xref: /netbsd-src/sys/netinet/tcp_var.h (revision 1ca5c1b28139779176bd5c13ad7c5f25c0bcd5f8)
1 /*	$NetBSD: tcp_var.h,v 1.87 2001/09/11 21:03:21 thorpej Exp $	*/
2 
3 /*
4 %%% portions-copyright-nrl-98
5 Portions of this software are Copyright 1998 by Randall Atkinson,
6 Ronald Lee, Daniel McDonald, Bao Phan, and Chris Winters. All Rights
7 Reserved. All rights under this copyright have been assigned to the US
8 Naval Research Laboratory (NRL). The NRL Copyright Notice and License
9 Agreement Version 1.1 (January 17, 1995) applies to these portions of the
10 software.
11 You should have received a copy of the license with this software. If you
12 didn't get a copy, you may request one from <license@ipv6.nrl.navy.mil>.
13 
14 */
15 
16 /*
17  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
18  * All rights reserved.
19  *
20  * Redistribution and use in source and binary forms, with or without
21  * modification, are permitted provided that the following conditions
22  * are met:
23  * 1. Redistributions of source code must retain the above copyright
24  *    notice, this list of conditions and the following disclaimer.
25  * 2. Redistributions in binary form must reproduce the above copyright
26  *    notice, this list of conditions and the following disclaimer in the
27  *    documentation and/or other materials provided with the distribution.
28  * 3. Neither the name of the project nor the names of its contributors
29  *    may be used to endorse or promote products derived from this software
30  *    without specific prior written permission.
31  *
32  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
33  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
34  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
35  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
36  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
37  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
38  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
39  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
40  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
41  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
42  * SUCH DAMAGE.
43  */
44 
45 /*-
46  * Copyright (c) 1997, 1998, 1999, 2001 The NetBSD Foundation, Inc.
47  * All rights reserved.
48  *
49  * This code is derived from software contributed to The NetBSD Foundation
50  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
51  * NASA Ames Research Center.
52  *
53  * Redistribution and use in source and binary forms, with or without
54  * modification, are permitted provided that the following conditions
55  * are met:
56  * 1. Redistributions of source code must retain the above copyright
57  *    notice, this list of conditions and the following disclaimer.
58  * 2. Redistributions in binary form must reproduce the above copyright
59  *    notice, this list of conditions and the following disclaimer in the
60  *    documentation and/or other materials provided with the distribution.
61  * 3. All advertising materials mentioning features or use of this software
62  *    must display the following acknowledgement:
63  *	This product includes software developed by the NetBSD
64  *	Foundation, Inc. and its contributors.
65  * 4. Neither the name of The NetBSD Foundation nor the names of its
66  *    contributors may be used to endorse or promote products derived
67  *    from this software without specific prior written permission.
68  *
69  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
70  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
71  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
72  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
73  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
74  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
75  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
76  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
77  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
78  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
79  * POSSIBILITY OF SUCH DAMAGE.
80  */
81 
82 /*
83  * Copyright (c) 1982, 1986, 1993, 1994, 1995
84  *	The Regents of the University of California.  All rights reserved.
85  *
86  * Redistribution and use in source and binary forms, with or without
87  * modification, are permitted provided that the following conditions
88  * are met:
89  * 1. Redistributions of source code must retain the above copyright
90  *    notice, this list of conditions and the following disclaimer.
91  * 2. Redistributions in binary form must reproduce the above copyright
92  *    notice, this list of conditions and the following disclaimer in the
93  *    documentation and/or other materials provided with the distribution.
94  * 3. All advertising materials mentioning features or use of this software
95  *    must display the following acknowledgement:
96  *	This product includes software developed by the University of
97  *	California, Berkeley and its contributors.
98  * 4. Neither the name of the University nor the names of its contributors
99  *    may be used to endorse or promote products derived from this software
100  *    without specific prior written permission.
