tcp_output.c revision 1.105
1 /* $NetBSD: tcp_output.c,v 1.105 2003/10/24 10:25:40 ragge 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 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 * 80 * Redistribution and use in source and binary forms, with or without 81 * modification, are permitted provided that the following conditions 82 * are met: 83 * 1. Redistributions of source code must retain the above copyright 84 * notice, this list of conditions and the following disclaimer. 85 * 2. Redistributions in binary form must reproduce the above copyright 86 * notice, this list of conditions and the following disclaimer in the 87 * documentation and/or other materials provided with the distribution. 88 * 3. All advertising materials mentioning features or use of this software 89 * must display the following acknowledgement: 90 * This product includes software developed by the NetBSD 91 * Foundation, Inc. and its contributors. 92 * 4. Neither the name of The NetBSD Foundation nor the names of its 93 * contributors may be used to endorse or promote products derived 94 * from this software without specific prior written permission. 95 * 96 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 97 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 98 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 99 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 100 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 101 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 102 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 103 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 104 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 105 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 106 * POSSIBILITY OF SUCH DAMAGE. 107 */ 108 109 /* 110 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995 111 * The Regents of the University of California. All rights reserved. 112 * 113 * Redistribution and use in source and binary forms, with or without 114 * modification, are permitted provided that the following conditions 115 * are met: 116 * 1. Redistributions of source code must retain the above copyright 117 * notice, this list of conditions and the following disclaimer. 118 * 2. Redistributions in binary form must reproduce the above copyright 119 * notice, this list of conditions and the following disclaimer in the 120 * documentation and/or other materials provided with the distribution. 121 * 3. Neither the name of the University nor the names of its contributors 122 * may be used to endorse or promote products derived from this software 123 * without specific prior written permission. 124 * 125 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 126 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 127 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 128 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 129 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 130 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 131 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 132 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 133 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 134 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 135 * SUCH DAMAGE. 136 * 137 * @(#)tcp_output.c 8.4 (Berkeley) 5/24/95 138 */ 139 140 #include <sys/cdefs.h> 141 __KERNEL_RCSID(0, "$NetBSD: tcp_output.c,v 1.105 2003/10/24 10:25:40 ragge Exp $"); 142 143 #include "opt_inet.h" 144 #include "opt_ipsec.h" 145 #include "opt_tcp_debug.h" 146 147 #include <sys/param.h> 148 #include <sys/systm.h> 149 #include <sys/malloc.h> 150 #include <sys/mbuf.h> 151 #include <sys/protosw.h> 152 #include <sys/socket.h> 153 #include <sys/socketvar.h> 154 #include <sys/errno.h> 155 #include <sys/domain.h> 156 #include <sys/kernel.h> 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 #endif /* FAST_IPSEC*/ 181 #ifdef IPSEC 182 #include <netinet6/ipsec.h> 183 #endif 184 185 #include <netinet/tcp.h> 186 #define TCPOUTFLAGS 187 #include <netinet/tcp_fsm.h> 188 #include <netinet/tcp_seq.h> 189 #include <netinet/tcp_timer.h> 190 #include <netinet/tcp_var.h> 191 #include <netinet/tcpip.h> 192 #include <netinet/tcp_debug.h> 193 194 #ifdef notyet 195 extern struct mbuf *m_copypack(); 196 #endif 197 198 #define MAX_TCPOPTLEN 32 /* max # bytes that go in options */ 199 200 /* 201 * Knob to enable Congestion Window Monitoring, and control the 202 * the burst size it allows. Default burst is 4 packets, per 203 * the Internet draft. 