xref: /netbsd-src/sys/netinet/tcp_var.h (revision bf1e9b32e27832f0c493206710fb8b58a980838a)
1 /*	$NetBSD: tcp_var.h,v 1.126 2005/05/29 21:41:23 christos 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, 2005 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  * This code is derived from software contributed to The NetBSD Foundation
80  * by Charles M. Hannum.
81  *
82  * Redistribution and use in source and binary forms, with or without
83  * modification, are permitted provided that the following conditions
84  * are met:
85  * 1. Redistributions of source code must retain the above copyright
86  *    notice, this list of conditions and the following disclaimer.
87  * 2. Redistributions in binary form must reproduce the above copyright
88  *    notice, this list of conditions and the following disclaimer in the
89  *    documentation and/or other materials provided with the distribution.
90  * 3. All advertising materials mentioning features or use of this software
91  *    must display the following acknowledgement:
92  *	This product includes software developed by the NetBSD
93  *	Foundation, Inc. and its contributors.
94  * 4. Neither the name of The NetBSD Foundation nor the names of its
95  *    contributors may be used to endorse or promote products derived
96  *    from this software without specific prior written permission.
97  *
98  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
99  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
100  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
101  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
102  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
103  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
104  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
105  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
106  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
107  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
108  * POSSIBILITY OF SUCH DAMAGE.
109  */
110 
111 /*
112  * Copyright (c) 1982, 1986, 1993, 1994, 1995
113  *	The Regents of the University of California.  All rights reserved.
114  *
115  * Redistribution and use in source and binary forms, with or without
116  * modification, are permitted provided that the following conditions
117  * are met:
118  * 1. Redistributions of source code must retain the above copyright
119  *    notice, this list of conditions and the following disclaimer.
120  * 2. Redistributions in binary form must reproduce the above copyright
121  *    notice, this list of conditions and the following disclaimer in the
122  *    documentation and/or other materials provided with the distribution.
123  * 3. Neither the name of the University nor the names of its contributors
124  *    may be used to endorse or promote products derived from this software
125  *    without specific prior written permission.
126  *
127  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
128  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
129  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
130  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
131  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
132  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
133  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
134  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
135  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
136  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
137  * SUCH DAMAGE.
138  *
139  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
140  */
141 
142 #ifndef _NETINET_TCP_VAR_H_
143 #define _NETINET_TCP_VAR_H_
144 
145 #if defined(_KERNEL_OPT)
146 #include "opt_inet.h"
147 #include "opt_mbuftrace.h"
148 #endif
149 
150 /*
151  * Kernel variables for tcp.
152  */
153 
154 #include <sys/callout.h>
155 
156 #ifdef TCP_SIGNATURE
157 /*
158  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
159  * for SADB verification and lookup.
160  */
161 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
162 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
163 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
164 /*
165  * Only a single SA per host may be specified at this time. An SPI is
166  * needed in order for the KEY_ALLOCSA() lookup to work.
167  */
168 #define	TCP_SIG_SPI	0x1000
169 #endif /* TCP_SIGNATURE */
170 
171 /*
172  * SACK option block.
