Home | History | Annotate | Line # | Download | only in netinet
tcp_var.h revision 1.150
      1 /*	$NetBSD: tcp_var.h,v 1.150 2007/08/02 02:42:42 rmind 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 #include "rnd.h"
    149 #endif
    150 
    151 /*
    152  * Kernel variables for tcp.
    153  */
    154 
    155 #include <sys/callout.h>
    156 
    157 #ifdef TCP_SIGNATURE
    158 /*
    159  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
    160  * for SADB verification and lookup.
    161  */
    162 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
    163 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
    164 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
    165 /*
    166  * Only a single SA per host may be specified at this time. An SPI is
    167  * needed in order for the KEY_ALLOCSA() lookup to work.
    168  */
    169 #define	TCP_SIG_SPI	0x1000
    170 #endif /* TCP_SIGNATURE */
    171 
    172 /*
    173  * SACK option block.
    174  */
    175 struct sackblk {
    176 	tcp_seq left;		/* Left edge of sack block. */
    177 	tcp_seq right;		/* Right edge of sack block. */
    178 };
    179 
    180 TAILQ_HEAD(sackhead, sackhole);
    181 struct sackhole {
    182 	tcp_seq start;
    183 	tcp_seq end;
    184 	tcp_seq rxmit;
    185 
    186 	TAILQ_ENTRY(sackhole) sackhole_q;
    187 };
    188 
    189 /*
    190  * Tcp control block, one per tcp; fields:
    191  */
    192 struct tcpcb {
    193 	int	t_family;		/* address family on the wire */
    194 	struct ipqehead segq;		/* sequencing queue */
    195 	int	t_segqlen;		/* length of the above */
    196 	callout_t t_timer[TCPT_NTIMERS];/* tcp timers */
    197 	short	t_state;		/* state of this connection */
    198 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
    199 	uint32_t t_rxtcur;		/* current retransmit value */
    200 	short	t_dupacks;		/* consecutive dup acks recd */
    201 	short	t_partialacks;		/* partials acks during fast rexmit */
    202 	u_short	t_peermss;		/* peer's maximum segment size */
    203 	u_short	t_ourmss;		/* our's maximum segment size */
    204 	u_short t_segsz;		/* current segment size in use */
    205 	char	t_force;		/* 1 if forcing out a byte */
    206 	u_int	t_flags;
    207 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
    208 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
    209 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
    210 #define	TF_NOOPT	0x0008		/* don't use tcp options */
    211 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
    212 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
    213 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
    214 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
    215 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
    216 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
    217 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
    218 #define	TF_REASSEMBLING	0x1000		/* we're busy reassembling */
    219 #define	TF_DEAD		0x2000		/* dead and to-be-released */
    220 #define	TF_PMTUD_PEND	0x4000		/* Path MTU Discovery pending */
    221 #define	TF_ECN_PERMIT	0x10000		/* other side said is ECN-ready */
    222 #define	TF_ECN_SND_CWR	0x20000		/* ECN CWR in queue */
    223 #define	TF_ECN_SND_ECE	0x40000		/* ECN ECE in queue */
    224 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
    225 
    226 
    227 	struct	mbuf *t_template;	/* skeletal packet for transmit */
    228 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
    229 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
    230 	callout_t t_delack_ch;		/* delayed ACK callout */
    231 /*
    232  * The following fields are used as in the protocol specification.
    233  * See RFC793, Dec. 1981, page 21.
    234  */
    235 /* send sequence variables */
    236 	tcp_seq	snd_una;		/* send unacknowledged */
    237 	tcp_seq	snd_nxt;		/* send next */
    238 	tcp_seq	snd_up;			/* send urgent pointer */
    239 	tcp_seq	snd_wl1;		/* window update seg seq number */
    240 	tcp_seq	snd_wl2;		/* window update seg ack number */
    241 	tcp_seq	iss;			/* initial send sequence number */
    242 	u_long	snd_wnd;		/* send window */
    243 	tcp_seq snd_recover;		/* for use in fast recovery */
    244 	tcp_seq	snd_high;		/* NewReno false fast rexmit seq */
    245 /* receive sequence variables */
    246 	u_long	rcv_wnd;		/* receive window */
    247 	tcp_seq	rcv_nxt;		/* receive next */
    248 	tcp_seq	rcv_up;			/* receive urgent pointer */
    249 	tcp_seq	irs;			/* initial receive sequence number */
    250 /*
    251  * Additional variables for this implementation.