101  *
102  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
103  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
104  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
105  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
106  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
107  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
108  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
109  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
110  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
111  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
112  * SUCH DAMAGE.
113  *
114  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
115  */
116 
117 #ifndef _NETINET_TCP_VAR_H_
118 #define _NETINET_TCP_VAR_H_
119 
120 #if defined(_KERNEL_OPT)
121 #include "opt_inet.h"
122 #endif
123 
124 /*
125  * Kernel variables for tcp.
126  */
127 
128 #include <sys/callout.h>
129 
130 /*
131  * Tcp control block, one per tcp; fields:
132  */
133 struct tcpcb {
134 	int	t_family;		/* address family on the wire */
135 	struct ipqehead segq;		/* sequencing queue */
136 	struct callout t_timer[TCPT_NTIMERS];/* tcp timers */
137 	short	t_state;		/* state of this connection */
138 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
139 	uint32_t t_rxtcur;		/* current retransmit value */
140 	short	t_dupacks;		/* consecutive dup acks recd */
141 	u_short	t_peermss;		/* peer's maximum segment size */
142 	u_short	t_ourmss;		/* our's maximum segment size */
143 	u_short t_segsz;		/* current segment size in use */
144 	char	t_force;		/* 1 if forcing out a byte */
145 	u_int	t_flags;
146 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
147 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
148 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
149 #define	TF_NOOPT	0x0008		/* don't use tcp options */
150 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
151 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
152 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
153 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
154 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
155 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
156 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
157 #define	TF_CANT_TXSACK	0x1000		/* other side said I could not SACK */
158 #define	TF_IGNR_RXSACK	0x2000		/* ignore received SACK blocks */
159 #define	TF_REASSEMBLING	0x4000		/* we're busy reassembling */
160 
161 
162 	struct	mbuf *t_template;	/* skeletal packet for transmit */
163 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
164 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
165 	struct	callout t_delack_ch;	/* delayed ACK callout */
166 /*
167  * The following fields are used as in the protocol specification.
168  * See RFC783, Dec. 1981, page 21.
169  */
170 /* send sequence variables */
171 	tcp_seq	snd_una;		/* send unacknowledged */
172 	tcp_seq	snd_nxt;		/* send next */
173 	tcp_seq	snd_up;			/* send urgent pointer */
174 	tcp_seq	snd_wl1;		/* window update seg seq number */
175 	tcp_seq	snd_wl2;		/* window update seg ack number */
176 	tcp_seq	iss;			/* initial send sequence number */
177 	u_long	snd_wnd;		/* send window */
178 	tcp_seq snd_recover;		/* for use in fast recovery */
179 /* receive sequence variables */
180 	u_long	rcv_wnd;		/* receive window */
181 	tcp_seq	rcv_nxt;		/* receive next */
182 	tcp_seq	rcv_up;			/* receive urgent pointer */
183 	tcp_seq	irs;			/* initial receive sequence number */
184 /*
185  * Additional variables for this implementation.
186  */
187 /* receive variables */
188 	tcp_seq	rcv_adv;		/* advertised window */
189 /* retransmit variables */
190 	tcp_seq	snd_max;		/* highest sequence number sent;
191 					 * used to recognize retransmits
192 					 */
193 /* congestion control (for slow start, source quench, retransmit after loss) */
194 	u_long	snd_cwnd;		/* congestion-controlled window */
195 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
196 					 * for slow start exponential to
197 					 * linear switch
198 					 */
199 /*
200  * transmit timing stuff.  See below for scale of srtt and rttvar.
201  * "Variance" is actually smoothed difference.
