| 1 | /* $Id: tcp_timer.c 37936 2011-07-14 03:54:41Z vboxsync $ */
|
|---|
| 2 | /** @file
|
|---|
| 3 | * NAT - TCP timers.
|
|---|
| 4 | */
|
|---|
| 5 |
|
|---|
| 6 | /*
|
|---|
| 7 | * Copyright (C) 2006-2010 Oracle Corporation
|
|---|
| 8 | *
|
|---|
| 9 | * This file is part of VirtualBox Open Source Edition (OSE), as
|
|---|
| 10 | * available from http://www.virtualbox.org. This file is free software;
|
|---|
| 11 | * you can redistribute it and/or modify it under the terms of the GNU
|
|---|
| 12 | * General Public License (GPL) as published by the Free Software
|
|---|
| 13 | * Foundation, in version 2 as it comes in the "COPYING" file of the
|
|---|
| 14 | * VirtualBox OSE distribution. VirtualBox OSE is distributed in the
|
|---|
| 15 | * hope that it will be useful, but WITHOUT ANY WARRANTY of any kind.
|
|---|
| 16 | */
|
|---|
| 17 |
|
|---|
| 18 | /*
|
|---|
| 19 | * This code is based on:
|
|---|
| 20 | *
|
|---|
| 21 | * Copyright (c) 1982, 1986, 1988, 1990, 1993
|
|---|
| 22 | * The Regents of the University of California. All rights reserved.
|
|---|
| 23 | *
|
|---|
| 24 | * Redistribution and use in source and binary forms, with or without
|
|---|
| 25 | * modification, are permitted provided that the following conditions
|
|---|
| 26 | * are met:
|
|---|
| 27 | * 1. Redistributions of source code must retain the above copyright
|
|---|
| 28 | * notice, this list of conditions and the following disclaimer.
|
|---|
| 29 | * 2. Redistributions in binary form must reproduce the above copyright
|
|---|
| 30 | * notice, this list of conditions and the following disclaimer in the
|
|---|
| 31 | * documentation and/or other materials provided with the distribution.
|
|---|
| 32 | * 3. All advertising materials mentioning features or use of this software
|
|---|
| 33 | * must display the following acknowledgement:
|
|---|
| 34 | * This product includes software developed by the University of
|
|---|
| 35 | * California, Berkeley and its contributors.
|
|---|
| 36 | * 4. Neither the name of the University nor the names of its contributors
|
|---|
| 37 | * may be used to endorse or promote products derived from this software
|
|---|
| 38 | * without specific prior written permission.
|
|---|
| 39 | *
|
|---|
| 40 | * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
|
|---|
| 41 | * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
|
|---|
| 42 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
|
|---|
| 43 | * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
|
|---|
| 44 | * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
|
|---|
| 45 | * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
|
|---|
| 46 | * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
|
|---|
| 47 | * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
|
|---|
| 48 | * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
|
|---|
| 49 | * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
|
|---|
| 50 | * SUCH DAMAGE.