173  */
174 struct sackblk {
175 	tcp_seq left;		/* Left edge of sack block. */
176 	tcp_seq right;		/* Right edge of sack block. */
177 };
178 
179 TAILQ_HEAD(sackhead, sackhole);
180 struct sackhole {
181 	tcp_seq start;
182 	tcp_seq end;
183 	tcp_seq rxmit;
184 
185 	TAILQ_ENTRY(sackhole) sackhole_q;
186 };
187 
188 /*
189  * Tcp control block, one per tcp; fields:
190  */
191 struct tcpcb {
192 	int	t_family;		/* address family on the wire */
193 	struct ipqehead segq;		/* sequencing queue */
194 	int	t_segqlen;		/* length of the above */
195 	struct callout t_timer[TCPT_NTIMERS];/* tcp timers */
196 	short	t_state;		/* state of this connection */
197 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
198 	uint32_t t_rxtcur;		/* current retransmit value */
199 	short	t_dupacks;		/* consecutive dup acks recd */
200 	short	t_partialacks;		/* partials acks during fast rexmit */
201 	u_short	t_peermss;		/* peer's maximum segment size */
202 	u_short	t_ourmss;		/* our's maximum segment size */
203 	u_short t_segsz;		/* current segment size in use */
204 	char	t_force;		/* 1 if forcing out a byte */
205 	u_int	t_flags;
206 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
207 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
208 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
209 #define	TF_NOOPT	0x0008		/* don't use tcp options */
210 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
211 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
212 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
213 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
214 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
215 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
216 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
217 #define	TF_REASSEMBLING	0x1000		/* we're busy reassembling */
218 #define	TF_DEAD		0x2000		/* dead and to-be-released */
219 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
220 
221 
222 	struct	mbuf *t_template;	/* skeletal packet for transmit */
223 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
224 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
225 	struct	callout t_delack_ch;	/* delayed ACK callout */
226 /*
227  * The following fields are used as in the protocol specification.
228  * See RFC783, Dec. 1981, page 21.
229  */
230 /* send sequence variables */
231 	tcp_seq	snd_una;		/* send unacknowledged */
232 	tcp_seq	snd_nxt;		/* send next */
233 	tcp_seq	snd_up;			/* send urgent pointer */
234 	tcp_seq	snd_wl1;		/* window update seg seq number */
235 	tcp_seq	snd_wl2;		/* window update seg ack number */
236 	tcp_seq	iss;			/* initial send sequence number */
237 	u_long	snd_wnd;		/* send window */
238 	tcp_seq snd_recover;		/* for use in fast recovery */
239 	tcp_seq	snd_high;		/* NewReno false fast rexmit seq */
240 /* receive sequence variables */
241 	u_long	rcv_wnd;		/* receive window */
242 	tcp_seq	rcv_nxt;		/* receive next */
243 	tcp_seq	rcv_up;			/* receive urgent pointer */
244 	tcp_seq	irs;			/* initial receive sequence number */
245 /*
246  * Additional variables for this implementation.
247  */
248 /* receive variables */
249 	tcp_seq	rcv_adv;		/* advertised window */
250 /* retransmit variables */
251 	tcp_seq	snd_max;		/* highest sequence number sent;
252 					 * used to recognize retransmits
253 					 */
254 /* congestion control (for slow start, source quench, retransmit after loss) */
255 	u_long	snd_cwnd;		/* congestion-controlled window */
256 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
257 					 * for slow start exponential to
258 					 * linear switch
259 					 */
260 /*
261  * transmit timing stuff.  See below for scale of srtt and rttvar.
262  * "Variance" is actually smoothed difference.
