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.
|
---|
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.
|
---|
20 | *
|
---|
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
|
---|
31 | * SUCH DAMAGE.
|
---|
32 | *
|
---|
33 | * @(#)tcp_timer.c 8.1 (Berkeley) 6/10/93
|
---|
34 | * tcp_timer.c,v 1.2 1994/08/02 07:49:10 davidg Exp
|
---|
35 | */
|
---|
36 |
|
---|
37 | #include <slirp.h>
|
---|
38 |
|
---|
39 | #ifndef VBOX
|
---|
40 | int tcp_keepidle = TCPTV_KEEP_IDLE;
|
---|
41 | int tcp_keepintvl = TCPTV_KEEPINTVL;
|
---|
42 | int tcp_maxidle;
|
---|
43 | int so_options = DO_KEEPALIVE;
|
---|
44 |
|
---|
45 | struct tcpstat tcpstat; /* tcp statistics */
|
---|
46 | u_int32_t tcp_now; /* for RFC 1323 timestamps */
|
---|
47 | #endif /* !VBOX */
|
---|
48 |
|
---|
49 | /*
|
---|
50 | * Fast timeout routine for processing delayed acks
|
---|
51 | */
|
---|
52 | void
|
---|
53 | #ifdef VBOX
|
---|
54 | tcp_fasttimo(PNATState pData)
|
---|
55 | #else /* !VBOX */
|
---|
56 | tcp_fasttimo()
|
---|
57 | #endif /* !VBOX */
|
---|
58 | {
|
---|
59 | register struct socket *so;
|
---|
60 | register struct tcpcb *tp;
|
---|
61 |
|
---|
62 | DEBUG_CALL("tcp_fasttimo");
|
---|
63 |
|
---|
64 | so = tcb.so_next;
|
---|
65 | if (so)
|
---|
66 | for (; so != &tcb; so = so->so_next)
|
---|
67 | if ((tp = (struct tcpcb *)so->so_tcpcb) &&
|
---|
68 | (tp->t_flags & TF_DELACK)) {
|
---|
69 | tp->t_flags &= ~TF_DELACK;
|
---|
70 | tp->t_flags |= TF_ACKNOW;
|
---|
71 | tcpstat.tcps_delack++;
|
---|
72 | #ifdef VBOX
|
---|
73 | (void) tcp_output(pData, tp);
|
---|
74 | #else /*! VBOX */
|
---|
75 | (void) tcp_output(tp);
|
---|
76 | #endif /* !VBOX */
|
---|
77 | }
|
---|
78 | }
|
---|
79 |
|
---|
80 | /*
|
---|
81 | * Tcp protocol timeout routine called every 500 ms.
|
---|
82 | * Updates the timers in all active tcb's and
|
---|
83 | * causes finite state machine actions if timers expire.
|
---|
84 | */
|
---|
85 | void
|
---|
86 | #ifdef VBOX
|
---|
87 | tcp_slowtimo(PNATState pData)
|
---|
88 | #else /* !VBOX */
|
---|
89 | tcp_slowtimo()
|
---|
90 | #endif /* !VBOX */
|
---|
91 | {
|
---|
92 | register struct socket *ip, *ipnxt;
|
---|
93 | register struct tcpcb *tp;
|
---|
94 | register int i;
|
---|
95 |
|
---|
96 | DEBUG_CALL("tcp_slowtimo");
|
---|
97 |
|
---|
98 | #ifndef VBOX
|
---|
99 | tcp_maxidle = TCPTV_KEEPCNT * tcp_keepintvl;
|
---|
100 | #endif /* !VBOX */
|
---|
101 | /*
|
---|
102 | * Search through tcb's and update active timers.
