tcp_var.h revision 1.38 1 /* $NetBSD: tcp_var.h,v 1.38 1998/04/03 07:54:01 thorpej Exp $ */
2
3 /*-
4 * Copyright (c) 1997, 1998 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
9 * NASA Ames Research Center.
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. All advertising materials mentioning features or use of this software
20 * must display the following acknowledgement:
21 * This product includes software developed by the NetBSD
22 * Foundation, Inc. and its contributors.
23 * 4. Neither the name of The NetBSD Foundation nor the names of its
24 * contributors may be used to endorse or promote products derived
25 * from this software without specific prior written permission.
26 *
27 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
28 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
31 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37 * POSSIBILITY OF SUCH DAMAGE.
38 */
39
40 /*
41 * Copyright (c) 1982, 1986, 1993, 1994, 1995
42 * The Regents of the University of California. All rights reserved.
43 *
44 * Redistribution and use in source and binary forms, with or without
45 * modification, are permitted provided that the following conditions
46 * are met:
47 * 1. Redistributions of source code must retain the above copyright
48 * notice, this list of conditions and the following disclaimer.
49 * 2. Redistributions in binary form must reproduce the above copyright
50 * notice, this list of conditions and the following disclaimer in the
51 * documentation and/or other materials provided with the distribution.
52 * 3. All advertising materials mentioning features or use of this software
53 * must display the following acknowledgement:
54 * This product includes software developed by the University of
55 * California, Berkeley and its contributors.
56 * 4. Neither the name of the University nor the names of its contributors
57 * may be used to endorse or promote products derived from this software
58 * without specific prior written permission.
59 *
60 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
61 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
62 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
63 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
64 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
65 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
66 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
67 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
68 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
69 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
70 * SUCH DAMAGE.
71 *
72 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
73 */
74
75 #ifndef _NETINET_TCP_VAR_H_
76 #define _NETINET_TCP_VAR_H_
77
78 /*
79 * Kernel variables for tcp.
80 */
81
82 /*
83 * Tcp control block, one per tcp; fields:
84 */
85 struct tcpcb {
86 struct ipqehead segq; /* sequencing queue */
87 short t_state; /* state of this connection */
88 short t_timer[TCPT_NTIMERS]; /* tcp timers */
89 short t_rxtshift; /* log(2) of rexmt exp. backoff */
90 short t_rxtcur; /* current retransmit value */
91 short t_dupacks; /* consecutive dup acks recd */
92 u_short t_peermss; /* peer's maximum segment size */
93 u_short t_ourmss; /* our's maximum segment size */
94 u_short t_segsz; /* current segment size in use */
95 char t_force; /* 1 if forcing out a byte */
96 u_short t_flags;
97 #define TF_ACKNOW 0x0001 /* ack peer immediately */
98 #define TF_DELACK 0x0002 /* ack, but try to delay it */
99 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */
100 #define TF_NOOPT 0x0008 /* don't use tcp options */
101 #define TF_SENTFIN 0x0010 /* have sent FIN */
102 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */
103 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */
104 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */
105 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */
106 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */
107 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */
108
109 struct tcpiphdr *t_template; /* skeletal packet for transmit */
110 struct inpcb *t_inpcb; /* back pointer to internet pcb */
111 LIST_ENTRY(tcpcb) t_delack; /* delayed ACK queue */
112 /*
113 * The following fields are used as in the protocol specification.
114 * See RFC783, Dec. 1981, page 21.
115 */
116 /* send sequence variables */
117 tcp_seq snd_una; /* send unacknowledged */
118 tcp_seq snd_nxt; /* send next */
119 tcp_seq snd_up; /* send urgent pointer */
120 tcp_seq snd_wl1; /* window update seg seq number */
121 tcp_seq snd_wl2; /* window update seg ack number */
122 tcp_seq iss; /* initial send sequence number */
123 u_long snd_wnd; /* send window */
124 /* receive sequence variables */
125 u_long rcv_wnd; /* receive window */
126 tcp_seq rcv_nxt; /* receive next */
127 tcp_seq rcv_up; /* receive urgent pointer */
128 tcp_seq irs; /* initial receive sequence number */
129 /*
130 * Additional variables for this implementation.
131 */
132 /* receive variables */
133 tcp_seq rcv_adv; /* advertised window */
134 /* retransmit variables */
135 tcp_seq snd_max; /* highest sequence number sent;
136 * used to recognize retransmits
137 */
138 /* congestion control (for slow start, source quench, retransmit after loss) */
139 u_long snd_cwnd; /* congestion-controlled window */
140 u_long snd_ssthresh; /* snd_cwnd size threshhold for
141 * for slow start exponential to
142 * linear switch
143 */
144 /*
145 * transmit timing stuff. See below for scale of srtt and rttvar.
