Home | History | Annotate | Line # | Download | only in netinet
tcp_var.h revision 1.56
      1 /*	$NetBSD: tcp_var.h,v 1.56 1998/12/18 21:38:03 thorpej Exp $	*/
      2 
      3 /*-
      4  * Copyright (c) 1997, 1998 The NetBSD Foundation, Inc.
      5  * All rights reserved.
      6  *
      7  * This code is derived from software contributed to The NetBSD Foundation
      8  * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
      9  * NASA Ames Research Center.
     10  *
     11  * Redistribution and use in source and binary forms, with or without
     12  * modification, are permitted provided that the following conditions
     13  * are met:
     14  * 1. Redistributions of source code must retain the above copyright
     15  *    notice, this list of conditions and the following disclaimer.
     16  * 2. Redistributions in binary form must reproduce the above copyright
     17  *    notice, this list of conditions and the following disclaimer in the
     18  *    documentation and/or other materials provided with the distribution.
     19  * 3. All advertising materials mentioning features or use of this software
     20  *    must display the following acknowledgement:
     21  *	This product includes software developed by the NetBSD
     22  *	Foundation, Inc. and its contributors.
     23  * 4. Neither the name of The NetBSD Foundation nor the names of its
     24  *    contributors may be used to endorse or promote products derived
     25  *    from this software without specific prior written permission.
     26  *
     27  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
     28  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
     29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
     30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
     31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
     32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
     33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
     34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
     35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
     36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
     37  * POSSIBILITY OF SUCH DAMAGE.
     38  */
     39 
     40 /*
     41  * Copyright (c) 1982, 1986, 1993, 1994, 1995
     42  *	The Regents of the University of California.  All rights reserved.
     43  *
     44  * Redistribution and use in source and binary forms, with or without
     45  * modification, are permitted provided that the following conditions
     46  * are met:
     47  * 1. Redistributions of source code must retain the above copyright
     48  *    notice, this list of conditions and the following disclaimer.
     49  * 2. Redistributions in binary form must reproduce the above copyright
     50  *    notice, this list of conditions and the following disclaimer in the
     51  *    documentation and/or other materials provided with the distribution.
     52  * 3. All advertising materials mentioning features or use of this software
     53  *    must display the following acknowledgement:
     54  *	This product includes software developed by the University of
     55  *	California, Berkeley and its contributors.
     56  * 4. Neither the name of the University nor the names of its contributors
     57  *    may be used to endorse or promote products derived from this software
     58  *    without specific prior written permission.
     59  *
     60  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     61  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     62  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     63  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     64  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     65  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     66  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     67  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     68  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     69  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     70  * SUCH DAMAGE.
     71  *
     72  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
     73  */
     74 
     75 #ifndef _NETINET_TCP_VAR_H_
     76 #define _NETINET_TCP_VAR_H_
     77 
     78 /*
     79  * Kernel variables for tcp.
     80  */
     81 
     82 /*
     83  * Tcp control block, one per tcp; fields:
     84  */
     85 struct tcpcb {
     86 	struct ipqehead segq;		/* sequencing queue */
     87 	u_int	t_timer[TCPT_NTIMERS];	/* tcp timers */
     88 	short	t_state;		/* state of this connection */
     89 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
     90 	short	t_rxtcur;		/* current retransmit value */
     91 	short	t_dupacks;		/* consecutive dup acks recd */
     92 	u_short	t_peermss;		/* peer's maximum segment size */
     93 	u_short	t_ourmss;		/* our's maximum segment size */
     94 	u_short t_segsz;		/* current segment size in use */
     95 	char	t_force;		/* 1 if forcing out a byte */
     96 	u_short	t_flags;
     97 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
     98 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
     99 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
    100 #define	TF_NOOPT	0x0008		/* don't use tcp options */
    101 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
    102 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
    103 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
    104 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
    105 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
    106 #define	TF_SYN_REXMT	0x0400		/* rexmit timer fired on SYN */
    107 #define	TF_WILL_SACK	0x0800		/* try to use SACK */
    108 #define	TF_CANT_TXSACK	0x1000		/* other side said I could not SACK */
    109 #define	TF_IGNR_RXSACK	0x2000		/* ignore received SACK blocks */
    110 #define	TF_REASSEMBLING	0x4000		/* we're busy reassembling */
    111 
    112 
    113 	struct	tcpiphdr *t_template;	/* skeletal packet for transmit */
    114 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
    115 	LIST_ENTRY(tcpcb) t_delack;	/* delayed ACK queue */
    116 /*
    117  * The following fields are used as in the protocol specification.
