Any pending ACK has now been sent. 1685 */ 1686 if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv)) 1687 tp->rcv_adv = tp->rcv_nxt + win; 1688 tp->last_ack_sent = tp->rcv_nxt; 1689 tp->t_flags &= ~TF_ACKNOW; 1690 TCP_CLEAR_DELACK(tp); 1691 #ifdef DIAGNOSTIC 1692 if (maxburst < 0) 1693 printf("tcp_output: maxburst exceeded by %d\n", -maxburst); 1694 #endif 1695 if (sendalot && (tp->t_congctl == &tcp_reno_ctl || --maxburst)) 1696 goto again; 1697 return (0); 1698 } 1699 1700 void 1701 tcp_setpersist(struct tcpcb *tp) 1702 { 1703 int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2); 1704 int nticks; 1705 1706 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT)) 1707 panic("tcp_output REXMT"); 1708 /* 1709 * Start/restart persistance timer. 1710 */ 1711 if (t < tp->t_rttmin) 1712 t = tp->t_rttmin; 1713 TCPT_RANGESET(nticks, t * tcp_backoff[tp->t_rxtshift], 1714 TCPTV_PERSMIN, TCPTV_PERSMAX); 1715 TCP_TIMER_ARM(tp, TCPT_PERSIST, nticks); 1716 if (tp->t_rxtshift < TCP_MAXRXTSHIFT) 1717 tp->t_rxtshift++; 1718 } 1719