[qemu.git] / slirp / tcp_timer.c

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)tcp_timer.c	8.1 (Berkeley) 6/10/93
 * tcp_timer.c,v 1.2 1994/08/02 07:49:10 davidg Exp
 */

#include <slirp.h>

static struct tcpcb *tcp_timers(register struct tcpcb *tp, int timer);

/*
 * Fast timeout routine for processing delayed acks
 */
void
tcp_fasttimo(Slirp *slirp)
{
	register struct socket *so;
	register struct tcpcb *tp;

	DEBUG_CALL("tcp_fasttimo");

	so = slirp->tcb.so_next;
	if (so)
	for (; so != &slirp->tcb; so = so->so_next)
		if ((tp = (struct tcpcb *)so->so_tcpcb) &&
		    (tp->t_flags & TF_DELACK)) {
			tp->t_flags &= ~TF_DELACK;
			tp->t_flags |= TF_ACKNOW;
			(void) tcp_output(tp);
		}
}

/*
 * Tcp protocol timeout routine called every 500 ms.
 * Updates the timers in all active tcb's and
 * causes finite state machine actions if timers expire.
 */
void
tcp_slowtimo(Slirp *slirp)
{
	register struct socket *ip, *ipnxt;
	register struct tcpcb *tp;
	register int i;

	DEBUG_CALL("tcp_slowtimo");

	/*
	 * Search through tcb's and update active timers.
	 */
	ip = slirp->tcb.so_next;
        if (ip == NULL) {
            return;
        }
	for (; ip != &slirp->tcb; ip = ipnxt) {
		ipnxt = ip->so_next;
		tp = sototcpcb(ip);
                if (tp == NULL) {
                        continue;
                }
		for (i = 0; i < TCPT_NTIMERS; i++) {
			if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
				tcp_timers(tp,i);
				if (ipnxt->so_prev != ip)
					goto tpgone;
			}
		}
		tp->t_idle++;
		if (tp->t_rtt)
		   tp->t_rtt++;
tpgone:
		;
	}
	slirp->tcp_iss += TCP_ISSINCR/PR_SLOWHZ;	/* increment iss */
	slirp->tcp_now++;				/* for timestamps */
}

/*
 * Cancel all timers for TCP tp.
 */
void
tcp_canceltimers(struct tcpcb *tp)
{
	register int i;

	for (i = 0; i < TCPT_NTIMERS; i++)
		tp->t_timer[i] = 0;
}

const int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
   { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };

/*
 * TCP timer processing.
 */
static struct tcpcb *
tcp_timers(register struct tcpcb *tp, int timer)
{
	register int rexmt;

	DEBUG_CALL("tcp_timers");

	switch (timer) {

	/*
	 * 2 MSL timeout in shutdown went off.  If we're closed but
	 * still waiting for peer to close and connection has been idle
	 * too long, or if 2MSL time is up from TIME_WAIT, delete connection
	 * control block.  Otherwise, check again in a bit.
	 */
	case TCPT_2MSL:
		if (tp->t_state != TCPS_TIME_WAIT &&
		    tp->t_idle <= TCP_MAXIDLE)
			tp->t_timer[TCPT_2MSL] = TCPTV_KEEPINTVL;
		else
			tp = tcp_close(tp);
		break;

	/*
	 * Retransmission timer went off.  Message has not
	 * been acked within retransmit interval.  Back off
	 * to a longer retransmit interval and retransmit one segment.
	 */
	case TCPT_REXMT:

		/*
		 * XXXXX If a packet has timed out, then remove all the queued
		 * packets for that session.
		 */

		if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
			/*
			 * This is a hack to suit our terminal server here at the uni of canberra
			 * since they have trouble with zeroes... It usually lets them through
			 * unharmed, but under some conditions, it'll eat the zeros.  If we
			 * keep retransmitting it, it'll keep eating the zeroes, so we keep
			 * retransmitting, and eventually the connection dies...
			 * (this only happens on incoming data)
			 *
			 * So, if we were gonna drop the connection from too many retransmits,
			 * don't... instead halve the t_maxseg, which might break up the NULLs and
			 * let them through
			 *
			 * *sigh*
			 */

			tp->t_maxseg >>= 1;
			if (tp->t_maxseg < 32) {
				/*
				 * We tried our best, now the connection must die!
				 */
				tp->t_rxtshift = TCP_MAXRXTSHIFT;
				tp = tcp_drop(tp, tp->t_softerror);
				/* tp->t_softerror : ETIMEDOUT); */ /* XXX */
				return (tp); /* XXX */
			}

