Home | History | Annotate | Line # | Download | only in netinet
tcp_var.h revision 1.73
      1 /*	$NetBSD: tcp_var.h,v 1.73 2000/07/27 11:34:07 itojun Exp $	*/
      2 
      3 /*
      4 %%% portions-copyright-nrl-98
      5 Portions of this software are Copyright 1998 by Randall Atkinson,
      6 Ronald Lee, Daniel McDonald, Bao Phan, and Chris Winters. All Rights
      7 Reserved. All rights under this copyright have been assigned to the US
      8 Naval Research Laboratory (NRL). The NRL Copyright Notice and License
      9 Agreement Version 1.1 (January 17, 1995) applies to these portions of the
     10 software.
     11 You should have received a copy of the license with this software. If you
     12 didn't get a copy, you may request one from <license (at) ipv6.nrl.navy.mil>.
     13 
     14 */
     15 
     16 /*
     17  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
     18  * All rights reserved.
     19  *
     20  * Redistribution and use in source and binary forms, with or without
     21  * modification, are permitted provided that the following conditions
     22  * are met:
     23  * 1. Redistributions of source code must retain the above copyright
     24  *    notice, this list of conditions and the following disclaimer.
     25  * 2. Redistributions in binary form must reproduce the above copyright
     26  *    notice, this list of conditions and the following disclaimer in the
     27  *    documentation and/or other materials provided with the distribution.
     28  * 3. Neither the name of the project nor the names of its contributors
     29  *    may be used to endorse or promote products derived from this software
     30  *    without specific prior written permission.
     31  *
     32  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
     33  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     34  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     35  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
     36  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     37  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     38  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     39  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     40  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     41  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     42  * SUCH DAMAGE.
     43  */
     44 
     45 /*-
     46  * Copyright (c) 1997, 1998, 1999 The NetBSD Foundation, Inc.
     47  * All rights reserved.
     48  *
     49  * This code is derived from software contributed to The NetBSD Foundation
     50  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
     51  * NASA Ames Research Center.
     52  *
     53  * Redistribution and use in source and binary forms, with or without
     54  * modification, are permitted provided that the following conditions
     55  * are met:
     56  * 1. Redistributions of source code must retain the above copyright
     57  *    notice, this list of conditions and the following disclaimer.
     58  * 2. Redistributions in binary form must reproduce the above copyright
     59  *    notice, this list of conditions and the following disclaimer in the
     60  *    documentation and/or other materials provided with the distribution.
     61  * 3. All advertising materials mentioning features or use of this software
     62  *    must display the following acknowledgement:
     63  *	This product includes software developed by the NetBSD
     64  *	Foundation, Inc. and its contributors.
     65  * 4. Neither the name of The NetBSD Foundation nor the names of its
     66  *    contributors may be used to endorse or promote products derived
     67  *    from this software without specific prior written permission.
     68  *
     69  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     70  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     71  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     72  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     73  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     74  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     75  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     76  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     77  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     78  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     79  * POSSIBILITY OF SUCH DAMAGE.
     80  */
     81 
     82 /*
     83  * Copyright (c) 1982, 1986, 1993, 1994, 1995
     84  *	The Regents of the University of California.  All rights reserved.
     85  *
     86  * Redistribution and use in source and binary forms, with or without
     87  * modification, are permitted provided that the following conditions
     88  * are met:
     89  * 1. Redistributions of source code must retain the above copyright
     90  *    notice, this list of conditions and the following disclaimer.
     91  * 2. Redistributions in binary form must reproduce the above copyright
     92  *    notice, this list of conditions and the following disclaimer in the
     93  *    documentation and/or other materials provided with the distribution.
     94  * 3. All advertising materials mentioning features or use of this software
     95  *    must display the following acknowledgement:
     96  *	This product includes software developed by the University of
     97  *	California, Berkeley and its contributors.
     98  * 4. Neither the name of the University nor the names of its contributors
     99  *    may be used to endorse or promote products derived from this software
    100  *    without specific prior written permission.
