tcp_output.c revision 1.63.2.5
1 /* $NetBSD: tcp_output.c,v 1.63.2.5 2001/11/14 19:17:57 nathanw Exp $ */ 2 3 /* 4 %%% portions-copyright-nrl-95 5 Portions of this software are Copyright 1995-1998 by Randall Atkinson, 6 Ronald Lee, Daniel McDonald, Bao Phan, and Chris Winters. All Rights 7 Reserved. All rights under this copyright have been assigned to the US 8 Naval Research Laboratory (NRL). The NRL Copyright Notice and License 9 Agreement Version 1.1 (January 17, 1995) applies to these portions of the 10 software. 11 You should have received a copy of the license with this software. If you 12 didn't get a copy, you may request one from <license (at) ipv6.nrl.navy.mil>. 13 14 */ 15 16 /* 17 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 18 * All rights reserved. 19 * 20 * Redistribution and use in source and binary forms, with or without 21 * modification, are permitted provided that the following conditions 22 * are met: 23 * 1. Redistributions of source code must retain the above copyright 24 * notice, this list of conditions and the following disclaimer. 25 * 2. Redistributions in binary form must reproduce the above copyright 26 * notice, this list of conditions and the following disclaimer in the 27 * documentation and/or other materials provided with the distribution. 28 * 3. Neither the name of the project nor the names of its contributors 29 * may be used to endorse or promote products derived from this software 30 * without specific prior written permission. 31 * 32 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 33 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 34 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 35 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 36 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 37 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 38 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 39 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 40 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 41 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 42 * SUCH DAMAGE. 43 */ 44 45 /*- 46 * Copyright (c) 1997, 1998, 2001 The NetBSD Foundation, Inc. 47 * All rights reserved. 48 * 49 * This code is derived from software contributed to The NetBSD Foundation 50 * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation 51 * Facility, NASA Ames Research Center. 52 * 53 * Redistribution and use in source and binary forms, with or without 54 * modification, are permitted provided that the following conditions 55 * are met: 56 * 1. Redistributions of source code must retain the above copyright 57 * notice, this list of conditions and the following disclaimer. 58 * 2. Redistributions in binary form must reproduce the above copyright 59 * notice, this list of conditions and the following disclaimer in the 60 * documentation and/or other materials provided with the distribution. 61 * 3. All advertising materials mentioning features or use of this software 62 * must display the following acknowledgement: 63 * This product includes software developed by the NetBSD 64 * Foundation, Inc. and its contributors. 65 * 4. Neither the name of The NetBSD Foundation nor the names of its 66 * contributors may be used to endorse or promote products derived 67 * from this software without specific prior written permission. 68 * 69 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 70 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 71 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 72 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 73 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 74 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 75 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 76 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 77 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 78 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 79 * POSSIBILITY OF SUCH DAMAGE. 80 */ 81 82 /* 83 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 84 * The Regents of the University of California. All rights reserved. 85 * 86 * Redistribution and use in source and binary forms, with or without 87 * modification, are permitted provided that the following conditions 88 * are met: 89 * 1. Redistributions of source code must retain the above copyright 90 * notice, this list of conditions and the following disclaimer. 