202  */
203 	uint32_t t_rcvtime;		/* time last segment received */
204 	uint32_t t_rtttime;		/* time we started measuring rtt */
205 	tcp_seq	t_rtseq;		/* sequence number being timed */
206 	int32_t	t_srtt;			/* smoothed round-trip time */
207 	int32_t	t_rttvar;		/* variance in round-trip time */
208 	uint32_t t_rttmin;		/* minimum rtt allowed */
209 	u_long	max_sndwnd;		/* largest window peer has offered */
210 
211 /* out-of-band data */
212 	char	t_oobflags;		/* have some */
213 	char	t_iobc;			/* input character */
214 #define	TCPOOB_HAVEDATA	0x01
215 #define	TCPOOB_HADDATA	0x02
216 	short	t_softerror;		/* possible error not yet reported */
217 
218 /* RFC 1323 variables */
219 	u_char	snd_scale;		/* window scaling for send window */
220 	u_char	rcv_scale;		/* window scaling for recv window */
221 	u_char	request_r_scale;	/* pending window scaling */
222 	u_char	requested_s_scale;
223 	u_int32_t ts_recent;		/* timestamp echo data */
224 	u_int32_t ts_recent_age;	/* when last updated */
225 	u_int32_t ts_timebase;		/* our timebase */
226 	tcp_seq	last_ack_sent;
227 
228 /* SACK stuff */
229 	struct ipqehead timeq;		/* time sequenced queue (for SACK) */
230 
231 /* pointer for syn cache entries*/
232 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
233 };
234 
235 #ifdef _KERNEL
236 /*
237  * TCP reassembly queue locks.
238  */
239 static __inline int tcp_reass_lock_try __P((struct tcpcb *))
240 	__attribute__((__unused__));
241 static __inline void tcp_reass_unlock __P((struct tcpcb *))
242 	__attribute__((__unused__));
243 
244 static __inline int
245 tcp_reass_lock_try(tp)
246 	struct tcpcb *tp;
247 {
248 	int s;
249 
250 	/*
251 	 * Use splvm() -- we're blocking things that would cause
252 	 * mbuf allocation.
253 	 */
254 	s = splvm();
255 	if (tp->t_flags & TF_REASSEMBLING) {
256 		splx(s);
257 		return (0);
258 	}
259 	tp->t_flags |= TF_REASSEMBLING;
260 	splx(s);
261 	return (1);
262 }
263 
264 static __inline void
265 tcp_reass_unlock(tp)
266 	struct tcpcb *tp;
267 {
268 	int s;
269 
270 	s = splvm();
271 	tp->t_flags &= ~TF_REASSEMBLING;
272 	splx(s);
273 }
274 
275 #ifdef DIAGNOSTIC
276 #define	TCP_REASS_LOCK(tp)						\
277 do {									\
278 	if (tcp_reass_lock_try(tp) == 0) {				\
279 		printf("%s:%d: tcpcb %p reass already locked\n",	\
280 		    __FILE__, __LINE__, tp);				\
281 		panic("tcp_reass_lock");				\
282 	}								\
283 } while (0)
284 #define	TCP_REASS_LOCK_CHECK(tp)					\
285 do {									\
286 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
287 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
288 		    __FILE__, __LINE__, tp);				\
289 		panic("tcp reass lock check");				\
290 	}								\
291 } while (0)
292 #else
293 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
294 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
295 #endif
296 
297 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
298 #endif /* _KERNEL */
299 
300 /*
301  * Queue for delayed ACK processing.
302  */
303 #ifdef _KERNEL
304 extern int tcp_delack_ticks;
305 void	tcp_delack(void *);
306 
307 #define TCP_RESTART_DELACK(tp)						\
308 	callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks,		\
309 	    tcp_delack, tp)
310 
311 #define	TCP_SET_DELACK(tp)						\
312 do {									\
313 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
314 		(tp)->t_flags |= TF_DELACK;				\
315 		TCP_RESTART_DELACK(tp);					\
316 	}								\
317 } while (/*CONSTCOND*/0)
318 
319 #define	TCP_CLEAR_DELACK(tp)						\
320 do {									\
321 	if ((tp)->t_flags & TF_DELACK) {				\
322 		(tp)->t_flags &= ~TF_DELACK;				\
323 		callout_stop(&(tp)->t_delack_ch);			\
324 	}								\
325 } while (/*CONSTCOND*/0)
326 #endif /* _KERNEL */
327 
328 /*
329  * Compute the current timestamp for a connection.
