source: rtems/cpukit/libnetworking/netinet/tcp_timer.c @ b3f8c9e1

4.104.11
Last change on this file since b3f8c9e1 was b3f8c9e1, checked in by Ralf Corsepius <ralf.corsepius@…>, on Dec 22, 2008 at 7:47:28 AM

Include <errno.h> (POSIX,C99) instead of <sys/errno.h> (BSD'ism).

  • Property mode set to 100644
File size: 10.5 KB
Line 
1/*
2 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
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 * 4. Neither the name of the University nor the names of its contributors
14 *    may be used to endorse or promote products derived from this software
15 *    without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 *      @(#)tcp_timer.c 8.2 (Berkeley) 5/24/95
30 *      $Id$
31 */
32
33#include "opt_tcpdebug.h"
34
35#ifndef TUBA_INCLUDE
36#include <sys/param.h>
37#include <rtems/bsd/sys/queue.h>
38#include <sys/systm.h>
39#include <sys/kernel.h>
40#include <sys/sysctl.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 <errno.h>
47
48#include <machine/cpu.h>        /* before tcp_seq.h, for tcp_random18() */
49
50#include <net/if.h>
51#include <net/route.h>
52
53#include <netinet/in.h>
54#include <netinet/in_systm.h>
55#include <netinet/ip.h>
56#include <netinet/in_pcb.h>
57#include <netinet/ip_var.h>
58#include <netinet/tcp.h>
59#include <netinet/tcp_fsm.h>
60#include <netinet/tcp_seq.h>
61#include <netinet/tcp_timer.h>
62#include <netinet/tcp_var.h>
63#include <netinet/tcpip.h>
64#ifdef TCPDEBUG
65#include <netinet/tcp_debug.h>
66#endif
67
68int     tcp_keepinit = TCPTV_KEEP_INIT;
69SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPINIT, keepinit,
70        CTLFLAG_RW, &tcp_keepinit , 0, "");
71
72int     tcp_keepidle = TCPTV_KEEP_IDLE;
73SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPIDLE, keepidle,
74        CTLFLAG_RW, &tcp_keepidle , 0, "");
75
76static int      tcp_keepintvl = TCPTV_KEEPINTVL;
77SYSCTL_INT(_net_inet_tcp, TCPCTL_KEEPINTVL, keepintvl,
78        CTLFLAG_RW, &tcp_keepintvl , 0, "");
79
80static int      always_keepalive = 0;
81SYSCTL_INT(_net_inet_tcp, OID_AUTO, always_keepalive,
82        CTLFLAG_RW, &always_keepalive , 0, "");
83
84static int      tcp_keepcnt = TCPTV_KEEPCNT;
85        /* max idle probes */
86static int      tcp_maxpersistidle = TCPTV_KEEP_IDLE;
87        /* max idle time in persist */
88int     tcp_maxidle;
89#else /* TUBA_INCLUDE */
90
91static  int tcp_maxpersistidle;
92#endif /* TUBA_INCLUDE */
93
94/*
95 * Fast timeout routine for processing delayed acks
96 */
97void
98tcp_fasttimo(void)
99{
100        register struct inpcb *inp;
101        register struct tcpcb *tp;
102        int s;
103
104        s = splnet();
105
106        for (inp = tcb.lh_first; inp != NULL; inp = inp->inp_list.le_next) {
107                if ((tp = (struct tcpcb *)inp->inp_ppcb) &&
108                    (tp->t_flags & TF_DELACK)) {
109                        tp->t_flags &= ~TF_DELACK;
110                        tp->t_flags |= TF_ACKNOW;
111                        tcpstat.tcps_delack++;
112                        (void) tcp_output(tp);
113                }
114        }
115        splx(s);
116}
117
118/*
119 * Tcp protocol timeout routine called every 500 ms.
120 * Updates the timers in all active tcb's and
121 * causes finite state machine actions if timers expire.