    118  * See RFC783, Dec. 1981, page 21.
    119  */
    120 /* send sequence variables */
    121 	tcp_seq	snd_una;		/* send unacknowledged */
    122 	tcp_seq	snd_nxt;		/* send next */
    123 	tcp_seq	snd_up;			/* send urgent pointer */
    124 	tcp_seq	snd_wl1;		/* window update seg seq number */
    125 	tcp_seq	snd_wl2;		/* window update seg ack number */
    126 	tcp_seq	iss;			/* initial send sequence number */
    127 	u_long	snd_wnd;		/* send window */
    128 	tcp_seq snd_recover;		/* for use in fast recovery */
    129 /* receive sequence variables */
    130 	u_long	rcv_wnd;		/* receive window */
    131 	tcp_seq	rcv_nxt;		/* receive next */
    132 	tcp_seq	rcv_up;			/* receive urgent pointer */
    133 	tcp_seq	irs;			/* initial receive sequence number */
    134 /*
    135  * Additional variables for this implementation.
    136  */
    137 /* receive variables */
    138 	tcp_seq	rcv_adv;		/* advertised window */
    139 /* retransmit variables */
    140 	tcp_seq	snd_max;		/* highest sequence number sent;
    141 					 * used to recognize retransmits
    142 					 */
    143 /* congestion control (for slow start, source quench, retransmit after loss) */
    144 	u_long	snd_cwnd;		/* congestion-controlled window */
    145 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
    146 					 * for slow start exponential to
    147 					 * linear switch
    148 					 */
    149 /*
    150  * transmit timing stuff.  See below for scale of srtt and rttvar.
    151  * "Variance" is actually smoothed difference.
    152  */
    153 	short	t_idle;			/* inactivity time */
    154 	short	t_rtt;			/* round trip time */
    155 	tcp_seq	t_rtseq;		/* sequence number being timed */
    156 	short	t_srtt;			/* smoothed round-trip time */
    157 	short	t_rttvar;		/* variance in round-trip time */
    158 	short	t_rttmin;		/* minimum rtt allowed */
    159 	u_long	max_sndwnd;		/* largest window peer has offered */
    160 
    161 /* out-of-band data */
    162 	char	t_oobflags;		/* have some */
    163 	char	t_iobc;			/* input character */
    164 #define	TCPOOB_HAVEDATA	0x01
    165 #define	TCPOOB_HADDATA	0x02
    166 	short	t_softerror;		/* possible error not yet reported */
    167 
    168 /* RFC 1323 variables */
    169 	u_char	snd_scale;		/* window scaling for send window */
    170 	u_char	rcv_scale;		/* window scaling for recv window */
    171 	u_char	request_r_scale;	/* pending window scaling */
    172 	u_char	requested_s_scale;
    173 	u_int32_t ts_recent;		/* timestamp echo data */
    174 	u_int32_t ts_recent_age;	/* when last updated */
    175 	tcp_seq	last_ack_sent;
    176 
    177 /* SACK stuff */
    178 	struct ipqehead timeq;		/* time sequenced queue (for SACK) */
    179 };
    180 
    181 #ifdef _KERNEL
    182 /*
    183  * TCP reassembly queue locks.