91 * 2. Redistributions in binary form must reproduce the above copyright 92 * notice, this list of conditions and the following disclaimer in the 93 * documentation and/or other materials provided with the distribution. 94 * 3. All advertising materials mentioning features or use of this software 95 * must display the following acknowledgement: 96 * This product includes software developed by the University of 97 * California, Berkeley and its contributors. 98 * 4. Neither the name of the University nor the names of its contributors 99 * may be used to endorse or promote products derived from this software 100 * without specific prior written permission. 101 * 102 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 103 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 104 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 105 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 106 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 107 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 108 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 109 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 110 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 111 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 112 * SUCH DAMAGE. 113 * 114 * @(#)tcp_output.c 8.4 (Berkeley) 5/24/95 115 */ 116 117 #include <sys/cdefs.h> 118 __KERNEL_RCSID(0, "$NetBSD: tcp_output.c,v 1.63.2.5 2001/11/14 19:17:57 nathanw Exp $"); 119 120 #include "opt_inet.h" 121 #include "opt_ipsec.h" 122 #include "opt_tcp_debug.h" 123 124 #include <sys/param.h> 125 #include <sys/systm.h> 126 #include <sys/malloc.h> 127 #include <sys/mbuf.h> 128 #include <sys/protosw.h> 129 #include <sys/socket.h> 130 #include <sys/socketvar.h> 131 #include <sys/errno.h> 132 #include <sys/domain.h> 133 #include <sys/kernel.h> 134 135 #include <net/if.h> 136 #include <net/route.h> 137 138 #include <netinet/in.h> 139 #include <netinet/in_systm.h> 140 #include <netinet/ip.h> 141 #include <netinet/in_pcb.h> 142 #include <netinet/ip_var.h> 143 144 #ifdef INET6 145 #ifndef INET 146 #include <netinet/in.h> 147 #endif 148 #include <netinet/ip6.h> 149 #include <netinet6/in6_pcb.h> 150 #include <netinet6/ip6_var.h> 151 #endif 152 153 #include <netinet/tcp.h> 154 #define TCPOUTFLAGS 155 #include <netinet/tcp_fsm.h> 156 #include <netinet/tcp_seq.h> 157 #include <netinet/tcp_timer.h> 158 #include <netinet/tcp_var.h> 159 #include <netinet/tcpip.h> 160 #include <netinet/tcp_debug.h> 161 162 #ifdef notyet 163 extern struct mbuf *m_copypack(); 164 #endif 165 166 #define MAX_TCPOPTLEN 32 /* max # bytes that go in options */ 167 168 /* 169 * Knob to enable Congestion Window Monitoring, and control the 170 * the burst size it allows. Default burst is 4 packets, per 171 * the Internet draft. 172 */ 173 int tcp_cwm = 1; 174 int tcp_cwm_burstsize = 4; 175 176 static 177 #ifndef GPROF 178 __inline 179 #endif 180 void 181 tcp_segsize(struct tcpcb *tp, int *txsegsizep, int *rxsegsizep) 182 { 183 #ifdef INET 184 struct inpcb *inp = tp->t_inpcb; 185 #endif 186 #ifdef INET6 187 struct in6pcb *in6p = tp->t_in6pcb; 188 #endif 189 struct rtentry *rt; 190 struct ifnet *ifp; 191 int size; 192 int iphlen; 193 194 #ifdef DIAGNOSTIC 195 if (tp->t_inpcb && tp->t_in6pcb) 196 panic("tcp_segsize: both t_inpcb and t_in6pcb are set"); 197 #endif 198 switch (tp->t_family) { 199 #ifdef INET 200 case AF_INET: 201 iphlen = sizeof(struct ip); 202 break; 203 #endif 204 #ifdef INET6 205 case AF_INET6: 206 iphlen = sizeof(struct ip6_hdr); 207 break; 208 #endif 209 default: 210 size = tcp_mssdflt; 211 goto out; 212 } 213 214 rt = NULL; 215 #ifdef INET 216 if (inp) 217 rt = in_pcbrtentry(inp); 218 #endif 219 #ifdef INET6 220 if (in6p) 221 rt = in6_pcbrtentry(in6p); 222 #endif 223 if (rt == NULL) { 224 size = tcp_mssdflt; 225 goto out; 226 } 227 228 ifp = rt->rt_ifp; 229 230 size = tcp_mssdflt; 231 if (rt->rt_rmx.rmx_mtu != 0) 232 size = rt->rt_rmx.rmx_mtu - iphlen - sizeof(struct tcphdr); 233 else if (ifp->if_flags & IFF_LOOPBACK) 234 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 235 #ifdef INET 236 else if (inp && ip_mtudisc) 237 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 238 else if (inp && in_localaddr(inp->inp_faddr)) 239 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 240 #endif 241 #ifdef INET6 242 else if (in6p) { 243 #ifdef INET 244 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 245 /* mapped addr case */ 246 struct in_addr d; 247 bcopy(&in6p->in6p_faddr.s6_addr32[3], &d, sizeof(d)); 248 if (ip_mtudisc || in_localaddr(d)) 249 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 250 } else 251 #endif 252 { 253 /* 254 * for IPv6, path MTU discovery is always turned on, 255 * or the node must use packet size <= 1280. 