    252  */
    253 /* receive variables */
    254 	tcp_seq	rcv_adv;		/* advertised window */
    255 /* retransmit variables */
    256 	tcp_seq	snd_max;		/* highest sequence number sent;
    257 					 * used to recognize retransmits
    258 					 */
    259 /* congestion control (for slow start, source quench, retransmit after loss) */
    260 	u_long	snd_cwnd;		/* congestion-controlled window */
    261 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
    262 					 * for slow start exponential to
    263 					 * linear switch
    264 					 */
    265 /* auto-sizing variables */
    266 	u_int rfbuf_cnt;		/* recv buffer autoscaling byte count */
    267 	uint32_t rfbuf_ts;		/* recv buffer autoscaling timestamp */
    268 
    269 /*
    270  * transmit timing stuff.  See below for scale of srtt and rttvar.
    271  * "Variance" is actually smoothed difference.
    272  */
    273 	uint32_t t_rcvtime;		/* time last segment received */
    274 	uint32_t t_rtttime;		/* time we started measuring rtt */
    275 	tcp_seq	t_rtseq;		/* sequence number being timed */
    276 	int32_t	t_srtt;			/* smoothed round-trip time */
    277 	int32_t	t_rttvar;		/* variance in round-trip time */
    278 	uint32_t t_rttmin;		/* minimum rtt allowed */
    279 	u_long	max_sndwnd;		/* largest window peer has offered */
    280 
    281 /* out-of-band data */
    282 	char	t_oobflags;		/* have some */
    283 	char	t_iobc;			/* input character */
    284 #define	TCPOOB_HAVEDATA	0x01
    285 #define	TCPOOB_HADDATA	0x02
    286 	short	t_softerror;		/* possible error not yet reported */
    287 
    288 /* RFC 1323 variables */
    289 	u_char	snd_scale;		/* window scaling for send window */
    290 	u_char	rcv_scale;		/* window scaling for recv window */
    291 	u_char	request_r_scale;	/* pending window scaling */
    292 	u_char	requested_s_scale;
    293 	u_int32_t ts_recent;		/* timestamp echo data */
    294 	u_int32_t ts_recent_age;	/* when last updated */
    295 	u_int32_t ts_timebase;		/* our timebase */
    296 	tcp_seq	last_ack_sent;
    297 
    298 /* RFC 3465 variables */
    299 	u_long	t_bytes_acked;		/* ABC "bytes_acked" parameter */
    300 
    301 /* SACK stuff */
    302 #define TCP_SACK_MAX 3
    303 #define TCPSACK_NONE 0
    304 #define TCPSACK_HAVED 1
    305 	u_char rcv_sack_flags;		/* SACK flags. */
    306 	struct sackblk rcv_dsack_block;	/* RX D-SACK block. */
    307 	struct ipqehead timeq;		/* time sequenced queue. */
    308 	struct sackhead snd_holes;	/* TX SACK holes. */
    309 	int	snd_numholes;		/* Number of TX SACK holes. */
    310 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
    311 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
    312 					   episode starts at this seq number*/
    313 	tcp_seq snd_fack;		/* FACK TCP.  Forward-most data held by
    314 					   peer. */
    315 
    316 /* pointer for syn cache entries*/
    317 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
    318 
    319 /* prediction of next mbuf when using large window sizes */
    320 	struct	mbuf *t_lastm;		/* last mbuf that data was sent from */
    321 	int	t_inoff;		/* data offset in previous mbuf */
    322 	int	t_lastoff;		/* last data address in mbuf chain */
    323 	int	t_lastlen;		/* last length read from mbuf chain */
    324 
    325 /* Path-MTU discovery blackhole detection */
    326 	int t_mtudisc;			/* perform mtudisc for this tcb */
    327 /* Path-MTU Discovery Information */
    328 	u_int	t_pmtud_mss_acked;	/* MSS acked, lower bound for MTU */
    329 	u_int	t_pmtud_mtu_sent;	/* MTU used, upper bound for MTU */
    330 	tcp_seq	t_pmtud_th_seq;		/* TCP SEQ from ICMP payload */
    331 	u_int	t_pmtud_nextmtu;	/* Advertised Next-Hop MTU from ICMP */
    332 	u_short	t_pmtud_ip_len;		/* IP length from ICMP payload */
    333 	u_short	t_pmtud_ip_hl;		/* IP header length from ICMP payload */
    334 
    335 	uint8_t t_ecn_retries;		/* # of ECN setup retries */
    336 
    337 	struct tcp_congctl *t_congctl;	/* per TCB congctl algorithm */
    338 
    339 	/* Keepalive per socket */
    340 	u_int	t_keepinit;
    341 	u_int	t_keepidle;
    342 	u_int	t_keepintvl;
    343 	u_int	t_keepcnt;
    344 	u_int	t_maxidle;		/* t_keepcnt * t_keepintvl */
    345 
    346 };
    347 
    348 /*
    349  * Macros to aid ECN TCP.