    101  *
    102  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
    103  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
    104  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
    105  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
    106  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
    107  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
    108  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
    109  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
    110  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
    111  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
    112  * SUCH DAMAGE.
    113  *
    114  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
    115  */
    116 
    117 #ifndef _NETINET_TCP_VAR_H_
    118 #define _NETINET_TCP_VAR_H_
    119 
    120 #if defined(_KERNEL) && !defined(_LKM)
    121 #include "opt_inet.h"
    122 #endif
    123 
    124 /*
    125  * Kernel variables for tcp.
    126  */
    127 
    128 /*
    129  * Tcp control block, one per tcp; fields:
    130  */
    131 struct tcpcb {
    132 	int	t_family;		/* address family on the wire */
    133 	struct ipqehead segq;		/* sequencing queue */
    134 	u_int	t_timer[TCPT_NTIMERS];	/* tcp timers */
    135 	short	t_state;		/* state of this connection */
    136 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
    137 	short	t_rxtcur;		/* current retransmit value */
    138 	short	t_dupacks;		/* consecutive dup acks recd */
    139 	u_short	t_peermss;		/* peer's maximum segment size */
    140 	u_short	t_ourmss;		/* our's maximum segment size */
    141 	u_short t_segsz;		/* current segment size in use */
    142 	char	t_force;		/* 1 if forcing out a byte */
    143 	u_short	t_flags;
    144 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
    145 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
    146 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
    147 #define	TF_NOOPT	0x0008		/* don't use tcp options */
    148 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
    149 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
    150 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
    151 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
    152 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
    153 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
    154 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
    155 #define	TF_CANT_TXSACK	0x1000		/* other side said I could not SACK */
    156 #define	TF_IGNR_RXSACK	0x2000		/* ignore received SACK blocks */
    157 #define	TF_REASSEMBLING	0x4000		/* we're busy reassembling */
    158 
    159 
    160 	struct	mbuf *t_template;	/* skeletal packet for transmit */
    161 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
    162 	struct	in6pcb *t_in6pcb;	/* back pointer to internet pcb */
    163 	LIST_ENTRY(tcpcb) t_delack;	/* delayed ACK queue */
    164 /*
    165  * The following fields are used as in the protocol specification.
    166  * See RFC783, Dec. 1981, page 21.
    167  */
    168 /* send sequence variables */
    169 	tcp_seq	snd_una;		/* send unacknowledged */
    170 	tcp_seq	snd_nxt;		/* send next */
    171 	tcp_seq	snd_up;			/* send urgent pointer */
    172 	tcp_seq	snd_wl1;		/* window update seg seq number */
    173 	tcp_seq	snd_wl2;		/* window update seg ack number */
    174 	tcp_seq	iss;			/* initial send sequence number */
    175 	u_long	snd_wnd;		/* send window */
    176 	tcp_seq snd_recover;		/* for use in fast recovery */
    177 /* receive sequence variables */
    178 	u_long	rcv_wnd;		/* receive window */
    179 	tcp_seq	rcv_nxt;		/* receive next */
    180 	tcp_seq	rcv_up;			/* receive urgent pointer */
    181 	tcp_seq	irs;			/* initial receive sequence number */
    182 /*
    183  * Additional variables for this implementation.
    184  */
    185 /* receive variables */
    186 	tcp_seq	rcv_adv;		/* advertised window */
    187 /* retransmit variables */
    188 	tcp_seq	snd_max;		/* highest sequence number sent;
    189 					 * used to recognize retransmits
    190 					 */
    191 /* congestion control (for slow start, source quench, retransmit after loss) */
    192 	u_long	snd_cwnd;		/* congestion-controlled window */
    193 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
    194 					 * for slow start exponential to
    195 					 * linear switch
    196 					 */
    197 /*
    198  * transmit timing stuff.  See below for scale of srtt and rttvar.
    199  * "Variance" is actually smoothed difference.