263  */
264 	uint32_t t_rcvtime;		/* time last segment received */
265 	uint32_t t_rtttime;		/* time we started measuring rtt */
266 	tcp_seq	t_rtseq;		/* sequence number being timed */
267 	int32_t	t_srtt;			/* smoothed round-trip time */
268 	int32_t	t_rttvar;		/* variance in round-trip time */
269 	uint32_t t_rttmin;		/* minimum rtt allowed */
270 	u_long	max_sndwnd;		/* largest window peer has offered */
271 
272 /* out-of-band data */
273 	char	t_oobflags;		/* have some */
274 	char	t_iobc;			/* input character */
275 #define	TCPOOB_HAVEDATA	0x01
276 #define	TCPOOB_HADDATA	0x02
277 	short	t_softerror;		/* possible error not yet reported */
278 
279 /* RFC 1323 variables */
280 	u_char	snd_scale;		/* window scaling for send window */
281 	u_char	rcv_scale;		/* window scaling for recv window */
282 	u_char	request_r_scale;	/* pending window scaling */
283 	u_char	requested_s_scale;
284 	u_int32_t ts_recent;		/* timestamp echo data */
285 	u_int32_t ts_recent_age;	/* when last updated */
286 	u_int32_t ts_timebase;		/* our timebase */
287 	tcp_seq	last_ack_sent;
288 
289 /* SACK stuff */
290 #define TCP_SACK_MAX 3
291 #define TCPSACK_NONE 0
292 #define TCPSACK_HAVED 1
293 	u_char rcv_sack_flags;		/* SACK flags. */
294 	struct sackblk rcv_dsack_block;	/* RX D-SACK block. */
295 	struct ipqehead timeq;		/* time sequenced queue. */
296 	struct sackhead snd_holes;	/* TX SACK holes. */
297 	int	snd_numholes;		/* Number of TX SACK holes. */
298 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
299 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
300 					   episode starts at this seq number*/
301 	tcp_seq snd_fack;		/* FACK TCP.  Forward-most data held by
302 					   peer. */
303 
304 /* path MTU discovery blackhole detection */
305 	int t_mtudisc;			/* perform mtudisc for this tcb */
306 
307 /* pointer for syn cache entries*/
308 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
309 
310 /* prediction of next mbuf when using large window sizes */
311 	struct	mbuf *t_lastm;		/* last mbuf that data was sent from */
312 	int	t_inoff;		/* data offset in previous mbuf */
313 	int	t_lastoff;		/* last data address in mbuf chain */
314 	int	t_lastlen;		/* last length read from mbuf chain */
315 };
316 
317 /*
318  * Macros to aid SACK/FACK TCP.
319  */
320 #define TCP_SACK_ENABLED(tp)	(tp->t_flags & TF_WILL_SACK)
321 #define TCP_FACK_FASTRECOV(tp)	\
322 	(TCP_SACK_ENABLED(tp) && \
323 	(SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz)))
324 
325 #ifdef _KERNEL
326 /*
327  * TCP reassembly queue locks.
328  */
329 static __inline int tcp_reass_lock_try (struct tcpcb *)
330 	__attribute__((__unused__));
331 static __inline void tcp_reass_unlock (struct tcpcb *)
332 	__attribute__((__unused__));
333 
334 static __inline int
335 tcp_reass_lock_try(tp)
336 	struct tcpcb *tp;
337 {
338 	int s;
339 
340 	/*
341 	 * Use splvm() -- we're blocking things that would cause
342 	 * mbuf allocation.
343 	 */
344 	s = splvm();
345 	if (tp->t_flags & TF_REASSEMBLING) {
346 		splx(s);
347 		return (0);
348 	}
349 	tp->t_flags |= TF_REASSEMBLING;
350 	splx(s);
351 	return (1);
352 }
353 
354 static __inline void
355 tcp_reass_unlock(tp)
356 	struct tcpcb *tp;
357 {
358 	int s;
359 
360 	s = splvm();
361 	tp->t_flags &= ~TF_REASSEMBLING;
362 	splx(s);
363 }
364 
365 #ifdef DIAGNOSTIC
366 #define	TCP_REASS_LOCK(tp)						\
367 do {									\
368 	if (tcp_reass_lock_try(tp) == 0) {				\
369 		printf("%s:%d: tcpcb %p reass already locked\n",	\
370 		    __FILE__, __LINE__, tp);				\
371 		panic("tcp_reass_lock");				\
372 	}								\
373 } while (/*CONSTCOND*/ 0)
374 #define	TCP_REASS_LOCK_CHECK(tp)					\
375 do {									\
376 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
377 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
378 		    __FILE__, __LINE__, tp);				\
379 		panic("tcp reass lock check");				\
380 	}								\
381 } while (/*CONSTCOND*/ 0)
382 #else
383 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
384 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
385 #endif
386 
387 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
388 #endif /* _KERNEL */
389 
390 /*
391  * Queue for delayed ACK processing.
