xref: /netbsd-src/sys/netinet/tcp_var.h (revision da5f4674a3fc214be3572d358b66af40ab9401e7)
1 /*	$NetBSD: tcp_var.h,v 1.104 2003/08/07 16:33:20 agc Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*
33  *      @(#)COPYRIGHT   1.1 (NRL) 17 January 1995
34  *
35  * NRL grants permission for redistribution and use in source and binary
36  * forms, with or without modification, of the software and documentation
37  * created at NRL provided that the following conditions are met:
38  *
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgements:
46  *      This product includes software developed by the University of
47  *      California, Berkeley and its contributors.
48  *      This product includes software developed at the Information
49  *      Technology Division, US Naval Research Laboratory.
50  * 4. Neither the name of the NRL nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
55  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
56  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
57  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
58  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
59  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
60  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
61  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
62  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
63  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
64  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
65  *
66  * The views and conclusions contained in the software and documentation
67  * are those of the authors and should not be interpreted as representing
68  * official policies, either expressed or implied, of the US Naval
69  * Research Laboratory (NRL).
70  */
71 
72 /*-
73  * Copyright (c) 1997, 1998, 1999, 2001 The NetBSD Foundation, Inc.
74  * All rights reserved.
75  *
76  * This code is derived from software contributed to The NetBSD Foundation
77  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
78  * NASA Ames Research Center.
79  *
80  * Redistribution and use in source and binary forms, with or without
81  * modification, are permitted provided that the following conditions
82  * are met:
83  * 1. Redistributions of source code must retain the above copyright
84  *    notice, this list of conditions and the following disclaimer.
85  * 2. Redistributions in binary form must reproduce the above copyright
86  *    notice, this list of conditions and the following disclaimer in the
87  *    documentation and/or other materials provided with the distribution.
88  * 3. All advertising materials mentioning features or use of this software
89  *    must display the following acknowledgement:
90  *	This product includes software developed by the NetBSD
91  *	Foundation, Inc. and its contributors.
92  * 4. Neither the name of The NetBSD Foundation nor the names of its
93  *    contributors may be used to endorse or promote products derived
94  *    from this software without specific prior written permission.
95  *
96  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
97  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
98  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
99  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
100  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
101  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
102  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
103  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
104  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
105  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
106  * POSSIBILITY OF SUCH DAMAGE.
107  */
108 
109 /*
110  * Copyright (c) 1982, 1986, 1993, 1994, 1995
111  *	The Regents of the University of California.  All rights reserved.
112  *
113  * Redistribution and use in source and binary forms, with or without
114  * modification, are permitted provided that the following conditions
115  * are met:
116  * 1. Redistributions of source code must retain the above copyright
117  *    notice, this list of conditions and the following disclaimer.
118  * 2. Redistributions in binary form must reproduce the above copyright
119  *    notice, this list of conditions and the following disclaimer in the
120  *    documentation and/or other materials provided with the distribution.
121  * 3. Neither the name of the University nor the names of its contributors
122  *    may be used to endorse or promote products derived from this software
123  *    without specific prior written permission.
124  *
125  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
126  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
127  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
128  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
129  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
130  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
131  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
132  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
133  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
134  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
135  * SUCH DAMAGE.
136  *
137  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
138  */
139 
140 #ifndef _NETINET_TCP_VAR_H_
141 #define _NETINET_TCP_VAR_H_
142 
143 #if defined(_KERNEL_OPT)
144 #include "opt_inet.h"
145 #include "opt_mbuftrace.h"
146 #endif
147 
148 /*
149  * Kernel variables for tcp.