|
|---|
| 51 | *
|
|---|
| 52 | * @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
|
|---|
| 53 | * tcp_timer.c,v 1.2 1994/08/02 07:49:10 davidg Exp
|
|---|
| 54 | */
|
|---|
| 55 |
|
|---|
| 56 | #include <slirp.h>
|
|---|
| 57 |
|
|---|
| 58 |
|
|---|
| 59 | /*
|
|---|
| 60 | * Fast timeout routine for processing delayed acks
|
|---|
| 61 | */
|
|---|
| 62 | void
|
|---|
| 63 | tcp_fasttimo(PNATState pData)
|
|---|
| 64 | {
|
|---|
| 65 | register struct socket *so, *so_next;
|
|---|
| 66 | register struct tcpcb *tp;
|
|---|
| 67 |
|
|---|
| 68 | LogFlowFuncEnter();
|
|---|
| 69 |
|
|---|
| 70 | so = tcb.so_next;
|
|---|
| 71 | if (so)
|
|---|
| 72 | QSOCKET_FOREACH (so, so_next, tcp)
|
|---|
| 73 | /* { */
|
|---|
| 74 | if ( (tp = (struct tcpcb *)so->so_tcpcb)
|
|---|
| 75 | && (tp->t_flags & TF_DELACK))
|
|---|
| 76 | {
|
|---|
| 77 | tp->t_flags &= ~TF_DELACK;
|
|---|
| 78 | tp->t_flags |= TF_ACKNOW;
|
|---|
| 79 | tcpstat.tcps_delack++;
|
|---|
| 80 | TCP_OUTPUT(pData, tp);
|
|---|
| 81 | }
|
|---|
| 82 | LOOP_LABEL(tcp, so, so_next);
|
|---|
| 83 | }
|
|---|
| 84 | }
|
|---|
| 85 |
|
|---|
| 86 | /*
|
|---|
| 87 | * Tcp protocol timeout routine called every 500 ms.
|
|---|
| 88 | * Updates the timers in all active tcb's and
|
|---|
| 89 | * causes finite state machine actions if timers expire.
|
|---|
| 90 | */
|
|---|
| 91 | void
|
|---|
| 92 | tcp_slowtimo(PNATState pData)
|
|---|
| 93 | {
|
|---|
| 94 | register struct socket *ip, *ipnxt;
|
|---|
| 95 | register struct tcpcb *tp;
|
|---|
| 96 | register int i;
|
|---|
| 97 |
|
|---|
| 98 | LogFlowFuncEnter();
|
|---|
| 99 |
|
|---|
| 100 | /*
|
|---|
| 101 | * Search through tcb's and update active timers.
|
|---|
| 102 | */
|
|---|
| 103 | ip = tcb.so_next;
|
|---|
| 104 | if (ip == 0)
|
|---|
| 105 | return;
|
|---|
| 106 | QSOCKET_FOREACH(ip, ipnxt, tcp)
|
|---|
| 107 | /* { */
|
|---|
| 108 | ipnxt = ip->so_next;
|
|---|
| 109 | tp = sototcpcb(ip);
|
|---|
| 110 | if (tp == 0)
|
|---|
| 111 | CONTINUE(tcp);
|
|---|
| 112 | for (i = 0; i < TCPT_NTIMERS; i++)
|
|---|
| 113 | {
|
|---|
| 114 | if (tp->t_timer[i] && --tp->t_timer[i] == 0)
|
|---|
| 115 | {
|
|---|
| 116 | tcp_timers(pData, tp, i);
|
|---|
| 117 | if (ipnxt->so_prev != ip)
|
|---|
| 118 | goto tpgone;
|
|---|
| 119 | }
|
|---|
| 120 | }
|
|---|
| 121 | tp->t_idle++;
|
|---|
| 122 | if (tp->t_rtt)
|
|---|
| 123 | tp->t_rtt++;
|
|---|
| 124 | tpgone:
|
|---|
| 125 | ;
|
|---|
| 126 | LOOP_LABEL(tcp, ip, ipnxt);
|
|---|
| 127 | }
|
|---|
| 128 | tcp_iss += TCP_ISSINCR / PR_SLOWHZ; /* increment iss */
|
|---|
| 129 | #ifdef TCP_COMPAT_42
|
|---|
| 130 | if ((int)tcp_iss < 0)
|
|---|
| 131 | tcp_iss = 0; /* XXX */
|
|---|
| 132 | #endif
|
|---|
| 133 | tcp_now++; /* for timestamps */
|
|---|
| 134 | }
|
|---|
| 135 |
|
|---|
| 136 | /*
|
|---|
| 137 | * Cancel all timers for TCP tp.