256 */ 257 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 258 } 259 } 260 #endif 261 size -= tcp_optlen(tp); 262 /* 263 * XXX tp->t_ourmss should have the right size, but without this code 264 * fragmentation will occur... need more investigation 265 */ 266 #ifdef INET 267 if (inp) { 268 #ifdef IPSEC 269 size -= ipsec4_hdrsiz_tcp(tp); 270 #endif 271 size -= ip_optlen(inp); 272 } 273 #endif 274 #ifdef INET6 275 #ifdef INET 276 if (in6p && tp->t_family == AF_INET) { 277 #ifdef IPSEC 278 size -= ipsec4_hdrsiz_tcp(tp); 279 #endif 280 /* XXX size -= ip_optlen(in6p); */ 281 } else 282 #endif 283 if (in6p && tp->t_family == AF_INET6) { 284 #ifdef IPSEC 285 size -= ipsec6_hdrsiz_tcp(tp); 286 #endif 287 size -= ip6_optlen(in6p); 288 } 289 #endif 290 291 out: 292 /* 293 * *rxsegsizep holds *estimated* inbound segment size (estimation 294 * assumes that path MTU is the same for both ways). this is only 295 * for silly window avoidance, do not use the value for other purposes. 296 * 297 * ipseclen is subtracted from both sides, this may not be right. 298 * I'm not quite sure about this (could someone comment). 299 */ 300 *txsegsizep = min(tp->t_peermss, size); 301 *rxsegsizep = min(tp->t_ourmss, size); 302 303 if (*txsegsizep != tp->t_segsz) { 304 /* 305 * If the new segment size is larger, we don't want to 306 * mess up the congestion window, but if it is smaller 307 * we'll have to reduce the congestion window to ensure 308 * that we don't get into trouble with initial windows 309 * and the rest. In any case, if the segment size 310 * has changed, chances are the path has, too, and 311 * our congestion window will be different. 312 */ 313 if (*txsegsizep < tp->t_segsz) { 314 tp->snd_cwnd = max((tp->snd_cwnd / tp->t_segsz) 315 * *txsegsizep, *txsegsizep); 316 tp->snd_ssthresh = max((tp->snd_ssthresh / tp->t_segsz) 317 * *txsegsizep, *txsegsizep); 318 } 319 tp->t_segsz = *txsegsizep; 320 } 321 } 322 323 static 324 #ifndef GPROF 325 __inline 326 #endif 327 int 328 tcp_build_datapkt(struct tcpcb *tp, struct socket *so, int off, 329 long len, int hdrlen, struct mbuf **mp) 330 { 331 struct mbuf *m; 332 333 if (tp->t_force && len == 1) 334 tcpstat.tcps_sndprobe++; 335 else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) { 336 tcpstat.tcps_sndrexmitpack++; 337 tcpstat.tcps_sndrexmitbyte += len; 338 } else { 339 tcpstat.tcps_sndpack++; 340 tcpstat.tcps_sndbyte += len; 341 } 342 #ifdef notyet 343 if ((m = m_copypack(so->so_snd.sb_mb, off, 344 (int)len, max_linkhdr + hdrlen)) == 0) 345 return (ENOBUFS); 346 /* 347 * m_copypack left space for our hdr; use it. 348 */ 349 m->m_len += hdrlen; 350 m->m_data -= hdrlen; 351 #else 352 MGETHDR(m, M_DONTWAIT, MT_HEADER); 353 if (m != NULL && 354 (max_linkhdr + hdrlen > MHLEN || 355 max_linkhdr + hdrlen + len <= MCLBYTES)) { 356 MCLGET(m, M_DONTWAIT); 357 if ((m->m_flags & M_EXT) == 0) { 358 m_freem(m); 359 m = NULL; 360 } 361 } 362 if (m == NULL) 363 return (ENOBUFS); 364 m->m_data += max_linkhdr; 365 m->m_len = hdrlen; 366 if (len <= M_TRAILINGSPACE(m)) { 367 m_copydata(so->so_snd.sb_mb, off, (int) len, 368 mtod(m, caddr_t) + hdrlen); 369 m->m_len += len; 370 } else { 371 m->m_next = m_copy(so->so_snd.sb_mb, off, (int) len); 372 if (m->m_next == NULL) { 373 m_freem(m); 374 return (ENOBUFS); 375 } 376 } 377 #endif 378 379 *mp = m; 380 return (0); 381 } 382 383 /* 384 * Tcp output routine: figure out what should be sent and send it. 385 */ 386 int 387 tcp_output(tp) 388 struct tcpcb *tp; 389 { 390 struct socket *so; 391 struct route *ro; 392 long len, win; 393 int off, flags, error; 394 struct mbuf *m; 395 struct ip *ip; 396 #ifdef INET6 397 struct ip6_hdr *ip6; 398 #endif 399 struct tcphdr *th; 400 u_char opt[MAX_TCPOPTLEN]; 401 unsigned