    184  */
    185 static __inline int tcp_reass_lock_try __P((struct tcpcb *))
    186 	__attribute__((__unused__));
    187 static __inline void tcp_reass_unlock __P((struct tcpcb *))
    188 	__attribute__((__unused__));
    189 
    190 static __inline int
    191 tcp_reass_lock_try(tp)
    192 	struct tcpcb *tp;
    193 {
    194 	int s;
    195 
    196 	s = splimp();
    197 	if (tp->t_flags & TF_REASSEMBLING) {
    198 		splx(s);
    199 		return (0);
    200 	}
    201 	tp->t_flags |= TF_REASSEMBLING;
    202 	splx(s);
    203 	return (1);
    204 }
    205 
    206 static __inline void
    207 tcp_reass_unlock(tp)
    208 	struct tcpcb *tp;
    209 {
    210 	int s;
    211 
    212 	s = splimp();
    213 	tp->t_flags &= ~TF_REASSEMBLING;
    214 	splx(s);
    215 }
    216 
    217 #ifdef DIAGNOSTIC
    218 #define	TCP_REASS_LOCK(tp)						\
    219 do {									\
    220 	if (tcp_reass_lock_try(tp) == 0) {				\
    221 		printf("%s:%d: tcpcb %p reass already locked\n",	\
    222 		    __FILE__, __LINE__, tp);				\
    223 		panic("tcp_reass_lock");				\
    224 	}								\
    225 } while (0)
    226 #define	TCP_REASS_LOCK_CHECK(tp)					\
    227 do {									\
    228 	if (((tp)->t_flags & TF_REASSEMBLING) == 0) {			\
    229 		printf("%s:%d: tcpcb %p reass lock not held\n",		\
    230 		    __FILE__, __LINE__, tp);				\
    231 		panic("tcp reass lock check");				\
    232 	}								\
    233 } while (0)
    234 #else
    235 #define	TCP_REASS_LOCK(tp)	(void) tcp_reass_lock_try((tp))
    236 #define	TCP_REASS_LOCK_CHECK(tp) /* nothing */
    237 #endif
    238 
    239 #define	TCP_REASS_UNLOCK(tp)	tcp_reass_unlock((tp))
    240 #endif /* _KERNEL */
    241 
    242 /*
    243  * Queue for delayed ACK processing.
    244  */
    245 LIST_HEAD(tcp_delack_head, tcpcb);
    246 #ifdef _KERNEL
    247 extern struct tcp_delack_head tcp_delacks;
    248 
    249 #define	TCP_SET_DELACK(tp) \
    250 do { \
    251 	if (((tp)->t_flags & TF_DELACK) == 0) { \
    252 		(tp)->t_flags |= TF_DELACK; \
    253 		LIST_INSERT_HEAD(&tcp_delacks, (tp), t_delack); \
    254 	} \
    255 } while (0)
    256 
    257 #define	TCP_CLEAR_DELACK(tp) \
    258 do { \
    259 	if ((tp)->t_flags & TF_DELACK) { \
    260 		(tp)->t_flags &= ~TF_DELACK; \
    261 		LIST_REMOVE((tp), t_delack); \
    262 	} \
    263 } while (0)
    264 #endif /* _KERNEL */
    265 
    266 /*
    267  * Handy way of passing around TCP option info.
    268  */
    269 struct tcp_opt_info {
    270 	int		ts_present;
    271 	u_int32_t	ts_val;
    272 	u_int32_t	ts_ecr;
    273 	u_int16_t	maxseg;
    274 };
    275 
    276 /*
    277  * Data for the TCP compressed state engine.
    278  */
    279 struct syn_cache {
    280 	TAILQ_ENTRY(syn_cache) sc_queue;
    281 	u_int sc_timer;
    282 	u_int32_t sc_hash;
    283 	struct in_addr sc_src;
    284 	struct in_addr sc_dst;
    285 	tcp_seq sc_irs;
    286 	tcp_seq sc_iss;
    287 	u_short sc_rexmt_count;			/* number of times SYN,ACK
    288 						   was retransmitted */
    289 	u_short sc_flags;
    290 
    291 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
    292 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
    293 
    294 	struct mbuf *sc_ipopts;			/* IP options */
    295 	u_int16_t sc_sport;
    296 	u_int16_t sc_dport;
    297 	u_int16_t sc_peermaxseg;
    298 	u_int16_t sc_ourmaxseg;
    299 	u_int8_t sc_request_r_scale	: 4,
    300 		 sc_requested_s_scale	: 4;
    301 };
    302 
    303 struct syn_cache_head {
    304 	TAILQ_HEAD(, syn_cache) sch_queue;	/* time-sorted entries */
    305 	LIST_ENTRY(syn_cache_head) sch_headq;	/* non-empty buckets */
    306 	u_short sch_length;			/* # entries in bucket */
    307 };
    308 
    309 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
    310 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
    311 
    312 /*
    313  * The smoothed round-trip time and estimated variance
    314  * are stored as fixed point numbers scaled by the values below.