204 */ 205 int tcp_cwm = 0; 206 int tcp_cwm_burstsize = 4; 207 208 #ifdef TCP_OUTPUT_COUNTERS 209 #include <sys/device.h> 210 211 extern struct evcnt tcp_output_bigheader; 212 extern struct evcnt tcp_output_predict_hit; 213 extern struct evcnt tcp_output_predict_miss; 214 extern struct evcnt tcp_output_copysmall; 215 extern struct evcnt tcp_output_copybig; 216 extern struct evcnt tcp_output_refbig; 217 218 #define TCP_OUTPUT_COUNTER_INCR(ev) (ev)->ev_count++ 219 #else 220 221 #define TCP_OUTPUT_COUNTER_INCR(ev) /* nothing */ 222 223 #endif /* TCP_OUTPUT_COUNTERS */ 224 225 static 226 #ifndef GPROF 227 __inline 228 #endif 229 void 230 tcp_segsize(struct tcpcb *tp, int *txsegsizep, int *rxsegsizep) 231 { 232 #ifdef INET 233 struct inpcb *inp = tp->t_inpcb; 234 #endif 235 #ifdef INET6 236 struct in6pcb *in6p = tp->t_in6pcb; 237 #endif 238 struct socket *so = NULL; 239 struct rtentry *rt; 240 struct ifnet *ifp; 241 int size; 242 int iphlen; 243 int optlen; 244 245 #ifdef DIAGNOSTIC 246 if (tp->t_inpcb && tp->t_in6pcb) 247 panic("tcp_segsize: both t_inpcb and t_in6pcb are set"); 248 #endif 249 switch (tp->t_family) { 250 #ifdef INET 251 case AF_INET: 252 iphlen = sizeof(struct ip); 253 break; 254 #endif 255 #ifdef INET6 256 case AF_INET6: 257 iphlen = sizeof(struct ip6_hdr); 258 break; 259 #endif 260 default: 261 size = tcp_mssdflt; 262 goto out; 263 } 264 265 rt = NULL; 266 #ifdef INET 267 if (inp) { 268 rt = in_pcbrtentry(inp); 269 so = inp->inp_socket; 270 } 271 #endif 272 #ifdef INET6 273 if (in6p) { 274 rt = in6_pcbrtentry(in6p); 275 so = in6p->in6p_socket; 276 } 277 #endif 278 if (rt == NULL) { 279 size = tcp_mssdflt; 280 goto out; 281 } 282 283 ifp = rt->rt_ifp; 284 285 size = tcp_mssdflt; 286 if (tp->t_mtudisc && rt->rt_rmx.rmx_mtu != 0) 287 size = rt->rt_rmx.rmx_mtu - iphlen - sizeof(struct tcphdr); 288 else if (ifp->if_flags & IFF_LOOPBACK) 289 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 290 #ifdef INET 291 else if (inp && tp->t_mtudisc) 292 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 293 else if (inp && in_localaddr(inp->inp_faddr)) 294 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 295 #endif 296 #ifdef INET6 297 else if (in6p) { 298 #ifdef INET 299 if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr)) { 300 /* mapped addr case */ 301 struct in_addr d; 302 bcopy(&in6p->in6p_faddr.s6_addr32[3], &d, sizeof(d)); 303 if (tp->t_mtudisc || in_localaddr(d)) 304 size = ifp->if_mtu - iphlen - sizeof(struct tcphdr); 305 } else 306 #endif 307 { 308 /* 309 * for IPv6, path MTU discovery is always turned on, 310 * or the node must use packet size <= 1280. 311 */ 312 size = tp->t_mtudisc ? IN6_LINKMTU(ifp) : IPV6_MMTU; 313 size -= (iphlen + sizeof(struct tcphdr)); 314 } 315 } 316 #endif 317 out: 318 /* 319 * Now we must make room for whatever extra TCP/IP options are in 320 * the packet. 321 */ 322 optlen = tcp_optlen(tp); 323 324 /* 325 * XXX tp->t_ourmss should have the right size, but without this code 326 * fragmentation will occur... need more investigation 327 */ 328 #ifdef INET 329 if (inp) { 330 #if defined(IPSEC) || defined(FAST_IPSEC) 331 optlen += ipsec4_hdrsiz_tcp(tp); 332 #endif 333 optlen += ip_optlen(inp); 334 } 335 #endif 336 #ifdef INET6 337 #ifdef INET 338 if (in6p && tp->t_family == AF_INET) { 339 #if defined(IPSEC) || defined(FAST_IPSEC) 340 optlen += ipsec4_hdrsiz_tcp(tp); 341 #endif 342 /* XXX size -= ip_optlen(in6p); */ 343 } else 344 #endif 345 if (in6p && tp->t_family == AF_INET6) { 346 #ifdef IPSEC 347 optlen += ipsec6_hdrsiz_tcp(tp); 348 #endif 349 optlen += ip6_optlen(in6p); 350 } 351 #endif 352 size -= optlen; 353 354 /* 355 * *rxsegsizep holds *estimated* inbound segment size (estimation 356 * assumes that path MTU is the same for both ways). this is only 357 * for silly window avoidance, do not use the value for other purposes. 358 * 359 * ipseclen is subtracted from both sides, this may not be right. 360 * I'm not quite sure about this (could someone comment). 361 */ 362 *txsegsizep = min(tp->t_peermss - optlen, size); 363 /* 364 * Never send more than half a buffer full. This insures that we can 365 * always keep 2 packets on the wire, no matter what SO_SNDBUF is, and 366 * therefore acks will never be delayed unless we run out of data to 367 * transmit. 