optlen, hdrlen; 402 int idle, sendalot, txsegsize, rxsegsize; 403 int maxburst = TCP_MAXBURST; 404 int af; /* address family on the wire */ 405 int iphdrlen; 406 407 #ifdef DIAGNOSTIC 408 if (tp->t_inpcb && tp->t_in6pcb) 409 panic("tcp_output: both t_inpcb and t_in6pcb are set"); 410 #endif 411 so = NULL; 412 ro = NULL; 413 if (tp->t_inpcb) { 414 so = tp->t_inpcb->inp_socket; 415 ro = &tp->t_inpcb->inp_route; 416 } 417 #ifdef INET6 418 else if (tp->t_in6pcb) { 419 so = tp->t_in6pcb->in6p_socket; 420 ro = (struct route *)&tp->t_in6pcb->in6p_route; 421 } 422 #endif 423 424 switch (af = tp->t_family) { 425 #ifdef INET 426 case AF_INET: 427 if (tp->t_inpcb) 428 break; 429 #ifdef INET6 430 /* mapped addr case */ 431 if (tp->t_in6pcb) 432 break; 433 #endif 434 return EINVAL; 435 #endif 436 #ifdef INET6 437 case AF_INET6: 438 if (tp->t_in6pcb) 439 break; 440 return EINVAL; 441 #endif 442 default: 443 return EAFNOSUPPORT; 444 } 445 446 tcp_segsize(tp, &txsegsize, &rxsegsize); 447 448 idle = (tp->snd_max == tp->snd_una); 449 450 /* 451 * Restart Window computation. From draft-floyd-incr-init-win-03: 452 * 453 * Optionally, a TCP MAY set the restart window to the 454 * minimum of the value used for the initial window and 455 * the current value of cwnd (in other words, using a 456 * larger value for the restart window should never increase 457 * the size of cwnd). 458 */ 459 if (tcp_cwm) { 460 /* 461 * Hughes/Touch/Heidemann Congestion Window Monitoring. 462 * Count the number of packets currently pending 463 * acknowledgement, and limit our congestion window 464 * to a pre-determined allowed burst size plus that count. 465 * This prevents bursting once all pending packets have 466 * been acknowledged (i.e. transmission is idle). 467 * 468 * XXX Link this to Initial Window? 469 */ 470 tp->snd_cwnd = min(tp->snd_cwnd, 471 (tcp_cwm_burstsize * txsegsize) + 472 (tp->snd_nxt - tp->snd_una)); 473 } else { 474 if (idle && (tcp_now - tp->t_rcvtime) >= tp->t_rxtcur) { 475 /* 476 * We have been idle for "a while" and no acks are 477 * expected to clock out any data we send -- 478 * slow start to get ack "clock" running again. 479 */ 480 tp->snd_cwnd = min(tp->snd_cwnd, 481 TCP_INITIAL_WINDOW(tcp_init_win, txsegsize)); 482 } 483 } 484 485 again: 486 /* 487 * Determine length of data that should be transmitted, and 488 * flags that should be used. If there is some data or critical 489 * controls (SYN, RST) to send, then transmit; otherwise, 490 * investigate further. 491 */ 492 sendalot = 0; 493 off = tp->snd_nxt - tp->snd_una; 494 win = min(tp->snd_wnd, tp->snd_cwnd); 495 496 flags = tcp_outflags[tp->t_state]; 497 /* 498 * If in persist timeout with window of 0, send 1 byte. 499 * Otherwise, if window is small but nonzero 500 * and timer expired, we will send what we can 501 * and go to transmit state. 502 */ 503 if (tp->t_force) { 504 if (win == 0) { 505 /* 506 * If we still have some data to send, then 507 * clear the FIN bit. Usually this would 508 * happen below when it realizes that we 509 * aren't sending all the data. However, 510 * if we have exactly 1 byte of unset data, 511 * then it won't clear the FIN bit below, 512 * and if we are in persist state, we wind 513 * up sending the packet without recording 514 * that we sent the FIN bit. 515 * 516 * We can't just blindly clear the FIN bit, 517 * because if we don't have any more data 518 * to send then the probe will be the FIN 519 * itself. 520 */ 521 if (off < so->so_snd.sb_cc) 522 flags &= ~TH_FIN; 523 win = 1; 524 } else { 525 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 526 tp->t_rxtshift = 0; 527 } 528 } 529 530 if (win < so->so_snd.sb_cc) { 531 len = win - off; 532 flags &= ~TH_FIN; 533 } else 534 len = so->so_snd.sb_cc - off; 535 536 if (len < 0) { 537 /* 538 * If FIN has been sent but not acked, 539 * but we haven't been called to retransmit, 540 * len will be -1. Otherwise, window shrank 541 * after we sent into it. If window shrank to 0, 542 * cancel pending retransmit, pull snd_nxt back 543 * to (closed) window, and set the persist timer 544 * if it isn't already going. If the window didn't 545 * close completely, just wait for an ACK. 546 * 547 * If we have a pending FIN, either it has already been 548 * transmitted or it is outside the window, so drop it. 549 * If the FIN has been transmitted, but this is not a 550 * retransmission, then len must be -1. Therefore we also 551 * prevent here the sending of `gratuitous FINs'. This 552 * eliminates the need to check for that case below (e.g. 553 * to back up snd_nxt before the FIN so that the sequence 554 * number is correct). 