|
---|
103 | */
|
---|
104 | ip = tcb.so_next;
|
---|
105 | if (ip == 0)
|
---|
106 | return;
|
---|
107 | for (; ip != &tcb; ip = ipnxt) {
|
---|
108 | ipnxt = ip->so_next;
|
---|
109 | tp = sototcpcb(ip);
|
---|
110 | if (tp == 0)
|
---|
111 | continue;
|
---|
112 | for (i = 0; i < TCPT_NTIMERS; i++) {
|
---|
113 | if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
|
---|
114 | #ifdef VBOX
|
---|
115 | tcp_timers(pData, tp,i);
|
---|
116 | #else /* !VBOX */
|
---|
117 | tcp_timers(tp,i);
|
---|
118 | #endif /* !VBOX */
|
---|
119 | if (ipnxt->so_prev != ip)
|
---|
120 | goto tpgone;
|
---|
121 | }
|
---|
122 | }
|
---|
123 | tp->t_idle++;
|
---|
124 | if (tp->t_rtt)
|
---|
125 | tp->t_rtt++;
|
---|
126 | tpgone:
|
---|
127 | ;
|
---|
128 | }
|
---|
129 | tcp_iss += TCP_ISSINCR/PR_SLOWHZ; /* increment iss */
|
---|
130 | #ifdef TCP_COMPAT_42
|
---|
131 | if ((int)tcp_iss < 0)
|
---|
132 | tcp_iss = 0; /* XXX */
|
---|
133 | #endif
|
---|
134 | tcp_now++; /* for timestamps */
|
---|
135 | }
|
---|
136 |
|
---|
137 | /*
|
---|
138 | * Cancel all timers for TCP tp.
|
---|
139 | */
|
---|
140 | void
|
---|
141 | tcp_canceltimers(tp)
|
---|
142 | struct tcpcb *tp;
|
---|
143 | {
|
---|
144 | register int i;
|
---|
145 |
|
---|
146 | for (i = 0; i < TCPT_NTIMERS; i++)
|
---|
147 | tp->t_timer[i] = 0;
|
---|
148 | }
|
---|
149 |
|
---|
150 | #ifdef VBOX
|
---|
151 | const int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
|
---|
152 | #else /* !VBOX */
|
---|
153 | int tcp_backoff[TCP_MAXRXTSHIFT + 1] =
|
---|
154 | #endif /* !VBOX */
|
---|
155 | { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
|
---|
156 |
|
---|
157 | /*
|
---|
158 | * TCP timer processing.
|
---|
159 | */
|
---|
160 | struct tcpcb *
|
---|
161 | #ifdef VBOX
|
---|
162 | tcp_timers(PNATState pData, register struct tcpcb *tp, int timer)
|
---|
163 | #else /* !VBOX */
|
---|
164 | tcp_timers(tp, timer)
|
---|
165 | register struct tcpcb *tp;
|
---|
166 | int timer;
|
---|
167 | #endif /* !VBOX */
|
---|
168 | {
|
---|
169 | register int rexmt;
|
---|
170 |
|
---|
171 | DEBUG_CALL("tcp_timers");
|
---|
172 |
|
---|
173 | switch (timer) {
|
---|
174 |
|
---|
175 | /*
|
---|
176 | * 2 MSL timeout in shutdown went off. If we're closed but
|
---|
177 | * still waiting for peer to close and connection has been idle
|
---|
178 | * too long, or if 2MSL time is up from TIME_WAIT, delete connection
|
---|
179 | * control block. Otherwise, check again in a bit.
|
---|
180 | */
|
---|
181 | case TCPT_2MSL:
|
---|
182 | if (tp->t_state != TCPS_TIME_WAIT &&
|
---|
183 | tp->t_idle <= tcp_maxidle)
|
---|
184 | tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
|
---|
185 | else
|
---|
186 | #ifdef VBOX
|
---|
187 | tp = tcp_close(pData, tp);
|
---|
188 | #else /* !VBOX */
|
---|
189 | tp = tcp_close(tp);
|
---|
190 | #endif /* !VBOX */
|
---|
191 | break;
|
---|
192 |
|
---|
193 | /*
|
---|
194 | * Retransmission timer went off. Message has not
|
---|
195 | * been acked within retransmit interval. Back off
|
---|
196 | * to a longer retransmit interval and retransmit one segment.
|
---|
197 | */
|
---|
198 | case TCPT_REXMT:
|
---|
199 |
|
---|
200 | /*
|
---|
201 | * XXXXX If a packet has timed out, then remove all the queued
|
---|
202 | * packets for that session.
|
---|
203 | */
|
---|
204 |
|
---|
205 | if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
|
---|
206 | /*
|
---|
207 | * This is a hack to suit our terminal server here at the uni of canberra
|
---|
208 | * since they have trouble with zeroes... It usually lets them through
|
---|
209 | * unharmed, but under some conditions, it'll eat the zeros. If we
|
---|
210 | * keep retransmitting it, it'll keep eating the zeroes, so we keep
|
---|
211 | * retransmitting, and eventually the connection dies...