    350  */
    351 #define TCP_ECN_ALLOWED(tp)	(tp->t_flags & TF_ECN_PERMIT)
    352 
    353 /*
    354  * Macros to aid SACK/FACK TCP.
    355  */
    356 #define TCP_SACK_ENABLED(tp)	(tp->t_flags & TF_WILL_SACK)
    357 #define TCP_FACK_FASTRECOV(tp)	\
    358 	(TCP_SACK_ENABLED(tp) && \
    359 	(SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz)))
    360 
    361 #ifdef _KERNEL
    362 /*
    363  * TCP reassembly queue locks.
    364  */
    365 static __inline int tcp_reass_lock_try (struct tcpcb *)
    366 	__attribute__((__unused__));
    367 static __inline void tcp_reass_unlock (struct tcpcb *)
    368 	__attribute__((__unused__));
    369 
    370 static __inline int
    371 tcp_reass_lock_try(tp)
    372 	struct tcpcb *tp;
    373 {
    374 	int s;
    375 
    376 	/*
    377 	 * Use splvm() -- we're blocking things that would cause
    378 	 * mbuf allocation.
    379 	 */
    380 	s = splvm();
    381 	if (tp->t_flags & TF_REASSEMBLING) {
    382 		splx(s);
    383 		return (0);
    384 	}
    385 	tp->t_flags |= TF_REASSEMBLING;
    386 	splx(s);
    387 	return (1);
    388 }
    389 
    390 static __inline void
    391 tcp_reass_unlock(tp)
    392 	struct tcpcb *tp;
    393 {
    394 	int s;
    395 
    396 	s = splvm();
    397 	tp->t_flags &= ~TF_REASSEMBLING;
    398 	splx(s);
    399 }
    400 
    401 #ifdef DIAGNOSTIC
    402 #define	TCP_REASS_LOCK(tp)						\
    403 do {									\
    404 	if (tcp_reass_lock_try(tp) == 0) {				\
    405 		printf("%s:%d: tcpcb %p reass already locked\n",	\
    406 		    __FILE__, __LINE__, tp);				\
    407 		panic("tcp_reass_lock");				\
    408 	}								\
    409 } while (/*CONSTCOND*/ 0)
    410 #define	TCP_REASS_LOCK_CHECK(tp)					\
    411 do {									\
    412 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
    413 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
    414 		    __FILE__, __LINE__, tp);				\
    415 		panic("tcp reass lock check");				\
    416 	}								\
    417 } while (/*CONSTCOND*/ 0)
    418 #else
    419 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
    420 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
    421 #endif
    422 
    423 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
    424 #endif /* _KERNEL */
    425 
    426 /*
    427  * Queue for delayed ACK processing.
    428  */
    429 #ifdef _KERNEL
    430 extern int tcp_delack_ticks;
    431 void	tcp_delack(void *);
    432 
    433 #define TCP_RESTART_DELACK(tp)						\
    434 	callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks,		\
    435 	    tcp_delack, tp)
    436 
    437 #define	TCP_SET_DELACK(tp)						\
    438 do {									\
    439 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
    440 		(tp)->t_flags |= TF_DELACK;				\
    441 		TCP_RESTART_DELACK(tp);					\
    442 	}								\
    443 } while (/*CONSTCOND*/0)
    444 
    445 #define	TCP_CLEAR_DELACK(tp)						\
    446 do {									\
    447 	if ((tp)->t_flags & TF_DELACK) {				\
    448 		(tp)->t_flags &= ~TF_DELACK;				\
    449 		callout_stop(&(tp)->t_delack_ch);			\
    450 	}								\
    451 } while (/*CONSTCOND*/0)
    452 #endif /* _KERNEL */
    453 
    454 /*
    455  * Compute the current timestamp for a connection.