555 */ 556 len = 0; 557 flags &= ~TH_FIN; 558 if (win == 0) { 559 TCP_TIMER_DISARM(tp, TCPT_REXMT); 560 tp->t_rxtshift = 0; 561 tp->snd_nxt = tp->snd_una; 562 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) 563 tcp_setpersist(tp); 564 } 565 } 566 if (len > txsegsize) { 567 len = txsegsize; 568 flags &= ~TH_FIN; 569 sendalot = 1; 570 } 571 572 win = sbspace(&so->so_rcv); 573 574 /* 575 * Sender silly window avoidance. If connection is idle 576 * and can send all data, a maximum segment, 577 * at least a maximum default-size segment do it, 578 * or are forced, do it; otherwise don't bother. 579 * If peer's buffer is tiny, then send 580 * when window is at least half open. 581 * If retransmitting (possibly after persist timer forced us 582 * to send into a small window), then must resend. 583 */ 584 if (len) { 585 if (len == txsegsize) 586 goto send; 587 if ((so->so_state & SS_MORETOCOME) == 0 && 588 ((idle || tp->t_flags & TF_NODELAY) && 589 len + off >= so->so_snd.sb_cc)) 590 goto send; 591 if (tp->t_force) 592 goto send; 593 if (len >= tp->max_sndwnd / 2) 594 goto send; 595 if (SEQ_LT(tp->snd_nxt, tp->snd_max)) 596 goto send; 597 } 598 599 /* 600 * Compare available window to amount of window known to peer 601 * (as advertised window less next expected input). If the 602 * difference is at least twice the size of the largest segment 603 * we expect to receive (i.e. two segments) or at least 50% of 604 * the maximum possible window, then want to send a window update 605 * to peer. 606 */ 607 if (win > 0) { 608 /* 609 * "adv" is the amount we can increase the window, 610 * taking into account that we are limited by 611 * TCP_MAXWIN << tp->rcv_scale. 612 */ 613 long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) - 614 (tp->rcv_adv - tp->rcv_nxt); 615 616 if (adv >= (long) (2 * rxsegsize)) 617 goto send; 618 if (2 * adv >= (long) so->so_rcv.sb_hiwat) 619 goto send; 620 } 621 622 /* 623 * Send if we owe peer an ACK. 624 */ 625 if (tp->t_flags & TF_ACKNOW) 626 goto send; 627 if (flags & (TH_SYN|TH_FIN|TH_RST)) 628 goto send; 629 if (SEQ_GT(tp->snd_up, tp->snd_una)) 630 goto send; 631 632 /* 633 * TCP window updates are not reliable, rather a polling protocol 634 * using ``persist'' packets is used to insure receipt of window 635 * updates. The three ``states'' for the output side are: 636 * idle not doing retransmits or persists 637 * persisting to move a small or zero window 638 * (re)transmitting and thereby not persisting 639 * 640 * tp->t_timer[TCPT_PERSIST] 641 * is set when we are in persist state. 642 * tp->t_force 643 * is set when we are called to send a persist packet. 644 * tp->t_timer[TCPT_REXMT] 645 * is set when we are retransmitting 646 * The output side is idle when both timers are zero. 647 * 648 * If send window is too small, there is data to transmit, and no 649 * retransmit or persist is pending, then go to persist state. 650 * If nothing happens soon, send when timer expires: 651 * if window is nonzero, transmit what we can, 652 * otherwise force out a byte. 653 */ 654 if (so->so_snd.sb_cc && TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 655 TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 656 tp->t_rxtshift = 0; 657 tcp_setpersist(tp); 658 } 659 660 /* 661 * No reason to send a segment, just return. 662 */ 663 return (0); 664 665 send: 666 /* 667 * Before ESTABLISHED, force sending of initial options 668 * unless TCP set not to do any options. 669 * NOTE: we assume that the IP/TCP header plus TCP options 670 * always fit in a single mbuf, leaving room for a maximum 671 * link header, i.e. 672 * max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MCLBYTES 673 */ 674 optlen = 0; 675 switch (af) { 676 #ifdef INET 677 case AF_INET: 678 iphdrlen = sizeof(struct ip) + sizeof(struct tcphdr); 679 break; 680 #endif 681 #ifdef INET6 682 case AF_INET6: 683 iphdrlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 684 break; 685 #endif 686 default: /*pacify gcc*/ 687 iphdrlen = 0; 688 break; 689 } 690 hdrlen = iphdrlen; 691 if (flags & TH_SYN) { 692 struct rtentry *rt; 693 694 rt = NULL; 695 #ifdef INET 696 if (tp->t_inpcb) 697 rt = in_pcbrtentry(tp->t_inpcb); 698 #endif 699 #ifdef INET6 700 if (tp->t_in6pcb) 701 rt = in6_pcbrtentry(tp->t_in6pcb); 702 #endif 703 704 tp->snd_nxt = tp->iss; 705 tp->t_ourmss = tcp_mss_to_advertise(rt != NULL ? 