150  */
151 
152 #include <sys/callout.h>
153 
154 /*
155  * Tcp control block, one per tcp; fields:
156  */
157 struct tcpcb {
158 	int	t_family;		/* address family on the wire */
159 	struct ipqehead segq;		/* sequencing queue */
160 	struct callout t_timer[TCPT_NTIMERS];/* tcp timers */
161 	short	t_state;		/* state of this connection */
162 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
163 	uint32_t t_rxtcur;		/* current retransmit value */
164 	short	t_dupacks;		/* consecutive dup acks recd */
165 	u_short	t_peermss;		/* peer's maximum segment size */
166 	u_short	t_ourmss;		/* our's maximum segment size */
167 	u_short t_segsz;		/* current segment size in use */
168 	char	t_force;		/* 1 if forcing out a byte */
169 	u_int	t_flags;
170 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
171 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
172 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
173 #define	TF_NOOPT	0x0008		/* don't use tcp options */
174 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
175 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
176 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
177 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
178 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
179 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
180 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
181 #define	TF_CANT_TXSACK	0x1000		/* other side said I could not SACK */
182 #define	TF_IGNR_RXSACK	0x2000		/* ignore received SACK blocks */
183 #define	TF_REASSEMBLING	0x4000		/* we're busy reassembling */
184 #define	TF_DEAD		0x8000		/* dead and to-be-released */
185 
186 
187 	struct	mbuf *t_template;	/* skeletal packet for transmit */
188 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
189 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
190 	struct	callout t_delack_ch;	/* delayed ACK callout */
191 /*
192  * The following fields are used as in the protocol specification.
193  * See RFC783, Dec. 1981, page 21.
194  */
195 /* send sequence variables */
196 	tcp_seq	snd_una;		/* send unacknowledged */
197 	tcp_seq	snd_nxt;		/* send next */
198 	tcp_seq	snd_up;			/* send urgent pointer */
199 	tcp_seq	snd_wl1;		/* window update seg seq number */
200 	tcp_seq	snd_wl2;		/* window update seg ack number */
201 	tcp_seq	iss;			/* initial send sequence number */
202 	u_long	snd_wnd;		/* send window */
203 	tcp_seq snd_recover;		/* for use in fast recovery */
204 /* receive sequence variables */
205 	u_long	rcv_wnd;		/* receive window */
206 	tcp_seq	rcv_nxt;		/* receive next */
207 	tcp_seq	rcv_up;			/* receive urgent pointer */
208 	tcp_seq	irs;			/* initial receive sequence number */
209 /*
210  * Additional variables for this implementation.
211  */
212 /* receive variables */
213 	tcp_seq	rcv_adv;		/* advertised window */
214 /* retransmit variables */
215 	tcp_seq	snd_max;		/* highest sequence number sent;
216 					 * used to recognize retransmits
217 					 */
218 /* congestion control (for slow start, source quench, retransmit after loss) */
219 	u_long	snd_cwnd;		/* congestion-controlled window */
220 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
221 					 * for slow start exponential to
222 					 * linear switch
223 					 */
224 /*
225  * transmit timing stuff.  See below for scale of srtt and rttvar.
226  * "Variance" is actually smoothed difference.
227  */
228 	uint32_t t_rcvtime;		/* time last segment received */
229 	uint32_t t_rtttime;		/* time we started measuring rtt */
230 	tcp_seq	t_rtseq;		/* sequence number being timed */
231 	int32_t	t_srtt;			/* smoothed round-trip time */
232 	int32_t	t_rttvar;		/* variance in round-trip time */
233 	uint32_t t_rttmin;		/* minimum rtt allowed */
234 	u_long	max_sndwnd;		/* largest window peer has offered */
235 
236 /* out-of-band data */
237 	char	t_oobflags;		/* have some */
238 	char	t_iobc;			/* input character */
239 #define	TCPOOB_HAVEDATA	0x01
240 #define	TCPOOB_HADDATA	0x02
241 	short	t_softerror;		/* possible error not yet reported */
242 
243 /* RFC 1323 variables */
244 	u_char	snd_scale;		/* window scaling for send window */
245 	u_char	rcv_scale;		/* window scaling for recv window */
246 	u_char	request_r_scale;	/* pending window scaling */
247 	u_char	requested_s_scale;
248 	u_int32_t ts_recent;		/* timestamp echo data */
249 	u_int32_t ts_recent_age;	/* when last updated */
250 	u_int32_t ts_timebase;		/* our timebase */
251 	tcp_seq	last_ack_sent;
252 
253 /* SACK stuff */
254 	struct ipqehead timeq;		/* time sequenced queue (for SACK) */
255 
256 /* path MTU discovery blackhole detection */
257 	int t_mtudisc;			/* perform mtudisc for this tcb */
258 
259 /* pointer for syn cache entries*/
260 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
261 
262 /* prediction of next mbuf when using large window sizes */
263 	struct	mbuf *t_lastm;		/* last mbuf that data was sent from */
264 	int	t_inoff;		/* data offset in previous mbuf */
265 	int	t_lastoff;		/* last data address in mbuf chain */
266 	int	t_lastlen;		/* last length read from mbuf chain */
267 };
268 
269 #ifdef _KERNEL
270 /*
271  * TCP reassembly queue locks.