368 */ 369 if (so) 370 *txsegsizep = min(so->so_snd.sb_hiwat >> 1, *txsegsizep); 371 *rxsegsizep = min(tp->t_ourmss - optlen, size); 372 373 if (*txsegsizep != tp->t_segsz) { 374 /* 375 * If the new segment size is larger, we don't want to 376 * mess up the congestion window, but if it is smaller 377 * we'll have to reduce the congestion window to ensure 378 * that we don't get into trouble with initial windows 379 * and the rest. In any case, if the segment size 380 * has changed, chances are the path has, too, and 381 * our congestion window will be different. 382 */ 383 if (*txsegsizep < tp->t_segsz) { 384 tp->snd_cwnd = max((tp->snd_cwnd / tp->t_segsz) 385 * *txsegsizep, *txsegsizep); 386 tp->snd_ssthresh = max((tp->snd_ssthresh / tp->t_segsz) 387 * *txsegsizep, *txsegsizep); 388 } 389 tp->t_segsz = *txsegsizep; 390 } 391 } 392 393 static 394 #ifndef GPROF 395 __inline 396 #endif 397 int 398 tcp_build_datapkt(struct tcpcb *tp, struct socket *so, int off, 399 long len, int hdrlen, struct mbuf **mp) 400 { 401 struct mbuf *m, *m0; 402 403 if (tp->t_force && len == 1) 404 tcpstat.tcps_sndprobe++; 405 else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) { 406 tcpstat.tcps_sndrexmitpack++; 407 tcpstat.tcps_sndrexmitbyte += len; 408 } else { 409 tcpstat.tcps_sndpack++; 410 tcpstat.tcps_sndbyte += len; 411 } 412 #ifdef notyet 413 if ((m = m_copypack(so->so_snd.sb_mb, off, 414 (int)len, max_linkhdr + hdrlen)) == 0) 415 return (ENOBUFS); 416 /* 417 * m_copypack left space for our hdr; use it. 418 */ 419 m->m_len += hdrlen; 420 m->m_data -= hdrlen; 421 #else 422 MGETHDR(m, M_DONTWAIT, MT_HEADER); 423 if (__predict_false(m == NULL)) 424 return (ENOBUFS); 425 MCLAIM(m, &tcp_tx_mowner); 426 427 /* 428 * XXX Because other code assumes headers will fit in 429 * XXX one header mbuf. 430 * 431 * (This code should almost *never* be run.) 432 */ 433 if (__predict_false((max_linkhdr + hdrlen) > MHLEN)) { 434 TCP_OUTPUT_COUNTER_INCR(&tcp_output_bigheader); 435 MCLGET(m, M_DONTWAIT); 436 if ((m->m_flags & M_EXT) == 0) { 437 m_freem(m); 438 return (ENOBUFS); 439 } 440 } 441 442 m->m_data += max_linkhdr; 443 m->m_len = hdrlen; 444 445 /* 446 * To avoid traversing the whole sb_mb chain for correct 447 * data to send, remember last sent mbuf, its offset and 448 * the sent size. When called the next time, see if the 449 * data to send is directly following the previous transfer. 450 * This is important for large TCP windows. 451 */ 452 if ( 453 #ifdef FAST_MBSEARCH 454 off == 0 || tp->t_lastm == NULL || 455 (tp->t_lastoff + tp->t_lastlen) != off 456 #else 457 1 458 #endif 459 ) 460 { 461 #ifdef FAST_MBSEARCH 462 TCP_OUTPUT_COUNTER_INCR(&tcp_output_predict_miss); 463 #endif 464 /* 465 * Either a new packet or a retransmit. 466 * Start from the beginning. 467 */ 468 tp->t_lastm = so->so_snd.sb_mb; 469 tp->t_inoff = off; 470 } else { 471 #ifdef FAST_MBSEARCH 472 TCP_OUTPUT_COUNTER_INCR(&tcp_output_predict_hit); 473 #endif 474 tp->t_inoff += tp->t_lastlen; 475 } 476 477 /* Traverse forward to next packet */ 478 while (tp->t_inoff > 0) { 479 if (tp->t_lastm == NULL) 480 panic("tp->t_lastm == NULL"); 481 if (tp->t_inoff < tp->t_lastm->m_len) 482 break; 483 tp->t_inoff -= tp->t_lastm->m_len; 484 tp->t_lastm = tp->t_lastm->m_next; 485 } 486 487 tp->t_lastoff = off; 488 tp->t_lastlen = len; 489 m0 = tp->t_lastm; 490 off = tp->t_inoff; 491 492 if (len <= M_TRAILINGSPACE(m)) { 493 m_copydata(m0, off, (int) len, mtod(m, caddr_t) + hdrlen); 494 m->m_len += len; 495 TCP_OUTPUT_COUNTER_INCR(&tcp_output_copysmall); 496 } else { 497 m->m_next = m_copy(m0, off, (int) len); 498 if (m->m_next == NULL) { 499 m_freem(m); 500 return (ENOBUFS); 501 } 502 #ifdef TCP_OUTPUT_COUNTERS 503 if (m->m_next->m_flags & M_EXT) 504 TCP_OUTPUT_COUNTER_INCR(&tcp_output_refbig); 505 else 506 TCP_OUTPUT_COUNTER_INCR(&tcp_output_copybig); 507 #endif /* TCP_OUTPUT_COUNTERS */ 508 } 509 #endif 510 511 *mp = m; 512 return (0); 513 } 514 515 /* 516 * Tcp output routine: figure out what should be sent and send it. 517 */ 518 int 519 tcp_output(tp) 520 struct tcpcb *tp; 521 { 522 struct socket *so; 523 struct route *ro; 524 long len, win; 525 int off, flags, error; 