			/*
			 * Set rxtshift to 6, which is still at the maximum
			 * backoff time
			 */
			tp->t_rxtshift = 6;
		}
		rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
		TCPT_RANGESET(tp->t_rxtcur, rexmt,
		    (short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */
		tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
		/*
		 * If losing, let the lower level know and try for
		 * a better route.  Also, if we backed off this far,
		 * our srtt estimate is probably bogus.  Clobber it
		 * so we'll take the next rtt measurement as our srtt;
		 * move the current srtt into rttvar to keep the current
		 * retransmit times until then.
		 */
		if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
			tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
			tp->t_srtt = 0;
		}
		tp->snd_nxt = tp->snd_una;
		/*
		 * If timing a segment in this window, stop the timer.
		 */
		tp->t_rtt = 0;
		/*
		 * Close the congestion window down to one segment
		 * (we'll open it by one segment for each ack we get).
		 * Since we probably have a window's worth of unacked
		 * data accumulated, this "slow start" keeps us from
		 * dumping all that data as back-to-back packets (which
		 * might overwhelm an intermediate gateway).
		 *
		 * There are two phases to the opening: Initially we
		 * open by one mss on each ack.  This makes the window
		 * size increase exponentially with time.  If the
		 * window is larger than the path can handle, this
		 * exponential growth results in dropped packet(s)
		 * almost immediately.  To get more time between
		 * drops but still "push" the network to take advantage
		 * of improving conditions, we switch from exponential
		 * to linear window opening at some threshold size.
		 * For a threshold, we use half the current window
		 * size, truncated to a multiple of the mss.
		 *
		 * (the minimum cwnd that will give us exponential
		 * growth is 2 mss.  We don't allow the threshold
		 * to go below this.)
		 */
		{
		u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
		if (win < 2)
			win = 2;
		tp->snd_cwnd = tp->t_maxseg;
		tp->snd_ssthresh = win * tp->t_maxseg;
		tp->t_dupacks = 0;
		}
		(void) tcp_output(tp);
		break;

	/*
	 * Persistence timer into zero window.
	 * Force a byte to be output, if possible.
	 */
	case TCPT_PERSIST:
		tcp_setpersist(tp);
		tp->t_force = 1;
		(void) tcp_output(tp);
		tp->t_force = 0;
		break;

	/*
	 * Keep-alive timer went off; send something
	 * or drop connection if idle for too long.
	 */
	case TCPT_KEEP:
		if (tp->t_state < TCPS_ESTABLISHED)
			goto dropit;

		if ((SO_OPTIONS) && tp->t_state <= TCPS_CLOSE_WAIT) {
		    	if (tp->t_idle >= TCPTV_KEEP_IDLE + TCP_MAXIDLE)
				goto dropit;
			/*
			 * Send a packet designed to force a response
			 * if the peer is up and reachable:
			 * either an ACK if the connection is still alive,
			 * or an RST if the peer has closed the connection
			 * due to timeout or reboot.
			 * Using sequence number tp->snd_una-1
			 * causes the transmitted zero-length segment
			 * to lie outside the receive window;
			 * by the protocol spec, this requires the
			 * correspondent TCP to respond.
			 */
			tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
			    tp->rcv_nxt, tp->snd_una - 1, 0);
			tp->t_timer[TCPT_KEEP] = TCPTV_KEEPINTVL;
		} else
			tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_IDLE;
		break;

