Annotation of qemu/slirp/tcp_timer.c, revision 1.1.1.2

1.1       root        1: /*
                      2:  * Copyright (c) 1982, 1986, 1988, 1990, 1993
                      3:  *     The Regents of the University of California.  All rights reserved.
                      4:  *
                      5:  * Redistribution and use in source and binary forms, with or without
                      6:  * modification, are permitted provided that the following conditions
                      7:  * are met:
                      8:  * 1. Redistributions of source code must retain the above copyright
                      9:  *    notice, this list of conditions and the following disclaimer.
                     10:  * 2. Redistributions in binary form must reproduce the above copyright
                     11:  *    notice, this list of conditions and the following disclaimer in the
                     12:  *    documentation and/or other materials provided with the distribution.
                     13:  * 3. All advertising materials mentioning features or use of this software
                     14:  *    must display the following acknowledgement:
                     15:  *     This product includes software developed by the University of
                     16:  *     California, Berkeley and its contributors.
                     17:  * 4. Neither the name of the University nor the names of its contributors
                     18:  *    may be used to endorse or promote products derived from this software
                     19:  *    without specific prior written permission.
                     20:  *
                     21:  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
                     22:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
                     23:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
                     24:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
                     25:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
                     26:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
                     27:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
                     28:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
                     29:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
                     30:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
                     31:  * SUCH DAMAGE.
                     32:  *
                     33:  *     @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
                     34:  * tcp_timer.c,v 1.2 1994/08/02 07:49:10 davidg Exp
                     35:  */
                     36: 
                     37: #include <slirp.h>
                     38: 
1.1.1.2 ! root       39: #ifdef LOG_ENABLED
1.1       root       40: struct   tcpstat tcpstat;        /* tcp statistics */
1.1.1.2 ! root       41: #endif
        !            42: 
1.1       root       43: u_int32_t        tcp_now;                /* for RFC 1323 timestamps */
                     44: 
1.1.1.2 ! root       45: static struct tcpcb *tcp_timers(register struct tcpcb *tp, int timer);
        !            46: 
1.1       root       47: /*
                     48:  * Fast timeout routine for processing delayed acks
                     49:  */
                     50: void
                     51: tcp_fasttimo()
                     52: {
                     53:        register struct socket *so;
                     54:        register struct tcpcb *tp;
                     55: 
                     56:        DEBUG_CALL("tcp_fasttimo");
1.1.1.2 ! root       57: 
1.1       root       58:        so = tcb.so_next;
                     59:        if (so)
                     60:        for (; so != &tcb; so = so->so_next)
                     61:                if ((tp = (struct tcpcb *)so->so_tcpcb) &&
                     62:                    (tp->t_flags & TF_DELACK)) {
                     63:                        tp->t_flags &= ~TF_DELACK;
                     64:                        tp->t_flags |= TF_ACKNOW;
1.1.1.2 ! root       65:                        STAT(tcpstat.tcps_delack++);
1.1       root       66:                        (void) tcp_output(tp);
                     67:                }
                     68: }
                     69: 
                     70: /*
                     71:  * Tcp protocol timeout routine called every 500 ms.
                     72:  * Updates the timers in all active tcb's and
                     73:  * causes finite state machine actions if timers expire.
                     74:  */
                     75: void
                     76: tcp_slowtimo()
                     77: {
                     78:        register struct socket *ip, *ipnxt;
                     79:        register struct tcpcb *tp;
                     80:        register int i;
                     81: 
                     82:        DEBUG_CALL("tcp_slowtimo");
1.1.1.2 ! root       83: 
1.1       root       84:        /*
                     85:         * Search through tcb's and update active timers.
                     86:         */
                     87:        ip = tcb.so_next;
                     88:        if (ip == 0)
                     89:           return;
                     90:        for (; ip != &tcb; ip = ipnxt) {
                     91:                ipnxt = ip->so_next;
                     92:                tp = sototcpcb(ip);
                     93:                if (tp == 0)
                     94:                        continue;
                     95:                for (i = 0; i < TCPT_NTIMERS; i++) {
                     96:                        if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
                     97:                                tcp_timers(tp,i);
                     98:                                if (ipnxt->so_prev != ip)
                     99:                                        goto tpgone;
                    100:                        }
                    101:                }
                    102:                tp->t_idle++;
                    103:                if (tp->t_rtt)
                    104:                   tp->t_rtt++;
                    105: tpgone:
                    106:                ;
                    107:        }
                    108:        tcp_iss += TCP_ISSINCR/PR_SLOWHZ;               /* increment iss */
                    109: #ifdef TCP_COMPAT_42
                    110:        if ((int)tcp_iss < 0)
                    111:                tcp_iss = 0;                            /* XXX */
                    112: #endif
                    113:        tcp_now++;                                      /* for timestamps */
                    114: }
                    115: 
                    116: /*
                    117:  * Cancel all timers for TCP tp.