526 struct mbuf *m; 527 struct ip *ip; 528 #ifdef INET6 529 struct ip6_hdr *ip6; 530 #endif 531 struct tcphdr *th; 532 u_char opt[MAX_TCPOPTLEN]; 533 unsigned optlen, hdrlen; 534 int idle, sendalot, txsegsize, rxsegsize; 535 int maxburst = TCP_MAXBURST; 536 int af; /* address family on the wire */ 537 int iphdrlen; 538 539 #ifdef DIAGNOSTIC 540 if (tp->t_inpcb && tp->t_in6pcb) 541 panic("tcp_output: both t_inpcb and t_in6pcb are set"); 542 #endif 543 so = NULL; 544 ro = NULL; 545 if (tp->t_inpcb) { 546 so = tp->t_inpcb->inp_socket; 547 ro = &tp->t_inpcb->inp_route; 548 } 549 #ifdef INET6 550 else if (tp->t_in6pcb) { 551 so = tp->t_in6pcb->in6p_socket; 552 ro = (struct route *)&tp->t_in6pcb->in6p_route; 553 } 554 #endif 555 556 switch (af = tp->t_family) { 557 #ifdef INET 558 case AF_INET: 559 if (tp->t_inpcb) 560 break; 561 #ifdef INET6 562 /* mapped addr case */ 563 if (tp->t_in6pcb) 564 break; 565 #endif 566 return EINVAL; 567 #endif 568 #ifdef INET6 569 case AF_INET6: 570 if (tp->t_in6pcb) 571 break; 572 return EINVAL; 573 #endif 574 default: 575 return EAFNOSUPPORT; 576 } 577 578 tcp_segsize(tp, &txsegsize, &rxsegsize); 579 580 idle = (tp->snd_max == tp->snd_una); 581 582 /* 583 * Restart Window computation. From draft-floyd-incr-init-win-03: 584 * 585 * Optionally, a TCP MAY set the restart window to the 586 * minimum of the value used for the initial window and 587 * the current value of cwnd (in other words, using a 588 * larger value for the restart window should never increase 589 * the size of cwnd). 590 */ 591 if (tcp_cwm) { 592 /* 593 * Hughes/Touch/Heidemann Congestion Window Monitoring. 594 * Count the number of packets currently pending 595 * acknowledgement, and limit our congestion window 596 * to a pre-determined allowed burst size plus that count. 597 * This prevents bursting once all pending packets have 598 * been acknowledged (i.e. transmission is idle). 599 * 600 * XXX Link this to Initial Window? 601 */ 602 tp->snd_cwnd = min(tp->snd_cwnd, 603 (tcp_cwm_burstsize * txsegsize) + 604 (tp->snd_nxt - tp->snd_una)); 605 } else { 606 if (idle && (tcp_now - tp->t_rcvtime) >= tp->t_rxtcur) { 607 /* 608 * We have been idle for "a while" and no acks are 609 * expected to clock out any data we send -- 610 * slow start to get ack "clock" running again. 611 */ 612 int ss = tcp_init_win; 613 #ifdef INET 614 if (tp->t_inpcb && 615 in_localaddr(tp->t_inpcb->inp_faddr)) 616 ss = tcp_init_win_local; 617 #endif 618 #ifdef INET6 619 if (tp->t_in6pcb && 620 in6_localaddr(&tp->t_in6pcb->in6p_faddr)) 621 ss = tcp_init_win_local; 622 #endif 623 tp->snd_cwnd = min(tp->snd_cwnd, 624 TCP_INITIAL_WINDOW(ss, txsegsize)); 625 } 626 } 627 628 again: 629 /* 630 * Determine length of data that should be transmitted, and 631 * flags that should be used. If there is some data or critical 632 * controls (SYN, RST) to send, then transmit; otherwise, 633 * investigate further. 634 */ 635 sendalot = 0; 636 off = tp->snd_nxt - tp->snd_una; 637 win = min(tp->snd_wnd, tp->snd_cwnd); 638 639 flags = tcp_outflags[tp->t_state]; 640 /* 641 * If in persist timeout with window of 0, send 1 byte. 642 * Otherwise, if window is small but nonzero 643 * and timer expired, we will send what we can 644 * and go to transmit state. 645 */ 646 if (tp->t_force) { 647 if (win == 0) { 648 /* 649 * If we still have some data to send, then 650 * clear the FIN bit. Usually this would 651 * happen below when it realizes that we 652 * aren't sending all the data. However, 653 * if we have exactly 1 byte of unset data, 654 * then it won't clear the FIN bit below, 655 * and if we are in persist state, we wind 656 * up sending the packet without recording 657 * that we sent the FIN bit. 658 * 659 * We can't just blindly clear the FIN bit, 660 * because if we don't have any more data 661 * to send then the probe will be the FIN 662 * itself. 