    315  * For convenience, these scales are also used in smoothing the average
    316  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
    317  * With these scales, srtt has 3 bits to the right of the binary point,
    318  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
    319  * binary point, and is smoothed with an ALPHA of 0.75.
    320  */
    321 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
    322 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
    323 
    324 /*
    325  * The initial retransmission should happen at rtt + 4 * rttvar.
    326  * Because of the way we do the smoothing, srtt and rttvar
    327  * will each average +1/2 tick of bias.  When we compute
    328  * the retransmit timer, we want 1/2 tick of rounding and
    329  * 1 extra tick because of +-1/2 tick uncertainty in the
    330  * firing of the timer.  The bias will give us exactly the
    331  * 1.5 tick we need.  But, because the bias is
    332  * statistical, we have to test that we don't drop below
    333  * the minimum feasible timer (which is 2 ticks).
    334  * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
    335  * is the same as the multiplier for rttvar.
    336  */
    337 #define	TCP_REXMTVAL(tp) \
    338 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
    339 
    340 /*
    341  * Compute the initial window for slow start.
    342  */
    343 #define	TCP_INITIAL_WINDOW(iw, segsz) \
    344 	(((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
    345 	 ((segsz) * (iw)))
    346 
    347 /*
    348  * TCP statistics.
    349  * Many of these should be kept per connection,
    350  * but that's inconvenient at the moment.
    351  */
    352 struct	tcpstat {
    353 	u_long	tcps_connattempt;	/* connections initiated */
    354 	u_long	tcps_accepts;		/* connections accepted */
    355 	u_long	tcps_connects;		/* connections established */
    356 	u_long	tcps_drops;		/* connections dropped */
    357 	u_long	tcps_conndrops;		/* embryonic connections dropped */
    358 	u_long	tcps_closed;		/* conn. closed (includes drops) */
    359 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
    360 	u_long	tcps_rttupdated;	/* times we succeeded */
    361 	u_long	tcps_delack;		/* delayed acks sent */
    362 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
    363 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
    364 	u_long	tcps_persisttimeo;	/* persist timeouts */
    365 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
    366 	u_long	tcps_keepprobe;		/* keepalive probes sent */
    367 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
    368 	u_long	tcps_persistdrops;	/* connections dropped in persist */
    369 	u_long	tcps_connsdrained;	/* connections drained due to memory
    370 					   shortage */
    371 
    372 	u_long	tcps_sndtotal;		/* total packets sent */
    373 	u_long	tcps_sndpack;		/* data packets sent */
    374 	u_long	tcps_sndbyte;		/* data bytes sent */
    375 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
    376 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
    377 	u_long	tcps_sndacks;		/* ack-only packets sent */
    378 	u_long	tcps_sndprobe;		/* window probes sent */
    379 	u_long	tcps_sndurg;		/* packets sent with URG only */
    380 	u_long	tcps_sndwinup;		/* window update-only packets sent */
    381 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
    382 
    383 	u_long	tcps_rcvtotal;		/* total packets received */
    384 	u_long	tcps_rcvpack;		/* packets received in sequence */
    385 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
    386 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
    387 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
    388 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
    389 	u_long	tcps_rcvshort;		/* packets received too short */
    390 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