392  */
393 #ifdef _KERNEL
394 extern int tcp_delack_ticks;
395 void	tcp_delack(void *);
396 
397 #define TCP_RESTART_DELACK(tp)						\
398 	callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks,		\
399 	    tcp_delack, tp)
400 
401 #define	TCP_SET_DELACK(tp)						\
402 do {									\
403 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
404 		(tp)->t_flags |= TF_DELACK;				\
405 		TCP_RESTART_DELACK(tp);					\
406 	}								\
407 } while (/*CONSTCOND*/0)
408 
409 #define	TCP_CLEAR_DELACK(tp)						\
410 do {									\
411 	if ((tp)->t_flags & TF_DELACK) {				\
412 		(tp)->t_flags &= ~TF_DELACK;				\
413 		callout_stop(&(tp)->t_delack_ch);			\
414 	}								\
415 } while (/*CONSTCOND*/0)
416 #endif /* _KERNEL */
417 
418 /*
419  * Compute the current timestamp for a connection.
420  */
421 #define	TCP_TIMESTAMP(tp)	(tcp_now - (tp)->ts_timebase)
422 
423 /*
424  * Handy way of passing around TCP option info.
425  */
426 struct tcp_opt_info {
427 	int		ts_present;
428 	u_int32_t	ts_val;
429 	u_int32_t	ts_ecr;
430 	u_int16_t	maxseg;
431 };
432 
433 #define	TOF_SIGNATURE	0x0040		/* signature option present */
434 #define	TOF_SIGLEN	0x0080		/* sigature length valid (RFC2385) */
435 
436 /*
437  * Data for the TCP compressed state engine.
438  */
439 union syn_cache_sa {
440 	struct sockaddr sa;
441 	struct sockaddr_in sin;
442 #if 1 /*def INET6*/
443 	struct sockaddr_in6 sin6;
444 #endif
445 };
446 
447 struct syn_cache {
448 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
449 	struct callout sc_timer;		/* rexmt timer */
450 	union {					/* cached route */
451 		struct route route4;
452 #ifdef INET6
453 		struct route_in6 route6;
454 #endif
455 	} sc_route_u;
456 #define sc_route4	sc_route_u.route4
457 #ifdef INET6
458 #define sc_route6	sc_route_u.route6
459 #endif
460 	long sc_win;				/* advertised window */
461 	int sc_bucketidx;			/* our bucket index */
462 	u_int32_t sc_hash;
463 	u_int32_t sc_timestamp;			/* timestamp from SYN */
464 	u_int32_t sc_timebase;			/* our local timebase */
465 	union syn_cache_sa sc_src;
466 	union syn_cache_sa sc_dst;
467 	tcp_seq sc_irs;
468 	tcp_seq sc_iss;
469 	u_int sc_rxtcur;			/* current rxt timeout */
470 	u_int sc_rxttot;			/* total time spend on queues */
471 	u_short sc_rxtshift;			/* for computing backoff */
472 	u_short sc_flags;
473 
474 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
475 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
476 #define	SCF_DEAD		0x0004		/* this entry to be released */
477 #define SCF_SACK_PERMIT		0x0008		/* peer will do SACK */
478 #define SCF_SIGNATURE	0x40			/* send MD5 digests */
479 
480 	struct mbuf *sc_ipopts;			/* IP options */
481 	u_int16_t sc_peermaxseg;
482 	u_int16_t sc_ourmaxseg;
483 	u_int8_t sc_request_r_scale	: 4,
484 		 sc_requested_s_scale	: 4;
485 
486 	struct tcpcb *sc_tp;			/* tcb for listening socket */
487 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
488 };
489 
490 struct syn_cache_head {
491 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
492 	u_short sch_length;			/* # entries in bucket */
493 };
494 
495 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
496 #ifdef INET6
497 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
498 #endif
499 #ifndef INET6
500 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
501 #else
502 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
503 				? intotcpcb(sotoinpcb(so)) \
504 				: in6totcpcb(sotoin6pcb(so)))
505 #endif
506 
507 /*
508  * The smoothed round-trip time and estimated variance
509  * are stored as fixed point numbers scaled by the values below.