272  */
273 static __inline int tcp_reass_lock_try __P((struct tcpcb *))
274 	__attribute__((__unused__));
275 static __inline void tcp_reass_unlock __P((struct tcpcb *))
276 	__attribute__((__unused__));
277 
278 static __inline int
279 tcp_reass_lock_try(tp)
280 	struct tcpcb *tp;
281 {
282 	int s;
283 
284 	/*
285 	 * Use splvm() -- we're blocking things that would cause
286 	 * mbuf allocation.
287 	 */
288 	s = splvm();
289 	if (tp->t_flags & TF_REASSEMBLING) {
290 		splx(s);
291 		return (0);
292 	}
293 	tp->t_flags |= TF_REASSEMBLING;
294 	splx(s);
295 	return (1);
296 }
297 
298 static __inline void
299 tcp_reass_unlock(tp)
300 	struct tcpcb *tp;
301 {
302 	int s;
303 
304 	s = splvm();
305 	tp->t_flags &= ~TF_REASSEMBLING;
306 	splx(s);
307 }
308 
309 #ifdef DIAGNOSTIC
310 #define	TCP_REASS_LOCK(tp)						\
311 do {									\
312 	if (tcp_reass_lock_try(tp) == 0) {				\
313 		printf("%s:%d: tcpcb %p reass already locked\n",	\
314 		    __FILE__, __LINE__, tp);				\
315 		panic("tcp_reass_lock");				\
316 	}								\
317 } while (/*CONSTCOND*/ 0)
318 #define	TCP_REASS_LOCK_CHECK(tp)					\
319 do {									\
320 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
321 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
322 		    __FILE__, __LINE__, tp);				\
323 		panic("tcp reass lock check");				\
324 	}								\
325 } while (/*CONSTCOND*/ 0)
326 #else
327 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
328 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
329 #endif
330 
331 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
332 #endif /* _KERNEL */
333 
334 /*
335  * Queue for delayed ACK processing.
336  */
337 #ifdef _KERNEL
338 extern int tcp_delack_ticks;
339 void	tcp_delack(void *);
340 
341 #define TCP_RESTART_DELACK(tp)						\
342 	callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks,		\
343 	    tcp_delack, tp)
344 
345 #define	TCP_SET_DELACK(tp)						\
346 do {									\
347 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
348 		(tp)->t_flags |= TF_DELACK;				\
349 		TCP_RESTART_DELACK(tp);					\
350 	}								\
351 } while (/*CONSTCOND*/0)
352 
353 #define	TCP_CLEAR_DELACK(tp)						\
354 do {									\
355 	if ((tp)->t_flags & TF_DELACK) {				\
356 		(tp)->t_flags &= ~TF_DELACK;				\
357 		callout_stop(&(tp)->t_delack_ch);			\
358 	}								\
359 } while (/*CONSTCOND*/0)
360 #endif /* _KERNEL */
361 
362 /*
363  * Compute the current timestamp for a connection.
364  */
365 #define	TCP_TIMESTAMP(tp)	(tcp_now - (tp)->ts_timebase)
366 
367 /*
368  * Handy way of passing around TCP option info.
369  */
370 struct tcp_opt_info {
371 	int		ts_present;
372 	u_int32_t	ts_val;
373 	u_int32_t	ts_ecr;
374 	u_int16_t	maxseg;
375 };
376 
377 /*
378  * Data for the TCP compressed state engine.