|
---|
212 | * (this only happens on incoming data)
|
---|
213 | *
|
---|
214 | * So, if we were gonna drop the connection from too many retransmits,
|
---|
215 | * don't... instead halve the t_maxseg, which might break up the NULLs and
|
---|
216 | * let them through
|
---|
217 | *
|
---|
218 | * *sigh*
|
---|
219 | */
|
---|
220 |
|
---|
221 | tp->t_maxseg >>= 1;
|
---|
222 | if (tp->t_maxseg < 32) {
|
---|
223 | /*
|
---|
224 | * We tried our best, now the connection must die!
|
---|
225 | */
|
---|
226 | tp->t_rxtshift = TCP_MAXRXTSHIFT;
|
---|
227 | tcpstat.tcps_timeoutdrop++;
|
---|
228 | #ifdef VBOX
|
---|
229 | tp = tcp_drop(pData, tp, tp->t_softerror);
|
---|
230 | #else /* !VBOX */
|
---|
231 | tp = tcp_drop(tp, tp->t_softerror);
|
---|
232 | #endif /* !VBOX */
|
---|
233 | /* tp->t_softerror : ETIMEDOUT); */ /* XXX */
|
---|
234 | return (tp); /* XXX */
|
---|
235 | }
|
---|
236 |
|
---|
237 | /*
|
---|
238 | * Set rxtshift to 6, which is still at the maximum
|
---|
239 | * backoff time
|
---|
240 | */
|
---|
241 | tp->t_rxtshift = 6;
|
---|
242 | }
|
---|
243 | tcpstat.tcps_rexmttimeo++;
|
---|
244 | rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
|
---|
245 | TCPT_RANGESET(tp->t_rxtcur, rexmt,
|
---|
246 | (short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */
|
---|
247 | tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
|
---|
248 | /*
|
---|
249 | * If losing, let the lower level know and try for
|
---|
250 | * a better route. Also, if we backed off this far,
|
---|
251 | * our srtt estimate is probably bogus. Clobber it
|
---|
252 | * so we'll take the next rtt measurement as our srtt;
|
---|
253 | * move the current srtt into rttvar to keep the current
|
---|
254 | * retransmit times until then.
|
---|
255 | */
|
---|
256 | if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
|
---|
257 | /* in_losing(tp->t_inpcb); */
|
---|
258 | tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
|
---|
259 | tp->t_srtt = 0;
|
---|
260 | }
|
---|
261 | tp->snd_nxt = tp->snd_una;
|
---|
262 | /*
|
---|
263 | * If timing a segment in this window, stop the timer.
|
---|
264 | */
|
---|
265 | tp->t_rtt = 0;
|
---|
266 | /*
|
---|
267 | * Close the congestion window down to one segment
|
---|
268 | * (we'll open it by one segment for each ack we get).
|
---|
269 | * Since we probably have a window's worth of unacked
|
---|
270 | * data accumulated, this "slow start" keeps us from
|
---|
271 | * dumping all that data as back-to-back packets (which
|
---|
272 | * might overwhelm an intermediate gateway).
|
---|
273 | *
|
---|
274 | * There are two phases to the opening: Initially we
|
---|
275 | * open by one mss on each ack. This makes the window
|
---|
276 | * size increase exponentially with time. If the
|
---|
277 | * window is larger than the path can handle, this
|
---|
278 | * exponential growth results in dropped packet(s)
|
---|
279 | * almost immediately. To get more time between
|
---|
280 | * drops but still "push" the network to take advantage
|
---|
281 | * of improving conditions, we switch from exponential
|
---|
282 | * to linear window opening at some threshold size.
|
---|
283 | * For a threshold, we use half the current window
|
---|
284 | * size, truncated to a multiple of the mss.
|
---|
285 | *
|
---|
286 | * (the minimum cwnd that will give us exponential
|
---|
287 | * growth is 2 mss. We don't allow the threshold
|
---|
288 | * to go below this.)
