workbench/network/stacks/AROSTCP/bsdsocket/netinet/tcp_timer.c

   1 /*
   2  * Copyright (c) 1982, 1986, 1988, 1990, 1993
   3  *      The Regents of the University of California.  All rights reserved.
   4  * Copyright (c) 2006
   5  *      Pavel Fedin
   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. All advertising materials mentioning features or use of this software
  16  *    must display the following acknowledgement:
  17  *      This product includes software developed by the University of
  18  *      California, Berkeley and its contributors.
  19  * 4. Neither the name of the University nor the names of its contributors
  20  *    may be used to endorse or promote products derived from this software
  21  *    without specific prior written permission.
  22  *
  23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  33  * SUCH DAMAGE.
  34  *
  35  *      @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
  36  * $Id$
  37  */
  38
  39 #ifndef TUBA_INCLUDE
  40 #include <sys/param.h>
  41 #include <sys/systm.h>
  42 #include <sys/malloc.h>
  43 #include <sys/mbuf.h>
  44 #include <sys/socket.h>
  45 #include <sys/socketvar.h>
  46 #include <sys/protosw.h>
  47 #include <sys/errno.h>
  48 #include <sys/queue.h>
  49 #include <sys/synch.h>
  50
  51 #include <net/route.h>
  52 #include <net/if.h>
  53
  54 #include <netinet/in.h>
  55 #include <netinet/in_systm.h>
  56 #include <netinet/ip.h>
  57 #include <netinet/in_pcb.h>
  58 #include <netinet/ip_var.h>
  59 #include <netinet/tcp.h>
  60 #include <netinet/tcp_fsm.h>
  61 #include <netinet/tcp_seq.h>
  62 #include <netinet/tcp_timer.h>
  63 #include <netinet/tcp_var.h>
  64 #include <netinet/tcpip.h>
  65
  66 int     tcp_keepidle = TCPTV_KEEP_IDLE;
  67 int     tcp_keepintvl = TCPTV_KEEPINTVL;
  68 int     tcp_maxidle;
  69 #endif /* TUBA_INCLUDE */
  70 /*
  71  * Fast timeout routine for processing delayed acks
  72  */
  73 void
  74 tcp_fasttimo()
  75 {
  76         register struct inpcb *inp;
  77         register struct tcpcb *tp;
  78         int s;
  79
  80         s = splnet();
  81
  82         for (inp = tcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
  83                 if ((tp = (struct tcpcb *)inp->inp_ppcb) &&
  84                     (tp->t_flags & TF_DELACK)) {
  85                         tp->t_flags &= ~TF_DELACK;
  86                         tp->t_flags |= TF_ACKNOW;
  87                         tcpstat.tcps_delack++;
  88                         (void) tcp_output(tp);
  89                 }
  90         }
  91         splx(s);
  92 }
  93
  94 /*
  95  * Tcp protocol timeout routine called every 500 ms.
  96  * Updates the timers in all active tcb's and
  97  * causes finite state machine actions if timers expire.
  98  */
  99 void
 100 tcp_slowtimo()
 101 {
 102         register struct inpcb *ip, *ipnxt;
 103         register struct tcpcb *tp;
 104         register long i;
 105         int s;
 106
 107         s = splnet();
 108
 109         tcp_maxidle = TCPTV_KEEPCNT * tcp_keepintvl;
 110
 111         ip = tcb.lh_first;
 112         if (ip == NULL) {
 113                 splx(s);
 114                 return;
 115         }
 116         /*
 117          * Search through tcb's and update active timers.
 118          */
 119         for (; ip != NULL; ip = ipnxt) {
 120                 ipnxt = ip->inp_list.le_next;
 121                 tp = intotcpcb(ip);
 122                 if (tp == 0)
 123                         continue;
 124                 for (i = 0; i < TCPT_NTIMERS; i++) {
 125                         if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
 126                                 if (tcp_usrreq(tp->t_inpcb->inp_socket,
 127                                     PRU_SLOWTIMO, (struct mbuf *)0,
 128                                     (struct mbuf *)i, (struct mbuf *)0) == 0)
 129                                         goto tpgone;
 130                         }
 131                 }
 132                 tp->t_idle++;
 133                 tp->t_duration++;
 134                 if (tp->t_rtt)
 135                         tp->t_rtt++;
 136 tpgone:
 137                 ;
 138         }
 139         tcp_iss += TCP_ISSINCR/PR_SLOWHZ;               /* increment iss */
 140 #ifdef TCP_COMPAT_42
 141         if ((int)tcp_iss < 0)
 142                 tcp_iss = 0;                            /* XXX */
 143 #endif
 144         tcp_now++;                                      /* for timestamps */
 145         splx(s);
 146 }
 147 #ifndef TUBA_INCLUDE
 148
 149 /*
 150  * Cancel all timers for TCP tp.
 151  */
 152 void
 153 tcp_canceltimers(tp)
 154         struct tcpcb *tp;
 155 {
 156         register int i;
 157
 158         for (i = 0; i < TCPT_NTIMERS; i++)
 159                 tp->t_timer[i] = 0;
 160 }
 161
 162 int     tcp_backoff[TCP_MAXRXTSHIFT + 1] =
 163     { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
 164
 165 /*
 166  * TCP timer processing.
 167  */
 168 struct tcpcb *
 169 tcp_timers(tp, timer)
 170         register struct tcpcb *tp;
 171         int timer;
 172 {
 173         register int rexmt;
 174
 175         switch (timer) {
 176
 177         /*
 178          * 2 MSL timeout in shutdown went off.  If we're closed but
 179          * still waiting for peer to close and connection has been idle
 180          * too long, or if 2MSL time is up from TIME_WAIT, delete connection
 181          * control block.  Otherwise, check again in a bit.