663 */ 664 if (off < so->so_snd.sb_cc) 665 flags &= ~TH_FIN; 666 win = 1; 667 } else { 668 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 669 tp->t_rxtshift = 0; 670 } 671 } 672 673 if (win < so->so_snd.sb_cc) { 674 len = win - off; 675 flags &= ~TH_FIN; 676 } else 677 len = so->so_snd.sb_cc - off; 678 679 if (len < 0) { 680 /* 681 * If FIN has been sent but not acked, 682 * but we haven't been called to retransmit, 683 * len will be -1. Otherwise, window shrank 684 * after we sent into it. If window shrank to 0, 685 * cancel pending retransmit, pull snd_nxt back 686 * to (closed) window, and set the persist timer 687 * if it isn't already going. If the window didn't 688 * close completely, just wait for an ACK. 689 * 690 * If we have a pending FIN, either it has already been 691 * transmitted or it is outside the window, so drop it. 692 * If the FIN has been transmitted, but this is not a 693 * retransmission, then len must be -1. Therefore we also 694 * prevent here the sending of `gratuitous FINs'. This 695 * eliminates the need to check for that case below (e.g. 696 * to back up snd_nxt before the FIN so that the sequence 697 * number is correct). 698 */ 699 len = 0; 700 flags &= ~TH_FIN; 701 if (win == 0) { 702 TCP_TIMER_DISARM(tp, TCPT_REXMT); 703 tp->t_rxtshift = 0; 704 tp->snd_nxt = tp->snd_una; 705 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) 706 tcp_setpersist(tp); 707 } 708 } 709 if (len > txsegsize) { 710 len = txsegsize; 711 flags &= ~TH_FIN; 712 sendalot = 1; 713 } 714 715 win = sbspace(&so->so_rcv); 716 717 /* 718 * Sender silly window avoidance. If connection is idle 719 * and can send all data, a maximum segment, 720 * at least a maximum default-size segment do it, 721 * or are forced, do it; otherwise don't bother. 722 * If peer's buffer is tiny, then send 723 * when window is at least half open. 724 * If retransmitting (possibly after persist timer forced us 725 * to send into a small window), then must resend. 726 */ 727 if (len) { 728 if (len == txsegsize) 729 goto send; 730 if ((so->so_state & SS_MORETOCOME) == 0 && 731 ((idle || tp->t_flags & TF_NODELAY) && 732 len + off >= so->so_snd.sb_cc)) 733 goto send; 734 if (tp->t_force) 735 goto send; 736 if (len >= tp->max_sndwnd / 2) 737 goto send; 738 if (SEQ_LT(tp->snd_nxt, tp->snd_max)) 739 goto send; 740 } 741 742 /* 743 * Compare available window to amount of window known to peer 744 * (as advertised window less next expected input). If the 745 * difference is at least twice the size of the largest segment 746 * we expect to receive (i.e. two segments) or at least 50% of 747 * the maximum possible window, then want to send a window update 748 * to peer. 749 */ 750 if (win > 0) { 751 /* 752 * "adv" is the amount we can increase the window, 753 * taking into account that we are limited by 754 * TCP_MAXWIN << tp->rcv_scale. 755 */ 756 long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) - 757 (tp->rcv_adv - tp->rcv_nxt); 758 759 if (adv >= (long) (2 * rxsegsize)) 760 goto send; 761 if (2 * adv >= (long) so->so_rcv.sb_hiwat) 762 goto send; 763 } 764 765 /* 766 * Send if we owe peer an ACK. 767 */ 768 if (tp->t_flags & TF_ACKNOW) 769 goto send; 770 if (flags & (TH_SYN|TH_FIN|TH_RST)) 771 goto send; 772 if (SEQ_GT(tp->snd_up, tp->snd_una)) 773 goto send; 774 775 /* 776 * TCP window updates are not reliable, rather a polling protocol 777 * using ``persist'' packets is used to insure receipt of window 778 * updates. The three ``states'' for the output side are: 779 * idle not doing retransmits or persists 780 * persisting to move a small or zero window 781 * (re)transmitting and thereby not persisting 782 * 783 * tp->t_timer[TCPT_PERSIST] 784 * is set when we are in persist state. 785 * tp->t_force 786 * is set when we are called to send a persist packet. 787 * tp->t_timer[TCPT_REXMT] 788 * is set when we are retransmitting 789 * The output side is idle when both timers are zero. 790 * 791 * If send window is too small, there is data to transmit, and no 792 * retransmit or persist is pending, then go to persist state. 793 * If nothing happens soon, send when timer expires: 794 * if window is nonzero, transmit what we can, 795 * otherwise force out a byte. 796 */ 797 if (so->so_snd.sb_cc && TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 798 TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 799 tp->t_rxtshift = 0; 800 tcp_setpersist(tp); 801 } 802 803 /* 804 * No reason to send a segment, just return. 805 */ 806 return (0); 807 808 send: 809 /* 810 * Before ESTABLISHED, force sending of initial options 811 * unless TCP set not to do any options. 