    200  */
    201 	short	t_idle;			/* inactivity time */
    202 	short	t_rtt;			/* round trip time */
    203 	tcp_seq	t_rtseq;		/* sequence number being timed */
    204 	short	t_srtt;			/* smoothed round-trip time */
    205 	short	t_rttvar;		/* variance in round-trip time */
    206 	short	t_rttmin;		/* minimum rtt allowed */
    207 	u_long	max_sndwnd;		/* largest window peer has offered */
    208 
    209 /* out-of-band data */
    210 	char	t_oobflags;		/* have some */
    211 	char	t_iobc;			/* input character */
    212 #define	TCPOOB_HAVEDATA	0x01
    213 #define	TCPOOB_HADDATA	0x02
    214 	short	t_softerror;		/* possible error not yet reported */
    215 
    216 /* RFC 1323 variables */
    217 	u_char	snd_scale;		/* window scaling for send window */
    218 	u_char	rcv_scale;		/* window scaling for recv window */
    219 	u_char	request_r_scale;	/* pending window scaling */
    220 	u_char	requested_s_scale;
    221 	u_int32_t ts_recent;		/* timestamp echo data */
    222 	u_int32_t ts_recent_age;	/* when last updated */
    223 	tcp_seq	last_ack_sent;
    224 
    225 /* SACK stuff */
    226 	struct ipqehead timeq;		/* time sequenced queue (for SACK) */
    227 
    228 /* pointer for syn cache entries*/
    229 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
    230 };
    231 
    232 #ifdef _KERNEL
    233 /*
    234  * TCP reassembly queue locks.
    235  */
    236 static __inline int tcp_reass_lock_try __P((struct tcpcb *))
    237 	__attribute__((__unused__));
    238 static __inline void tcp_reass_unlock __P((struct tcpcb *))
    239 	__attribute__((__unused__));
    240 
    241 static __inline int
    242 tcp_reass_lock_try(tp)
    243 	struct tcpcb *tp;
    244 {
    245 	int s;
    246 
    247 	s = splimp();
    248 	if (tp->t_flags & TF_REASSEMBLING) {
    249 		splx(s);
    250 		return (0);
    251 	}
    252 	tp->t_flags |= TF_REASSEMBLING;
    253 	splx(s);
    254 	return (1);
    255 }
    256 
    257 static __inline void
    258 tcp_reass_unlock(tp)
    259 	struct tcpcb *tp;
    260 {
    261 	int s;
    262 
    263 	s = splimp();
    264 	tp->t_flags &= ~TF_REASSEMBLING;
    265 	splx(s);
    266 }
    267 
    268 #ifdef DIAGNOSTIC
    269 #define	TCP_REASS_LOCK(tp)						\
    270 do {									\
    271 	if (tcp_reass_lock_try(tp) == 0) {				\
    272 		printf("%s:%d: tcpcb %p reass already locked\n",	\
    273 		    __FILE__, __LINE__, tp);				\
    274 		panic("tcp_reass_lock");				\
    275 	}								\
    276 } while (0)
    277 #define	TCP_REASS_LOCK_CHECK(tp)					\
    278 do {									\
    279 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
    280 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
    281 		    __FILE__, __LINE__, tp);				\
    282 		panic("tcp reass lock check");				\
    283 	}								\
    284 } while (0)
    285 #else
    286 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
    287 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
    288 #endif
    289 
    290 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
    291 #endif /* _KERNEL */
    292 
    293 /*
    294  * Queue for delayed ACK processing.
    295  */
    296 LIST_HEAD(tcp_delack_head, tcpcb);
    297 #ifdef _KERNEL
    298 extern struct tcp_delack_head tcp_delacks;
    299 
    300 #define	TCP_SET_DELACK(tp) \
    301 do { \
    302 	if (((tp)->t_flags & TF_DELACK) == 0) { \
    303 		(tp)->t_flags |= TF_DELACK; \
    304 		LIST_INSERT_HEAD(&tcp_delacks, (tp), t_delack); \
    305 	} \
    306 } while (0)
    307 
    308 #define	TCP_CLEAR_DELACK(tp) \
    309 do { \
    310 	if ((tp)->t_flags & TF_DELACK) { \
    311 		(tp)->t_flags &= ~TF_DELACK; \
    312 		LIST_REMOVE((tp), t_delack); \
    313 	} \
    314 } while (0)
    315 #endif /* _KERNEL */
    316 
    317 /*
    318  * Handy way of passing around TCP option info.