122 */
123void
124tcp_slowtimo(void)
125{
126        register struct inpcb *ip, *ipnxt;
127        register struct tcpcb *tp;
128        register int i;
129        int s;
130#ifdef TCPDEBUG
131        int ostate;
132#endif
133
134        s = splnet();
135
136        tcp_maxidle = tcp_keepcnt * tcp_keepintvl;
137
138        ip = tcb.lh_first;
139        if (ip == NULL) {
140                splx(s);
141                return;
142        }
143        /*
144         * Search through tcb's and update active timers.
145         */
146        for (; ip != NULL; ip = ipnxt) {
147                ipnxt = ip->inp_list.le_next;
148                tp = intotcpcb(ip);
149                if (tp == 0 || tp->t_state == TCPS_LISTEN)
150                        continue;
151                for (i = 0; i < TCPT_NTIMERS; i++) {
152                        if (tp->t_timer[i] && --tp->t_timer[i] == 0) {
153#ifdef TCPDEBUG
154                                ostate = tp->t_state;
155#endif
156                                tp = tcp_timers(tp, i);
157                                if (tp == NULL)
158                                        goto tpgone;
159#ifdef TCPDEBUG
160                                if (tp->t_inpcb->inp_socket->so_options
161                                    & SO_DEBUG)
162                                        tcp_trace(TA_USER, ostate, tp,
163                                                  (struct tcpiphdr *)0,
164                                                  PRU_SLOWTIMO);
165#endif
166                        }
167                }
168                tp->t_idle++;
169                tp->t_duration++;
170                if (tp->t_rtt)
171                        tp->t_rtt++;
172tpgone:
173                ;
174        }
175        tcp_iss += TCP_ISSINCR/PR_SLOWHZ;               /* increment iss */
176#ifdef TCP_COMPAT_42
177        if ((int)tcp_iss < 0)
178                tcp_iss = TCP_ISSINCR;                  /* XXX */
179#endif
180        tcp_now++;                                      /* for timestamps */
181        splx(s);
182}
183#ifndef TUBA_INCLUDE
184
185/*
186 * Cancel all timers for TCP tp.
187 */
188void
189tcp_canceltimers(struct tcpcb *tp)
190{
191        register int i;
192
193        for (i = 0; i < TCPT_NTIMERS; i++)
194                tp->t_timer[i] = 0;
195}
196
197int     tcp_backoff[TCP_MAXRXTSHIFT + 1] =
198    { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
199
200static int tcp_totbackoff = 511;        /* sum of tcp_backoff[] */
201
202/*
203 * TCP timer processing.
204 */
205struct tcpcb *
206tcp_timers(struct tcpcb *tp, int timer)
207{
208        register int rexmt;
209
210        switch (timer) {
211
212        /*
213         * 2 MSL timeout in shutdown went off.  If we're closed but
214         * still waiting for peer to close and connection has been idle
215         * too long, or if 2MSL time is up from TIME_WAIT, delete connection
216         * control block.  Otherwise, check again in a bit.
217         */
218        case TCPT_2MSL:
219                if (tp->t_state != TCPS_TIME_WAIT &&
220                    tp->t_idle <= tcp_maxidle)
221                        tp->t_timer[TCPT_2MSL] = tcp_keepintvl;
222                else
223                        tp = tcp_close(tp);
224                break;
225
226        /*
227         * Retransmission timer went off.  Message has not
228         * been acked within retransmit interval.  Back off
229         * to a longer retransmit interval and retransmit one segment.
230         */
231        case TCPT_REXMT:
232                if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
233                        tp->t_rxtshift = TCP_MAXRXTSHIFT;
234                        tcpstat.tcps_timeoutdrop++;
235                        tp = tcp_drop(tp, tp->t_softerror ?
236                            tp->t_softerror : ETIMEDOUT);
237                        break;
238                }
239                tcpstat.tcps_rexmttimeo++;
240                rexmt = TCP_REXMTVAL(tp) * tcp_backoff[tp->t_rxtshift];
241                TCPT_RANGESET(tp->t_rxtcur, rexmt,
242                    tp->t_rttmin, TCPTV_REXMTMAX);
243                tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
244                /*
245                 * If losing, let the lower level know and try for
246                 * a better route.  Also, if we backed off this far,
247                 * our srtt estimate is probably bogus.  Clobber it
248                 * so we'll take the next rtt measurement as our srtt;
249                 * move the current srtt into rttvar to keep the current
250                 * retransmit times until then.