146 * "Variance" is actually smoothed difference.
147 */
148 short t_idle; /* inactivity time */
149 short t_rtt; /* round trip time */
150 tcp_seq t_rtseq; /* sequence number being timed */
151 short t_srtt; /* smoothed round-trip time */
152 short t_rttvar; /* variance in round-trip time */
153 short t_rttmin; /* minimum rtt allowed */
154 u_long max_sndwnd; /* largest window peer has offered */
155
156 /* out-of-band data */
157 char t_oobflags; /* have some */
158 char t_iobc; /* input character */
159 #define TCPOOB_HAVEDATA 0x01
160 #define TCPOOB_HADDATA 0x02
161 short t_softerror; /* possible error not yet reported */
162
163 /* RFC 1323 variables */
164 u_char snd_scale; /* window scaling for send window */
165 u_char rcv_scale; /* window scaling for recv window */
166 u_char request_r_scale; /* pending window scaling */
167 u_char requested_s_scale;
168 u_int32_t ts_recent; /* timestamp echo data */
169 u_int32_t ts_recent_age; /* when last updated */
170 tcp_seq last_ack_sent;
171
172 /* TUBA stuff */
173 caddr_t t_tuba_pcb; /* next level down pcb for TCP over z */
174 };
175
176 /*
177 * Queue for delayed ACK processing.
178 */
179 LIST_HEAD(tcp_delack_head, tcpcb);
180 #ifdef _KERNEL
181 extern struct tcp_delack_head tcp_delacks;
182
183 #define TCP_SET_DELACK(tp) \
184 do { \
185 if (((tp)->t_flags & TF_DELACK) == 0) { \
186 (tp)->t_flags |= TF_DELACK; \
187 LIST_INSERT_HEAD(&tcp_delacks, (tp), t_delack); \
188 } \
189 } while (0)
190
191 #define TCP_CLEAR_DELACK(tp) \
192 do { \
193 if ((tp)->t_flags & TF_DELACK) { \
194 (tp)->t_flags &= ~TF_DELACK; \
195 LIST_REMOVE((tp), t_delack); \
196 } \
197 } while (0)
198 #endif /* _KERNEL */
199
200 /*
201 * Handy way of passing around TCP option info.
202 */
203 struct tcp_opt_info {
204 int ts_present;
205 u_int32_t ts_val;
206 u_int32_t ts_ecr;
207 u_int16_t maxseg;
208 };
209
210 /*
211 * Data for the TCP compressed state engine.
212 */
213 struct syn_cache {
214 struct syn_cache *sc_next;
215 u_int32_t sc_tstmp : 1,
216 sc_hash : 31;
217 struct in_addr sc_src;
218 struct in_addr sc_dst;
219 tcp_seq sc_irs;
220 tcp_seq sc_iss;
221 int sc_flags;
222
223 #define SCF_SYNACK_REXMT 0x0001 /* SYN,ACK was retransmitted */
224
225 u_int16_t sc_sport;
226 u_int16_t sc_dport;
227 u_int16_t sc_peermaxseg;
228 u_int16_t sc_ourmaxseg;
229 u_int8_t sc_timer;
230 u_int8_t sc_request_r_scale : 4,
231 sc_requested_s_scale : 4;
232 };
233
234 struct syn_cache_head {
235 struct syn_cache *sch_first; /* First entry in the bucket */
236 struct syn_cache *sch_last; /* Last entry in the bucket */
237 struct syn_cache_head *sch_headq; /* The next non-empty bucket */
238 int16_t sch_timer_sum; /* Total time in this bucket */
239 u_int16_t sch_length; /* @ # elements in bucket */
240 };
241
242 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
243 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
244
245 /*
246 * The smoothed round-trip time and estimated variance
247 * are stored as fixed point numbers scaled by the values below.
248 * For convenience, these scales are also used in smoothing the average
249 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
250 * With these scales, srtt has 3 bits to the right of the binary point,
251 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
252 * binary point, and is smoothed with an ALPHA of 0.75.
253 */
254 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */
255 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */
256
257 /*
258 * The initial retransmission should happen at rtt + 4 * rttvar.
259 * Because of the way we do the smoothing, srtt and rttvar
260 * will each average +1/2 tick of bias. When we compute
261 * the retransmit timer, we want 1/2 tick of rounding and
262 * 1 extra tick because of +-1/2 tick uncertainty in the
263 * firing of the timer. The bias will give us exactly the
264 * 1.5 tick we need. But, because the bias is
265 * statistical, we have to test that we don't drop below
266 * the minimum feasible timer (which is 2 ticks).