                    118:  */
                    119: void
                    120: tcp_canceltimers(tp)
                    121:        struct tcpcb *tp;
                    122: {
                    123:        register int i;
                    124: 
                    125:        for (i = 0; i < TCPT_NTIMERS; i++)
                    126:                tp->t_timer[i] = 0;
                    127: }
                    128: 
1.1.1.2 ! root      129: const int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
1.1       root      130:    { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
                    131: 
                    132: /*
                    133:  * TCP timer processing.
                    134:  */
1.1.1.2 ! root      135: static struct tcpcb *
        !           136: tcp_timers(register struct tcpcb *tp, int timer)
1.1       root      137: {
                    138:        register int rexmt;
1.1.1.2 ! root      139: 
1.1       root      140:        DEBUG_CALL("tcp_timers");
1.1.1.2 ! root      141: 
1.1       root      142:        switch (timer) {
                    143: 
                    144:        /*
                    145:         * 2 MSL timeout in shutdown went off.  If we're closed but
                    146:         * still waiting for peer to close and connection has been idle
                    147:         * too long, or if 2MSL time is up from TIME_WAIT, delete connection
                    148:         * control block.  Otherwise, check again in a bit.
                    149:         */
                    150:        case TCPT_2MSL:
                    151:                if (tp->t_state != TCPS_TIME_WAIT &&
1.1.1.2 ! root      152:                    tp->t_idle <= TCP_MAXIDLE)
        !           153:                        tp->t_timer[TCPT_2MSL] = TCPTV_KEEPINTVL;
1.1       root      154:                else
                    155:                        tp = tcp_close(tp);
                    156:                break;
                    157: 
                    158:        /*
                    159:         * Retransmission timer went off.  Message has not
                    160:         * been acked within retransmit interval.  Back off
                    161:         * to a longer retransmit interval and retransmit one segment.
                    162:         */
                    163:        case TCPT_REXMT:
1.1.1.2 ! root      164: 
1.1       root      165:                /*
                    166:                 * XXXXX If a packet has timed out, then remove all the queued
                    167:                 * packets for that session.
                    168:                 */
1.1.1.2 ! root      169: 
1.1       root      170:                if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
                    171:                        /*
                    172:                         * This is a hack to suit our terminal server here at the uni of canberra
                    173:                         * since they have trouble with zeroes... It usually lets them through
                    174:                         * unharmed, but under some conditions, it'll eat the zeros.  If we
                    175:                         * keep retransmitting it, it'll keep eating the zeroes, so we keep
                    176:                         * retransmitting, and eventually the connection dies...
                    177:                         * (this only happens on incoming data)
1.1.1.2 ! root      178:                         *
1.1       root      179:                         * So, if we were gonna drop the connection from too many retransmits,
                    180:                         * don't... instead halve the t_maxseg, which might break up the NULLs and
                    181:                         * let them through
1.1.1.2 ! root      182:                         *
1.1       root      183:                         * *sigh*
                    184:                         */
1.1.1.2 ! root      185: 
1.1       root      186:                        tp->t_maxseg >>= 1;
                    187:                        if (tp->t_maxseg < 32) {
                    188:                                /*
                    189:                                 * We tried our best, now the connection must die!
                    190:                                 */
                    191:                                tp->t_rxtshift = TCP_MAXRXTSHIFT;
1.1.1.2 ! root      192:                                STAT(tcpstat.tcps_timeoutdrop++);
1.1       root      193:                                tp = tcp_drop(tp, tp->t_softerror);
                    194:                                /* tp->t_softerror : ETIMEDOUT); */ /* XXX */
                    195:                                return (tp); /* XXX */
                    196:                        }
1.1.1.2 ! root      197: 
1.1       root      198:                        /*
                    199:                         * Set rxtshift to 6, which is still at the maximum
                    200:                         * backoff time
                    201:                         */
                    202:                        tp->t_rxtshift = 6;
                    203:                }
1.1.1.2 ! root      204:                STAT(tcpstat.tcps_rexmttimeo++);
1.1       root      205:                rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
                    206:                TCPT_RANGESET(tp->t_rxtcur, rexmt,
                    207:                    (short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */
                    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:                 * If timing a segment in this window, stop the timer.