379  */
380 union syn_cache_sa {
381 	struct sockaddr sa;
382 	struct sockaddr_in sin;
383 #if 1 /*def INET6*/
384 	struct sockaddr_in6 sin6;
385 #endif
386 };
387 
388 struct syn_cache {
389 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
390 	struct callout sc_timer;		/* rexmt timer */
391 	union {					/* cached route */
392 		struct route route4;
393 #ifdef INET6
394 		struct route_in6 route6;
395 #endif
396 	} sc_route_u;
397 #define sc_route4	sc_route_u.route4
398 #ifdef INET6
399 #define sc_route6	sc_route_u.route6
400 #endif
401 	long sc_win;				/* advertised window */
402 	int sc_bucketidx;			/* our bucket index */
403 	u_int32_t sc_hash;
404 	u_int32_t sc_timestamp;			/* timestamp from SYN */
405 	u_int32_t sc_timebase;			/* our local timebase */
406 	union syn_cache_sa sc_src;
407 	union syn_cache_sa sc_dst;
408 	tcp_seq sc_irs;
409 	tcp_seq sc_iss;
410 	u_int sc_rxtcur;			/* current rxt timeout */
411 	u_int sc_rxttot;			/* total time spend on queues */
412 	u_short sc_rxtshift;			/* for computing backoff */
413 	u_short sc_flags;
414 
415 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
416 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
417 #define	SCF_DEAD		0x0004		/* this entry to be released */
418 
419 	struct mbuf *sc_ipopts;			/* IP options */
420 	u_int16_t sc_peermaxseg;
421 	u_int16_t sc_ourmaxseg;
422 	u_int8_t sc_request_r_scale	: 4,
423 		 sc_requested_s_scale	: 4;
424 
425 	struct tcpcb *sc_tp;			/* tcb for listening socket */
426 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
427 };
428 
429 struct syn_cache_head {
430 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
431 	u_short sch_length;			/* # entries in bucket */
432 };
433 
434 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
435 #ifdef INET6
436 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
437 #endif
438 #ifndef INET6
439 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
440 #else
441 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
442 				? intotcpcb(sotoinpcb(so)) \
443 				: in6totcpcb(sotoin6pcb(so)))
444 #endif
445 
446 /*
447  * The smoothed round-trip time and estimated variance
448  * are stored as fixed point numbers scaled by the values below.
449  * For convenience, these scales are also used in smoothing the average
450  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
451  * With these scales, srtt has 3 bits to the right of the binary point,
452  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
453  * binary point, and is smoothed with an ALPHA of 0.75.
454  */
455 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
456 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
457 
458 /*
459  * The initial retransmission should happen at rtt + 4 * rttvar.
460  * Because of the way we do the smoothing, srtt and rttvar
461  * will each average +1/2 tick of bias.  When we compute
462  * the retransmit timer, we want 1/2 tick of rounding and
463  * 1 extra tick because of +-1/2 tick uncertainty in the
464  * firing of the timer.  The bias will give us exactly the
465  * 1.5 tick we need.  But, because the bias is
466  * statistical, we have to test that we don't drop below
467  * the minimum feasible timer (which is 2 ticks).
468  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
469  * is the same as the multiplier for rttvar.
470  */
471 #define	TCP_REXMTVAL(tp) \
472 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
473 
474 /*
475  * Compute the initial window for slow start.
476  */
477 #define	TCP_INITIAL_WINDOW(iw, segsz) \
478 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
479 	 ((segsz) * (iw)))
480 
481 /*
482  * TCP statistics.
483  * Many of these should be kept per connection,
484  * but that's inconvenient at the moment.
