tcp_timer.c revision 1.20 
1/*	$NetBSD: tcp_timer.c,v 1.20 1997/10/10 01:51:10 explorer 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_now++;					/* for timestamps */
142	if (++syn_cache_last >= tcp_syn_cache_interval) {
143		syn_cache_timer(syn_cache_last);
144		syn_cache_last = 0;
145	}
146	splx(s);
147}
148#ifndef TUBA_INCLUDE
149
150/*
151 * Cancel all timers for TCP tp.
152 */
153void
154tcp_canceltimers(tp)
155	struct tcpcb *tp;
156{
157	register int i;
158
159	for (i = 0; i < TCPT_NTIMERS; i++)
160		tp->t_timer[i] = 0;
161}
162
163int	tcp_backoff[TCP_MAXRXTSHIFT + 1] =
164    { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
165
166/*
167 * TCP timer processing.
168 */
169struct tcpcb *
170tcp_timers(tp, timer)
171	register struct tcpcb *tp;
172	int timer;
173{
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		TCPT_RANGESET(tp->t_rxtcur,
206		    TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift],
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		 * 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)
240		 * almost immediately.  To get more time between
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 threshhold size.
244		 * For a threshhold, 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 threshhold
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	 * Persistance timer into zero window.
264	 * Force a byte to be output, if possible.
265	 */
266	case TCPT_PERSIST:
267		tcpstat.tcps_persisttimeo++;
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:
279		tcpstat.tcps_keeptimeo++;
280		if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
281			goto dropit;
282		if (tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE &&
283		    tp->t_state <= TCPS_CLOSE_WAIT) {
284		    	if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
285				goto dropit;
286			/*
287			 * Send a packet designed to force a response
288			 * if the peer is up and reachable:
289			 * either an ACK if the connection is still alive,
290			 * or an RST if the peer has closed the connection
291			 * due to timeout or reboot.
292			 * Using sequence number tp->snd_una-1
293			 * causes the transmitted zero-length segment
294			 * to lie outside the receive window;
295			 * by the protocol spec, this requires the
296			 * correspondent TCP to respond.
297			 */
298			tcpstat.tcps_keepprobe++;
299#ifdef TCP_COMPAT_42
300			/*
301			 * The keepalive packet must have nonzero length
302			 * to get a 4.2 host to respond.
303			 */
304			(void)tcp_respond(tp, tp->t_template,
305			    (struct mbuf *)NULL, tp->rcv_nxt - 1,
306			    tp->snd_una - 1, 0);
307#else
308			(void)tcp_respond(tp, tp->t_template,
309			    (struct mbuf *)NULL, tp->rcv_nxt,
310			    tp->snd_una - 1, 0);
311#endif
312			tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
313		} else
314			tp->t_timer[TCPT_KEEP] = tcp_keepidle;
315		break;
316	dropit:
317		tcpstat.tcps_keepdrops++;
318		tp = tcp_drop(tp, ETIMEDOUT);
319		break;
320	}
321	return (tp);
322}
323#endif /* TUBA_INCLUDE */
324