    456  */
    457 #define	TCP_TIMESTAMP(tp)	(tcp_now - (tp)->ts_timebase)
    458 
    459 /*
    460  * Handy way of passing around TCP option info.
    461  */
    462 struct tcp_opt_info {
    463 	int		ts_present;
    464 	u_int32_t	ts_val;
    465 	u_int32_t	ts_ecr;
    466 	u_int16_t	maxseg;
    467 };
    468 
    469 #define	TOF_SIGNATURE	0x0040		/* signature option present */
    470 #define	TOF_SIGLEN	0x0080		/* sigature length valid (RFC2385) */
    471 
    472 /*
    473  * Data for the TCP compressed state engine.
    474  */
    475 union syn_cache_sa {
    476 	struct sockaddr sa;
    477 	struct sockaddr_in sin;
    478 #if 1 /*def INET6*/
    479 	struct sockaddr_in6 sin6;
    480 #endif
    481 };
    482 
    483 struct syn_cache {
    484 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
    485 	callout_t sc_timer;			/* rexmt timer */
    486 	struct route sc_route;
    487 	long sc_win;				/* advertised window */
    488 	int sc_bucketidx;			/* our bucket index */
    489 	u_int32_t sc_hash;
    490 	u_int32_t sc_timestamp;			/* timestamp from SYN */
    491 	u_int32_t sc_timebase;			/* our local timebase */
    492 	union syn_cache_sa sc_src;
    493 	union syn_cache_sa sc_dst;
    494 	tcp_seq sc_irs;
    495 	tcp_seq sc_iss;
    496 	u_int sc_rxtcur;			/* current rxt timeout */
    497 	u_int sc_rxttot;			/* total time spend on queues */
    498 	u_short sc_rxtshift;			/* for computing backoff */
    499 	u_short sc_flags;
    500 
    501 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
    502 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
    503 #define	SCF_DEAD		0x0004		/* this entry to be released */
    504 #define SCF_SACK_PERMIT		0x0008		/* peer will do SACK */
    505 #define SCF_ECN_PERMIT		0x0010		/* peer will do ECN */
    506 #define SCF_SIGNATURE	0x40			/* send MD5 digests */
    507 
    508 	struct mbuf *sc_ipopts;			/* IP options */
    509 	u_int16_t sc_peermaxseg;
    510 	u_int16_t sc_ourmaxseg;
    511 	u_int8_t sc_request_r_scale	: 4,
    512 		 sc_requested_s_scale	: 4;
    513 
    514 	struct tcpcb *sc_tp;			/* tcb for listening socket */
    515 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
    516 };
    517 
    518 struct syn_cache_head {
    519 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
    520 	u_short sch_length;			/* # entries in bucket */
    521 };
    522 
    523 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
    524 #ifdef INET6
    525 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
    526 #endif
    527 #ifndef INET6
    528 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
    529 #else
    530 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
    531 				? intotcpcb(sotoinpcb(so)) \
    532 				: in6totcpcb(sotoin6pcb(so)))
    533 #endif
    534 
    535 /*
    536  * The smoothed round-trip time and estimated variance
    537  * are stored as fixed point numbers scaled by the values below.
    538  * For convenience, these scales are also used in smoothing the average
    539  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
    540  * With these scales, srtt has 3 bits to the right of the binary point,
    541  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
    542  * binary point, and is smoothed with an ALPHA of 0.75.
    543  */
    544 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
    545 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
    546 
    547 /*
    548  * The initial retransmission should happen at rtt + 4 * rttvar.
    549  * Because of the way we do the smoothing, srtt and rttvar
    550  * will each average +1/2 tick of bias.  When we compute
    551  * the retransmit timer, we want 1/2 tick of rounding and
    552  * 1 extra tick because of +-1/2 tick uncertainty in the
    553  * firing of the timer.  The bias will give us exactly the
    554  * 1.5 tick we need.  But, because the bias is
    555  * statistical, we have to test that we don't drop below
    556  * the minimum feasible timer (which is 2 ticks).