812 * NOTE: we assume that the IP/TCP header plus TCP options 813 * always fit in a single mbuf, leaving room for a maximum 814 * link header, i.e. 815 * max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MCLBYTES 816 */ 817 optlen = 0; 818 switch (af) { 819 #ifdef INET 820 case AF_INET: 821 iphdrlen = sizeof(struct ip) + sizeof(struct tcphdr); 822 break; 823 #endif 824 #ifdef INET6 825 case AF_INET6: 826 iphdrlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr); 827 break; 828 #endif 829 default: /*pacify gcc*/ 830 iphdrlen = 0; 831 break; 832 } 833 hdrlen = iphdrlen; 834 if (flags & TH_SYN) { 835 struct rtentry *rt; 836 837 rt = NULL; 838 #ifdef INET 839 if (tp->t_inpcb) 840 rt = in_pcbrtentry(tp->t_inpcb); 841 #endif 842 #ifdef INET6 843 if (tp->t_in6pcb) 844 rt = in6_pcbrtentry(tp->t_in6pcb); 845 #endif 846 847 tp->snd_nxt = tp->iss; 848 tp->t_ourmss = tcp_mss_to_advertise(rt != NULL ? 849 rt->rt_ifp : NULL, af); 850 if ((tp->t_flags & TF_NOOPT) == 0) { 851 opt[0] = TCPOPT_MAXSEG; 852 opt[1] = 4; 853 opt[2] = (tp->t_ourmss >> 8) & 0xff; 854 opt[3] = tp->t_ourmss & 0xff; 855 optlen = 4; 856 857 if ((tp->t_flags & TF_REQ_SCALE) && 858 ((flags & TH_ACK) == 0 || 859 (tp->t_flags & TF_RCVD_SCALE))) { 860 *((u_int32_t *) (opt + optlen)) = htonl( 861 TCPOPT_NOP << 24 | 862 TCPOPT_WINDOW << 16 | 863 TCPOLEN_WINDOW << 8 | 864 tp->request_r_scale); 865 optlen += 4; 866 } 867 } 868 } 869 870 /* 871 * Send a timestamp and echo-reply if this is a SYN and our side 872 * wants to use timestamps (TF_REQ_TSTMP is set) or both our side 873 * and our peer have sent timestamps in our SYN's. 874 */ 875 if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP && 876 (flags & TH_RST) == 0 && 877 ((flags & (TH_SYN|TH_ACK)) == TH_SYN || 878 (tp->t_flags & TF_RCVD_TSTMP))) { 879 u_int32_t *lp = (u_int32_t *)(opt + optlen); 880 881 /* Form timestamp option as shown in appendix A of RFC 1323. */ 882 *lp++ = htonl(TCPOPT_TSTAMP_HDR); 883 *lp++ = htonl(TCP_TIMESTAMP(tp)); 884 *lp = htonl(tp->ts_recent); 885 optlen += TCPOLEN_TSTAMP_APPA; 886 } 887 888 hdrlen += optlen; 889 890 #ifdef DIAGNOSTIC 891 if (len > txsegsize) 892 panic("tcp data to be sent is larger than segment"); 893 if (max_linkhdr + hdrlen > MCLBYTES) 894 panic("tcphdr too big"); 895 #endif 896 897 /* 898 * Grab a header mbuf, attaching a copy of data to 899 * be transmitted, and initialize the header from 900 * the template for sends on this connection. 901 */ 902 if (len) { 903 error = tcp_build_datapkt(tp, so, off, len, hdrlen, &m); 904 if (error) 905 goto out; 906 /* 907 * If we're sending everything we've got, set PUSH. 908 * (This will keep happy those implementations which only 909 * give data to the user when a buffer fills or 910 * a PUSH comes in.) 911 */ 912 if (off + len == so->so_snd.sb_cc) 913 flags |= TH_PUSH; 914 } else { 915 if (tp->t_flags & TF_ACKNOW) 916 tcpstat.tcps_sndacks++; 917 else if (flags & (TH_SYN|TH_FIN|TH_RST)) 918 tcpstat.tcps_sndctrl++; 919 else if (SEQ_GT(tp->snd_up, tp->snd_una)) 920 tcpstat.tcps_sndurg++; 921 else 922 tcpstat.tcps_sndwinup++; 923 924 MGETHDR(m, M_DONTWAIT, MT_HEADER); 925 if (m != NULL && max_linkhdr + hdrlen > MHLEN) { 926 MCLGET(m, M_DONTWAIT); 927 if ((m->m_flags & M_EXT) == 0) { 928 m_freem(m); 929 m = NULL; 930 } 931 } 932 if (m == NULL) { 933 error = ENOBUFS; 934 goto out; 935 } 936 MCLAIM(m, &tcp_tx_mowner); 937 m->m_data += max_linkhdr; 938 m->m_len = hdrlen; 939 } 940 m->m_pkthdr.rcvif = (struct ifnet *)0; 941 switch (af) { 942 #ifdef INET 943 case AF_INET: 944 ip = mtod(m, struct ip *); 945 #ifdef INET6 946 ip6 = NULL; 947 #endif 948 th = (struct tcphdr *)(ip + 1); 949 break; 950 #endif 951 #ifdef INET6 952 case AF_INET6: 953 ip = NULL; 954 ip6 = mtod(m, struct ip6_hdr *); 955 th = (struct tcphdr *)(ip6 + 1); 956 break; 957 #endif 958 default: /*pacify gcc*/ 959 ip = NULL; 960 #ifdef INET6 961 ip6 = NULL; 962 #endif 963 th = NULL; 964 break; 965 } 966 if (tp->t_template == 0) 967 panic("tcp_output"); 968 if (tp->t_template->m_len < iphdrlen) 969 panic("tcp_output"); 970 bcopy(mtod(tp->t_template, caddr_t), mtod(m, caddr_t), iphdrlen); 971 972 /* 973 * If we are doing retransmissions, then snd_nxt will 974 * not reflect the first unsent octet. For ACK only 975 * packets, we do not want the sequence number of the 976 * retransmitted packet, we want the sequence number 977 * of the next unsent octet. So, if there is no data 978 * (and no SYN or FIN), use snd_max instead of snd_nxt 979 * when filling in ti_seq. But if we are in persist 980 * state, snd_max might reflect one byte beyond the 981 * right edge of the window, so use snd_nxt in that 982 * case, since we know we aren't doing a retransmission. 