|
|---|
| 138 | */
|
|---|
| 139 | void
|
|---|
| 140 | tcp_canceltimers(struct tcpcb *tp)
|
|---|
| 141 | {
|
|---|
| 142 | register int i;
|
|---|
| 143 |
|
|---|
| 144 | for (i = 0; i < TCPT_NTIMERS; i++)
|
|---|
| 145 | tp->t_timer[i] = 0;
|
|---|
| 146 | }
|
|---|
| 147 |
|
|---|
| 148 | const int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
|
|---|
| 149 | {
|
|---|
| 150 | 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64
|
|---|
| 151 | };
|
|---|
| 152 |
|
|---|
| 153 | /*
|
|---|
| 154 | * TCP timer processing.
|
|---|
| 155 | */
|
|---|
| 156 | struct tcpcb *
|
|---|
| 157 | tcp_timers(PNATState pData, register struct tcpcb *tp, int timer)
|
|---|
| 158 | {
|
|---|
| 159 | register int rexmt;
|
|---|
| 160 |
|
|---|
| 161 | LogFlowFunc(("ENTER: tp:%R[tcpcb793], timer:%d\n", tp, timer));
|
|---|
| 162 |
|
|---|
| 163 | switch (timer)
|
|---|
| 164 | {
|
|---|
| 165 | /*
|
|---|
| 166 | * 2 MSL timeout in shutdown went off. If we're closed but
|
|---|
| 167 | * still waiting for peer to close and connection has been idle
|
|---|
| 168 | * too long, or if 2MSL time is up from TIME_WAIT, delete connection
|
|---|
| 169 | * control block. Otherwise, check again in a bit.
|
|---|
| 170 | */
|
|---|
| 171 | case TCPT_2MSL:
|
|---|
| 172 | if (tp->t_state != TCPS_TIME_WAIT &&
|
|---|
| 173 | tp->t_idle <= tcp_maxidle)
|
|---|
| 174 | tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
|
|---|
| 175 | else
|
|---|
| 176 | tp = tcp_close(pData, tp);
|
|---|
| 177 | break;
|
|---|
| 178 |
|
|---|
| 179 | /*
|
|---|
| 180 | * Retransmission timer went off. Message has not
|
|---|
| 181 | * been acked within retransmit interval. Back off
|
|---|
| 182 | * to a longer retransmit interval and retransmit one segment.
|
|---|
| 183 | */
|
|---|
| 184 | case TCPT_REXMT:
|
|---|
| 185 | STAM_COUNTER_INC(&pData->StatTCP_retransmit);
|
|---|
| 186 | /*
|
|---|
| 187 | * XXX If a packet has timed out, then remove all the queued
|
|---|
| 188 | * packets for that session.
|
|---|
| 189 | */
|
|---|
| 190 | if (++tp->t_rxtshift > TCP_MAXRXTSHIFT)
|
|---|
| 191 | {
|
|---|
| 192 | /*
|
|---|
| 193 | * This is a hack to suit our terminal server here at the uni of canberra
|
|---|
| 194 | * since they have trouble with zeroes... It usually lets them through
|
|---|
| 195 | * unharmed, but under some conditions, it'll eat the zeros. If we
|
|---|
| 196 | * keep retransmitting it, it'll keep eating the zeroes, so we keep
|
|---|
| 197 | * retransmitting, and eventually the connection dies...
|
|---|
| 198 | * (this only happens on incoming data)
|
|---|
| 199 | *
|
|---|
| 200 | * So, if we were gonna drop the connection from too many retransmits,
|
|---|
| 201 | * don't... instead halve the t_maxseg, which might break up the NULLs and
|
|---|
| 202 | * let them through
|
|---|
| 203 | *
|
|---|
| 204 | * *sigh*
|
|---|
| 205 | */
|
|---|
| 206 | tp->t_maxseg >>= 1;
|
|---|
| 207 | if (tp->t_maxseg < 32)
|
|---|
| 208 | {
|
|---|
| 209 | /*
|
|---|
| 210 | * We tried our best, now the connection must die!