 182          */
 183         case TCPT_2MSL:
 184                 if (tp->t_state != TCPS_TIME_WAIT &&
 185                     tp->t_idle <= tcp_maxidle)
 186                         tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
 187                 else
 188                         tp = tcp_close(tp);
 189                 break;
 190
 191         /*
 192          * Retransmission timer went off.  Message has not
 193          * been acked within retransmit interval.  Back off
 194          * to a longer retransmit interval and retransmit one segment.
 195          */
 196         case TCPT_REXMT:
 197                 if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
 198                         tp->t_rxtshift = TCP_MAXRXTSHIFT;
 199                         tcpstat.tcps_timeoutdrop++;
 200                         tp = tcp_drop(tp, tp->t_softerror ?
 201                             tp->t_softerror : ETIMEDOUT);
 202                         break;
 203                 }
 204                 tcpstat.tcps_rexmttimeo++;
 205                 rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
 206                 TCPT_RANGESET(tp->t_rxtcur, rexmt,
 207                     tp->t_rttmin, TCPTV_REXMTMAX);
 208                 tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
 209                 /*
 210                  * If losing, let the lower level know and try for
 211                  * a better route.  Also, if we backed off this far,
 212                  * our srtt estimate is probably bogus.  Clobber it
 213                  * so we'll take the next rtt measurement as our srtt;
 214                  * move the current srtt into rttvar to keep the current
 215                  * retransmit times until then.
 216                  */
 217                 if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
 218                         in_losing(tp->t_inpcb);
 219                         tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
 220                         tp->t_srtt = 0;
 221                 }
 222                 tp->snd_nxt = tp->snd_una;
 223                 /*
 224                  * Force a segment to be sent.
 225                  */
 226                 tp->t_flags |= TF_ACKNOW;
 227                 /*
 228                  * If timing a segment in this window, stop the timer.
 229                  */
 230                 tp->t_rtt = 0;
 231                 /*
 232                  * Close the congestion window down to one segment
 233                  * (we'll open it by one segment for each ack we get).
 234                  * Since we probably have a window's worth of unacked
 235                  * data accumulated, this "slow start" keeps us from
 236                  * dumping all that data as back-to-back packets (which
 237                  * might overwhelm an intermediate gateway).
 238                  *
 239                  * There are two phases to the opening: Initially we
 240                  * open by one mss on each ack.  This makes the window
 241                  * size increase exponentially with time.  If the
 242                  * window is larger than the path can handle, this
 243                  * exponential growth results in dropped packet(s)
 244                  * almost immediately.  To get more time between
 245                  * drops but still "push" the network to take advantage
 246                  * of improving conditions, we switch from exponential
 247                  * to linear window opening at some threshhold size.
 248                  * For a threshhold, we use half the current window
 249                  * size, truncated to a multiple of the mss.
 250                  *
 251                  * (the minimum cwnd that will give us exponential
 252                  * growth is 2 mss.  We don't allow the threshhold
 253                  * to go below this.)
 254                  */
 255                 {
 256                 u_int win = MIN(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
 257                 if (win < 2)
 258                         win = 2;
 259                 tp->snd_cwnd = tp->t_maxseg;
 260                 tp->snd_ssthresh = win * tp->t_maxseg;
 261                 tp->t_dupacks = 0;
 262                 }
 263                 (void) tcp_output(tp);
 264                 break;
 265
 266         /*
 267          * Persistance timer into zero window.
 268          * Force a byte to be output, if possible.
 269          */
 270         case TCPT_PERSIST:
 271                 tcpstat.tcps_persisttimeo++;
 272                 tcp_setpersist(tp);
 273                 tp->t_force = 1;
 274                 (void) tcp_output(tp);
 275                 tp->t_force = 0;
 276                 break;
 277
 278         /*
 279          * Keep-alive timer went off; send something
 280          * or drop connection if idle for too long.
 281          */
 282         case TCPT_KEEP:
 283                 tcpstat.tcps_keeptimeo++;
 284                 if (tp->t_state < TCPS_ESTABLISHED)
 285                         goto dropit;
 286                 if (tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE &&
 287                     tp->t_state <= TCPS_CLOSE_WAIT) {
 288                         if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
 289                                 goto dropit;
 290                         /*
 291                          * Send a packet designed to force a response
 292                          * if the peer is up and reachable:
 293                          * either an ACK if the connection is still alive,
 294                          * or an RST if the peer has closed the connection
 295                          * due to timeout or reboot.
 296                          * Using sequence number tp->snd_una-1
 297                          * causes the transmitted zero-length segment
 298                          * to lie outside the receive window;
 299                          * by the protocol spec, this requires the
 300                          * correspondent TCP to respond.
 301                          */
 302                         tcpstat.tcps_keepprobe++;
 303 #ifdef TCP_COMPAT_42
 304                         /*
 305                          * The keepalive packet must have nonzero length
 306                          * to get a 4.2 host to respond.
 307                          */
 308                         tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
 309                             tp->rcv_nxt - 1, tp->snd_una - 1, 0);
 310 #else
 311                         tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
 312                             tp->rcv_nxt, tp->snd_una - 1, 0);
 313 #endif
 314                         tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
 315                 } else
 316                         tp->t_timer[TCPT_KEEP] = tcp_keepidle;
 317                 break;
 318         dropit:
 319                 tcpstat.tcps_keepdrops++;
 320                 tp = tcp_drop(tp, ETIMEDOUT);
 321                 break;
 322         }
 323         return (tp);
 324 }
 325 #endif /* TUBA_INCLUDE */