983 * (retransmit and persist are mutually exclusive...) 984 */ 985 if (len || (flags & (TH_SYN|TH_FIN)) || 986 TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) 987 th->th_seq = htonl(tp->snd_nxt); 988 else 989 th->th_seq = htonl(tp->snd_max); 990 th->th_ack = htonl(tp->rcv_nxt); 991 if (optlen) { 992 bcopy((caddr_t)opt, (caddr_t)(th + 1), optlen); 993 th->th_off = (sizeof (struct tcphdr) + optlen) >> 2; 994 } 995 th->th_flags = flags; 996 /* 997 * Calculate receive window. Don't shrink window, 998 * but avoid silly window syndrome. 999 */ 1000 if (win < (long)(so->so_rcv.sb_hiwat / 4) && win < (long)rxsegsize) 1001 win = 0; 1002 if (win > (long)TCP_MAXWIN << tp->rcv_scale) 1003 win = (long)TCP_MAXWIN << tp->rcv_scale; 1004 if (win < (long)(tp->rcv_adv - tp->rcv_nxt)) 1005 win = (long)(tp->rcv_adv - tp->rcv_nxt); 1006 th->th_win = htons((u_int16_t) (win>>tp->rcv_scale)); 1007 if (SEQ_GT(tp->snd_up, tp->snd_nxt)) { 1008 u_int32_t urp = tp->snd_up - tp->snd_nxt; 1009 if (urp > IP_MAXPACKET) 1010 urp = IP_MAXPACKET; 1011 th->th_urp = htons((u_int16_t)urp); 1012 th->th_flags |= TH_URG; 1013 } else 1014 /* 1015 * If no urgent pointer to send, then we pull 1016 * the urgent pointer to the left edge of the send window 1017 * so that it doesn't drift into the send window on sequence 1018 * number wraparound. 1019 */ 1020 tp->snd_up = tp->snd_una; /* drag it along */ 1021 1022 /* 1023 * Set ourselves up to be checksummed just before the packet 1024 * hits the wire. 1025 */ 1026 switch (af) { 1027 #ifdef INET 1028 case AF_INET: 1029 m->m_pkthdr.csum_flags = M_CSUM_TCPv4; 1030 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 1031 if (len + optlen) { 1032 /* Fixup the pseudo-header checksum. */ 1033 /* XXXJRT Not IP Jumbogram safe. */ 1034 th->th_sum = in_cksum_addword(th->th_sum, 1035 htons((u_int16_t) (len + optlen))); 1036 } 1037 break; 1038 #endif 1039 #ifdef INET6 1040 case AF_INET6: 1041 /* 1042 * XXX Actually delaying the checksum is Hard 1043 * XXX (well, maybe not for Itojun, but it is 1044 * XXX for me), but we can still take advantage 1045 * XXX of the cached pseudo-header checksum. 1046 */ 1047 /* equals to hdrlen + len */ 1048 m->m_pkthdr.len = sizeof(struct ip6_hdr) 1049 + sizeof(struct tcphdr) + optlen + len; 1050 #ifdef notyet 1051 m->m_pkthdr.csum_flags = M_CSUM_TCPv6; 1052 m->m_pkthdr.csum_data = offsetof(struct tcphdr, th_sum); 1053 #endif 1054 if (len + optlen) { 1055 /* Fixup the pseudo-header checksum. */ 1056 /* XXXJRT: Not IPv6 Jumbogram safe. */ 1057 th->th_sum = in_cksum_addword(th->th_sum, 1058 htons((u_int16_t) (len + optlen))); 1059 } 1060 #ifndef notyet 1061 th->th_sum = in6_cksum(m, 0, sizeof(struct ip6_hdr), 1062 sizeof(struct tcphdr) + optlen + len); 1063 #endif 1064 break; 1065 #endif 1066 } 1067 1068 /* 1069 * In transmit state, time the transmission and arrange for 1070 * the retransmit. In persist state, just set snd_max. 1071 */ 1072 if (tp->t_force == 0 || TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0) { 1073 tcp_seq startseq = tp->snd_nxt; 1074 1075 /* 1076 * Advance snd_nxt over sequence space of this segment. 1077 * There are no states in which we send both a SYN and a FIN, 1078 * so we collapse the tests for these flags. 1079 */ 1080 if (flags & (TH_SYN|TH_FIN)) 1081 tp->snd_nxt++; 1082 tp->snd_nxt += len; 1083 if (SEQ_GT(tp->snd_nxt, tp->snd_max)) { 1084 tp->snd_max = tp->snd_nxt; 1085 /* 1086 * Time this transmission if not a retransmission and 1087 * not currently timing anything. 1088 */ 1089 if (tp->t_rtttime == 0) { 1090 tp->t_rtttime = tcp_now; 1091 tp->t_rtseq = startseq; 1092 tcpstat.tcps_segstimed++; 1093 } 1094 } 1095 1096 /* 1097 * Set retransmit timer if not currently set, 1098 * and not doing an ack or a keep-alive probe. 1099 * Initial value for retransmit timer is smoothed 1100 * round-trip time + 2 * round-trip time variance. 