tcp_timer.c revision 1.19 
1/*	$NetBSD: tcp_timer.c,v 1.19 1997/07/28 22:07:40 thorpej Exp $	*/
2
3/*
4 * Copyright (c) 1982, 1986, 1988, 1990, 1993
5 *	The Regents of the University of California.  All rights reserved.
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 */
37
38#ifndef TUBA_INCLUDE
39#include <sys/param.h>
40#include <sys/systm.h>
41#include <sys/malloc.h>
42#include <sys/mbuf.h>
43#include <sys/socket.h>
44#include <sys/socketvar.h>
45#include <sys/protosw.h>
46#include <sys/errno.h>
47
48#include <net/if.h>
49#include <net/route.h>
50
51#include <netinet/in.h>
52#include <netinet/in_systm.h>
53#include <netinet/ip.h>
54#include <netinet/in_pcb.h>
55#include <netinet/ip_var.h>
56#include <netinet/tcp.h>
57#include <netinet/tcp_fsm.h>
58#include <netinet/tcp_seq.h>
59#include <netinet/tcp_timer.h>
60#include <netinet/tcp_var.h>
61#include <netinet/tcpip.h>
62
63int	tcp_keepidle = TCPTV_KEEP_IDLE;
64int	tcp_keepintvl = TCPTV_KEEPINTVL;
65int	tcp_maxidle;
66#endif /* TUBA_INCLUDE */
67/*
68 * Fast timeout routine for processing delayed acks
69 */
70void
71tcp_fasttimo()
72{
73	register struct inpcb *inp;
74	register struct tcpcb *tp;
75	int s;
76
77	s = splsoftnet();
78	inp = tcbtable.inpt_queue.cqh_first;
79	if (inp)						/* XXX */
80	for (; inp != (struct inpcb *)&tcbtable.inpt_queue;
81	    inp = inp->inp_queue.cqe_next) {
82		if ((tp = (struct tcpcb *)inp->inp_ppcb) &&
83		    (tp->t_flags & TF_DELACK)) {
84			tp->t_flags &= ~TF_DELACK;
85			tp->t_flags |= TF_ACKNOW;
86			tcpstat.tcps_delack++;
87			(void) tcp_output(tp);
88		}
89	}
90	splx(s);
91}
92
93/*
94 * Tcp protocol timeout routine called every 500 ms.
95 * Updates the timers in all active tcb's and
96 * causes finite state machine actions if timers expire.
97 */
98void
99tcp_slowtimo()
100{
101	register struct inpcb *inp, *ninp;
102	register struct tcpcb *tp;
103	int s;
104	register long i;
105	static int syn_cache_last = 0;
106
107	s = splsoftnet();
108	tcp_maxidle = TCPTV_KEEPCNT * tcp_keepintvl;
109	/*
110	 * Search through tcb's and update active timers.
111	 */
112	inp = tcbtable.inpt_queue.cqh_first;
113	if (inp == (struct inpcb *)0) {				/* XXX */
114		splx(s);
115		return;
116	}
117	for (; inp != (struct inpcb *)&tcbtable.inpt_queue; inp = ninp) {
118		ninp = inp->inp_queue.cqe_next;
119		tp = intotcpcb(inp);
120		if (tp == 0)
121			continue;
122		for (i = 0; i < TCPT_NTIMERS; i++) {
123			if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
124				(void) tcp_usrreq(tp->t_inpcb->inp_socket,
125				    PRU_SLOWTIMO, (struct mbuf *)0,
126				    (struct mbuf *)i, (struct mbuf *)0,
127				    (struct proc *)0);
128				/* XXX NOT MP SAFE */
129				if ((ninp == (void *)&tcbtable.inpt_queue &&
130				    tcbtable.inpt_queue.cqh_last != inp) ||
131				    ninp->inp_queue.cqe_prev != inp)
132					goto tpgone;
133			}
134		}
135		tp->t_idle++;
136		if (tp->t_rtt)
137			tp->t_rtt++;
138tpgone:
139		;
140	}
141	tcp_iss += TCP_ISSINCR/PR_SLOWHZ;		/* increment iss */
142#ifdef TCP_COMPAT_42
143	if ((int)tcp_iss < 0)
144		tcp_iss = 0;				/* XXX */
145#endif
146	tcp_now++;					/* for timestamps */
147	if (++syn_cache_last >= tcp_syn_cache_interval) {
148		syn_cache_timer(syn_cache_last);
149		syn_cache_last = 0;
150	}
151	splx(s);
152}
153#ifndef TUBA_INCLUDE
154
155/*
156 * Cancel all timers for TCP tp.
157 */
158void
159tcp_canceltimers(tp)
160	struct tcpcb *tp;
161{
162	register int i;
163
164	for (i = 0; i < TCPT_NTIMERS; i++)
165		tp->t_timer[i] = 0;
166}
167
168int	tcp_backoff[TCP_MAXRXTSHIFT + 1] =
169    { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
170
171/*
172 * TCP timer processing.
173 */
174struct tcpcb *
175tcp_timers(tp, timer)
176	register struct tcpcb *tp;
177	int timer;
178{
179
180	switch (timer) {
181
182	/*
183	 * 2 MSL timeout in shutdown went off.  If we're closed but
184	 * still waiting for peer to close and connection has been idle
185	 * too long, or if 2MSL time is up from TIME_WAIT, delete connection
186	 * control block.  Otherwise, check again in a bit.