330  */
331 #define	TCP_TIMESTAMP(tp)	(tcp_now - (tp)->ts_timebase)
332 
333 /*
334  * Handy way of passing around TCP option info.
335  */
336 struct tcp_opt_info {
337 	int		ts_present;
338 	u_int32_t	ts_val;
339 	u_int32_t	ts_ecr;
340 	u_int16_t	maxseg;
341 };
342 
343 /*
344  * Data for the TCP compressed state engine.
345  */
346 union syn_cache_sa {
347 	struct sockaddr sa;
348 	struct sockaddr_in sin;
349 #if 1 /*def INET6*/
350 	struct sockaddr_in6 sin6;
351 #endif
352 };
353 
354 struct syn_cache {
355 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
356 	struct callout sc_timer;		/* rexmt timer */
357 	union {					/* cached route */
358 		struct route route4;
359 #ifdef INET6
360 		struct route_in6 route6;
361 #endif
362 	} sc_route_u;
363 #define sc_route4	sc_route_u.route4
364 #ifdef INET6
365 #define sc_route6	sc_route_u.route6
366 #endif
367 	long sc_win;				/* advertised window */
368 	int sc_bucketidx;			/* our bucket index */
369 	u_int32_t sc_hash;
370 	u_int32_t sc_timestamp;			/* timestamp from SYN */
371 	u_int32_t sc_timebase;			/* our local timebase */
372 	union syn_cache_sa sc_src;
373 	union syn_cache_sa sc_dst;
374 	tcp_seq sc_irs;
375 	tcp_seq sc_iss;
376 	u_int sc_rxtcur;			/* current rxt timeout */
377 	u_int sc_rxttot;			/* total time spend on queues */
378 	u_short sc_rxtshift;			/* for computing backoff */
379 	u_short sc_flags;
380 
381 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
382 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
383 
384 	struct mbuf *sc_ipopts;			/* IP options */
385 	u_int16_t sc_peermaxseg;
386 	u_int16_t sc_ourmaxseg;
387 	u_int8_t sc_request_r_scale	: 4,
388 		 sc_requested_s_scale	: 4;
389 
390 	struct tcpcb *sc_tp;			/* tcb for listening socket */
391 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
392 };
393 
394 struct syn_cache_head {
395 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
396 	u_short sch_length;			/* # entries in bucket */
397 };
398 
399 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
400 #ifdef INET6
401 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
402 #endif
403 #ifndef INET6
404 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
405 #else
406 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
407 				? intotcpcb(sotoinpcb(so)) \
408 				: in6totcpcb(sotoin6pcb(so)))
409 #endif
410 
411 /*
412  * The smoothed round-trip time and estimated variance
413  * are stored as fixed point numbers scaled by the values below.
414  * For convenience, these scales are also used in smoothing the average
415  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
416  * With these scales, srtt has 3 bits to the right of the binary point,
417  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
418  * binary point, and is smoothed with an ALPHA of 0.75.
419  */
420 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
421 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
422 
423 /*
424  * The initial retransmission should happen at rtt + 4 * rttvar.
425  * Because of the way we do the smoothing, srtt and rttvar
426  * will each average +1/2 tick of bias.  When we compute
427  * the retransmit timer, we want 1/2 tick of rounding and
428  * 1 extra tick because of +-1/2 tick uncertainty in the
429  * firing of the timer.  The bias will give us exactly the
430  * 1.5 tick we need.  But, because the bias is
431  * statistical, we have to test that we don't drop below
432  * the minimum feasible timer (which is 2 ticks).
433  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
434  * is the same as the multiplier for rttvar.