485  */
486 struct	tcpstat {
487 	u_quad_t tcps_connattempt;	/* connections initiated */
488 	u_quad_t tcps_accepts;		/* connections accepted */
489 	u_quad_t tcps_connects;		/* connections established */
490 	u_quad_t tcps_drops;		/* connections dropped */
491 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
492 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
493 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
494 	u_quad_t tcps_rttupdated;	/* times we succeeded */
495 	u_quad_t tcps_delack;		/* delayed acks sent */
496 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
497 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
498 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
499 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
500 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
501 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
502 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
503 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
504 					   shortage */
505 	u_quad_t tcps_pmtublackhole;	/* PMTUD blackhole detected */
506 
507 	u_quad_t tcps_sndtotal;		/* total packets sent */
508 	u_quad_t tcps_sndpack;		/* data packets sent */
509 	u_quad_t tcps_sndbyte;		/* data bytes sent */
510 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
511 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
512 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
513 	u_quad_t tcps_sndprobe;		/* window probes sent */
514 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
515 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
516 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
517 
518 	u_quad_t tcps_rcvtotal;		/* total packets received */
519 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
520 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
521 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
522 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
523 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
524 	u_quad_t tcps_rcvshort;		/* packets received too short */
525 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
526 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
527 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
528 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
529 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
530 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
531 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
532 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
533 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
534 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
535 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
536 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
537 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
538 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
539 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
540 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
541 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
542 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
543 
544 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
545 	u_quad_t tcps_noport;		/* no socket on port */
546 	u_quad_t tcps_badsyn;		/* received ack for which we have
547 					   no SYN in compressed state */
548 	u_quad_t tcps_delayed_free;	/* delayed pool_put() of tcpcb */
549 
550 	/* These statistics deal with the SYN cache. */
551 	u_quad_t tcps_sc_added;		/* # of entries added */
552 	u_quad_t tcps_sc_completed;	/* # of connections completed */
553 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
554 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
555 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
556 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
557 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
558 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
559 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
560 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
561 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
562 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
563 	u_quad_t tcps_sc_delayed_free;	/* # of delayed pool_put()s */
564 
565 	u_quad_t tcps_selfquench;	/* # of ENOBUFS we get on output */
566 };
567 
568 /*
569  * Names for TCP sysctl objects.
570  */
571 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