    557  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
    558  * is the same as the multiplier for rttvar.
    559  */
    560 #define	TCP_REXMTVAL(tp) \
    561 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
    562 
    563 /*
    564  * Compute the initial window for slow start.
    565  */
    566 #define	TCP_INITIAL_WINDOW(iw, segsz) \
    567 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
    568 	 ((segsz) * (iw)))
    569 
    570 /*
    571  * TCP statistics.
    572  * Many of these should be kept per connection,
    573  * but that's inconvenient at the moment.
    574  */
    575 struct	tcpstat {
    576 	u_quad_t tcps_connattempt;	/* connections initiated */
    577 	u_quad_t tcps_accepts;		/* connections accepted */
    578 	u_quad_t tcps_connects;		/* connections established */
    579 	u_quad_t tcps_drops;		/* connections dropped */
    580 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
    581 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
    582 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
    583 	u_quad_t tcps_rttupdated;	/* times we succeeded */
    584 	u_quad_t tcps_delack;		/* delayed acks sent */
    585 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
    586 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
    587 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
    588 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
    589 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
    590 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
    591 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
    592 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
    593 					   shortage */
    594 	u_quad_t tcps_pmtublackhole;	/* PMTUD blackhole detected */
    595 
    596 	u_quad_t tcps_sndtotal;		/* total packets sent */
    597 	u_quad_t tcps_sndpack;		/* data packets sent */
    598 	u_quad_t tcps_sndbyte;		/* data bytes sent */
    599 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
    600 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
    601 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
    602 	u_quad_t tcps_sndprobe;		/* window probes sent */
    603 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
    604 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
    605 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
    606 
    607 	u_quad_t tcps_rcvtotal;		/* total packets received */
    608 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
    609 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
    610 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
    611 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
    612 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
    613 	u_quad_t tcps_rcvshort;		/* packets received too short */
    614 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
    615 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
    616 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
    617 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
    618 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
    619 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
    620 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
    621 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
    622 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
    623 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
    624 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
    625 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
    626 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
    627 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
    628 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
    629 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
    630 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
    631 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
    632 
    633 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
    634 	u_quad_t tcps_noport;		/* no socket on port */
    635 	u_quad_t tcps_badsyn;		/* received ack for which we have
    636 					   no SYN in compressed state */
    637 	u_quad_t tcps_delayed_free;	/* delayed pool_put() of tcpcb */
    638 
    639 	/* These statistics deal with the SYN cache. */
    640 	u_quad_t tcps_sc_added;		/* # of entries added */
    641 	u_quad_t tcps_sc_completed;	/* # of connections completed */
    642 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
    643 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
    644 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
    645 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
    646 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
    647 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
    648 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
    649 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
    650 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
    651 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
    652 	u_quad_t tcps_sc_delayed_free;	/* # of delayed pool_put()s */
    653 
    654 	u_quad_t tcps_selfquench;	/* # of ENOBUFS we get on output */
    655 	u_quad_t tcps_badsig;		/* # of drops due to bad signature */
    656 	u_quad_t tcps_goodsig;		/* # of packets with good signature */
    657 
    658 	u_quad_t tcps_ecn_shs;		/* # of sucessful ECN handshakes */
    659 	u_quad_t tcps_ecn_ce;		/* # of packets with CE bit */
    660 	u_quad_t tcps_ecn_ect;		/* # of packets with ECT(0) bit */
    661 };
    662 
    663 /*
    664  * Names for TCP sysctl objects.