|
---|
289 | */
|
---|
290 | {
|
---|
291 | u_int win = min(tp->snd_wnd, tp->snd_cwnd) / 2 / tp->t_maxseg;
|
---|
292 | if (win < 2)
|
---|
293 | win = 2;
|
---|
294 | tp->snd_cwnd = tp->t_maxseg;
|
---|
295 | tp->snd_ssthresh = win * tp->t_maxseg;
|
---|
296 | tp->t_dupacks = 0;
|
---|
297 | }
|
---|
298 | #ifdef VBOX
|
---|
299 | (void) tcp_output(pData, tp);
|
---|
300 | #else /* !VBOX */
|
---|
301 | (void) tcp_output(tp);
|
---|
302 | #endif /* !VBOX */
|
---|
303 | break;
|
---|
304 |
|
---|
305 | /*
|
---|
306 | * Persistence timer into zero window.
|
---|
307 | * Force a byte to be output, if possible.
|
---|
308 | */
|
---|
309 | case TCPT_PERSIST:
|
---|
310 | tcpstat.tcps_persisttimeo++;
|
---|
311 | tcp_setpersist(tp);
|
---|
312 | tp->t_force = 1;
|
---|
313 | #ifdef VBOX
|
---|
314 | (void) tcp_output(pData, tp);
|
---|
315 | #else /* !VBOX */
|
---|
316 | (void) tcp_output(tp);
|
---|
317 | #endif /* !VBOX */
|
---|
318 | tp->t_force = 0;
|
---|
319 | break;
|
---|
320 |
|
---|
321 | /*
|
---|
322 | * Keep-alive timer went off; send something
|
---|
323 | * or drop connection if idle for too long.
|
---|
324 | */
|
---|
325 | case TCPT_KEEP:
|
---|
326 | tcpstat.tcps_keeptimeo++;
|
---|
327 | if (tp->t_state < TCPS_ESTABLISHED)
|
---|
328 | goto dropit;
|
---|
329 |
|
---|
330 | /* if (tp->t_socket->so_options & SO_KEEPALIVE && */
|
---|
331 | if ((so_options) && tp->t_state <= TCPS_CLOSE_WAIT) {
|
---|
332 | if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
|
---|
333 | goto dropit;
|
---|
334 | /*
|
---|
335 | * Send a packet designed to force a response
|
---|
336 | * if the peer is up and reachable:
|
---|
337 | * either an ACK if the connection is still alive,
|
---|
338 | * or an RST if the peer has closed the connection
|
---|
339 | * due to timeout or reboot.
|
---|
340 | * Using sequence number tp->snd_una-1
|
---|
341 | * causes the transmitted zero-length segment
|
---|
342 | * to lie outside the receive window;
|
---|
343 | * by the protocol spec, this requires the
|
---|
344 | * correspondent TCP to respond.
|
---|
345 | */
|
---|
346 | tcpstat.tcps_keepprobe++;
|
---|
347 | #ifdef TCP_COMPAT_42
|
---|
348 | /*
|
---|
349 | * The keepalive packet must have nonzero length
|
---|
350 | * to get a 4.2 host to respond.
|
---|
351 | */
|
---|
352 | tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
|
---|
353 | tp->rcv_nxt - 1, tp->snd_una - 1, 0);
|
---|
354 | #else
|
---|
355 | #ifdef VBOX
|
---|
356 | tcp_respond(pData, tp, &tp->t_template, (struct mbuf *)NULL,
|
---|
357 | tp->rcv_nxt, tp->snd_una - 1, 0);
|
---|
358 | #else /* !VBOX */
|
---|
359 | tcp_respond(tp, &tp->t_template, (struct mbuf *)NULL,
|
---|
360 | tp->rcv_nxt, tp->snd_una - 1, 0);
|
---|
361 | #endif /* !VBOX */
|
---|
362 | #endif
|
---|
363 | tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
|
---|
364 | } else
|
---|
365 | tp->t_timer[TCPT_KEEP] = tcp_keepidle;
|
---|
366 | break;
|
---|
367 |
|
---|
368 | dropit:
|
---|
369 | tcpstat.tcps_keepdrops++;
|
---|
370 | #ifdef VBOX
|
---|
371 | tp = tcp_drop(pData, tp, 0); /* ETIMEDOUT); */
|
---|
372 | #else /* !VBOX */
|
---|
373 | tp = tcp_drop(tp, 0); /* ETIMEDOUT); */
|
---|
374 | #endif /* !VBOX */
|
---|
375 | break;
|
---|
376 | }
|
---|
377 |
|
---|
378 | return (tp);
|
---|
379 | }
|
---|