	dropit:
		tp = tcp_drop(tp, 0);
		break;
	}

	return (tp);
}
Commit	Line	Data
f0cbd3ec FB	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.
2f5f8996	13	* 3. Neither the name of the University nor the names of its contributors
f0cbd3ec FB	14	* may be used to endorse or promote products derived from this software
	15	* without specific prior written permission.
	16	*
	17	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
	18	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	19	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	20	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
	21	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	22	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	23	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	24	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	25	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	26	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	27	* SUCH DAMAGE.
	28	*
	29	* @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
	30	* tcp_timer.c,v 1.2 1994/08/02 07:49:10 davidg Exp
	31	*/
	32
	33	#include <slirp.h>
	34
9634d903 BS	35	static struct tcpcb tcp_timers(register struct tcpcb tp, int timer);
9634d903 BS	36
f0cbd3ec FB	37	/*
	38	* Fast timeout routine for processing delayed acks
	39	*/
	40	void
460fec67	41	tcp_fasttimo(Slirp *slirp)
f0cbd3ec FB	42	{
	43	register struct socket *so;
	44	register struct tcpcb *tp;
	45
	46	DEBUG_CALL("tcp_fasttimo");
5fafdf24	47
460fec67	48	so = slirp->tcb.so_next;
f0cbd3ec	49	if (so)
460fec67	50	for (; so != &slirp->tcb; so = so->so_next)
f0cbd3ec FB	51	if ((tp = (struct tcpcb *)so->so_tcpcb) &&
	52	(tp->t_flags & TF_DELACK)) {
	53	tp->t_flags &= ~TF_DELACK;
	54	tp->t_flags \|= TF_ACKNOW;
f0cbd3ec FB	55	(void) tcp_output(tp);
	56	}
	57	}
	58
	59	/*
	60	* Tcp protocol timeout routine called every 500 ms.
	61	* Updates the timers in all active tcb's and
	62	* causes finite state machine actions if timers expire.
	63	*/
	64	void
460fec67	65	tcp_slowtimo(Slirp *slirp)
f0cbd3ec FB	66	{
	67	register struct socket ip, ipnxt;
	68	register struct tcpcb *tp;
	69	register int i;
	70
	71	DEBUG_CALL("tcp_slowtimo");
5fafdf24	72
f0cbd3ec FB	73	/*
	74	* Search through tcb's and update active timers.
	75	*/
460fec67	76	ip = slirp->tcb.so_next;
7cba04f6 BS	77	if (ip == NULL) {
	78	return;
	79	}
460fec67	80	for (; ip != &slirp->tcb; ip = ipnxt) {
f0cbd3ec FB	81	ipnxt = ip->so_next;
f0cbd3ec FB	82	tp = sototcpcb(ip);
7cba04f6 BS	83	if (tp == NULL) {
	84	continue;
	85	}
f0cbd3ec FB	86	for (i = 0; i < TCPT_NTIMERS; i++) {
	87	if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
	88	tcp_timers(tp,i);
	89	if (ipnxt->so_prev != ip)
	90	goto tpgone;
	91	}
	92	}
	93	tp->t_idle++;
	94	if (tp->t_rtt)
	95	tp->t_rtt++;
	96	tpgone:
	97	;
	98	}
460fec67 JK	99	slirp->tcp_iss += TCP_ISSINCR/PR_SLOWHZ; /* increment iss */
460fec67 JK	100	slirp->tcp_now++; /* for timestamps */
f0cbd3ec FB	101	}
	102
	103	/*
	104	* Cancel all timers for TCP tp.
	105	*/
	106	void
aeed97c4	107	tcp_canceltimers(struct tcpcb *tp)
f0cbd3ec FB	108	{
	109	register int i;
	110
	111	for (i = 0; i < TCPT_NTIMERS; i++)
	112	tp->t_timer[i] = 0;
	113	}
	114
9634d903	115	const int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
f0cbd3ec FB	116	{ 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
	117
	118	/*
	119	* TCP timer processing.
	120	*/
9634d903 BS	121	static struct tcpcb *
9634d903 BS	122	tcp_timers(register struct tcpcb *tp, int timer)
f0cbd3ec FB	123	{
f0cbd3ec FB	124	register int rexmt;
5fafdf24	125
f0cbd3ec	126	DEBUG_CALL("tcp_timers");
5fafdf24	127
f0cbd3ec FB	128	switch (timer) {
	129
	130	/*
	131	* 2 MSL timeout in shutdown went off. If we're closed but
	132	* still waiting for peer to close and connection has been idle
	133	* too long, or if 2MSL time is up from TIME_WAIT, delete connection
	134	* control block. Otherwise, check again in a bit.
	135	*/
	136	case TCPT_2MSL:
	137	if (tp->t_state != TCPS_TIME_WAIT &&
9634d903 BS	138	tp->t_idle <= TCP_MAXIDLE)
9634d903 BS	139	tp->t_timer[TCPT_2MSL] = TCPTV_KEEPINTVL;
f0cbd3ec FB	140	else
	141	tp = tcp_close(tp);
	142	break;
	143
	144	/*
	145	* Retransmission timer went off. Message has not
	146	* been acked within retransmit interval. Back off
	147	* to a longer retransmit interval and retransmit one segment.
	148	*/
	149	case TCPT_REXMT:
3b46e624	150
f0cbd3ec FB	151	/*
	152	* XXXXX If a packet has timed out, then remove all the queued
	153	* packets for that session.
	154	*/
3b46e624	155
f0cbd3ec FB	156	if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
	157	/*
	158	* This is a hack to suit our terminal server here at the uni of canberra
	159	* since they have trouble with zeroes... It usually lets them through