    665  */
    666 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
    667 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
    668 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
    669 #define	TCPCTL_MSSDFLT		4	/* default seg size */
    670 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
    671 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
    672 #if 0	/*obsoleted*/
    673 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
    674 #endif
    675 #define	TCPCTL_INIT_WIN		8	/* initial window */
    676 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
    677 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
    678 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
    679 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
    680 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
    681 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
    682 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
    683 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
    684 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
    685 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
    686 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
    687 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
    688 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
    689 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
    690 #if 0	/*obsoleted*/
    691 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
    692 #endif
    693 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
    694 #define	TCPCTL_DELACK_TICKS	25	/* # ticks to delay ACK */
    695 #define	TCPCTL_INIT_WIN_LOCAL	26	/* initial window for local nets */
    696 #define	TCPCTL_IDENT		27	/* rfc 931 identd */
    697 #define	TCPCTL_ACKDROPRATELIMIT	28	/* SYN/RST -> ACK rate limit */
    698 #define	TCPCTL_LOOPBACKCKSUM	29	/* do TCP checksum on loopback */
    699 #define	TCPCTL_STATS		30	/* TCP statistics */
    700 #define	TCPCTL_DEBUG		31	/* TCP debug sockets */
    701 #define	TCPCTL_DEBX		32	/* # of tcp debug sockets */
    702 #define	TCPCTL_DROP		33	/* drop tcp connection */
    703 #define	TCPCTL_MAXID		34
    704 
    705 #define	TCPCTL_NAMES { \
    706 	{ 0, 0 }, \
    707 	{ "rfc1323",	CTLTYPE_INT }, \
    708 	{ "sendspace",	CTLTYPE_INT }, \
    709 	{ "recvspace",	CTLTYPE_INT }, \
    710 	{ "mssdflt",	CTLTYPE_INT }, \
    711 	{ "syn_cache_limit", CTLTYPE_INT }, \
    712 	{ "syn_bucket_limit", CTLTYPE_INT }, \
    713 	{ 0, 0 },\
    714 	{ "init_win", CTLTYPE_INT }, \
    715 	{ "mss_ifmtu", CTLTYPE_INT }, \
    716 	{ "sack", CTLTYPE_INT }, \
    717 	{ "win_scale", CTLTYPE_INT }, \
    718 	{ "timestamps", CTLTYPE_INT }, \
    719 	{ "compat_42", CTLTYPE_INT }, \
    720 	{ "cwm", CTLTYPE_INT }, \
    721 	{ "cwm_burstsize", CTLTYPE_INT }, \
    722 	{ "ack_on_push", CTLTYPE_INT }, \
    723 	{ "keepidle",	CTLTYPE_INT }, \
    724 	{ "keepintvl",	CTLTYPE_INT }, \
    725 	{ "keepcnt",	CTLTYPE_INT }, \
    726 	{ "slowhz",	CTLTYPE_INT }, \
    727 	{ 0, 0 }, \
    728 	{ "log_refused",CTLTYPE_INT }, \
    729 	{ 0, 0 }, \
    730 	{ "rstppslimit", CTLTYPE_INT }, \
    731 	{ "delack_ticks", CTLTYPE_INT }, \
    732 	{ "init_win_local", CTLTYPE_INT }, \
    733 	{ "ident", CTLTYPE_STRUCT }, \
    734 	{ "ackdropppslimit", CTLTYPE_INT }, \
    735 	{ "do_loopback_cksum", CTLTYPE_INT }, \
    736 	{ "stats", CTLTYPE_STRUCT }, \
    737 	{ "debug", CTLTYPE_STRUCT }, \
    738 	{ "debx", CTLTYPE_INT }, \
    739 	{ "drop", CTLTYPE_STRUCT }, \
    740 }
    741 
    742 #ifdef _KERNEL
    743 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
    744 extern	struct tcpstat tcpstat;	/* tcp statistics */
    745 extern	u_int32_t tcp_now;	/* for RFC 1323 timestamps */
    746 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
    747 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
    748 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
    749 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
