Return to tcp_timer.c CVS log

Up to [Apple XNU] / XNU / bsd / netinet

Annotation of XNU/bsd/netinet/tcp_timer.c, revision 1.1

1.1     ! root        1: /*
        !             2:  * Copyright (c) 2000 Apple Computer, Inc. All rights reserved.
        !             3:  *
        !             4:  * @APPLE_LICENSE_HEADER_START@
        !             5:  * 
        !             6:  * The contents of this file constitute Original Code as defined in and
        !             7:  * are subject to the Apple Public Source License Version 1.1 (the
        !             8:  * "License").  You may not use this file except in compliance with the
        !             9:  * License.  Please obtain a copy of the License at
        !            10:  * http://www.apple.com/publicsource and read it before using this file.
        !            11:  * 
        !            12:  * This Original Code and all software distributed under the License are
        !            13:  * distributed on an "AS IS" basis, WITHOUT WARRANTY OF ANY KIND, EITHER
        !            14:  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
        !            15:  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
        !            16:  * FITNESS FOR A PARTICULAR PURPOSE OR NON-INFRINGEMENT.  Please see the
        !            17:  * License for the specific language governing rights and limitations
        !            18:  * under the License.
        !            19:  * 
        !            20:  * @APPLE_LICENSE_HEADER_END@
        !            21:  */
        !            22: /*
        !            23:  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
        !            24:  *     The Regents of the University of California.  All rights reserved.
        !            25:  *
        !            26:  * Redistribution and use in source and binary forms, with or without
        !            27:  * modification, are permitted provided that the following conditions
        !            28:  * are met:
        !            29:  * 1. Redistributions of source code must retain the above copyright
        !            30:  *    notice, this list of conditions and the following disclaimer.
        !            31:  * 2. Redistributions in binary form must reproduce the above copyright
        !            32:  *    notice, this list of conditions and the following disclaimer in the
        !            33:  *    documentation and/or other materials provided with the distribution.
        !            34:  * 3. All advertising materials mentioning features or use of this software
        !            35:  *    must display the following acknowledgement:
        !            36:  *     This product includes software developed by the University of
        !            37:  *     California, Berkeley and its contributors.
        !            38:  * 4. Neither the name of the University nor the names of its contributors
        !            39:  *    may be used to endorse or promote products derived from this software
        !            40:  *    without specific prior written permission.
        !            41:  *
        !            42:  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
        !            43:  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
        !            44:  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
        !            45:  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
        !            46:  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
        !            47:  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
        !            48:  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
        !            49:  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
        !            50:  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
        !            51:  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
        !            52:  * SUCH DAMAGE.
        !            53:  *
        !            54:  *     @(#)tcp_timer.c 8.2 (Berkeley) 5/24/95
        !            55:  */
        !            56: 
        !            57: #if ISFB31
        !            58: #include "opt_compat.h"
        !            59: #include "opt_tcpdebug.h"
        !            60: #endif
        !            61: 
        !            62: #include <sys/param.h>
        !            63: #include <sys/systm.h>
        !            64: #include <sys/kernel.h>
        !            65: #include <sys/sysctl.h>
        !            66: #include <sys/socket.h>
        !            67: #include <sys/socketvar.h>
        !            68: #include <sys/protosw.h>
        !            69: 
        !            70: #include <kern/cpu_number.h>   /* before tcp_seq.h, for tcp_random18() */
        !            71: 
        !            72: #include <net/route.h>
        !            73: 
        !            74: #include <netinet/in.h>
        !            75: #include <netinet/in_systm.h>
        !            76: #include <netinet/in_pcb.h>
        !            77: #include <netinet/ip_var.h>
        !            78: #include <netinet/tcp.h>
        !            79: #include <netinet/tcp_fsm.h>
        !            80: #include <netinet/tcp_seq.h>
        !            81: #include <netinet/tcp_timer.h>
        !            82: #include <netinet/tcp_var.h>
        !            83: #include <netinet/tcpip.h>
        !            84: #if TCPDEBUG
        !            85: #include <netinet/tcp_debug.h>
        !            86: #endif
        !            87: 
        !            88: int    tcp_keepinit = TCPTV_KEEP_INIT;
        !            89: SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPINIT, keepinit,
        !            90:        CTLFLAG_RW, &tcp_keepinit , 0, "");
        !            91: 
        !            92: int    tcp_keepidle = TCPTV_KEEP_IDLE;
        !            93: SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPIDLE, keepidle,
