tcp_timer.c revision 1.21 
1/*	$NetBSD: tcp_timer.c,v 1.21 1997/10/13 00:48:16 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#if NRND == 0 /* Do we need to do this when using random() ? */
142	tcp_iss_seq += TCP_ISSINCR;			/* increment iss */
143#ifdef TCP_COMPAT_42
144	if ((int)tcp_iss_seq < 0)
145		tcp_iss_seq = 0;			/* XXX */
146#endif
147#endif
148	tcp_now++;					/* for timestamps */
149	if (++syn_cache_last >= tcp_syn_cache_interval) {
150		syn_cache_timer(syn_cache_last);
151		syn_cache_last = 0;
152	}
153	splx(s);
154}
155#ifndef TUBA_INCLUDE
156
157/*
158 * Cancel all timers for TCP tp.
159 */
160void
161tcp_canceltimers(tp)
162	struct tcpcb *tp;
163{
164	register int i;
165
166	for (i = 0; i < TCPT_NTIMERS; i++)
167		tp->t_timer[i] = 0;
168}
169
170int	tcp_backoff[TCP_MAXRXTSHIFT + 1] =
171    { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
172
173/*
174 * TCP timer processing.
175 */
176struct tcpcb *
177tcp_timers(tp, timer)
178	register struct tcpcb *tp;
179	int timer;
180{
181
182	switch (timer) {
183
184	/*
185	 * 2 MSL timeout in shutdown went off.  If we're closed but
186	 * still waiting for peer to close and connection has been idle
187	 * too long, or if 2MSL time is up from TIME_WAIT, delete connection
188	 * control block.  Otherwise, check again in a bit.
189	 */
190	case TCPT_2MSL:
191		if (tp->t_state != TCPS_TIME_WAIT &&
192		    tp->t_idle <= tcp_maxidle)
193			tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
194		else
195			tp = tcp_close(tp);
196		break;
197
198	/*
199	 * Retransmission timer went off.  Message has not
200	 * been acked within retransmit interval.  Back off
201	 * to a longer retransmit interval and retransmit one segment.
202	 */
203	case TCPT_REXMT:
204		if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
205			tp->t_rxtshift = TCP_MAXRXTSHIFT;
206			tcpstat.tcps_timeoutdrop++;
207			tp = tcp_drop(tp, tp->t_softerror ?
208			    tp->t_softerror : ETIMEDOUT);
209			break;
210		}
211		tcpstat.tcps_rexmttimeo++;
212		TCPT_RANGESET(tp->t_rxtcur,
213		    TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift],
214		    tp->t_rttmin, TCPTV_REXMTMAX);
215		tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
216		/*
217		 * If losing, let the lower level know and try for
218		 * a better route.  Also, if we backed off this far,
219		 * our srtt estimate is probably bogus.  Clobber it
220		 * so we'll take the next rtt measurement as our srtt;
221		 * move the current srtt into rttvar to keep the current
222		 * retransmit times until then.
223		 */
224		if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
225			in_losing(tp->t_inpcb);
226			tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
227			tp->t_srtt = 0;
228		}
229		tp->snd_nxt = tp->snd_una;
230		/*
231		 * If timing a segment in this window, stop the timer.
232		 */
233		tp->t_rtt = 0;
234		/*
235		 * Close the congestion window down to one segment
236		 * (we'll open it by one segment for each ack we get).
237		 * Since we probably have a window's worth of unacked
238		 * data accumulated, this "slow start" keeps us from
239		 * dumping all that data as back-to-back packets (which
240		 * might overwhelm an intermediate gateway).
241		 *
242		 * There are two phases to the opening: Initially we
243		 * open by one mss on each ack.  This makes the window
244		 * size increase exponentially with time.  If the
245		 * window is larger than the path can handle, this
246		 * exponential growth results in dropped packet(s)
247		 * almost immediately.  To get more time between
248		 * drops but still "push" the network to take advantage
249		 * of improving conditions, we switch from exponential
250		 * to linear window opening at some threshhold size.
251		 * For a threshhold, we use half the current window
252		 * size, truncated to a multiple of the mss.
253		 *
254		 * (the minimum cwnd that will give us exponential
255		 * growth is 2 mss.  We don't allow the threshhold
256		 * to go below this.)
257		 */
258		{
259		u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
260		if (win < 2)
261			win = 2;
262		tp->snd_cwnd = tp->t_maxseg;
263		tp->snd_ssthresh = win * tp->t_maxseg;
264		tp->t_dupacks = 0;
265		}
266		(void) tcp_output(tp);
267		break;
268
269	/*
270	 * Persistance timer into zero window.
271	 * Force a byte to be output, if possible.
272	 */
273	case TCPT_PERSIST:
274		tcpstat.tcps_persisttimeo++;
275		tcp_setpersist(tp);
276		tp->t_force = 1;
277		(void) tcp_output(tp);
278		tp->t_force = 0;
279		break;
280
281	/*
282	 * Keep-alive timer went off; send something
283	 * or drop connection if idle for too long.
284	 */
285	case TCPT_KEEP:
286		tcpstat.tcps_keeptimeo++;
287		if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
288			goto dropit;
289		if (tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE &&
290		    tp->t_state <= TCPS_CLOSE_WAIT) {
291		    	if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
292				goto dropit;
293			/*
294			 * Send a packet designed to force a response
295			 * if the peer is up and reachable:
296			 * either an ACK if the connection is still alive,
297			 * or an RST if the peer has closed the connection
298			 * due to timeout or reboot.
299			 * Using sequence number tp->snd_una-1
300			 * causes the transmitted zero-length segment
301			 * to lie outside the receive window;
302			 * by the protocol spec, this requires the
303			 * correspondent TCP to respond.
304			 */
305			tcpstat.tcps_keepprobe++;
306#ifdef TCP_COMPAT_42
307			/*
308			 * The keepalive packet must have nonzero length
309			 * to get a 4.2 host to respond.
310			 */
311			(void)tcp_respond(tp, tp->t_template,
312			    (struct mbuf *)NULL, tp->rcv_nxt - 1,
313			    tp->snd_una - 1, 0);
314#else
315			(void)tcp_respond(tp, tp->t_template,
316			    (struct mbuf *)NULL, tp->rcv_nxt,
317			    tp->snd_una - 1, 0);
318#endif
319			tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
320		} else
321			tp->t_timer[TCPT_KEEP] = tcp_keepidle;
322		break;
323	dropit:
324		tcpstat.tcps_keepdrops++;
325		tp = tcp_drop(tp, ETIMEDOUT);
326		break;
327	}
328	return (tp);
329}
330#endif /* TUBA_INCLUDE */
331