435  */
436 #define	TCP_REXMTVAL(tp) \
437 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
438 
439 /*
440  * Compute the initial window for slow start.
441  */
442 #define	TCP_INITIAL_WINDOW(iw, segsz) \
443 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
444 	 ((segsz) * (iw)))
445 
446 /*
447  * TCP statistics.
448  * Many of these should be kept per connection,
449  * but that's inconvenient at the moment.
450  */
451 struct	tcpstat {
452 	u_quad_t tcps_connattempt;	/* connections initiated */
453 	u_quad_t tcps_accepts;		/* connections accepted */
454 	u_quad_t tcps_connects;		/* connections established */
455 	u_quad_t tcps_drops;		/* connections dropped */
456 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
457 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
458 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
459 	u_quad_t tcps_rttupdated;	/* times we succeeded */
460 	u_quad_t tcps_delack;		/* delayed acks sent */
461 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
462 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
463 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
464 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
465 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
466 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
467 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
468 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
469 					   shortage */
470 
471 	u_quad_t tcps_sndtotal;		/* total packets sent */
472 	u_quad_t tcps_sndpack;		/* data packets sent */
473 	u_quad_t tcps_sndbyte;		/* data bytes sent */
474 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
475 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
476 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
477 	u_quad_t tcps_sndprobe;		/* window probes sent */
478 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
479 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
480 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
481 
482 	u_quad_t tcps_rcvtotal;		/* total packets received */
483 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
484 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
485 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
486 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
487 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
488 	u_quad_t tcps_rcvshort;		/* packets received too short */
489 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
490 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
491 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
492 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
493 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
494 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
495 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
496 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
497 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
498 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
499 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
500 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
501 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
502 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
503 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
504 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
505 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
506 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
507 
508 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
509 	u_quad_t tcps_noport;		/* no socket on port */
510 	u_quad_t tcps_badsyn;		/* received ack for which we have
511 					   no SYN in compressed state */
512 
513 	/* These statistics deal with the SYN cache. */
514 	u_quad_t tcps_sc_added;		/* # of entries added */
515 	u_quad_t tcps_sc_completed;	/* # of connections completed */
516 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
517 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
518 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
519 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
520 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
521 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