        !            94:        CTLFLAG_RW, &tcp_keepidle , 0, "");
        !            95: 
        !            96: static int     tcp_keepintvl = TCPTV_KEEPINTVL;
        !            97: SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPINTVL, keepintvl,
        !            98:        CTLFLAG_RW, &tcp_keepintvl , 0, "");
        !            99: 
        !           100: static int     always_keepalive = 0;
        !           101: SYSCTL_INT(_net_inet_tcp, OID_AUTO, always_keepalive,
        !           102:        CTLFLAG_RW, &always_keepalive , 0, "");
        !           103: 
        !           104: static int     tcp_keepcnt = TCPTV_KEEPCNT;
        !           105:        /* max idle probes */
        !           106: static int     tcp_maxpersistidle = TCPTV_KEEP_IDLE;
        !           107:        /* max idle time in persist */
        !           108: int    tcp_maxidle;
        !           109: 
        !           110: /*
        !           111:  * Fast timeout routine for processing delayed acks
        !           112:  */
        !           113: void
        !           114: tcp_fasttimo()
        !           115: {
        !           116:        register struct inpcb *inp;
        !           117:        register struct tcpcb *tp;
        !           118:        int s;
        !           119: 
        !           120:        if (tcp_delack_enabled) {
        !           121:                s = splnet();
        !           122:                for (inp = tcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
        !           123:                        if ((tp = (struct tcpcb *)inp->inp_ppcb) &&
        !           124:                            (tp->t_flags & TF_DELACK)) {
        !           125:                                tp->t_flags &= ~TF_DELACK;
        !           126:                                tp->t_flags |= TF_ACKNOW;
        !           127:                                tcpstat.tcps_delack++;
        !           128:                                (void) tcp_output(tp);
        !           129:                        }
        !           130:                }
        !           131:                splx(s);
        !           132:        }
        !           133: }
        !           134: 
        !           135: /*
        !           136:  * Tcp protocol timeout routine called every 500 ms.
        !           137:  * Updates the timers in all active tcb's and
        !           138:  * causes finite state machine actions if timers expire.
        !           139:  */
        !           140: void
        !           141: tcp_slowtimo()
        !           142: {
        !           143:        register struct inpcb *ip, *ipnxt;
        !           144:        register struct tcpcb *tp;
        !           145:        register int i;
        !           146:        int s;
        !           147: #if TCPDEBUG
        !           148:        int ostate;
        !           149: #endif
        !           150: 
        !           151:        s = splnet();
        !           152: 
        !           153:        tcp_maxidle = tcp_keepcnt * tcp_keepintvl;
        !           154: 
        !           155:        ip = tcb.lh_first;
        !           156:        if (ip == NULL) {
        !           157:                splx(s);
        !           158:                return;
        !           159:        }
        !           160:        /*
        !           161:         * Search through tcb's and update active timers.
        !           162:         */
        !           163:        for (; ip != NULL; ip = ipnxt) {
        !           164:                ipnxt = ip->inp_list.le_next;
        !           165:                tp = intotcpcb(ip);
        !           166:                if (tp == 0 || tp->t_state == TCPS_LISTEN)
        !           167:                        continue;
        !           168:                for (i = 0; i < TCPT_NTIMERS; i++) {
        !           169:                        if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
        !           170: #if TCPDEBUG
        !           171:                                ostate = tp->t_state;
        !           172: #endif
        !           173:                                tp = tcp_timers(tp, i);
        !           174:                                if (tp == NULL)
        !           175:                                        goto tpgone;
        !           176: #if TCPDEBUG
        !           177:                                if (tp->t_inpcb->inp_socket->so_options
        !           178:                                    & SO_DEBUG)
        !           179:                                        tcp_trace(TA_USER, ostate, tp,
        !           180:                                                  (struct tcpiphdr *)0,
        !           181:                                                  PRU_SLOWTIMO);
        !           182: #endif
        !           183:                        }
        !           184:                }
        !           185:                tp->t_idle++;
        !           186:                tp->t_duration++;
        !           187:                if (tp->t_rtt)
        !           188:                        tp->t_rtt++;
        !           189: tpgone:
        !           190:                ;
        !           191:        }
        !           192:        tcp_iss += TCP_ISSINCR/PR_SLOWHZ;               /* increment iss */
        !           193: #if TCP_COMPAT_42
        !           194:        if ((int)tcp_iss < 0)
        !           195:                tcp_iss = TCP_ISSINCR;                  /* XXX */
        !           196: #endif
        !           197:        tcp_now++;                                      /* for timestamps */
        !           198:        splx(s);
        !           199: }
        !           200: 
        !           201: /*
        !           202:  * Cancel all timers for TCP tp.