    750 extern	int tcp_mssdflt;	/* default seg size */
    751 extern	int tcp_minmss;		/* minimal seg size */
    752 extern	int tcp_init_win;	/* initial window */
    753 extern	int tcp_init_win_local;	/* initial window for local nets */
    754 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
    755 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
    756 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
    757 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
    758 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
    759 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
    760 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
    761 extern	int tcp_log_refused;	/* log refused connections */
    762 extern	int tcp_do_ecn;		/* TCP ECN enabled/disabled? */
    763 extern	int tcp_ecn_maxretries;	/* Max ECN setup retries */
    764 #if NRND > 0
    765 extern	int tcp_do_rfc1948;	/* ISS by cryptographic hash */
    766 #endif
    767 extern int tcp_sack_tp_maxholes;	/* Max holes per connection. */
    768 extern int tcp_sack_globalmaxholes;	/* Max holes per system. */
    769 extern int tcp_sack_globalholes;	/* Number of holes present. */
    770 extern int tcp_do_abc;			/* RFC3465 ABC enabled/disabled? */
    771 extern int tcp_abc_aggressive;		/* 1: L=2*SMSS  0: L=1*SMSS */
    772 
    773 extern	int tcp_rst_ppslim;
    774 extern	int tcp_ackdrop_ppslim;
    775 
    776 extern	int tcp_syn_cache_size;
    777 extern	struct syn_cache_head tcp_syn_cache[];
    778 extern	u_long syn_cache_count;
    779 
    780 #ifdef MBUFTRACE
    781 extern	struct mowner tcp_rx_mowner;
    782 extern	struct mowner tcp_tx_mowner;
    783 extern	struct mowner tcp_reass_mowner;
    784 extern	struct mowner tcp_sock_mowner;
    785 extern	struct mowner tcp_sock_rx_mowner;
    786 extern	struct mowner tcp_sock_tx_mowner;
    787 extern	struct mowner tcp_mowner;
    788 #endif
    789 
    790 extern int tcp_do_autorcvbuf;
    791 extern int tcp_autorcvbuf_inc;
    792 extern int tcp_autorcvbuf_max;
    793 extern int tcp_do_autosndbuf;
    794 extern int tcp_autosndbuf_inc;
    795 extern int tcp_autosndbuf_max;
    796 
    797 
    798 #define	TCPCTL_VARIABLES { \
    799 	{ 0 },					\
    800 	{ 1, 0, &tcp_do_rfc1323 },		\
    801 	{ 1, 0, &tcp_sendspace },		\
    802 	{ 1, 0, &tcp_recvspace },		\
    803 	{ 1, 0, &tcp_mssdflt },			\
    804 	{ 1, 0, &tcp_syn_cache_limit },		\
    805 	{ 1, 0, &tcp_syn_bucket_limit },	\
    806 	{ 0 },					\
    807 	{ 1, 0, &tcp_init_win },		\
    808 	{ 1, 0, &tcp_mss_ifmtu },		\
    809 	{ 1, 0, &tcp_do_sack },			\
    810 	{ 1, 0, &tcp_do_win_scale },		\
    811 	{ 1, 0, &tcp_do_timestamps },		\
    812 	{ 1, 0, &tcp_compat_42 },		\
    813 	{ 1, 0, &tcp_cwm },			\
    814 	{ 1, 0, &tcp_cwm_burstsize },		\
    815 	{ 1, 0, &tcp_ack_on_push },		\
    816 	{ 1, 0, &tcp_keepidle },		\
    817 	{ 1, 0, &tcp_keepintvl },		\
    818 	{ 1, 0, &tcp_keepcnt },			\
    819 	{ 1, 1, 0, PR_SLOWHZ },			\
    820 	{ 0 },					\
    821 	{ 1, 0, &tcp_log_refused },		\
    822 	{ 0 },					\
    823 	{ 1, 0, &tcp_rst_ppslim },		\
    824 	{ 1, 0, &tcp_delack_ticks },		\
    825 	{ 1, 0, &tcp_init_win_local },		\
    826 	{ 1, 0, &tcp_ackdrop_ppslim },		\
    827 }
    828 
    829 #ifdef __NO_STRICT_ALIGNMENT
    830 #define	TCP_HDR_ALIGNED_P(th)	1
    831 #else
    832 #define	TCP_HDR_ALIGNED_P(th)	((((vaddr_t)(th)) & 3) == 0)
    833 #endif
    834 
    835 struct secasvar;
    836 
    837 int	 tcp_attach(struct socket *);
    838 void	 tcp_canceltimers(struct tcpcb *);
    839 int	 tcp_timers_invoking(struct tcpcb*);
    840 struct tcpcb *
    841 	 tcp_close(struct tcpcb *);
    842 int	 tcp_isdead(struct tcpcb *);
    843 #ifdef INET6
    844 void	 tcp6_ctlinput(int, const struct sockaddr *, void *);
    845 #endif
    846 void	 *tcp_ctlinput(int, const struct sockaddr *, void *);
    847 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf **);
    848 struct tcpcb *
    849 	 tcp_disconnect(struct tcpcb *);
    850 struct tcpcb *