522 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
523 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
524 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
525 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
526 
527 	u_quad_t tcps_selfquench;	/* # of ENOBUFS we get on output */
528 };
529 
530 /*
531  * Names for TCP sysctl objects.
532  */
533 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
534 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
535 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
536 #define	TCPCTL_MSSDFLT		4	/* default seg size */
537 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
538 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
539 #if 0	/*obsoleted*/
540 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
541 #endif
542 #define	TCPCTL_INIT_WIN		8	/* initial window */
543 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
544 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
545 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
546 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
547 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
548 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
549 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
550 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
551 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
552 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
553 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
554 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
555 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
556 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
557 #if 0	/*obsoleted*/
558 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
559 #endif
560 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
561 #define	TCPCTL_DELACK_TICKS	25	/* # ticks to delay ACK */
562 #define	TCPCTL_MAXID		26
563 
564 #define	TCPCTL_NAMES { \
565 	{ 0, 0 }, \
566 	{ "rfc1323",	CTLTYPE_INT }, \
567 	{ "sendspace",	CTLTYPE_INT }, \
568 	{ "recvspace",	CTLTYPE_INT }, \
569 	{ "mssdflt",	CTLTYPE_INT }, \
570 	{ "syn_cache_limit", CTLTYPE_INT }, \
571 	{ "syn_bucket_limit", CTLTYPE_INT }, \
572 	{ 0, 0 },\
573 	{ "init_win", CTLTYPE_INT }, \
574 	{ "mss_ifmtu", CTLTYPE_INT }, \
575 	{ "sack", CTLTYPE_INT }, \
576 	{ "win_scale", CTLTYPE_INT }, \
577 	{ "timestamps", CTLTYPE_INT }, \
578 	{ "compat_42", CTLTYPE_INT }, \
579 	{ "cwm", CTLTYPE_INT }, \
580 	{ "cwm_burstsize", CTLTYPE_INT }, \
581 	{ "ack_on_push", CTLTYPE_INT }, \
582 	{ "keepidle",	CTLTYPE_INT }, \
583 	{ "keepintvl",	CTLTYPE_INT }, \
584 	{ "keepcnt",	CTLTYPE_INT }, \
585 	{ "slowhz",	CTLTYPE_INT }, \
586 	{ "newreno",	CTLTYPE_INT }, \
587 	{ "log_refused",CTLTYPE_INT }, \
588 	{ 0, 0 }, \
589 	{ "rstppslimit", CTLTYPE_INT }, \
590 	{ "delack_ticks", CTLTYPE_INT }, \
591 }
592 
593 #ifdef _KERNEL
594 struct	inpcbtable tcbtable;	/* head of queue of active tcpcb's */
595 #ifdef INET6
596 extern struct in6pcb tcb6;
597 #endif
598 struct	tcpstat tcpstat;	/* tcp statistics */
599 u_int32_t tcp_now;		/* for RFC 1323 timestamps */
600 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
601 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
602 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
603 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
604 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
605 extern	int tcp_mssdflt;	/* default seg size */
606 extern	int tcp_init_win;	/* initial window */
607 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
608 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
609 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
610 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
611 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
612 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
613 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
614 extern	int tcp_log_refused;	/* log refused connections */
615 
616 extern	int tcp_rst_ppslim;
617 
618 extern	int tcp_syn_cache_size;
619 extern	struct syn_cache_head tcp_syn_cache[];
620 extern	u_long syn_cache_count;
621 
622 #define	TCPCTL_VARIABLES { \
623 	{ 0 },					\
624 	{ 1, 0, &tcp_do_rfc1323 },		\