510  * For convenience, these scales are also used in smoothing the average
511  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
512  * With these scales, srtt has 3 bits to the right of the binary point,
513  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
514  * binary point, and is smoothed with an ALPHA of 0.75.
515  */
516 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
517 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
518 
519 /*
520  * The initial retransmission should happen at rtt + 4 * rttvar.
521  * Because of the way we do the smoothing, srtt and rttvar
522  * will each average +1/2 tick of bias.  When we compute
523  * the retransmit timer, we want 1/2 tick of rounding and
524  * 1 extra tick because of +-1/2 tick uncertainty in the
525  * firing of the timer.  The bias will give us exactly the
526  * 1.5 tick we need.  But, because the bias is
527  * statistical, we have to test that we don't drop below
528  * the minimum feasible timer (which is 2 ticks).
529  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
530  * is the same as the multiplier for rttvar.
531  */
532 #define	TCP_REXMTVAL(tp) \
533 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
534 
535 /*
536  * Compute the initial window for slow start.
537  */
538 #define	TCP_INITIAL_WINDOW(iw, segsz) \
539 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
540 	 ((segsz) * (iw)))
541 
542 /*
543  * TCP statistics.
544  * Many of these should be kept per connection,
545  * but that's inconvenient at the moment.
546  */
547 struct	tcpstat {
548 	u_quad_t tcps_connattempt;	/* connections initiated */
549 	u_quad_t tcps_accepts;		/* connections accepted */
550 	u_quad_t tcps_connects;		/* connections established */
551 	u_quad_t tcps_drops;		/* connections dropped */
552 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
553 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
554 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
555 	u_quad_t tcps_rttupdated;	/* times we succeeded */
556 	u_quad_t tcps_delack;		/* delayed acks sent */
557 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
558 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
559 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
560 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
561 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
562 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
563 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
564 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
565 					   shortage */
566 	u_quad_t tcps_pmtublackhole;	/* PMTUD blackhole detected */
567 
568 	u_quad_t tcps_sndtotal;		/* total packets sent */
569 	u_quad_t tcps_sndpack;		/* data packets sent */
570 	u_quad_t tcps_sndbyte;		/* data bytes sent */
571 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
572 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
573 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
574 	u_quad_t tcps_sndprobe;		/* window probes sent */
575 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
576 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
577 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
578 
579 	u_quad_t tcps_rcvtotal;		/* total packets received */
580 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
581 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
582 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
583 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
584 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
585 	u_quad_t tcps_rcvshort;		/* packets received too short */
586 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
587 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
588 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
589 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
590 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
591 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
592 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
593 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
594 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
595 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
596 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
597 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
598 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
599 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
600 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
601 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
602 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
603 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
604 
605 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
606 	u_quad_t tcps_noport;		/* no socket on port */
607 	u_quad_t tcps_badsyn;		/* received ack for which we have
608 					   no SYN in compressed state */
609 	u_quad_t tcps_delayed_free;	/* delayed pool_put() of tcpcb */
610 
611 	/* These statistics deal with the SYN cache. */
612 	u_quad_t tcps_sc_added;		/* # of entries added */
613 	u_quad_t tcps_sc_completed;	/* # of connections completed */
614 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
615 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
616 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
617 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
618 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
619 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
620 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
621 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
622 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
623 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
624 	u_quad_t tcps_sc_delayed_free;	/* # of delayed pool_put()s */
625 
626 	u_quad_t tcps_selfquench;	/* # of ENOBUFS we get on output */
627 	u_quad_t tcps_badsig;		/* # of drops due to bad signature */
628 	u_quad_t tcps_goodsig;		/* # of packets with good signature */
629 };
630 
631 /*
632  * Names for TCP sysctl objects.