    851 	 tcp_drop(struct tcpcb *, int);
    852 #ifdef TCP_SIGNATURE
    853 int	 tcp_signature_apply(void *, void *, u_int);
    854 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *);
    855 int	 tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *,
    856 	    char *);
    857 #endif
    858 void	 tcp_drain(void);
    859 void	 tcp_established(struct tcpcb *);
    860 void	 tcp_init(void);
    861 #ifdef INET6
    862 int	 tcp6_input(struct mbuf **, int *, int);
    863 #endif
    864 void	 tcp_input(struct mbuf *, ...);
    865 u_int	 tcp_hdrsz(struct tcpcb *);
    866 u_long	 tcp_mss_to_advertise(const struct ifnet *, int);
    867 void	 tcp_mss_from_peer(struct tcpcb *, int);
    868 void	 tcp_tcpcb_template(void);
    869 struct tcpcb *
    870 	 tcp_newtcpcb(int, void *);
    871 void	 tcp_notify(struct inpcb *, int);
    872 #ifdef INET6
    873 void	 tcp6_notify(struct in6pcb *, int);
    874 #endif
    875 u_int	 tcp_optlen(struct tcpcb *);
    876 int	 tcp_output(struct tcpcb *);
    877 void	 tcp_pulloutofband(struct socket *,
    878 	    struct tcphdr *, struct mbuf *, int);
    879 void	 tcp_quench(struct inpcb *, int);
    880 #ifdef INET6
    881 void	 tcp6_quench(struct in6pcb *, int);
    882 #endif
    883 void	 tcp_mtudisc(struct inpcb *, int);
    884 
    885 struct ipqent *tcpipqent_alloc(void);
    886 void	 tcpipqent_free(struct ipqent *);
    887 
    888 int	 tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *,
    889 	    struct tcphdr *, tcp_seq, tcp_seq, int);
    890 void	 tcp_rmx_rtt(struct tcpcb *);
    891 void	 tcp_setpersist(struct tcpcb *);
    892 #ifdef TCP_SIGNATURE
    893 int	 tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int,
    894 	    int, u_char *, u_int);
    895 #endif
    896 void	 tcp_slowtimo(void);
    897 struct mbuf *
    898 	 tcp_template(struct tcpcb *);
    899 void	 tcp_trace(short, short, struct tcpcb *, struct mbuf *, int);
    900 struct tcpcb *
    901 	 tcp_usrclosed(struct tcpcb *);
    902 int	 tcp_usrreq(struct socket *,
    903 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *);
    904 void	 tcp_xmit_timer(struct tcpcb *, uint32_t);
    905 tcp_seq	 tcp_new_iss(struct tcpcb *, tcp_seq);
    906 tcp_seq  tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t,
    907 	    tcp_seq);
    908 
    909 void	 tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t);
    910 void	 tcp_sack_option(struct tcpcb *, const struct tcphdr *,
    911 	    const u_char *, int);
    912 void	 tcp_del_sackholes(struct tcpcb *, const struct tcphdr *);
    913 void	 tcp_free_sackholes(struct tcpcb *);
    914 void	 tcp_sack_adjust(struct tcpcb *tp);
    915 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
    916 void	 tcp_sack_newack(struct tcpcb *, const struct tcphdr *);
    917 int	 tcp_sack_numblks(const struct tcpcb *);
    918 #define	TCP_SACK_OPTLEN(nblks)	((nblks) * 8 + 2 + 2)
    919 
    920 int	 syn_cache_add(struct sockaddr *, struct sockaddr *,
    921 		struct tcphdr *, unsigned int, struct socket *,
    922 		struct mbuf *, u_char *, int, struct tcp_opt_info *);
    923 void	 syn_cache_unreach(const struct sockaddr *, const struct sockaddr *,
    924 	   struct tcphdr *);
    925 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
    926 		struct tcphdr *, unsigned int, unsigned int,
    927 		struct socket *so, struct mbuf *);
    928 void	 syn_cache_init(void);
    929 void	 syn_cache_insert(struct syn_cache *, struct tcpcb *);
    930 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *,
    931 		struct syn_cache_head **);
    932 void	 syn_cache_reset(struct sockaddr *, struct sockaddr *,
    933 		struct tcphdr *);
    934 int	 syn_cache_respond(struct syn_cache *, struct mbuf *);
    935 void	 syn_cache_timer(void *);
    936 void	 syn_cache_cleanup(struct tcpcb *);
    937 
    938 int	 tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int,
    939     int);
    940 #endif
    941 
    942 #endif /* !_NETINET_TCP_VAR_H_ */
    943