|
|---|
| 211 | */
|
|---|
| 212 | tp->t_rxtshift = TCP_MAXRXTSHIFT;
|
|---|
| 213 | tcpstat.tcps_timeoutdrop++;
|
|---|
| 214 | tp = tcp_drop(pData, tp, tp->t_softerror);
|
|---|
| 215 | /* tp->t_softerror : ETIMEDOUT); */ /* XXX */
|
|---|
| 216 | return (tp); /* XXX */
|
|---|
| 217 | }
|
|---|
| 218 |
|
|---|
| 219 | /*
|
|---|
| 220 | * Set rxtshift to 6, which is still at the maximum
|
|---|
| 221 | * backoff time
|
|---|
| 222 | */
|
|---|
| 223 | tp->t_rxtshift = 6;
|
|---|
| 224 | }
|
|---|
| 225 | tcpstat.tcps_rexmttimeo++;
|
|---|
| 226 | rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
|
|---|
| 227 | TCPT_RANGESET(tp->t_rxtcur, rexmt,
|
|---|
| 228 | (short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */
|
|---|
| 229 | tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
|
|---|
| 230 | /*
|
|---|
| 231 | * If losing, let the lower level know and try for
|
|---|
| 232 | * a better route. Also, if we backed off this far,
|
|---|
| 233 | * our srtt estimate is probably bogus. Clobber it
|
|---|
| 234 | * so we'll take the next rtt measurement as our srtt;
|
|---|
| 235 | * move the current srtt into rttvar to keep the current
|
|---|
| 236 | * retransmit times until then.
|
|---|
| 237 | */
|
|---|
| 238 | if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4)
|
|---|
| 239 | {
|
|---|
| 240 | /* in_losing(tp->t_inpcb); */
|
|---|
| 241 | tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
|
|---|
| 242 | tp->t_srtt = 0;
|
|---|
| 243 | }
|
|---|
| 244 | tp->snd_nxt = tp->snd_una;
|
|---|
| 245 | /*
|
|---|
| 246 | * If timing a segment in this window, stop the timer.
|
|---|
| 247 | */
|
|---|
| 248 | tp->t_rtt = 0;
|
|---|
| 249 | /*
|
|---|
| 250 | * Close the congestion window down to one segment
|
|---|
| 251 | * (we'll open it by one segment for each ack we get).
|
|---|
| 252 | * Since we probably have a window's worth of unacked
|
|---|
| 253 | * data accumulated, this "slow start" keeps us from
|
|---|
| 254 | * dumping all that data as back-to-back packets (which
|
|---|
| 255 | * might overwhelm an intermediate gateway).
|
|---|
| 256 | *
|
|---|
| 257 | * There are two phases to the opening: Initially we
|
|---|
| 258 | * open by one mss on each ack. This makes the window
|
|---|
| 259 | * size increase exponentially with time. If the
|
|---|
| 260 | * window is larger than the path can handle, this
|
|---|
| 261 | * exponential growth results in dropped packet(s)
|
|---|
| 262 | * almost immediately. To get more time between
|
|---|
| 263 | * drops but still "push" the network to take advantage
|
|---|
| 264 | * of improving conditions, we switch from exponential
|
|---|
| 265 | * to linear window opening at some threshold size.
|
|---|
| 266 | * For a threshold, we use half the current window
|
|---|
| 267 | * size, truncated to a multiple of the mss.
|
|---|
| 268 | *
|
|---|
| 269 | * (the minimum cwnd that will give us exponential
|
|---|
| 270 | * growth is 2 mss. We don't allow the threshold
|
|---|
| 271 | * to go below this.)