633  */
634 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
635 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
636 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
637 #define	TCPCTL_MSSDFLT		4	/* default seg size */
638 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
639 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
640 #if 0	/*obsoleted*/
641 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
642 #endif
643 #define	TCPCTL_INIT_WIN		8	/* initial window */
644 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
645 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
646 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
647 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
648 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
649 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
650 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
651 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
652 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
653 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
654 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
655 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
656 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
657 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
658 #if 0	/*obsoleted*/
659 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
660 #endif
661 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
662 #define	TCPCTL_DELACK_TICKS	25	/* # ticks to delay ACK */
663 #define	TCPCTL_INIT_WIN_LOCAL	26	/* initial window for local nets */
664 #define	TCPCTL_IDENT		27	/* rfc 931 identd */
665 #define	TCPCTL_ACKDROPRATELIMIT	28	/* SYN/RST -> ACK rate limit */
666 #define	TCPCTL_LOOPBACKCKSUM	29	/* do TCP checksum on loopback */
667 #define	TCPCTL_MAXID		30
668 
669 #define	TCPCTL_NAMES { \
670 	{ 0, 0 }, \
671 	{ "rfc1323",	CTLTYPE_INT }, \
672 	{ "sendspace",	CTLTYPE_INT }, \
673 	{ "recvspace",	CTLTYPE_INT }, \
674 	{ "mssdflt",	CTLTYPE_INT }, \
675 	{ "syn_cache_limit", CTLTYPE_INT }, \
676 	{ "syn_bucket_limit", CTLTYPE_INT }, \
677 	{ 0, 0 },\
678 	{ "init_win", CTLTYPE_INT }, \
679 	{ "mss_ifmtu", CTLTYPE_INT }, \
680 	{ "sack", CTLTYPE_INT }, \
681 	{ "win_scale", CTLTYPE_INT }, \
682 	{ "timestamps", CTLTYPE_INT }, \
683 	{ "compat_42", CTLTYPE_INT }, \
684 	{ "cwm", CTLTYPE_INT }, \
685 	{ "cwm_burstsize", CTLTYPE_INT }, \
686 	{ "ack_on_push", CTLTYPE_INT }, \
687 	{ "keepidle",	CTLTYPE_INT }, \
688 	{ "keepintvl",	CTLTYPE_INT }, \
689 	{ "keepcnt",	CTLTYPE_INT }, \
690 	{ "slowhz",	CTLTYPE_INT }, \
691 	{ "newreno",	CTLTYPE_INT }, \
692 	{ "log_refused",CTLTYPE_INT }, \
693 	{ 0, 0 }, \
694 	{ "rstppslimit", CTLTYPE_INT }, \
695 	{ "delack_ticks", CTLTYPE_INT }, \
696 	{ "init_win_local", CTLTYPE_INT }, \
697 	{ "ident", CTLTYPE_STRUCT }, \
698 	{ "ackdropppslimit", CTLTYPE_INT }, \
699 	{ "do_loopback_cksum", CTLTYPE_INT }, \
700 }
701 
702 #ifdef _KERNEL
703 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
704 extern	struct tcpstat tcpstat;	/* tcp statistics */
705 extern	u_int32_t tcp_now;	/* for RFC 1323 timestamps */
706 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
707 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
708 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
709 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
710 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