572 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
573 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
574 #define	TCPCTL_MSSDFLT		4	/* default seg size */
575 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
576 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
577 #if 0	/*obsoleted*/
578 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
579 #endif
580 #define	TCPCTL_INIT_WIN		8	/* initial window */
581 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
582 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
583 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
584 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
585 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
586 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
587 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
588 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
589 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
590 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
591 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
592 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
593 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
594 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
595 #if 0	/*obsoleted*/
596 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
597 #endif
598 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
599 #define	TCPCTL_DELACK_TICKS	25	/* # ticks to delay ACK */
600 #define	TCPCTL_INIT_WIN_LOCAL	26	/* initial window for local nets */
601 #define	TCPCTL_IDENT		27	/* rfc 931 identd */
602 #define	TCPCTL_MAXID		28
603 
604 #define	TCPCTL_NAMES { \
605 	{ 0, 0 }, \
606 	{ "rfc1323",	CTLTYPE_INT }, \
607 	{ "sendspace",	CTLTYPE_INT }, \
608 	{ "recvspace",	CTLTYPE_INT }, \
609 	{ "mssdflt",	CTLTYPE_INT }, \
610 	{ "syn_cache_limit", CTLTYPE_INT }, \
611 	{ "syn_bucket_limit", CTLTYPE_INT }, \
612 	{ 0, 0 },\
613 	{ "init_win", CTLTYPE_INT }, \
614 	{ "mss_ifmtu", CTLTYPE_INT }, \
615 	{ "sack", CTLTYPE_INT }, \
616 	{ "win_scale", CTLTYPE_INT }, \
617 	{ "timestamps", CTLTYPE_INT }, \
618 	{ "compat_42", CTLTYPE_INT }, \
619 	{ "cwm", CTLTYPE_INT }, \
620 	{ "cwm_burstsize", CTLTYPE_INT }, \
621 	{ "ack_on_push", CTLTYPE_INT }, \
622 	{ "keepidle",	CTLTYPE_INT }, \
623 	{ "keepintvl",	CTLTYPE_INT }, \
624 	{ "keepcnt",	CTLTYPE_INT }, \
625 	{ "slowhz",	CTLTYPE_INT }, \
626 	{ "newreno",	CTLTYPE_INT }, \
627 	{ "log_refused",CTLTYPE_INT }, \
628 	{ 0, 0 }, \
629 	{ "rstppslimit", CTLTYPE_INT }, \
630 	{ "delack_ticks", CTLTYPE_INT }, \
631 	{ "init_win_local", CTLTYPE_INT }, \
632 	{ "ident", CTLTYPE_STRUCT }, \
633 }
634 
635 #ifdef _KERNEL
636 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
637 #ifdef INET6
638 extern	struct in6pcb tcb6;
639 #endif
640 extern	struct tcpstat tcpstat;	/* tcp statistics */
641 extern	u_int32_t tcp_now;	/* for RFC 1323 timestamps */
642 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
643 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
644 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
645 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
646 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
647 extern	int tcp_mssdflt;	/* default seg size */
648 extern	int tcp_init_win;	/* initial window */
649 extern	int tcp_init_win_local;	/* initial window for local nets */
650 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
651 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
652 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
653 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
654 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
655 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
656 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
657 extern	int tcp_log_refused;	/* log refused connections */
658 
659 extern	int tcp_rst_ppslim;
660 
661 extern	int tcp_syn_cache_size;
662 extern	struct syn_cache_head tcp_syn_cache[];
663 extern	u_long syn_cache_count;
664 
665 #ifdef MBUFTRACE
666 extern	struct mowner tcp_rx_mowner;
667 extern	struct mowner tcp_tx_mowner;
668 extern	struct mowner tcp_mowner;
669 #endif
670 
671 #define	TCPCTL_VARIABLES { \
672 	{ 0 },					\
673 	{ 1, 0, &tcp_do_rfc1323 },		\
674 	{ 1, 0, &tcp_sendspace },		\
675 	{ 1, 0, &tcp_recvspace },		\
676 	{ 1, 0, &tcp_mssdflt },			\
677 	{ 1, 0, &tcp_syn_cache_limit },		\
678 	{ 1, 0, &tcp_syn_bucket_limit },	\