706 rt->rt_ifp : NULL, af); 707 if ((tp->t_flags & TF_NOOPT) == 0) { 708 opt[0] = TCPOPT_MAXSEG; 709 opt[1] = 4; 710 opt[2] = (tp->t_ourmss >> 8) & 0xff; 711 opt[3] = tp->t_ourmss & 0xff; 712 optlen = 4; 713 714 if ((tp->t_flags & TF_REQ_SCALE) && 715 ((flags & TH_ACK) == 0 || 716 (tp->t_flags & TF_RCVD_SCALE))) { 717 *((u_int32_t *) (opt + optlen)) = htonl( 718 TCPOPT_NOP << 24 | 719 TCPOPT_WINDOW << 16 | 720 TCPOLEN_WINDOW << 8 | 721 tp->request_r_scale); 722 optlen += 4; 723 } 724 } 725 } 726 727 /* 728 * Send a timestamp and echo-reply if this is a SYN and our side 729 * wants to use timestamps (TF_REQ_TSTMP is set) or both our side 730 * and our peer have sent timestamps in our SYN's. 731 */ 732 if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && 733 (flags & TH_RST) == 0 && 734 ((flags & (TH_SYN|TH_ACK)) == TH_SYN || 735 (tp->t_flags & TF_RCVD_TSTMP))) { 736 u_int32_t *lp = (u_int32_t *)(opt + optlen); 737 738 /* Form timestamp option as shown in appendix A of RFC 1323. */ 739 *lp++ = htonl(TCPOPT_TSTAMP_HDR); 740 *lp++ = htonl(TCP_TIMESTAMP(tp)); 741 *lp = htonl(tp->ts_recent); 742 optlen += TCPOLEN_TSTAMP_APPA; 743 } 744 745 hdrlen += optlen; 746 747 #ifdef DIAGNOSTIC 748 if (len > txsegsize) 749 panic("tcp data to be sent is larger than segment"); 750 if (max_linkhdr + hdrlen > MCLBYTES) 751 panic("tcphdr too big"); 752 #endif 753 754 /* 755 * Grab a header mbuf, attaching a copy of data to 756 * be transmitted, and initialize the header from 757 * the template for sends on this connection. 758 */ 759 if (len) { 760 error = tcp_build_datapkt(tp, so, off, len, hdrlen, &m); 761 if (error) 762 goto out; 763 /* 764 * If we're sending everything we've got, set PUSH. 765 * (This will keep happy those implementations which only 766 * give data to the user when a buffer fills or 767 * a PUSH comes in.) 768 */ 769 if (off + len == so->so_snd.sb_cc) 770 flags |= TH_PUSH; 771 } else { 772 if (tp->t_flags & TF_ACKNOW) 773 tcpstat.tcps_sndacks++; 774 else if (flags & (TH_SYN|TH_FIN|TH_RST)) 775 tcpstat.tcps_sndctrl++; 776 else if (SEQ_GT(tp->snd_up, tp->snd_una)) 777 tcpstat.tcps_sndurg++; 778 else 779 tcpstat.tcps_sndwinup++; 780 781 MGETHDR(m, M_DONTWAIT, MT_HEADER); 782 if (m != NULL && max_linkhdr + hdrlen > MHLEN) { 783 MCLGET(m, M_DONTWAIT); 784 if ((m->m_flags & M_EXT) == 0) { 785 m_freem(m); 786 m = NULL; 787 } 788 } 789 if (m == NULL) { 790 error = ENOBUFS; 791 goto out; 792 } 793 m->m_data += max_linkhdr; 794 m->m_len = hdrlen; 795 } 796 m->m_pkthdr.rcvif = (struct ifnet *)0; 797 switch (af) { 798 #ifdef INET 799 case AF_INET: 800 ip = mtod(m, struct ip *); 801 #ifdef INET6 802 ip6 = NULL; 803 #endif 804 th = (struct tcphdr *)(ip + 1); 805 break; 806 #endif 807 #ifdef INET6 808 case AF_INET6: 809 ip = NULL; 810 ip6 = mtod(m, struct ip6_hdr *); 811 th = (struct tcphdr *)(ip6 + 1); 812 break; 813 #endif 814 default: /*pacify gcc*/ 815 ip = NULL; 816 #ifdef INET6 817 ip6 = NULL; 818 #endif 819 th = NULL; 820 break; 821 } 822 if (tp->t_template == 0) 823 panic("tcp_output"); 824 if (tp->t_template->m_len < iphdrlen) 825 panic("tcp_output"); 826 bcopy(mtod(tp->t_template, caddr_t), mtod(m, caddr_t), iphdrlen); 827 828 /* 829 * If we are doing retransmissions, then snd_nxt will 830 * not reflect the first unsent octet. For ACK only 831 * packets, we do not want the sequence number of the 832 * retransmitted packet, we want the sequence number 833 * of the next unsent octet. So, if there is no data 834 * (and no SYN or FIN), use snd_max instead of snd_nxt 835 * when filling in ti_seq. But if we are in persist 836 * state, snd_max might reflect one byte beyond the 837 * right edge of the window, so use snd_nxt in that 838 * case, since we know we aren't doing a retransmission. 839 * (retransmit and persist are mutually exclusive...) 