187	 */
188	case TCPT_2MSL:
189		if (tp->t_state != TCPS_TIME_WAIT &&
190		    tp->t_idle <= tcp_maxidle)
191			tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
192		else
193			tp = tcp_close(tp);
194		break;
195
196	/*
197	 * Retransmission timer went off.  Message has not
198	 * been acked within retransmit interval.  Back off
199	 * to a longer retransmit interval and retransmit one segment.
200	 */
201	case TCPT_REXMT:
202		if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
203			tp->t_rxtshift = TCP_MAXRXTSHIFT;
204			tcpstat.tcps_timeoutdrop++;
205			tp = tcp_drop(tp, tp->t_softerror ?
206			    tp->t_softerror : ETIMEDOUT);
207			break;
208		}
209		tcpstat.tcps_rexmttimeo++;
210		TCPT_RANGESET(tp->t_rxtcur,
211		    TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift],
212		    tp->t_rttmin, TCPTV_REXMTMAX);
213		tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
214		/*
215		 * If losing, let the lower level know and try for
216		 * a better route.  Also, if we backed off this far,
217		 * our srtt estimate is probably bogus.  Clobber it
218		 * so we'll take the next rtt measurement as our srtt;
219		 * move the current srtt into rttvar to keep the current
220		 * retransmit times until then.
221		 */
222		if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
223			in_losing(tp->t_inpcb);
224			tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
225			tp->t_srtt = 0;
226		}
227		tp->snd_nxt = tp->snd_una;
228		/*
229		 * If timing a segment in this window, stop the timer.
230		 */
231		tp->t_rtt = 0;
232		/*
233		 * Close the congestion window down to one segment
234		 * (we'll open it by one segment for each ack we get).
235		 * Since we probably have a window's worth of unacked
236		 * data accumulated, this "slow start" keeps us from
237		 * dumping all that data as back-to-back packets (which
238		 * might overwhelm an intermediate gateway).
239		 *
240		 * There are two phases to the opening: Initially we
241		 * open by one mss on each ack.  This makes the window
242		 * size increase exponentially with time.  If the
243		 * window is larger than the path can handle, this
244		 * exponential growth results in dropped packet(s)
245		 * almost immediately.  To get more time between
246		 * drops but still "push" the network to take advantage
247		 * of improving conditions, we switch from exponential
248		 * to linear window opening at some threshhold size.
249		 * For a threshhold, we use half the current window
250		 * size, truncated to a multiple of the mss.
251		 *
252		 * (the minimum cwnd that will give us exponential
253		 * growth is 2 mss.  We don't allow the threshhold
254		 * to go below this.)
255		 */
256		{
257		u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
258		if (win < 2)
259			win = 2;
260		tp->snd_cwnd = tp->t_maxseg;
261		tp->snd_ssthresh = win * tp->t_maxseg;
262		tp->t_dupacks = 0;
263		}
264		(void) tcp_output(tp);
265		break;
266
267	/*
268	 * Persistance timer into zero window.
269	 * Force a byte to be output, if possible.
270	 */
271	case TCPT_PERSIST:
272		tcpstat.tcps_persisttimeo++;
273		tcp_setpersist(tp);
274		tp->t_force = 1;
275		(void) tcp_output(tp);
276		tp->t_force = 0;
277		break;
278
279	/*
280	 * Keep-alive timer went off; send something
281	 * or drop connection if idle for too long.
282	 */
283	case TCPT_KEEP:
284		tcpstat.tcps_keeptimeo++;
285		if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
286			goto dropit;
287		if (tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE &&
288		    tp->t_state <= TCPS_CLOSE_WAIT) {
289		    	if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
290				goto dropit;
291			/*
292			 * Send a packet designed to force a response
293			 * if the peer is up and reachable:
294			 * either an ACK if the connection is still alive,
295			 * or an RST if the peer has closed the connection
296			 * due to timeout or reboot.
297			 * Using sequence number tp->snd_una-1
298			 * causes the transmitted zero-length segment
299			 * to lie outside the receive window;
300			 * by the protocol spec, this requires the
301			 * correspondent TCP to respond.
302			 */
303			tcpstat.tcps_keepprobe++;
304#ifdef TCP_COMPAT_42
305			/*
306			 * The keepalive packet must have nonzero length
307			 * to get a 4.2 host to respond.
308			 */
309			(void)tcp_respond(tp, tp->t_template,
310			    (struct mbuf *)NULL, tp->rcv_nxt - 1,
311			    tp->snd_una - 1, 0);
312#else
313			(void)tcp_respond(tp, tp->t_template,
314			    (struct mbuf *)NULL, tp->rcv_nxt,
315			    tp->snd_una - 1, 0);
316#endif
317			tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
318		} else
319			tp->t_timer[TCPT_KEEP] = tcp_keepidle;
320		break;
321	dropit:
322		tcpstat.tcps_keepdrops++;
323		tp = tcp_drop(tp, ETIMEDOUT);
324		break;
325	}
326	return (tp);
327}
328#endif /* TUBA_INCLUDE */
329