711 extern	int tcp_mssdflt;	/* default seg size */
712 extern	int tcp_init_win;	/* initial window */
713 extern	int tcp_init_win_local;	/* initial window for local nets */
714 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
715 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
716 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
717 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
718 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
719 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
720 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
721 extern	int tcp_log_refused;	/* log refused connections */
722 extern	int tcp_do_loopback_cksum;/* do TCP checksum on loopback? */
723 extern int tcp_sack_tp_maxholes;	/* Max holes per connection. */
724 extern int tcp_sack_globalmaxholes;	/* Max holes per system. */
725 extern int tcp_sack_globalholes;	/* Number of holes present. */
726 
727 extern	int tcp_rst_ppslim;
728 extern	int tcp_ackdrop_ppslim;
729 
730 extern	int tcp_syn_cache_size;
731 extern	struct syn_cache_head tcp_syn_cache[];
732 extern	u_long syn_cache_count;
733 
734 #ifdef MBUFTRACE
735 extern	struct mowner tcp_rx_mowner;
736 extern	struct mowner tcp_tx_mowner;
737 extern	struct mowner tcp_mowner;
738 #endif
739 
740 #define	TCPCTL_VARIABLES { \
741 	{ 0 },					\
742 	{ 1, 0, &tcp_do_rfc1323 },		\
743 	{ 1, 0, &tcp_sendspace },		\
744 	{ 1, 0, &tcp_recvspace },		\
745 	{ 1, 0, &tcp_mssdflt },			\
746 	{ 1, 0, &tcp_syn_cache_limit },		\
747 	{ 1, 0, &tcp_syn_bucket_limit },	\
748 	{ 0 },					\
749 	{ 1, 0, &tcp_init_win },		\
750 	{ 1, 0, &tcp_mss_ifmtu },		\
751 	{ 1, 0, &tcp_do_sack },			\
752 	{ 1, 0, &tcp_do_win_scale },		\
753 	{ 1, 0, &tcp_do_timestamps },		\
754 	{ 1, 0, &tcp_compat_42 },		\
755 	{ 1, 0, &tcp_cwm },			\
756 	{ 1, 0, &tcp_cwm_burstsize },		\
757 	{ 1, 0, &tcp_ack_on_push },		\
758 	{ 1, 0, &tcp_keepidle },		\
759 	{ 1, 0, &tcp_keepintvl },		\
760 	{ 1, 0, &tcp_keepcnt },			\
761 	{ 1, 1, 0, PR_SLOWHZ },			\
762 	{ 1, 0, &tcp_do_newreno },		\
763 	{ 1, 0, &tcp_log_refused },		\
764 	{ 0 },					\
765 	{ 1, 0, &tcp_rst_ppslim },		\
766 	{ 1, 0, &tcp_delack_ticks },		\
767 	{ 1, 0, &tcp_init_win_local },		\
768 	{ 1, 0, &tcp_ackdrop_ppslim },		\
769 }
770 
771 #ifdef __NO_STRICT_ALIGNMENT
772 #define	TCP_HDR_ALIGNED_P(th)	1
773 #else
774 #define	TCP_HDR_ALIGNED_P(th)	((((vaddr_t)(th)) & 3) == 0)
775 #endif
776 
777 struct secasvar;
778 
779 int	 tcp_attach(struct socket *);
780 void	 tcp_canceltimers(struct tcpcb *);
781 int	 tcp_timers_invoking(struct tcpcb*);
782 struct tcpcb *
783 	 tcp_close(struct tcpcb *);
784 int	 tcp_isdead(struct tcpcb *);
785 #ifdef INET6
786 void	 tcp6_ctlinput(int, struct sockaddr *, void *);
787 #endif
788 void	 *tcp_ctlinput(int, struct sockaddr *, void *);
789 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf **);
790 struct tcpcb *
791 	 tcp_disconnect(struct tcpcb *);
792 struct tcpcb *
793 	 tcp_drop(struct tcpcb *, int);
794 #ifdef TCP_SIGNATURE
795 int	 tcp_signature_apply(void *, caddr_t, u_int);
796 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *);
797 int	 tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *,
798 	    char *);
799 #endif
800 int	 tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
801 	    struct mbuf *, int, struct tcp_opt_info *);
802 void	 tcp_drain(void);
803 void	 tcp_established(struct tcpcb *);
