2 * Copyright (c) 1982, 1986, 1988, 1990, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
13 * 3. All advertising materials mentioning features or use of this software
14 * must display the following acknowledgement:
15 * This product includes software developed by the University of
16 * California, Berkeley and its contributors.
17 * 4. Neither the name of the University nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33 * @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
37 #include <sys/param.h>
38 #include <sys/systm.h>
39 #include <sys/malloc.h>
41 #include <sys/socket.h>
42 #include <sys/socketvar.h>
43 #include <sys/protosw.h>
44 #include <sys/errno.h>
45 #include <sys/queue.h>
48 #include <net/route.h>
50 #include <netinet/in.h>
51 #include <netinet/in_systm.h>
52 #include <netinet/ip.h>
53 #include <netinet/in_pcb.h>
54 #include <netinet/ip_var.h>
55 #include <netinet/tcp.h>
56 #include <netinet/tcp_fsm.h>
57 #include <netinet/tcp_seq.h>
58 #include <netinet/tcp_timer.h>
59 #include <netinet/tcp_var.h>
60 #include <netinet/tcpip.h>
62 #include <netinet/tcp_debug.h>
65 int tcp_keepidle
= TCPTV_KEEP_IDLE
;
66 int tcp_keepintvl
= TCPTV_KEEPINTVL
;
67 int tcp_keepinit
= TCPTV_KEEP_INIT
;
69 int tcp_maxpersistidle
= TCPTV_KEEP_IDLE
;
70 int tcp_totbackoff
= 511;
71 #endif /* TUBA_INCLUDE */
73 * Fast timeout routine for processing delayed acks
78 register struct inpcb
*inp
;
79 register struct tcpcb
*tp
;
84 for (inp
= tcb
.lh_first
; inp
!= NULL
; inp
= inp
->inp_list
.le_next
) {
85 if ((tp
= (struct tcpcb
*)inp
->inp_ppcb
) &&
86 (tp
->t_flags
& TF_DELACK
)) {
87 tp
->t_flags
&= ~TF_DELACK
;
88 tp
->t_flags
|= TF_ACKNOW
;
89 tcpstat
.tcps_delack
++;
90 (void) tcp_output(tp
);
97 * Tcp protocol timeout routine called every 500 ms.
98 * Updates the timers in all active tcb's and
99 * causes finite state machine actions if timers expire.
104 register struct inpcb
*ip
, *ipnxt
;
105 register struct tcpcb
*tp
;
114 tcp_maxidle
= TCPTV_KEEPCNT
* tcp_keepintvl
;
122 * Search through tcb's and update active timers.
124 for (; ip
!= NULL
; ip
= ipnxt
) {
125 ipnxt
= ip
->inp_list
.le_next
;
129 for (i
= 0; i
< TCPT_NTIMERS
; i
++) {
130 if (tp
->t_timer
[i
] && --tp
->t_timer
[i
] == 0) {
132 ostate
= tp
->t_state
;
134 tp
= tcp_timers(tp
, i
);
138 if (tp
->t_inpcb
->inp_socket
->so_options
140 tcp_trace(TA_USER
, ostate
, tp
,
141 (struct tcpiphdr
*)0,
153 tcp_iss
+= TCP_ISSINCR
/PR_SLOWHZ
; /* increment iss */
155 if ((int)tcp_iss
< 0)
156 tcp_iss
= 0; /* XXX */
158 tcp_now
++; /* for timestamps */
164 * Cancel all timers for TCP tp.
172 for (i
= 0; i
< TCPT_NTIMERS
; i
++)
176 int tcp_backoff
[TCP_MAXRXTSHIFT
+ 1] =
177 { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
180 * TCP timer processing.
183 tcp_timers(tp
, timer
)
184 register struct tcpcb
*tp
;
192 * 2 MSL timeout in shutdown went off. If we're closed but
193 * still waiting for peer to close and connection has been idle
194 * too long, or if 2MSL time is up from TIME_WAIT, delete connection
195 * control block. Otherwise, check again in a bit.
198 if (tp
->t_state
!= TCPS_TIME_WAIT
&&
199 tp
->t_idle
<= tcp_maxidle
)
200 tp
->t_timer
[TCPT_2MSL
] = tcp_keepintvl
;
206 * Retransmission timer went off. Message has not
207 * been acked within retransmit interval. Back off
208 * to a longer retransmit interval and retransmit one segment.
211 if (++tp
->t_rxtshift
> TCP_MAXRXTSHIFT
) {
212 tp
->t_rxtshift
= TCP_MAXRXTSHIFT
;
213 tcpstat
.tcps_timeoutdrop
++;
214 tp
= tcp_drop(tp
, tp
->t_softerror
?