                    225:                 */
                    226:                tp->t_rtt = 0;
                    227:                /*
                    228:                 * Close the congestion window down to one segment
                    229:                 * (we'll open it by one segment for each ack we get).
                    230:                 * Since we probably have a window's worth of unacked
                    231:                 * data accumulated, this "slow start" keeps us from
                    232:                 * dumping all that data as back-to-back packets (which
                    233:                 * might overwhelm an intermediate gateway).
                    234:                 *
                    235:                 * There are two phases to the opening: Initially we
                    236:                 * open by one mss on each ack.  This makes the window
                    237:                 * size increase exponentially with time.  If the
                    238:                 * window is larger than the path can handle, this
                    239:                 * exponential growth results in dropped packet(s)
1.1.1.2 ! root      240:                 * almost immediately.  To get more time between
1.1       root      241:                 * drops but still "push" the network to take advantage
                    242:                 * of improving conditions, we switch from exponential
                    243:                 * to linear window opening at some threshold size.
                    244:                 * For a threshold, we use half the current window
                    245:                 * size, truncated to a multiple of the mss.
                    246:                 *
                    247:                 * (the minimum cwnd that will give us exponential
                    248:                 * growth is 2 mss.  We don't allow the threshold
                    249:                 * to go below this.)
                    250:                 */
                    251:                {
                    252:                u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
                    253:                if (win < 2)
                    254:                        win = 2;
                    255:                tp->snd_cwnd = tp->t_maxseg;
                    256:                tp->snd_ssthresh = win * tp->t_maxseg;
                    257:                tp->t_dupacks = 0;
                    258:                }
                    259:                (void) tcp_output(tp);
                    260:                break;
                    261: 
                    262:        /*
                    263:         * Persistence timer into zero window.
                    264:         * Force a byte to be output, if possible.
                    265:         */
                    266:        case TCPT_PERSIST:
1.1.1.2 ! root      267:                STAT(tcpstat.tcps_persisttimeo++);
1.1       root      268:                tcp_setpersist(tp);
                    269:                tp->t_force = 1;
                    270:                (void) tcp_output(tp);
                    271:                tp->t_force = 0;
                    272:                break;
                    273: 
                    274:        /*
                    275:         * Keep-alive timer went off; send something
                    276:         * or drop connection if idle for too long.
                    277:         */
                    278:        case TCPT_KEEP:
1.1.1.2 ! root      279:                STAT(tcpstat.tcps_keeptimeo++);
1.1       root      280:                if (tp->t_state < TCPS_ESTABLISHED)
                    281:                        goto dropit;
                    282: 
                    283: /*             if (tp->t_socket->so_options & SO_KEEPALIVE && */
1.1.1.2 ! root      284:                if ((SO_OPTIONS) && tp->t_state <= TCPS_CLOSE_WAIT) {
        !           285:                        if (tp->t_idle >= TCPTV_KEEP_IDLE + TCP_MAXIDLE)
1.1       root      286:                                goto dropit;
                    287:                        /*
                    288:                         * Send a packet designed to force a response
                    289:                         * if the peer is up and reachable:
                    290:                         * either an ACK if the connection is still alive,
                    291:                         * or an RST if the peer has closed the connection
                    292:                         * due to timeout or reboot.
                    293:                         * Using sequence number tp->snd_una-1
                    294:                         * causes the transmitted zero-length segment
                    295:                         * to lie outside the receive window;
                    296:                         * by the protocol spec, this requires the
                    297:                         * correspondent TCP to respond.
                    298:                         */
1.1.1.2 ! root      299:                        STAT(tcpstat.tcps_keepprobe++);
1.1       root      300: #ifdef TCP_COMPAT_42
                    301:                        /*
                    302:                         * The keepalive packet must have nonzero length
                    303:                         * to get a 4.2 host to respond.
                    304:                         */
                    305:                        tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
                    306:                            tp->rcv_nxt - 1, tp->snd_una - 1, 0);
                    307: #else
                    308:                        tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
                    309:                            tp->rcv_nxt, tp->snd_una - 1, 0);
                    310: #endif
1.1.1.2 ! root      311:                        tp->t_timer[TCPT_KEEP] = TCPTV_KEEPINTVL;
1.1       root      312:                } else
1.1.1.2 ! root      313:                        tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_IDLE;
1.1       root      314:                break;
                    315: 
                    316:        dropit:
1.1.1.2 ! root      317:                STAT(tcpstat.tcps_keepdrops++);
1.1       root      318:                tp = tcp_drop(tp, 0); /* ETIMEDOUT); */
                    319:                break;
                    320:        }
                    321: 
                    322:        return (tp);
                    323: }

unix.superglobalmegacorp.com