267 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
268 * is the same as the multiplier for rttvar.
269 */
270 #define TCP_REXMTVAL(tp) \
271 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
272
273 #ifdef _KERNEL
274 /*
275 * Compute the initial window for slow start.
276 */
277 extern int tcp_init_win;
278 #define TCP_INITIAL_WINDOW(iw, segsz) \
279 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
280 ((segsz) * (iw)))
281 #endif /* _KERNEL */
282
283 /*
284 * TCP statistics.
285 * Many of these should be kept per connection,
286 * but that's inconvenient at the moment.
287 */
288 struct tcpstat {
289 u_long tcps_connattempt; /* connections initiated */
290 u_long tcps_accepts; /* connections accepted */
291 u_long tcps_connects; /* connections established */
292 u_long tcps_drops; /* connections dropped */
293 u_long tcps_conndrops; /* embryonic connections dropped */
294 u_long tcps_closed; /* conn. closed (includes drops) */
295 u_long tcps_segstimed; /* segs where we tried to get rtt */
296 u_long tcps_rttupdated; /* times we succeeded */
297 u_long tcps_delack; /* delayed acks sent */
298 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
299 u_long tcps_rexmttimeo; /* retransmit timeouts */
300 u_long tcps_persisttimeo; /* persist timeouts */
301 u_long tcps_keeptimeo; /* keepalive timeouts */
302 u_long tcps_keepprobe; /* keepalive probes sent */
303 u_long tcps_keepdrops; /* connections dropped in keepalive */
304 u_long tcps_persistdrops; /* connections dropped in persist */
305 u_long tcps_connsdrained; /* connections drained due to memory
306 shortage */
307
308 u_long tcps_sndtotal; /* total packets sent */
309 u_long tcps_sndpack; /* data packets sent */
310 u_long tcps_sndbyte; /* data bytes sent */
311 u_long tcps_sndrexmitpack; /* data packets retransmitted */
312 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
313 u_long tcps_sndacks; /* ack-only packets sent */
314 u_long tcps_sndprobe; /* window probes sent */
315 u_long tcps_sndurg; /* packets sent with URG only */
316 u_long tcps_sndwinup; /* window update-only packets sent */
317 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
318
319 u_long tcps_rcvtotal; /* total packets received */
320 u_long tcps_rcvpack; /* packets received in sequence */
321 u_long tcps_rcvbyte; /* bytes received in sequence */
322 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
323 u_long tcps_rcvbadoff; /* packets received with bad offset */
324 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
325 u_long tcps_rcvshort; /* packets received too short */
326 u_long tcps_rcvduppack; /* duplicate-only packets received */
327 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
328 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
329 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
330 u_long tcps_rcvoopack; /* out-of-order packets received */
331 u_long tcps_rcvoobyte; /* out-of-order bytes received */
332 u_long tcps_rcvpackafterwin; /* packets with data after window */
333 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
334 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
335 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
336 u_long tcps_rcvdupack; /* rcvd duplicate acks */
337 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
338 u_long tcps_rcvackpack; /* rcvd ack packets */
339 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
340 u_long tcps_rcvwinupd; /* rcvd window update packets */
341 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
342 u_long tcps_predack; /* times hdr predict ok for acks */
343 u_long tcps_preddat; /* times hdr predict ok for data pkts */
344
345 u_long tcps_pcbhashmiss; /* input packets missing pcb hash */
346 u_long tcps_noport; /* no socket on port */
347 u_long tcps_badsyn; /* received ack for which we have
348 no SYN in compressed state */
349
350 /* These statistics deal with the SYN cache. */
351 u_long tcps_sc_added; /* # of entries added */
352 u_long tcps_sc_completed; /* # of connections completed */
353 u_long tcps_sc_timed_out; /* # of entries timed out */
354 u_long tcps_sc_overflowed; /* # dropped due to overflow */
355 u_long tcps_sc_reset; /* # dropped due to RST */
356 u_long tcps_sc_unreach; /* # dropped due to ICMP unreach */
357 u_long tcps_sc_bucketoverflow; /* # dropped due to bucket overflow */
358 u_long tcps_sc_aborted; /* # of entries aborted (no mem) */
359 u_long tcps_sc_dupesyn; /* # of duplicate SYNs received */
360 u_long tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */
361 u_long tcps_sc_collisions; /* # of hash collisions */
362 };
363
364 /*
365 * Names for TCP sysctl objects.