215 tp
->t_softerror
: ETIMEDOUT
);
218 tcpstat
.tcps_rexmttimeo
++;
219 rexmt
= TCP_REXMTVAL(tp
) * tcp_backoff
[tp
->t_rxtshift
];
220 TCPT_RANGESET(tp
->t_rxtcur
, rexmt
,
221 tp
->t_rttmin
, TCPTV_REXMTMAX
);
222 tp
->t_timer
[TCPT_REXMT
] = tp
->t_rxtcur
;
224 * If losing, let the lower level know and try for
225 * a better route. Also, if we backed off this far,
226 * our srtt estimate is probably bogus. Clobber it
227 * so we'll take the next rtt measurement as our srtt;
228 * move the current srtt into rttvar to keep the current
229 * retransmit times until then.
231 if (tp
->t_rxtshift
> TCP_MAXRXTSHIFT
/ 4) {
232 in_losing(tp
->t_inpcb
);
233 tp
->t_rttvar
+= (tp
->t_srtt
>> TCP_RTT_SHIFT
);
236 tp
->snd_nxt
= tp
->snd_una
;
238 * Force a segment to be sent.
240 tp
->t_flags
|= TF_ACKNOW
;
242 * If timing a segment in this window, stop the timer.
246 * Close the congestion window down to one segment
247 * (we'll open it by one segment for each ack we get).
248 * Since we probably have a window's worth of unacked
249 * data accumulated, this "slow start" keeps us from
250 * dumping all that data as back-to-back packets (which
251 * might overwhelm an intermediate gateway).
253 * There are two phases to the opening: Initially we
254 * open by one mss on each ack. This makes the window
255 * size increase exponentially with time. If the
256 * window is larger than the path can handle, this
257 * exponential growth results in dropped packet(s)
258 * almost immediately. To get more time between
259 * drops but still "push" the network to take advantage
260 * of improving conditions, we switch from exponential
261 * to linear window opening at some threshhold size.
262 * For a threshhold, we use half the current window
263 * size, truncated to a multiple of the mss.
265 * (the minimum cwnd that will give us exponential
266 * growth is 2 mss. We don't allow the threshhold
270 u_int win
= min(tp
->snd_wnd
, tp
->snd_cwnd
) / 2 / tp
->t_maxseg
;
273 tp
->snd_cwnd
= tp
->t_maxseg
;
274 tp
->snd_ssthresh
= win
* tp
->t_maxseg
;
277 (void) tcp_output(tp
);
281 * Persistance timer into zero window.
282 * Force a byte to be output, if possible.
285 tcpstat
.tcps_persisttimeo
++;
287 * Hack: if the peer is dead/unreachable, we do not
288 * time out if the window is closed. After a full
289 * backoff, drop the connection if the idle time
290 * (no responses to probes) reaches the maximum
291 * backoff that we would use if retransmitting.
293 if (tp
->t_rxtshift
== TCP_MAXRXTSHIFT
) {
294 u_long maxidle
= TCP_REXMTVAL(tp
);
295 if (maxidle
< tp
->t_rttmin
)
296 maxidle
= tp
->t_rttmin
;
297 maxidle
*= tcp_totbackoff
;
298 if (tp
->t_idle
>= tcp_maxpersistidle
||
299 tp
->t_idle
>= maxidle
) {
300 tcpstat
.tcps_persistdrop
++;
301 tp
= tcp_drop(tp
, ETIMEDOUT
);
307 (void) tcp_output(tp
);
312 * Keep-alive timer went off; send something
313 * or drop connection if idle for too long.
316 tcpstat
.tcps_keeptimeo
++;
317 if (tp
->t_state
< TCPS_ESTABLISHED
)
319 if (tp
->t_inpcb
->inp_socket
->so_options
& SO_KEEPALIVE
&&
320 tp
->t_state
<= TCPS_CLOSING
) {
321 if (tp
->t_idle
>= tcp_keepidle
+ tcp_maxidle
)
324 * Send a packet designed to force a response
325 * if the peer is up and reachable:
326 * either an ACK if the connection is still alive,
327 * or an RST if the peer has closed the connection
328 * due to timeout or reboot.
329 * Using sequence number tp->snd_una-1
330 * causes the transmitted zero-length segment
331 * to lie outside the receive window;
332 * by the protocol spec, this requires the
333 * correspondent TCP to respond.
335 tcpstat
.tcps_keepprobe
++;
338 * The keepalive packet must have nonzero length
339 * to get a 4.2 host to respond.
341 tcp_respond(tp
, tp
->t_template
, (struct mbuf
*)NULL
,
342 tp
->rcv_nxt
- 1, tp
->snd_una
- 1, 0);
344 tcp_respond(tp
, tp
->t_template
, (struct mbuf
*)NULL
,
345 tp
->rcv_nxt
, tp
->snd_una
- 1, 0);
347 tp
->t_timer
[TCPT_KEEP
] = tcp_keepintvl
;
349 tp
->t_timer
[TCPT_KEEP
] = tcp_keepidle
;
352 tcpstat
.tcps_keepdrops
++;
353 tp
= tcp_drop(tp
, ETIMEDOUT
);
358 #endif /* TUBA_INCLUDE */