    319  */
    320 struct tcp_opt_info {
    321 	int		ts_present;
    322 	u_int32_t	ts_val;
    323 	u_int32_t	ts_ecr;
    324 	u_int16_t	maxseg;
    325 };
    326 
    327 /*
    328  * Data for the TCP compressed state engine.
    329  */
    330 union syn_cache_sa {
    331 	struct sockaddr sa;
    332 	struct sockaddr_in sin;
    333 #if 1 /*def INET6*/
    334 	struct sockaddr_in6 sin6;
    335 #endif
    336 };
    337 
    338 struct syn_cache {
    339 	LIST_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
    340 	TAILQ_ENTRY(syn_cache) sc_timeq;	/* link on timer queue */
    341 	union {					/* cached route */
    342 		struct route route4;
    343 #ifdef INET6
    344 		struct route_in6 route6;
    345 #endif
    346 	} sc_route_u;
    347 #define sc_route4	sc_route_u.route4
    348 #ifdef INET6
    349 #define sc_route6	sc_route_u.route6
    350 #endif
    351 	long sc_win;				/* advertised window */
    352 	int sc_bucketidx;			/* our bucket index */
    353 	u_int32_t sc_hash;
    354 	u_int32_t sc_timestamp;			/* timestamp from SYN */
    355 	union syn_cache_sa sc_src;
    356 	union syn_cache_sa sc_dst;
    357 	tcp_seq sc_irs;
    358 	tcp_seq sc_iss;
    359 	u_int sc_rexmt;				/* retransmit timer */
    360 	u_int sc_rxtcur;			/* current rxt timeout */
    361 	u_int sc_rxttot;			/* total time spend on queues */
    362 	u_short sc_rxtshift;			/* for computing backoff */
    363 	u_short sc_flags;
    364 
    365 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
    366 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
    367 
    368 	struct mbuf *sc_ipopts;			/* IP options */
    369 	u_int16_t sc_peermaxseg;
    370 	u_int16_t sc_ourmaxseg;
    371 	u_int8_t sc_request_r_scale	: 4,
    372 		 sc_requested_s_scale	: 4;
    373 
    374 	struct tcpcb *sc_tp;			/* tcb for listening socket */
    375 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
    376 };
    377 
    378 struct syn_cache_head {
    379 	LIST_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
    380 	u_short sch_length;			/* # entries in bucket */
    381 };
    382 
    383 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
    384 #ifdef INET6
    385 #define	in6totcpcb(ip)	((struct tcpcb *)(ip)->in6p_ppcb)
    386 #endif
    387 #ifndef INET6
    388 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
    389 #else
    390 #define	sototcpcb(so)	(((so)->so_proto->pr_domain->dom_family == AF_INET) \
    391 				? intotcpcb(sotoinpcb(so)) \
    392 				: in6totcpcb(sotoin6pcb(so)))
    393 #endif
    394 
    395 /*
    396  * The smoothed round-trip time and estimated variance
    397  * are stored as fixed point numbers scaled by the values below.
    398  * For convenience, these scales are also used in smoothing the average
    399  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
    400  * With these scales, srtt has 3 bits to the right of the binary point,
    401  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
    402  * binary point, and is smoothed with an ALPHA of 0.75.
    403  */
    404 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
    405 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
    406 
    407 /*
    408  * The initial retransmission should happen at rtt + 4 * rttvar.