366 */
367 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */
368 #define TCPCTL_SENDSPACE 2 /* default send buffer */
369 #define TCPCTL_RECVSPACE 3 /* default recv buffer */
370 #define TCPCTL_MSSDFLT 4 /* default seg size */
371 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */
372 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */
373 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */
374 #define TCPCTL_INIT_WIN 8 /* initial window */
375 #define TCPCTL_MAXID 9
376
377 #define TCPCTL_NAMES { \
378 { 0, 0 }, \
379 { "rfc1323", CTLTYPE_INT }, \
380 { "sendspace", CTLTYPE_INT }, \
381 { "recvspace", CTLTYPE_INT }, \
382 { "mssdflt", CTLTYPE_INT }, \
383 { "syn_cache_limit", CTLTYPE_INT }, \
384 { "syn_bucket_limit", CTLTYPE_INT }, \
385 { "syn_cache_interval", CTLTYPE_INT },\
386 { "init_win", CTLTYPE_INT }, \
387 }
388
389 #ifdef _KERNEL
390 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */
391 struct tcpstat tcpstat; /* tcp statistics */
392 u_int32_t tcp_now; /* for RFC 1323 timestamps */
393 extern int tcp_do_rfc1323; /* enabled/disabled? */
394 extern int tcp_mssdflt; /* default seg size */
395 extern int tcp_init_win; /* initial window */
396 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */
397 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */
398 extern int tcp_syn_cache_interval; /* compressed state timer */
399
400 extern int tcp_syn_cache_size;
401 extern int tcp_syn_cache_timeo;
402 extern struct syn_cache_head tcp_syn_cache[], *tcp_syn_cache_first;
403 extern u_long syn_cache_count;
404
405 int tcp_attach __P((struct socket *));
406 void tcp_canceltimers __P((struct tcpcb *));
407 struct tcpcb *
408 tcp_close __P((struct tcpcb *));
409 void *tcp_ctlinput __P((int, struct sockaddr *, void *));
410 int tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
411 struct tcpcb *
412 tcp_disconnect __P((struct tcpcb *));
413 struct tcpcb *
414 tcp_drop __P((struct tcpcb *, int));
415 void tcp_dooptions __P((struct tcpcb *,
416 u_char *, int, struct tcpiphdr *, struct tcp_opt_info *));
417 void tcp_drain __P((void));
418 void tcp_established __P((struct tcpcb *));
419 void tcp_fasttimo __P((void));
420 void tcp_init __P((void));
421 void tcp_input __P((struct mbuf *, ...));
422 int tcp_mss_to_advertise __P((const struct tcpcb *));
423 void tcp_mss_from_peer __P((struct tcpcb *, int));
424 void tcp_mtudisc __P((struct inpcb *, int));
425 struct tcpcb *
426 tcp_newtcpcb __P((struct inpcb *));
427 void tcp_notify __P((struct inpcb *, int));
428 u_int tcp_optlen __P((struct tcpcb *));
429 int tcp_output __P((struct tcpcb *));
430 void tcp_pulloutofband __P((struct socket *,
431 struct tcpiphdr *, struct mbuf *));
432 void tcp_quench __P((struct inpcb *, int));
433 int tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *));
434 int tcp_respond __P((struct tcpcb *,
435 struct tcpiphdr *, struct mbuf *, tcp_seq, tcp_seq, int));
436 void tcp_rmx_rtt __P((struct tcpcb *));
437 void tcp_setpersist __P((struct tcpcb *));
438 void tcp_slowtimo __P((void));
439 struct tcpiphdr *
440 tcp_template __P((struct tcpcb *));
441 struct tcpcb *
442 tcp_timers __P((struct tcpcb *, int));
443 void tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int));
444 struct tcpcb *
445 tcp_usrclosed __P((struct tcpcb *));
446 int tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
447 int tcp_usrreq __P((struct socket *,
448 int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
449 void tcp_xmit_timer __P((struct tcpcb *, int));
450 tcp_seq tcp_new_iss __P((void *, u_long, tcp_seq));
451
452 int syn_cache_add __P((struct socket *, struct mbuf *, u_char *,
453 int, struct tcp_opt_info *));
454 void syn_cache_unreach __P((struct ip *, struct tcphdr *));
455 struct socket *
456 syn_cache_get __P((struct socket *so, struct mbuf *));
457 void syn_cache_insert __P((struct syn_cache *, struct syn_cache ***,
458 struct syn_cache_head **));
459 struct syn_cache *
460 syn_cache_lookup __P((struct tcpiphdr *, struct syn_cache ***,
461 struct syn_cache_head **));
462 void syn_cache_reset __P((struct tcpiphdr *));
463 int syn_cache_respond __P((struct syn_cache *, struct mbuf *,
464 register struct tcpiphdr *, long, u_long));
465 void syn_cache_timer __P((int));
466
467 #endif
468
469 #endif /* _NETINET_TCP_VAR_H_ */
470