625 	{ 1, 0, &tcp_sendspace },		\
626 	{ 1, 0, &tcp_recvspace },		\
627 	{ 1, 0, &tcp_mssdflt },			\
628 	{ 1, 0, &tcp_syn_cache_limit },		\
629 	{ 1, 0, &tcp_syn_bucket_limit },	\
630 	{ 0 },					\
631 	{ 1, 0, &tcp_init_win },		\
632 	{ 1, 0, &tcp_mss_ifmtu },		\
633 	{ 1, 0, &tcp_do_sack },			\
634 	{ 1, 0, &tcp_do_win_scale },		\
635 	{ 1, 0, &tcp_do_timestamps },		\
636 	{ 1, 0, &tcp_compat_42 },		\
637 	{ 1, 0, &tcp_cwm },			\
638 	{ 1, 0, &tcp_cwm_burstsize },		\
639 	{ 1, 0, &tcp_ack_on_push },		\
640 	{ 1, 0, &tcp_keepidle },		\
641 	{ 1, 0, &tcp_keepintvl },		\
642 	{ 1, 0, &tcp_keepcnt },			\
643 	{ 1, 1, 0, PR_SLOWHZ },			\
644 	{ 1, 0, &tcp_do_newreno },		\
645 	{ 1, 0, &tcp_log_refused },		\
646 	{ 0 },					\
647 	{ 1, 0, &tcp_rst_ppslim },		\
648 	{ 1, 0, &tcp_delack_ticks },		\
649 }
650 
651 int	 tcp_attach __P((struct socket *));
652 void	 tcp_canceltimers __P((struct tcpcb *));
653 struct tcpcb *
654 	 tcp_close __P((struct tcpcb *));
655 #ifdef INET6
656 void	 tcp6_ctlinput __P((int, struct sockaddr *, void *));
657 #endif
658 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
659 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
660 struct tcpcb *
661 	 tcp_disconnect __P((struct tcpcb *));
662 struct tcpcb *
663 	 tcp_drop __P((struct tcpcb *, int));
664 void	 tcp_dooptions __P((struct tcpcb *,
665 	    u_char *, int, struct tcphdr *, struct tcp_opt_info *));
666 void	 tcp_drain __P((void));
667 void	 tcp_established __P((struct tcpcb *));
668 void	 tcp_init __P((void));
669 #ifdef INET6
670 int	 tcp6_input __P((struct mbuf **, int *, int));
671 #endif
672 void	 tcp_input __P((struct mbuf *, ...));
673 u_long	 tcp_mss_to_advertise __P((const struct ifnet *, int));
674 void	 tcp_mss_from_peer __P((struct tcpcb *, int));
675 struct tcpcb *
676 	 tcp_newtcpcb __P((int, void *));
677 void	 tcp_notify __P((struct inpcb *, int));
678 #ifdef INET6
679 void	 tcp6_notify __P((struct in6pcb *, int));
680 #endif
681 u_int	 tcp_optlen __P((struct tcpcb *));
682 int	 tcp_output __P((struct tcpcb *));
683 void	 tcp_pulloutofband __P((struct socket *,
684 	    struct tcphdr *, struct mbuf *, int));
685 void	 tcp_quench __P((struct inpcb *, int));
686 #ifdef INET6
687 void	 tcp6_quench __P((struct in6pcb *, int));
688 #endif
689 int	 tcp_reass __P((struct tcpcb *, struct tcphdr *, struct mbuf *, int *));
690 int	 tcp_respond __P((struct tcpcb *, struct mbuf *, struct mbuf *,
691 	    struct tcphdr *, tcp_seq, tcp_seq, int));
692 void	 tcp_rmx_rtt __P((struct tcpcb *));
693 void	 tcp_setpersist __P((struct tcpcb *));
694 void	 tcp_slowtimo __P((void));
695 struct mbuf *
696 	 tcp_template __P((struct tcpcb *));
697 void	 tcp_trace __P((int, int, struct tcpcb *, struct mbuf *, int));
698 struct tcpcb *
699 	 tcp_usrclosed __P((struct tcpcb *));
700 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
701 int	 tcp_usrreq __P((struct socket *,
702 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
703 void	 tcp_xmit_timer __P((struct tcpcb *, uint32_t));
704 tcp_seq	 tcp_new_iss __P((struct tcpcb *, tcp_seq));
705 tcp_seq  tcp_new_iss1 __P((void *, void *, u_int16_t, u_int16_t, size_t,
706 	    tcp_seq));
707 
708 int	 syn_cache_add __P((struct sockaddr *, struct sockaddr *,
709 		struct tcphdr *, unsigned int, struct socket *,
710 		struct mbuf *, u_char *, int, struct tcp_opt_info *));
711 void	 syn_cache_unreach __P((struct sockaddr *, struct sockaddr *,
712 	   struct tcphdr *));
713 struct socket *syn_cache_get __P((struct sockaddr *, struct sockaddr *,
714 		struct tcphdr *, unsigned int, unsigned int,
715 		struct socket *so, struct mbuf *));
716 void	 syn_cache_init __P((void));
717 void	 syn_cache_insert __P((struct syn_cache *, struct tcpcb *));
718 struct syn_cache *syn_cache_lookup __P((struct sockaddr *, struct sockaddr *,
719 		struct syn_cache_head **));
720 void	 syn_cache_reset __P((struct sockaddr *, struct sockaddr *,
721 		struct tcphdr *));
722 int	 syn_cache_respond __P((struct syn_cache *, struct mbuf *));
723 void	 syn_cache_timer __P((void *));
724 void	 syn_cache_cleanup __P((struct tcpcb *));
725 
726 int	tcp_newreno __P((struct tcpcb *, struct tcphdr *));
727 #endif
728 
729 #endif /* _NETINET_TCP_VAR_H_ */
730