    409  * Because of the way we do the smoothing, srtt and rttvar
    410  * will each average +1/2 tick of bias.  When we compute
    411  * the retransmit timer, we want 1/2 tick of rounding and
    412  * 1 extra tick because of +-1/2 tick uncertainty in the
    413  * firing of the timer.  The bias will give us exactly the
    414  * 1.5 tick we need.  But, because the bias is
    415  * statistical, we have to test that we don't drop below
    416  * the minimum feasible timer (which is 2 ticks).
    417  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
    418  * is the same as the multiplier for rttvar.
    419  */
    420 #define	TCP_REXMTVAL(tp) \
    421 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
    422 
    423 /*
    424  * Compute the initial window for slow start.
    425  */
    426 #define	TCP_INITIAL_WINDOW(iw, segsz) \
    427 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
    428 	 ((segsz) * (iw)))
    429 
    430 /*
    431  * TCP statistics.
    432  * Many of these should be kept per connection,
    433  * but that's inconvenient at the moment.
    434  */
    435 struct	tcpstat {
    436 	u_quad_t tcps_connattempt;	/* connections initiated */
    437 	u_quad_t tcps_accepts;		/* connections accepted */
    438 	u_quad_t tcps_connects;		/* connections established */
    439 	u_quad_t tcps_drops;		/* connections dropped */
    440 	u_quad_t tcps_conndrops;	/* embryonic connections dropped */
    441 	u_quad_t tcps_closed;		/* conn. closed (includes drops) */
    442 	u_quad_t tcps_segstimed;	/* segs where we tried to get rtt */
    443 	u_quad_t tcps_rttupdated;	/* times we succeeded */
    444 	u_quad_t tcps_delack;		/* delayed acks sent */
    445 	u_quad_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
    446 	u_quad_t tcps_rexmttimeo;	/* retransmit timeouts */
    447 	u_quad_t tcps_persisttimeo;	/* persist timeouts */
    448 	u_quad_t tcps_keeptimeo;	/* keepalive timeouts */
    449 	u_quad_t tcps_keepprobe;	/* keepalive probes sent */
    450 	u_quad_t tcps_keepdrops;	/* connections dropped in keepalive */
    451 	u_quad_t tcps_persistdrops;	/* connections dropped in persist */
    452 	u_quad_t tcps_connsdrained;	/* connections drained due to memory
    453 					   shortage */
    454 
    455 	u_quad_t tcps_sndtotal;		/* total packets sent */
    456 	u_quad_t tcps_sndpack;		/* data packets sent */
    457 	u_quad_t tcps_sndbyte;		/* data bytes sent */
    458 	u_quad_t tcps_sndrexmitpack;	/* data packets retransmitted */
    459 	u_quad_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
    460 	u_quad_t tcps_sndacks;		/* ack-only packets sent */
    461 	u_quad_t tcps_sndprobe;		/* window probes sent */
    462 	u_quad_t tcps_sndurg;		/* packets sent with URG only */
    463 	u_quad_t tcps_sndwinup;		/* window update-only packets sent */
    464 	u_quad_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
    465 
    466 	u_quad_t tcps_rcvtotal;		/* total packets received */
    467 	u_quad_t tcps_rcvpack;		/* packets received in sequence */
    468 	u_quad_t tcps_rcvbyte;		/* bytes received in sequence */
    469 	u_quad_t tcps_rcvbadsum;	/* packets received with ccksum errs */
    470 	u_quad_t tcps_rcvbadoff;	/* packets received with bad offset */
    471 	u_quad_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
    472 	u_quad_t tcps_rcvshort;		/* packets received too short */
    473 	u_quad_t tcps_rcvduppack;	/* duplicate-only packets received */
    474 	u_quad_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
    475 	u_quad_t tcps_rcvpartduppack;	/* packets with some duplicate data */
    476 	u_quad_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
    477 	u_quad_t tcps_rcvoopack;	/* out-of-order packets received */