        !           203:  */
        !           204: void
        !           205: tcp_canceltimers(tp)
        !           206:        struct tcpcb *tp;
        !           207: {
        !           208:        register int i;
        !           209: 
        !           210:        for (i = 0; i < TCPT_NTIMERS; i++)
        !           211:                tp->t_timer[i] = 0;
        !           212: }
        !           213: 
        !           214: int    tcp_backoff[TCP_MAXRXTSHIFT + 1] =
        !           215:     { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
        !           216: 
        !           217: static int tcp_totbackoff = 511;       /* sum of tcp_backoff[] */
        !           218: 
        !           219: /*
        !           220:  * TCP timer processing.
        !           221:  */
        !           222: struct tcpcb *
        !           223: tcp_timers(tp, timer)
        !           224:        register struct tcpcb *tp;
        !           225:        int timer;
        !           226: {
        !           227:        register int rexmt;
        !           228:        struct socket *so_tmp;
        !           229: 
        !           230:        switch (timer) {
        !           231: 
        !           232:        /*
        !           233:         * 2 MSL timeout in shutdown went off.  If we're closed but
        !           234:         * still waiting for peer to close and connection has been idle
        !           235:         * too long, or if 2MSL time is up from TIME_WAIT, delete connection
        !           236:         * control block.  Otherwise, check again in a bit.
        !           237:         */
        !           238:        case TCPT_2MSL:
        !           239:                if (tp->t_state != TCPS_TIME_WAIT &&
        !           240:                    tp->t_idle <= tcp_maxidle)
        !           241:                        tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
        !           242:                else
        !           243:                        tp = tcp_close(tp);
        !           244:                break;
        !           245: 
        !           246:        /*
        !           247:         * Retransmission timer went off.  Message has not
        !           248:         * been acked within retransmit interval.  Back off
        !           249:         * to a longer retransmit interval and retransmit one segment.
        !           250:         */
        !           251:        case TCPT_REXMT:
        !           252:                if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
        !           253:                        tp->t_rxtshift = TCP_MAXRXTSHIFT;
        !           254:                        tcpstat.tcps_timeoutdrop++;
        !           255:                        so_tmp = tp->t_inpcb->inp_socket;
        !           256:                        tp = tcp_drop(tp, tp->t_softerror ?
        !           257:                            tp->t_softerror : ETIMEDOUT);
        !           258:                        postevent(so_tmp, 0, EV_TIMEOUT);                       
        !           259:                        break;
        !           260:                }
        !           261:                tcpstat.tcps_rexmttimeo++;
        !           262:                rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
        !           263:                TCPT_RANGESET(tp->t_rxtcur, rexmt,
        !           264:                    tp->t_rttmin, TCPTV_REXMTMAX);
        !           265:                tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
        !           266:                /*
        !           267:                 * If losing, let the lower level know and try for
        !           268:                 * a better route.  Also, if we backed off this far,
        !           269:                 * our srtt estimate is probably bogus.  Clobber it
        !           270:                 * so we'll take the next rtt measurement as our srtt;
        !           271:                 * move the current srtt into rttvar to keep the current
        !           272:                 * retransmit times until then.
        !           273:                 */
        !           274:                if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
        !           275:                        in_losing(tp->t_inpcb);
        !           276:                        tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
        !           277:                        tp->t_srtt = 0;
        !           278:                }
        !           279:                tp->snd_nxt = tp->snd_una;
        !           280:                /*
        !           281:                 * Force a segment to be sent.
        !           282:                 */
        !           283:                tp->t_flags |= TF_ACKNOW;
        !           284:                /*
        !           285:                 * If timing a segment in this window, stop the timer.
        !           286:                 */
        !           287:                tp->t_rtt = 0;
        !           288:                /*
        !           289:                 * Close the congestion window down to one segment
        !           290:                 * (we'll open it by one segment for each ack we get).
        !           291:                 * Since we probably have a window's worth of unacked
        !           292:                 * data accumulated, this "slow start" keeps us from
        !           293:                 * dumping all that data as back-to-back packets (which
        !           294:                 * might overwhelm an intermediate gateway).
        !           295:                 *
        !           296:                 * There are two phases to the opening: Initially we
        !           297:                 * open by one mss on each ack.  This makes the window
        !           298:                 * size increase exponentially with time.  If the
        !           299:                 * window is larger than the path can handle, this
        !           300:                 * exponential growth results in dropped packet(s)
        !           301:                 * almost immediately.  To get more time between
        !           302:                 * drops but still "push" the network to take advantage
        !           303:                 * of improving conditions, we switch from exponential
        !           304:                 * to linear window opening at some threshhold size.