840 */ 841 if (len || (flags & (TH_SYN|TH_FIN)) || 842 TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) 843 th->th_seq = htonl(tp->snd_nxt); 844 else 845 th->th_seq = htonl(tp->snd_max); 846 th->th_ack = htonl(tp->rcv_nxt); 847 if (optlen) { 848 bcopy((caddr_t)opt, (caddr_t)(th + 1), optlen); 849 th->th_off = (sizeof (struct tcphdr) + optlen) >> 2; 850 } 851 th->th_flags = flags; 852 /* 853 * Calculate receive window. Don't shrink window, 854 * but avoid silly window syndrome. 855 */ 856 if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)rxsegsize) 857 win = 0; 858 if (win > (long)TCP_MAXWIN << tp->rcv_scale) 859 win = (long)TCP_MAXWIN << tp->rcv_scale; 860 if (win < (long)(tp->rcv_adv - tp->rcv_nxt)) 861 win = (long)(tp->rcv_adv - tp->rcv_nxt); 862 th->th_win = htons((u_int16_t) (win>>tp->rcv_scale)); 863 if (SEQ_GT(tp->snd_up, tp->snd_nxt)) { 864 u_int32_t urp = tp->snd_up - tp->snd_nxt; 865 if (urp > IP_MAXPACKET) 866 urp = IP_MAXPACKET; 867 th->th_urp = htons((u_int16_t)urp); 868 th->th_flags |= TH_URG; 869 } else 870 /* 871 * If no urgent pointer to send, then we pull 872 * the urgent pointer to the left edge of the send window 873 * so that it doesn't drift into the send window on sequence 874 * number wraparound. 875 */ 876 tp->snd_up = tp->snd_una; /* drag it along */ 877 878 /* 879 * Set ourselves up to be checksummed just before the packet 880 * hits the wire. 881 */ 882 switch (af) { 883 #ifdef INET 884 case AF_INET: 885 m->m_pkthdr.csum_flags = M_CSUM_TCPv4; 886 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 887 if (len + optlen) { 888 /* Fixup the pseudo-header checksum. */ 889 /* XXXJRT Not IP Jumbogram safe. */ 890 th->th_sum = in_cksum_addword(th->th_sum, 891 htons((u_int16_t) (len + optlen))); 892 } 893 break; 894 #endif 895 #ifdef INET6 896 case AF_INET6: 897 /* 898 * XXX Actually delaying the checksum is Hard 899 * XXX (well, maybe not for Itojun, but it is 900 * XXX for me), but we can still take advantage 901 * XXX of the cached pseudo-header checksum. 902 */ 903 /* equals to hdrlen + len */ 904 m->m_pkthdr.len = sizeof(struct ip6_hdr) 905 + sizeof(struct tcphdr) + optlen + len; 906 #ifdef notyet 907 m->m_pkthdr.csum_flags = M_CSUM_TCPv6; 908 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 909 #endif 910 if (len + optlen) { 911 /* Fixup the pseudo-header checksum. */ 912 /* XXXJRT: Not IPv6 Jumbogram safe. */ 913 th->th_sum = in_cksum_addword(th->th_sum, 914 htons((u_int16_t) (len + optlen))); 915 } 916 #ifndef notyet 917 th->th_sum = in6_cksum(m, 0, sizeof(struct ip6_hdr), 918 sizeof(struct tcphdr) + optlen + len); 919 #endif 920 break; 921 #endif 922 } 923 924 /* 925 * In transmit state, time the transmission and arrange for 926 * the retransmit. In persist state, just set snd_max. 927 */ 928 if (tp->t_force == 0 || TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 929 tcp_seq startseq = tp->snd_nxt; 930 931 /* 932 * Advance snd_nxt over sequence space of this segment. 933 * There are no states in which we send both a SYN and a FIN, 934 * so we collapse the tests for these flags. 935 */ 936 if (flags & (TH_SYN|TH_FIN)) 937 tp->snd_nxt++; 938 tp->snd_nxt += len; 939 if (SEQ_GT(tp->snd_nxt, tp->snd_max)) { 940 tp->snd_max = tp->snd_nxt; 941 /* 942 * Time this transmission if not a retransmission and 943 * not currently timing anything. 944 */ 945 if (tp->t_rtttime == 0) { 946 tp->t_rtttime = tcp_now; 947 tp->t_rtseq = startseq; 948 tcpstat.tcps_segstimed++; 949 } 950 } 951 952 /* 953 * Set retransmit timer if not currently set, 954 * and not doing an ack or a keep-alive probe. 955 * Initial value for retransmit timer is smoothed 956 * round-trip time + 2 * round-trip time variance. 957 * Initialize shift counter which is used for backoff 958 * of retransmit time. 959 */ 960 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 961 tp->snd_nxt != tp->snd_una) { 962 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 963 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) { 964 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 965 tp->t_rxtshift = 0; 966 } 967 } 968 } else 969 if (SEQ_GT(tp->snd_nxt + len, tp->snd_max)) 970 tp->snd_max = tp->snd_nxt + len; 971 972 #ifdef TCP_DEBUG 973 /* 974 * Trace. 975 */ 976 if (so->so_options & SO_DEBUG) { 977 /* 978 * need to recover version # field, which was overwritten 979 * on ip_cksum computation. 