251                 */
252                if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
253                        in_losing(tp->t_inpcb);
254                        tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
255                        tp->t_srtt = 0;
256                }
257                tp->snd_nxt = tp->snd_una;
258                /*
259                 * Force a segment to be sent.
260                 */
261                tp->t_flags |= TF_ACKNOW;
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 threshhold size.
283                 * For a threshhold, 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 threshhold
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                (void) tcp_output(tp);
299                break;
300
301        /*
302         * Persistance timer into zero window.
303         * Force a byte to be output, if possible.
304         */
305        case TCPT_PERSIST:
306                tcpstat.tcps_persisttimeo++;
307                /*
308                 * Hack: if the peer is dead/unreachable, we do not
309                 * time out if the window is closed.  After a full
310                 * backoff, drop the connection if the idle time
311                 * (no responses to probes) reaches the maximum
312                 * backoff that we would use if retransmitting.
313                 */
314                if (tp->t_rxtshift == TCP_MAXRXTSHIFT) {
315                        u_long maxidle = TCP_REXMTVAL(tp);
316                        if (maxidle < tp->t_rttmin)
317                                maxidle = tp->t_rttmin;
318                        maxidle *= tcp_totbackoff;
319                        if (tp->t_idle >= tcp_maxpersistidle ||
320                            tp->t_idle >= maxidle) {
321                                tcpstat.tcps_persistdrop++;
322                                tp = tcp_drop(tp, ETIMEDOUT);
323                                break;
324                        }
325                }
326                tcp_setpersist(tp);
327                tp->t_force = 1;
328                (void) tcp_output(tp);
329                tp->t_force = 0;
330                break;
331
332        /*
333         * Keep-alive timer went off; send something
334         * or drop connection if idle for too long.
335         */
336        case TCPT_KEEP:
337                tcpstat.tcps_keeptimeo++;
338                if (tp->t_state < TCPS_ESTABLISHED)
339                        goto dropit;
340                if ((always_keepalive ||
341                    tp->t_inpcb->inp_socket->so_options & SO_KEEPALIVE) &&
342                    tp->t_state <= TCPS_CLOSING) {
343                        if (tp->t_idle >= tcp_keepidle + tcp_maxidle)
344                                goto dropit;
345                        /*
346                         * Send a packet designed to force a response
347                         * if the peer is up and reachable:
348                         * either an ACK if the connection is still alive,
349                         * or an RST if the peer has closed the connection
350                         * due to timeout or reboot.
351                         * Using sequence number tp->snd_una-1
352                         * causes the transmitted zero-length segment
353                         * to lie outside the receive window;
354                         * by the protocol spec, this requires the
355                         * correspondent TCP to respond.
356                         */
357                        tcpstat.tcps_keepprobe++;
358#ifdef TCP_COMPAT_42
359                        /*
360                         * The keepalive packet must have nonzero length
361                         * to get a 4.2 host to respond.
362                         */
363                        tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
364                            tp->rcv_nxt - 1, tp->snd_una - 1, 0);
365#else
366                        tcp_respond(tp, tp->t_template, (struct mbuf *)NULL,
367                            tp->rcv_nxt, tp->snd_una - 1, 0);
368#endif
369                        tp->t_timer[TCPT_KEEP] = tcp_keepintvl;
370                } else
371                        tp->t_timer[TCPT_KEEP] = tcp_keepidle;
372                break;
373        dropit:
374                tcpstat.tcps_keepdrops++;
375                tp = tcp_drop(tp, ETIMEDOUT);
376                break;
377        }
378        return (tp);
379}
380#endif /* TUBA_INCLUDE */
Note: See TracBrowser for help on using the repository browser.