    478 	u_quad_t tcps_rcvoobyte;	/* out-of-order bytes received */
    479 	u_quad_t tcps_rcvpackafterwin;	/* packets with data after window */
    480 	u_quad_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
    481 	u_quad_t tcps_rcvafterclose;	/* packets rcvd after "close" */
    482 	u_quad_t tcps_rcvwinprobe;	/* rcvd window probe packets */
    483 	u_quad_t tcps_rcvdupack;	/* rcvd duplicate acks */
    484 	u_quad_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
    485 	u_quad_t tcps_rcvackpack;	/* rcvd ack packets */
    486 	u_quad_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
    487 	u_quad_t tcps_rcvwinupd;	/* rcvd window update packets */
    488 	u_quad_t tcps_pawsdrop;		/* segments dropped due to PAWS */
    489 	u_quad_t tcps_predack;		/* times hdr predict ok for acks */
    490 	u_quad_t tcps_preddat;		/* times hdr predict ok for data pkts */
    491 
    492 	u_quad_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
    493 	u_quad_t tcps_noport;		/* no socket on port */
    494 	u_quad_t tcps_badsyn;		/* received ack for which we have
    495 					   no SYN in compressed state */
    496 
    497 	/* These statistics deal with the SYN cache. */
    498 	u_quad_t tcps_sc_added;		/* # of entries added */
    499 	u_quad_t tcps_sc_completed;	/* # of connections completed */
    500 	u_quad_t tcps_sc_timed_out;	/* # of entries timed out */
    501 	u_quad_t tcps_sc_overflowed;	/* # dropped due to overflow */
    502 	u_quad_t tcps_sc_reset;		/* # dropped due to RST */
    503 	u_quad_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
    504 	u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
    505 	u_quad_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
    506 	u_quad_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
    507 	u_quad_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
    508 	u_quad_t tcps_sc_collisions;	/* # of hash collisions */
    509 	u_quad_t tcps_sc_retransmitted;	/* # of retransmissions */
    510 };
    511 
    512 /*
    513  * Names for TCP sysctl objects.
    514  */
    515 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
    516 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
    517 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
    518 #define	TCPCTL_MSSDFLT		4	/* default seg size */
    519 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
    520 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
    521 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
    522 #define	TCPCTL_INIT_WIN		8	/* initial window */
    523 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
    524 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
    525 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
    526 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
    527 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
    528 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
    529 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
    530 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
    531 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
    532 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
    533 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
    534 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
    535 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
    536 #define TCPCTL_LOG_REFUSED	22	/* Log refused connections */
    537 #define	TCPCTL_RSTRATELIMIT	23	/* RST rate limit */
    538 #define	TCPCTL_RSTPPSLIMIT	24	/* RST pps limit */
    539 #define	TCPCTL_MAXID		25
    540 
    541 #define	TCPCTL_NAMES { \
    542 	{ 0, 0 }, \
    543 	{ "rfc1323",	CTLTYPE_INT }, \