679 	{ 0 },					\
680 	{ 1, 0, &tcp_init_win },		\
681 	{ 1, 0, &tcp_mss_ifmtu },		\
682 	{ 1, 0, &tcp_do_sack },			\
683 	{ 1, 0, &tcp_do_win_scale },		\
684 	{ 1, 0, &tcp_do_timestamps },		\
685 	{ 1, 0, &tcp_compat_42 },		\
686 	{ 1, 0, &tcp_cwm },			\
687 	{ 1, 0, &tcp_cwm_burstsize },		\
688 	{ 1, 0, &tcp_ack_on_push },		\
689 	{ 1, 0, &tcp_keepidle },		\
690 	{ 1, 0, &tcp_keepintvl },		\
691 	{ 1, 0, &tcp_keepcnt },			\
692 	{ 1, 1, 0, PR_SLOWHZ },			\
693 	{ 1, 0, &tcp_do_newreno },		\
694 	{ 1, 0, &tcp_log_refused },		\
695 	{ 0 },					\
696 	{ 1, 0, &tcp_rst_ppslim },		\
697 	{ 1, 0, &tcp_delack_ticks },		\
698 	{ 1, 0, &tcp_init_win_local },		\
699 }
700 
701 #ifdef __NO_STRICT_ALIGNMENT
702 #define	TCP_HDR_ALIGNED_P(th)	1
703 #else
704 #define	TCP_HDR_ALIGNED_P(th)	((((vaddr_t) (th)) & 3) == 0)
705 #endif
706 
707 int	 tcp_attach __P((struct socket *));
708 void	 tcp_canceltimers __P((struct tcpcb *));
709 int	 tcp_timers_invoking __P((struct tcpcb*));
710 struct tcpcb *
711 	 tcp_close __P((struct tcpcb *));
712 int	 tcp_isdead __P((struct tcpcb *));
713 #ifdef INET6
714 void	 tcp6_ctlinput __P((int, struct sockaddr *, void *));
715 #endif
716 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
717 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
718 struct tcpcb *
719 	 tcp_disconnect __P((struct tcpcb *));
720 struct tcpcb *
721 	 tcp_drop __P((struct tcpcb *, int));
722 void	 tcp_dooptions __P((struct tcpcb *,
723 	    u_char *, int, struct tcphdr *, struct tcp_opt_info *));
724 void	 tcp_drain __P((void));
725 #ifdef INET6
726 void	 tcp6_drain __P((void));
727 #endif
728 void	 tcp_established __P((struct tcpcb *));
729 void	 tcp_init __P((void));
730 #ifdef INET6
731 int	 tcp6_input __P((struct mbuf **, int *, int));
732 #endif
733 void	 tcp_input __P((struct mbuf *, ...));
734 u_long	 tcp_mss_to_advertise __P((const struct ifnet *, int));
735 void	 tcp_mss_from_peer __P((struct tcpcb *, int));
736 struct tcpcb *
737 	 tcp_newtcpcb __P((int, void *));
738 void	 tcp_notify __P((struct inpcb *, int));
739 #ifdef INET6
740 void	 tcp6_notify __P((struct in6pcb *, int));
741 #endif
742 u_int	 tcp_optlen __P((struct tcpcb *));
743 int	 tcp_output __P((struct tcpcb *));
744 void	 tcp_pulloutofband __P((struct socket *,
745 	    struct tcphdr *, struct mbuf *, int));
746 void	 tcp_quench __P((struct inpcb *, int));
747 #ifdef INET6
748 void	 tcp6_quench __P((struct in6pcb *, int));
749 #endif
750 int	 tcp_reass __P((struct tcpcb *, struct tcphdr *, struct mbuf *, int *));
751 int	 tcp_respond __P((struct tcpcb *, struct mbuf *, struct mbuf *,
752 	    struct tcphdr *, tcp_seq, tcp_seq, int));
753 void	 tcp_rmx_rtt __P((struct tcpcb *));
754 void	 tcp_setpersist __P((struct tcpcb *));
755 void	 tcp_slowtimo __P((void));
756 struct mbuf *
757 	 tcp_template __P((struct tcpcb *));
758 void	 tcp_trace __P((int, int, struct tcpcb *, struct mbuf *, int));
759 struct tcpcb *
760 	 tcp_usrclosed __P((struct tcpcb *));
761 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
762 int	 tcp_usrreq __P((struct socket *,
763 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
764 void	 tcp_xmit_timer __P((struct tcpcb *, uint32_t));
765 tcp_seq	 tcp_new_iss __P((struct tcpcb *, tcp_seq));
766 tcp_seq  tcp_new_iss1 __P((void *, void *, u_int16_t, u_int16_t, size_t,
767 	    tcp_seq));
768 
769 int	 syn_cache_add __P((struct sockaddr *, struct sockaddr *,
770 		struct tcphdr *, unsigned int, struct socket *,
771 		struct mbuf *, u_char *, int, struct tcp_opt_info *));
772 void	 syn_cache_unreach __P((struct sockaddr *, struct sockaddr *,
773 	   struct tcphdr *));
774 struct socket *syn_cache_get __P((struct sockaddr *, struct sockaddr *,
775 		struct tcphdr *, unsigned int, unsigned int,
776 		struct socket *so, struct mbuf *));
777 void	 syn_cache_init __P((void));
778 void	 syn_cache_insert __P((struct syn_cache *, struct tcpcb *));
779 struct syn_cache *syn_cache_lookup __P((struct sockaddr *, struct sockaddr *,
780 		struct syn_cache_head **));
781 void	 syn_cache_reset __P((struct sockaddr *, struct sockaddr *,
782 		struct tcphdr *));
783 int	 syn_cache_respond __P((struct syn_cache *, struct mbuf *));
784 void	 syn_cache_timer __P((void *));
785 void	 syn_cache_cleanup __P((struct tcpcb *));
786 
787 int	tcp_newreno __P((struct tcpcb *, struct tcphdr *));
788 #endif
789 
790 #endif /* _NETINET_TCP_VAR_H_ */
791