804 void	 tcp_init(void);
805 #ifdef INET6
806 int	 tcp6_input(struct mbuf **, int *, int);
807 #endif
808 void	 tcp_input(struct mbuf *, ...);
809 u_long	 tcp_mss_to_advertise(const struct ifnet *, int);
810 void	 tcp_mss_from_peer(struct tcpcb *, int);
811 void	 tcp_tcpcb_template(void);
812 struct tcpcb *
813 	 tcp_newtcpcb(int, void *);
814 void	 tcp_notify(struct inpcb *, int);
815 #ifdef INET6
816 void	 tcp6_notify(struct in6pcb *, int);
817 #endif
818 u_int	 tcp_optlen(struct tcpcb *);
819 int	 tcp_output(struct tcpcb *);
820 void	 tcp_pulloutofband(struct socket *,
821 	    struct tcphdr *, struct mbuf *, int);
822 void	 tcp_quench(struct inpcb *, int);
823 #ifdef INET6
824 void	 tcp6_quench(struct in6pcb *, int);
825 #endif
826 
827 struct ipqent *tcpipqent_alloc(void);
828 void	 tcpipqent_free(struct ipqent *);
829 
830 int	 tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
831 int	 tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *,
832 	    struct tcphdr *, tcp_seq, tcp_seq, int);
833 void	 tcp_rmx_rtt(struct tcpcb *);
834 void	 tcp_setpersist(struct tcpcb *);
835 #ifdef TCP_SIGNATURE
836 int	 tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int,
837 	    int, u_char *, u_int);
838 #endif
839 void	 tcp_slowtimo(void);
840 struct mbuf *
841 	 tcp_template(struct tcpcb *);
842 void	 tcp_trace(short, short, struct tcpcb *, struct mbuf *, int);
843 struct tcpcb *
844 	 tcp_usrclosed(struct tcpcb *);
845 int	 tcp_usrreq(struct socket *,
846 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *);
847 void	 tcp_xmit_timer(struct tcpcb *, uint32_t);
848 tcp_seq	 tcp_new_iss(struct tcpcb *, tcp_seq);
849 tcp_seq  tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t,
850 	    tcp_seq);
851 
852 void	 tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t);
853 void	 tcp_sack_option(struct tcpcb *, struct tcphdr *, u_char *, int);
854 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
855 void	 tcp_free_sackholes(struct tcpcb *);
856 void	 tcp_sack_adjust(struct tcpcb *tp);
857 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
858 void	 tcp_sack_newack(struct tcpcb *, struct tcphdr *);
859 int	 tcp_sack_numblks(const struct tcpcb *);
860 #define	TCP_SACK_OPTLEN(nblks)	((nblks) * 8 + 2 + 2)
861 
862 int	 syn_cache_add(struct sockaddr *, struct sockaddr *,
863 		struct tcphdr *, unsigned int, struct socket *,
864 		struct mbuf *, u_char *, int, struct tcp_opt_info *);
865 void	 syn_cache_unreach(const struct sockaddr *, const struct sockaddr *,
866 	   struct tcphdr *);
867 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
868 		struct tcphdr *, unsigned int, unsigned int,
869 		struct socket *so, struct mbuf *);
870 void	 syn_cache_init(void);
871 void	 syn_cache_insert(struct syn_cache *, struct tcpcb *);
872 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *,
873 		struct syn_cache_head **);
874 void	 syn_cache_reset(struct sockaddr *, struct sockaddr *,
875 		struct tcphdr *);
876 int	 syn_cache_respond(struct syn_cache *, struct mbuf *);
877 void	 syn_cache_timer(void *);
878 void	 syn_cache_cleanup(struct tcpcb *);
879 
880 void	 tcp_reno_newack(struct tcpcb *, struct tcphdr *);
881 void	 tcp_newreno_newack(struct tcpcb *, struct tcphdr *);
882 
883 int	 tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int,
884     int);
885 #endif
886 
887 #endif /* _NETINET_TCP_VAR_H_ */
888