    544 	{ "sendspace",	CTLTYPE_INT }, \
    545 	{ "recvspace",	CTLTYPE_INT }, \
    546 	{ "mssdflt",	CTLTYPE_INT }, \
    547 	{ "syn_cache_limit", CTLTYPE_INT }, \
    548 	{ "syn_bucket_limit", CTLTYPE_INT }, \
    549 	{ "syn_cache_interval", CTLTYPE_INT },\
    550 	{ "init_win", CTLTYPE_INT }, \
    551 	{ "mss_ifmtu", CTLTYPE_INT }, \
    552 	{ "sack", CTLTYPE_INT }, \
    553 	{ "win_scale", CTLTYPE_INT }, \
    554 	{ "timestamps", CTLTYPE_INT }, \
    555 	{ "compat_42", CTLTYPE_INT }, \
    556 	{ "cwm", CTLTYPE_INT }, \
    557 	{ "cwm_burstsize", CTLTYPE_INT }, \
    558 	{ "ack_on_push", CTLTYPE_INT }, \
    559 	{ "keepidle",	CTLTYPE_INT }, \
    560 	{ "keepintvl",	CTLTYPE_INT }, \
    561 	{ "keepcnt",	CTLTYPE_INT }, \
    562 	{ "slowhz",	CTLTYPE_INT }, \
    563 	{ "newreno",	CTLTYPE_INT }, \
    564 	{ "log_refused",CTLTYPE_INT }, \
    565 	{ "rstratelimit", CTLTYPE_INT }, \
    566 	{ "rstppslimit", CTLTYPE_INT }, \
    567 }
    568 
    569 #ifdef _KERNEL
    570 struct	inpcbtable tcbtable;	/* head of queue of active tcpcb's */
    571 #ifdef INET6
    572 extern struct in6pcb tcb6;
    573 #endif
    574 struct	tcpstat tcpstat;	/* tcp statistics */
    575 u_int32_t tcp_now;		/* for RFC 1323 timestamps */
    576 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
    577 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
    578 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
    579 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
    580 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
    581 extern	int tcp_mssdflt;	/* default seg size */
    582 extern	int tcp_init_win;	/* initial window */
    583 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
    584 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
    585 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
    586 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
    587 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
    588 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
    589 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
    590 extern	int tcp_syn_cache_interval; /* compressed state timer */
    591 extern	int tcp_log_refused;	/* log refused connections */
    592 
    593 extern	struct timeval tcp_rst_ratelim;
    594 extern	int tcp_rst_ppslim;
    595 
    596 extern	int tcp_syn_cache_size;
    597 extern	struct syn_cache_head tcp_syn_cache[];
    598 extern	u_long syn_cache_count;
    599 
    600 #define	TCPCTL_VARIABLES { \
    601 	{ 0 },					\
    602 	{ 1, 0, &tcp_do_rfc1323 },		\
    603 	{ 1, 0, &tcp_sendspace },		\
    604 	{ 1, 0, &tcp_recvspace },		\
    605 	{ 1, 0, &tcp_mssdflt },			\
    606 	{ 1, 0, &tcp_syn_cache_limit },		\
    607 	{ 1, 0, &tcp_syn_bucket_limit },	\
    608 	{ 1, 0, &tcp_syn_cache_interval },	\
    609 	{ 1, 0, &tcp_init_win },		\
    610 	{ 1, 0, &tcp_mss_ifmtu },		\
    611 	{ 1, 0, &tcp_do_sack },			\
    612 	{ 1, 0, &tcp_do_win_scale },		\
    613 	{ 1, 0, &tcp_do_timestamps },		\
    614 	{ 1, 0, &tcp_compat_42 },		\
    615 	{ 1, 0, &tcp_cwm },			\
    616 	{ 1, 0, &tcp_cwm_burstsize },		\
    617 	{ 1, 0, &tcp_ack_on_push },		\
    618 	{ 1, 0, &tcp_keepidle },		\
    619 	{ 1, 0, &tcp_keepintvl },		\
    620 	{ 1, 0, &tcp_keepcnt },			\
    621 	{ 1, 1, 0, PR_SLOWHZ },			\
    622 	{ 1, 0, &tcp_do_newreno },		\
    623 	{ 1, 0, &tcp_log_refused },		\
    624 	{ 1, 1, 0, 0 },				\
    625 	{ 1, 0, &tcp_rst_ppslim },		\
    626 }
    627 
    628 int	 tcp_attach __P((struct socket *));
    629 void	 tcp_canceltimers __P((struct tcpcb *));
    630 struct tcpcb *
    631 	 tcp_close __P((struct tcpcb *));
    632 #if defined(INET6) && !defined(TCP6)
    633 void	 tcp6_ctlinput __P((int, struct sockaddr *, void *));