980 */ 981 struct ip *sip; 982 sip = mtod(m, struct ip *); 983 switch (af) { 984 #ifdef INET 985 case AF_INET: 986 sip->ip_v = 4; 987 break; 988 #endif 989 #ifdef INET6 990 case AF_INET6: 991 sip->ip_v = 6; 992 break; 993 #endif 994 } 995 tcp_trace(TA_OUTPUT, tp->t_state, tp, m, 0); 996 } 997 #endif 998 999 /* 1000 * Fill in IP length and desired time to live and 1001 * send to IP level. There should be a better way 1002 * to handle ttl and tos; we could keep them in 1003 * the template, but need a way to checksum without them. 1004 */ 1005 m->m_pkthdr.len = hdrlen + len; 1006 1007 switch (af) { 1008 #ifdef INET 1009 case AF_INET: 1010 ip->ip_len = m->m_pkthdr.len; 1011 if (tp->t_inpcb) { 1012 ip->ip_ttl = tp->t_inpcb->inp_ip.ip_ttl; 1013 ip->ip_tos = tp->t_inpcb->inp_ip.ip_tos; 1014 } 1015 #ifdef INET6 1016 else if (tp->t_in6pcb) { 1017 ip->ip_ttl = in6_selecthlim(tp->t_in6pcb, NULL); /*XXX*/ 1018 ip->ip_tos = 0; /*XXX*/ 1019 } 1020 #endif 1021 break; 1022 #endif 1023 #ifdef INET6 1024 case AF_INET6: 1025 ip6->ip6_nxt = IPPROTO_TCP; 1026 if (tp->t_in6pcb) { 1027 /* 1028 * we separately set hoplimit for every segment, since 1029 * the user might want to change the value via 1030 * setsockopt. Also, desired default hop limit might 1031 * be changed via Neighbor Discovery. 1032 */ 1033 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, 1034 ro->ro_rt ? ro->ro_rt->rt_ifp : NULL); 1035 } 1036 /* ip6->ip6_flow = ??? */ 1037 /* ip6_plen will be filled in ip6_output(). */ 1038 break; 1039 #endif 1040 } 1041 1042 #ifdef IPSEC 1043 if (ipsec_setsocket(m, so) != 0) { 1044 m_freem(m); 1045 error = ENOBUFS; 1046 goto out; 1047 } 1048 #endif /*IPSEC*/ 1049 1050 switch (af) { 1051 #ifdef INET 1052 case AF_INET: 1053 { 1054 struct mbuf *opts; 1055 1056 if (tp->t_inpcb) 1057 opts = tp->t_inpcb->inp_options; 1058 else 1059 opts = NULL; 1060 error = ip_output(m, opts, ro, 1061 (ip_mtudisc ? IP_MTUDISC : 0) | 1062 (so->so_options & SO_DONTROUTE), 1063 0); 1064 break; 1065 } 1066 #endif 1067 #ifdef INET6 1068 case AF_INET6: 1069 { 1070 struct ip6_pktopts *opts; 1071 1072 if (tp->t_in6pcb) 1073 opts = tp->t_in6pcb->in6p_outputopts; 1074 else 1075 opts = NULL; 1076 error = ip6_output(m, opts, (struct route_in6 *)ro, 1077 so->so_options & SO_DONTROUTE, 0, NULL); 1078 break; 1079 } 1080 #endif 1081 default: 1082 error = EAFNOSUPPORT; 1083 break; 1084 } 1085 if (error) { 1086 out: 1087 if (error == ENOBUFS) { 1088 tcpstat.tcps_selfquench++; 1089 #ifdef INET 1090 if (tp->t_inpcb) 1091 tcp_quench(tp->t_inpcb, 0); 1092 #endif 1093 #ifdef INET6 1094 if (tp->t_in6pcb) 1095 tcp6_quench(tp->t_in6pcb, 0); 1096 #endif 1097 error = 0; 1098 } else if ((error == EHOSTUNREACH || error == ENETDOWN) && 1099 TCPS_HAVERCVDSYN(tp->t_state)) { 1100 tp->t_softerror = error; 1101 error = 0; 1102 } 1103 1104 /* Restart the delayed ACK timer, if necessary. */ 1105 if (tp->t_flags & TF_DELACK) 1106 TCP_RESTART_DELACK(tp); 1107 1108 return (error); 1109 } 1110 tcpstat.tcps_sndtotal++; 1111 if (tp->t_flags & TF_DELACK) 1112 tcpstat.tcps_delack++; 1113 1114 /* 1115 * Data sent (as far as we can tell). 1116 * If this advertises a larger window than any other segment, 1117 * then remember the size of the advertised window. 1118 * Any pending ACK has now been sent. 1119 */ 1120 if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv)) 1121 tp->rcv_adv = tp->rcv_nxt + win; 1122 tp->last_ack_sent = tp->rcv_nxt; 1123 tp->t_flags &= ~TF_ACKNOW; 1124 TCP_CLEAR_DELACK(tp); 1125 #ifdef DIAGNOSTIC 1126 if (maxburst < 0) 1127 printf("tcp_output: maxburst exceeded by %d\n", -maxburst); 1128 #endif 1129 if (sendalot && (!tcp_do_newreno || --maxburst)) 1130 goto again; 1131 return (0); 1132 } 1133 1134 void 1135 tcp_setpersist(tp) 1136 struct tcpcb *tp; 1137 { 1138 int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2); 1139 int nticks; 1140 1141 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT)) 1142 panic("tcp_output REXMT"); 1143 /* 1144 * Start/restart persistance timer. 1145 */ 1146 if (t < tp->t_rttmin) 1147 t = tp->t_rttmin; 1148 TCPT_RANGESET(nticks, t * tcp_backoff[tp->t_rxtshift], 1149 TCPTV_PERSMIN, TCPTV_PERSMAX); 1150 TCP_TIMER_ARM(tp, TCPT_PERSIST, nticks); 1151 if (tp->t_rxtshift < TCP_MAXRXTSHIFT) 1152 tp->t_rxtshift++; 1153 } 1154
Indexes created Tue Sep 23 07:09:52 GMT 2025