1101 * Initialize shift counter which is used for backoff 1102 * of retransmit time. 1103 */ 1104 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 && 1105 tp->snd_nxt != tp->snd_una) { 1106 TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur); 1107 if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST)) { 1108 TCP_TIMER_DISARM(tp, TCPT_PERSIST); 1109 tp->t_rxtshift = 0; 1110 } 1111 } 1112 } else 1113 if (SEQ_GT(tp->snd_nxt + len, tp->snd_max)) 1114 tp->snd_max = tp->snd_nxt + len; 1115 1116 #ifdef TCP_DEBUG 1117 /* 1118 * Trace. 1119 */ 1120 if (so->so_options & SO_DEBUG) 1121 tcp_trace(TA_OUTPUT, tp->t_state, tp, m, 0); 1122 #endif 1123 1124 /* 1125 * Fill in IP length and desired time to live and 1126 * send to IP level. There should be a better way 1127 * to handle ttl and tos; we could keep them in 1128 * the template, but need a way to checksum without them. 1129 */ 1130 m->m_pkthdr.len = hdrlen + len; 1131 1132 switch (af) { 1133 #ifdef INET 1134 case AF_INET: 1135 ip->ip_len = htons(m->m_pkthdr.len); 1136 if (tp->t_inpcb) { 1137 ip->ip_ttl = tp->t_inpcb->inp_ip.ip_ttl; 1138 ip->ip_tos = tp->t_inpcb->inp_ip.ip_tos; 1139 } 1140 #ifdef INET6 1141 else if (tp->t_in6pcb) { 1142 ip->ip_ttl = in6_selecthlim(tp->t_in6pcb, NULL); /*XXX*/ 1143 ip->ip_tos = 0; /*XXX*/ 1144 } 1145 #endif 1146 break; 1147 #endif 1148 #ifdef INET6 1149 case AF_INET6: 1150 ip6->ip6_nxt = IPPROTO_TCP; 1151 if (tp->t_in6pcb) { 1152 /* 1153 * we separately set hoplimit for every segment, since 1154 * the user might want to change the value via 1155 * setsockopt. Also, desired default hop limit might 1156 * be changed via Neighbor Discovery. 1157 */ 1158 ip6->ip6_hlim = in6_selecthlim(tp->t_in6pcb, 1159 ro->ro_rt ? ro->ro_rt->rt_ifp : NULL); 1160 } 1161 /* ip6->ip6_flow = ??? */ 1162 /* ip6_plen will be filled in ip6_output(). */ 1163 break; 1164 #endif 1165 } 1166 1167 switch (af) { 1168 #ifdef INET 1169 case AF_INET: 1170 { 1171 struct mbuf *opts; 1172 1173 if (tp->t_inpcb) 1174 opts = tp->t_inpcb->inp_options; 1175 else 1176 opts = NULL; 1177 error = ip_output(m, opts, ro, 1178 (tp->t_mtudisc ? IP_MTUDISC : 0) | 1179 (so->so_options & SO_DONTROUTE), 1180 (struct ip_moptions *)0, so); 1181 break; 1182 } 1183 #endif 1184 #ifdef INET6 1185 case AF_INET6: 1186 { 1187 struct ip6_pktopts *opts; 1188 1189 if (tp->t_in6pcb) 1190 opts = tp->t_in6pcb->in6p_outputopts; 1191 else 1192 opts = NULL; 1193 error = ip6_output(m, opts, (struct route_in6 *)ro, 1194 so->so_options & SO_DONTROUTE, 1195 (struct ip6_moptions *)0, so, NULL); 1196 break; 1197 } 1198 #endif 1199 default: 1200 error = EAFNOSUPPORT; 1201 break; 1202 } 1203 if (error) { 1204 out: 1205 if (error == ENOBUFS) { 1206 tcpstat.tcps_selfquench++; 1207 #ifdef INET 1208 if (tp->t_inpcb) 1209 tcp_quench(tp->t_inpcb, 0); 1210 #endif 1211 #ifdef INET6 1212 if (tp->t_in6pcb) 1213 tcp6_quench(tp->t_in6pcb, 0); 1214 #endif 1215 error = 0; 1216 } else if ((error == EHOSTUNREACH || error == ENETDOWN) && 1217 TCPS_HAVERCVDSYN(tp->t_state)) { 1218 tp->t_softerror = error; 1219 error = 0; 1220 } 1221 1222 /* Restart the delayed ACK timer, if necessary. */ 1223 if (tp->t_flags & TF_DELACK) 1224 TCP_RESTART_DELACK(tp); 1225 1226 return (error); 1227 } 1228 tcpstat.tcps_sndtotal++; 1229 if (tp->t_flags & TF_DELACK) 1230 tcpstat.tcps_delack++; 1231 1232 /* 1233 * Data sent (as far as we can tell). 1234 * If this advertises a larger window than any other segment, 1235 * then remember the size of the advertised window. 1236 * Any pending ACK has now been sent. 1237 */ 1238 if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv)) 1239 tp->rcv_adv = tp->rcv_nxt + win; 1240 tp->last_ack_sent = tp->rcv_nxt; 1241 tp->t_flags &= ~TF_ACKNOW; 1242 TCP_CLEAR_DELACK(tp); 1243 #ifdef DIAGNOSTIC 1244 if (maxburst < 0) 1245 printf("tcp_output: maxburst exceeded by %d\n", -maxburst); 1246 #endif 1247 if (sendalot && (!tcp_do_newreno || --maxburst)) 1248 goto again; 1249 return (0); 1250 } 1251 1252 void 1253 tcp_setpersist(tp) 1254 struct tcpcb *tp; 1255 { 1256 int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> (1 + 2); 1257 int nticks; 1258 1259 if (TCP_TIMER_ISARMED(tp, TCPT_REXMT)) 1260 panic("tcp_output REXMT"); 1261 /* 1262 * Start/restart persistance timer. 1263 */ 1264 if (t < tp->t_rttmin) 1265 t = tp->t_rttmin; 1266 TCPT_RANGESET(nticks, t * tcp_backoff[tp->t_rxtshift], 1267 TCPTV_PERSMIN, TCPTV_PERSMAX); 1268 TCP_TIMER_ARM(tp, TCPT_PERSIST, nticks); 1269 if (tp->t_rxtshift < TCP_MAXRXTSHIFT) 1270 tp->t_rxtshift++; 1271 } 1272
Indexes created Tue Sep 23 14:10:03 GMT 2025