    634 #endif
    635 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
    636 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
    637 struct tcpcb *
    638 	 tcp_disconnect __P((struct tcpcb *));
    639 struct tcpcb *
    640 	 tcp_drop __P((struct tcpcb *, int));
    641 void	 tcp_dooptions __P((struct tcpcb *,
    642 	    u_char *, int, struct tcphdr *, struct tcp_opt_info *));
    643 void	 tcp_drain __P((void));
    644 void	 tcp_established __P((struct tcpcb *));
    645 void	 tcp_fasttimo __P((void));
    646 void	 tcp_init __P((void));
    647 #if defined(INET6) && !defined(TCP6)
    648 int	 tcp6_input __P((struct mbuf **, int *, int));
    649 #endif
    650 void	 tcp_input __P((struct mbuf *, ...));
    651 u_long	 tcp_mss_to_advertise __P((const struct ifnet *, int));
    652 void	 tcp_mss_from_peer __P((struct tcpcb *, int));
    653 void	 tcp_mtudisc __P((struct inpcb *, int));
    654 #if defined(INET6) && !defined(TCP6)
    655 void	 tcp6_mtudisc __P((struct in6pcb *, int));
    656 #endif
    657 struct tcpcb *
    658 	 tcp_newtcpcb __P((int, void *));
    659 void	 tcp_notify __P((struct inpcb *, int));
    660 #if defined(INET6) && !defined(TCP6)
    661 void	 tcp6_notify __P((struct in6pcb *, int));
    662 #endif
    663 u_int	 tcp_optlen __P((struct tcpcb *));
    664 int	 tcp_output __P((struct tcpcb *));
    665 void	 tcp_pulloutofband __P((struct socket *,
    666 	    struct tcphdr *, struct mbuf *, int));
    667 void	 tcp_quench __P((struct inpcb *, int));
    668 #if defined(INET6) && !defined(TCP6)
    669 void	 tcp6_quench __P((struct in6pcb *, int));
    670 #endif
    671 int	 tcp_reass __P((struct tcpcb *, struct tcphdr *, struct mbuf *, int *));
    672 int	 tcp_respond __P((struct tcpcb *, struct mbuf *, struct mbuf *,
    673 	    struct tcphdr *, tcp_seq, tcp_seq, int));
    674 void	 tcp_rmx_rtt __P((struct tcpcb *));
    675 void	 tcp_setpersist __P((struct tcpcb *));
    676 void	 tcp_slowtimo __P((void));
    677 struct mbuf *
    678 	 tcp_template __P((struct tcpcb *));
    679 struct tcpcb *
    680 	 tcp_timers __P((struct tcpcb *, int));
    681 void	 tcp_trace __P((int, int, struct tcpcb *, struct mbuf *, int));
    682 struct tcpcb *
    683 	 tcp_usrclosed __P((struct tcpcb *));
    684 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
    685 int	 tcp_usrreq __P((struct socket *,
    686 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
    687 void	 tcp_xmit_timer __P((struct tcpcb *, int));
    688 tcp_seq  tcp_new_iss __P((void *, u_long, tcp_seq));
    689 
    690 int	 syn_cache_add __P((struct sockaddr *, struct sockaddr *,
    691 		struct tcphdr *, unsigned int, struct socket *,
    692 		struct mbuf *, u_char *, int, struct tcp_opt_info *));
    693 void	 syn_cache_unreach __P((struct sockaddr *, struct sockaddr *,
    694 	   struct tcphdr *));
    695 struct socket *syn_cache_get __P((struct sockaddr *, struct sockaddr *,
    696 		struct tcphdr *, unsigned int, unsigned int,
    697 		struct socket *so, struct mbuf *));
    698 void	 syn_cache_init __P((void));
    699 void	 syn_cache_insert __P((struct syn_cache *, struct tcpcb *));
    700 struct syn_cache *syn_cache_lookup __P((struct sockaddr *, struct sockaddr *,
    701 		struct syn_cache_head **));
    702 void	 syn_cache_reset __P((struct sockaddr *, struct sockaddr *,
    703 		struct tcphdr *));
    704 int	 syn_cache_respond __P((struct syn_cache *, struct mbuf *));
    705 void	 syn_cache_timer __P((void));
    706 void	 syn_cache_cleanup __P((struct tcpcb *));
    707 
    708 int	tcp_newreno __P((struct tcpcb *, struct tcphdr *));
    709 #endif
    710 
    711 #endif /* _NETINET_TCP_VAR_H_ */
    712