|
|---|
| 272 | */
|
|---|
| 273 | {
|
|---|
| 274 | u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
|
|---|
| 275 | if (win < 2)
|
|---|
| 276 | win = 2;
|
|---|
| 277 | tp->snd_cwnd = tp->t_maxseg;
|
|---|
| 278 | tp->snd_ssthresh = win * tp->t_maxseg;
|
|---|
| 279 | tp->t_dupacks = 0;
|
|---|
| 280 | }
|
|---|
| 281 | (void) tcp_output(pData, tp);
|
|---|
| 282 | break;
|
|---|
| 283 |
|
|---|
| 284 | /*
|
|---|
| 285 | * Persistence timer into zero window.
|
|---|
| 286 | * Force a byte to be output, if possible.
|
|---|
| 287 | */
|
|---|
| 288 | case TCPT_PERSIST:
|
|---|
| 289 | tcpstat.tcps_persisttimeo++;
|
|---|
| 290 | tcp_setpersist(tp);
|
|---|
| 291 | tp->t_force = 1;
|
|---|
| 292 | (void) tcp_output(pData, tp);
|
|---|
| 293 | tp->t_force = 0;
|
|---|
| 294 | break;
|
|---|
| 295 |
|
|---|
| 296 | /*
|
|---|
| 297 | * Keep-alive timer went off; send something
|
|---|
| 298 | * or drop connection if idle for too long.
|
|---|
| 299 | */
|
|---|
| 300 | case TCPT_KEEP:
|
|---|
| 301 | tcpstat.tcps_keeptimeo++;
|
|---|
| 302 | if (tp->t_state < TCPS_ESTABLISHED)
|
|---|
| 303 | goto dropit;
|
|---|
| 304 | /* if (tp->t_socket->so_options & SO_KEEPALIVE && */
|
|---|
| 305 | if ((so_options) && tp->t_state <= TCPS_CLOSE_WAIT)
|
|---|
| 306 | {
|
|---|
| 307 | if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
|
|---|
| 308 | goto dropit;
|
|---|
| 309 | /*
|
|---|
| 310 | * Send a packet designed to force a response
|
|---|
| 311 | * if the peer is up and reachable:
|
|---|
| 312 | * either an ACK if the connection is still alive,
|
|---|
| 313 | * or an RST if the peer has closed the connection
|
|---|
| 314 | * due to timeout or reboot.
|
|---|
| 315 | * Using sequence number tp->snd_una-1
|
|---|
| 316 | * causes the transmitted zero-length segment
|
|---|
| 317 | * to lie outside the receive window;
|
|---|
| 318 | * by the protocol spec, this requires the
|
|---|
| 319 | * correspondent TCP to respond.
|
|---|
| 320 | */
|
|---|
| 321 | tcpstat.tcps_keepprobe++;
|
|---|
| 322 | #ifdef TCP_COMPAT_42
|
|---|
| 323 | /*
|
|---|
| 324 | * The keepalive packet must have nonzero length
|
|---|
| 325 | * to get a 4.2 host to respond.
|
|---|
| 326 | */
|
|---|
| 327 | tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
|
|---|
| 328 | tp->rcv_nxt - 1, tp->snd_una - 1, 0);
|
|---|
| 329 | #else
|
|---|
| 330 | tcp_respond(pData, tp, &tp->t_template, (struct mbuf *)NULL,
|
|---|
| 331 | tp->rcv_nxt, tp->snd_una - 1, 0);
|
|---|
| 332 | #endif
|
|---|
| 333 | tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
|
|---|
| 334 | }
|
|---|
| 335 | else
|
|---|
| 336 | tp->t_timer[TCPT_KEEP] = tcp_keepidle;
|
|---|
| 337 | break;
|
|---|
| 338 |
|
|---|
| 339 | dropit:
|
|---|
| 340 | tcpstat.tcps_keepdrops++;
|
|---|
| 341 | tp = tcp_drop(pData, tp, 0); /* ETIMEDOUT); */
|
|---|
| 342 | break;
|
|---|
| 343 | }
|
|---|
| 344 |
|
|---|
| 345 | return tp;
|
|---|
| 346 | }
|
|---|