        !           305:                 * For a threshhold, we use half the current window
        !           306:                 * size, truncated to a multiple of the mss.
        !           307:                 *
        !           308:                 * (the minimum cwnd that will give us exponential
        !           309:                 * growth is 2 mss.  We don't allow the threshhold
        !           310:                 * to go below this.)
        !           311:                 */
        !           312:                {
        !           313:                u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
        !           314:                if (win < 2)
        !           315:                        win = 2;
        !           316:                tp->snd_cwnd = tp->t_maxseg;
        !           317:                tp->snd_ssthresh = win * tp->t_maxseg;
        !           318:                tp->t_dupacks = 0;
        !           319:                }
        !           320:                (void) tcp_output(tp);
        !           321:                break;
        !           322: 
        !           323:        /*
        !           324:         * Persistance timer into zero window.
        !           325:         * Force a byte to be output, if possible.
        !           326:         */
        !           327:        case TCPT_PERSIST:
        !           328:                tcpstat.tcps_persisttimeo++;
        !           329:                /*
        !           330:                 * Hack: if the peer is dead/unreachable, we do not
        !           331:                 * time out if the window is closed.  After a full
        !           332:                 * backoff, drop the connection if the idle time
        !           333:                 * (no responses to probes) reaches the maximum
        !           334:                 * backoff that we would use if retransmitting.
        !           335:                 */
        !           336:                if (tp->t_rxtshift == TCP_MAXRXTSHIFT &&
        !           337:                    (tp->t_idle >= tcp_maxpersistidle ||
        !           338:                    tp->t_idle >= TCP_REXMTVAL(tp) * tcp_totbackoff)) {
        !           339:                        tcpstat.tcps_persistdrop++;
        !           340:                        so_tmp = tp->t_inpcb->inp_socket;
        !           341:                        tp = tcp_drop(tp, ETIMEDOUT);
        !           342:                        postevent(so_tmp, 0, EV_TIMEOUT);
        !           343:                        break;
        !           344:                }
        !           345:                tcp_setpersist(tp);
        !           346:                tp->t_force = 1;
        !           347:                (void) tcp_output(tp);
        !           348:                tp->t_force = 0;
        !           349:                break;
        !           350: 
        !           351:        /*
        !           352:         * Keep-alive timer went off; send something
        !           353:         * or drop connection if idle for too long.
        !           354:         */
        !           355:        case TCPT_KEEP:
        !           356:                tcpstat.tcps_keeptimeo++;
        !           357:                if (tp->t_state < TCPS_ESTABLISHED)
        !           358:                        goto dropit;
        !           359:                if ((always_keepalive ||
        !           360:                    tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE) &&
        !           361:                    tp->t_state <= TCPS_CLOSING) {
        !           362:                        if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
        !           363:                                goto dropit;
        !           364:                        /*
        !           365:                         * Send a packet designed to force a response
        !           366:                         * if the peer is up and reachable:
        !           367:                         * either an ACK if the connection is still alive,
        !           368:                         * or an RST if the peer has closed the connection
        !           369:                         * due to timeout or reboot.
        !           370:                         * Using sequence number tp->snd_una-1
        !           371:                         * causes the transmitted zero-length segment
        !           372:                         * to lie outside the receive window;
        !           373:                         * by the protocol spec, this requires the
        !           374:                         * correspondent TCP to respond.
        !           375:                         */
        !           376:                        tcpstat.tcps_keepprobe++;
        !           377: #if TCP_COMPAT_42
        !           378:                        /*
        !           379:                         * The keepalive packet must have nonzero length
        !           380:                         * to get a 4.2 host to respond.
        !           381:                         */
        !           382:                        tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
        !           383:                            tp->rcv_nxt - 1, tp->snd_una - 1, 0);
        !           384: #else
        !           385:                        tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
        !           386:                            tp->rcv_nxt, tp->snd_una - 1, 0);
        !           387: #endif
        !           388:                        tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
        !           389:                } else
        !           390:                        tp->t_timer[TCPT_KEEP] = tcp_keepidle;
        !           391:                break;
        !           392:        dropit:
        !           393:                tcpstat.tcps_keepdrops++;
        !           394:                so_tmp = tp->t_inpcb->inp_socket;
        !           395:                tp = tcp_drop(tp, ETIMEDOUT);
        !           396:                postevent(so_tmp, 0, EV_TIMEOUT);
        !           397:                break;
        !           398:        }
        !           399:        return (tp);
        !           400: }

This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.