	160	* unharmed, but under some conditions, it'll eat the zeros. If we
	161	* keep retransmitting it, it'll keep eating the zeroes, so we keep
	162	* retransmitting, and eventually the connection dies...
	163	* (this only happens on incoming data)
5fafdf24	164	*
f0cbd3ec FB	165	* So, if we were gonna drop the connection from too many retransmits,
	166	* don't... instead halve the t_maxseg, which might break up the NULLs and
	167	* let them through
5fafdf24	168	*
f0cbd3ec FB	169	* sigh
f0cbd3ec FB	170	*/
3b46e624	171
f0cbd3ec FB	172	tp->t_maxseg >>= 1;
	173	if (tp->t_maxseg < 32) {
	174	/*
	175	* We tried our best, now the connection must die!
	176	*/
	177	tp->t_rxtshift = TCP_MAXRXTSHIFT;
f0cbd3ec FB	178	tp = tcp_drop(tp, tp->t_softerror);
	179	/* tp->t_softerror : ETIMEDOUT); / / XXX */
	180	return (tp); /* XXX */
	181	}
3b46e624	182
f0cbd3ec FB	183	/*
	184	* Set rxtshift to 6, which is still at the maximum
	185	* backoff time
	186	*/
	187	tp->t_rxtshift = 6;
	188	}
f0cbd3ec FB	189	rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
	190	TCPT_RANGESET(tp->t_rxtcur, rexmt,
	191	(short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */
	192	tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
	193	/*
	194	* If losing, let the lower level know and try for
	195	* a better route. Also, if we backed off this far,
	196	* our srtt estimate is probably bogus. Clobber it
	197	* so we'll take the next rtt measurement as our srtt;
	198	* move the current srtt into rttvar to keep the current
	199	* retransmit times until then.
	200	*/
	201	if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
f0cbd3ec FB	202	tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
	203	tp->t_srtt = 0;
	204	}
	205	tp->snd_nxt = tp->snd_una;
	206	/*
	207	* If timing a segment in this window, stop the timer.
	208	*/
	209	tp->t_rtt = 0;
	210	/*
	211	* Close the congestion window down to one segment
	212	* (we'll open it by one segment for each ack we get).
	213	* Since we probably have a window's worth of unacked
	214	* data accumulated, this "slow start" keeps us from
	215	* dumping all that data as back-to-back packets (which
	216	* might overwhelm an intermediate gateway).
	217	*
	218	* There are two phases to the opening: Initially we
	219	* open by one mss on each ack. This makes the window
	220	* size increase exponentially with time. If the
	221	* window is larger than the path can handle, this
	222	* exponential growth results in dropped packet(s)
5fafdf24	223	* almost immediately. To get more time between
f0cbd3ec FB	224	* drops but still "push" the network to take advantage
	225	* of improving conditions, we switch from exponential
	226	* to linear window opening at some threshold size.
	227	* For a threshold, we use half the current window
	228	* size, truncated to a multiple of the mss.
	229	*
	230	* (the minimum cwnd that will give us exponential
	231	* growth is 2 mss. We don't allow the threshold
	232	* to go below this.)
	233	*/
	234	{
	235	u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
	236	if (win < 2)
	237	win = 2;
	238	tp->snd_cwnd = tp->t_maxseg;
	239	tp->snd_ssthresh = win * tp->t_maxseg;
	240	tp->t_dupacks = 0;
	241	}
	242	(void) tcp_output(tp);
	243	break;
	244
	245	/*
	246	* Persistence timer into zero window.
	247	* Force a byte to be output, if possible.
	248	*/
	249	case TCPT_PERSIST:
f0cbd3ec FB	250	tcp_setpersist(tp);
	251	tp->t_force = 1;
	252	(void) tcp_output(tp);
	253	tp->t_force = 0;
	254	break;
	255
	256	/*
	257	* Keep-alive timer went off; send something
	258	* or drop connection if idle for too long.
	259	*/
	260	case TCPT_KEEP:
f0cbd3ec FB	261	if (tp->t_state < TCPS_ESTABLISHED)
	262	goto dropit;
	263
9634d903 BS	264	if ((SO_OPTIONS) && tp->t_state <= TCPS_CLOSE_WAIT) {
9634d903 BS	265	if (tp->t_idle >= TCPTV_KEEP_IDLE + TCP_MAXIDLE)
f0cbd3ec FB	266	goto dropit;
	267	/*
	268	* Send a packet designed to force a response
	269	* if the peer is up and reachable:
	270	* either an ACK if the connection is still alive,
	271	* or an RST if the peer has closed the connection
	272	* due to timeout or reboot.
	273	* Using sequence number tp->snd_una-1
	274	* causes the transmitted zero-length segment
	275	* to lie outside the receive window;
	276	* by the protocol spec, this requires the
	277	* correspondent TCP to respond.
	278	*/
f0cbd3ec FB	279	tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
f0cbd3ec FB	280	tp->rcv_nxt, tp->snd_una - 1, 0);
9634d903	281	tp->t_timer[TCPT_KEEP] = TCPTV_KEEPINTVL;
f0cbd3ec	282	} else
9634d903	283	tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_IDLE;
f0cbd3ec FB	284	break;
	285
	286	dropit:
0d62c4cf	287	tp = tcp_drop(tp, 0);
f0cbd3ec FB	288	break;
	289	}
	290
	291	return (tp);
	292	}