    391 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
    392 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
    393 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
    394 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
    395 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
    396 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
    397 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
    398 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
    399 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
    400 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
    401 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
    402 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
    403 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
    404 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
    405 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
    406 	u_long	tcps_predack;		/* times hdr predict ok for acks */
    407 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
    408 
    409 	u_long	tcps_pcbhashmiss;	/* input packets missing pcb hash */
    410 	u_long	tcps_noport;		/* no socket on port */
    411 	u_long	tcps_badsyn;		/* received ack for which we have
    412 					   no SYN in compressed state */
    413 
    414 	/* These statistics deal with the SYN cache. */
    415 	u_long	tcps_sc_added;		/* # of entries added */
    416 	u_long	tcps_sc_completed;	/* # of connections completed */
    417 	u_long	tcps_sc_timed_out;	/* # of entries timed out */
    418 	u_long	tcps_sc_overflowed;	/* # dropped due to overflow */
    419 	u_long	tcps_sc_reset;		/* # dropped due to RST */
    420 	u_long	tcps_sc_unreach;	/* # dropped due to ICMP unreach */
    421 	u_long	tcps_sc_bucketoverflow;	/* # dropped due to bucket overflow */
    422 	u_long	tcps_sc_aborted;	/* # of entries aborted (no mem) */
    423 	u_long	tcps_sc_dupesyn;	/* # of duplicate SYNs received */
    424 	u_long	tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
    425 	u_long	tcps_sc_collisions;	/* # of hash collisions */
    426 };
    427 
    428 /*
    429  * Names for TCP sysctl objects.
    430  */
    431 #define	TCPCTL_RFC1323		1	/* RFC1323 timestamps/scaling */
    432 #define	TCPCTL_SENDSPACE	2	/* default send buffer */
    433 #define	TCPCTL_RECVSPACE	3	/* default recv buffer */
    434 #define	TCPCTL_MSSDFLT		4	/* default seg size */
    435 #define	TCPCTL_SYN_CACHE_LIMIT	5	/* max size of comp. state engine */
    436 #define	TCPCTL_SYN_BUCKET_LIMIT	6	/* max size of hash bucket */
    437 #define	TCPCTL_SYN_CACHE_INTER	7	/* interval of comp. state timer */
    438 #define	TCPCTL_INIT_WIN		8	/* initial window */
    439 #define	TCPCTL_MSS_IFMTU	9	/* mss from interface, not in_maxmtu */
    440 #define	TCPCTL_SACK		10	/* RFC2018 selective acknowledgement */
    441 #define	TCPCTL_WSCALE		11	/* RFC1323 window scaling */
    442 #define	TCPCTL_TSTAMP		12	/* RFC1323 timestamps */
    443 #define	TCPCTL_COMPAT_42	13	/* 4.2BSD TCP bug work-arounds */
    444 #define	TCPCTL_CWM		14	/* Congestion Window Monitoring */
    445 #define	TCPCTL_CWM_BURSTSIZE	15	/* burst size allowed by CWM */
    446 #define	TCPCTL_ACK_ON_PUSH	16	/* ACK immediately on PUSH */
    447 #define	TCPCTL_KEEPIDLE		17	/* keepalive idle time */
    448 #define	TCPCTL_KEEPINTVL	18	/* keepalive probe interval */
    449 #define	TCPCTL_KEEPCNT		19	/* keepalive count */
    450 #define	TCPCTL_SLOWHZ		20	/* PR_SLOWHZ (read-only) */
    451 #define	TCPCTL_NEWRENO		21	/* NewReno Congestion Control */
    452 #define	TCPCTL_MAXID		22
    453 
    454 #define	TCPCTL_NAMES { \
    455 	{ 0, 0 }, \
    456 	{ "rfc1323",	CTLTYPE_INT }, \
    457 	{ "sendspace",	CTLTYPE_INT }, \
    458 	{ "recvspace",	CTLTYPE_INT }, \
    459 	{ "mssdflt",	CTLTYPE_INT }, \
    460 	{ "syn_cache_limit", CTLTYPE_INT }, \
    461 	{ "syn_bucket_limit", CTLTYPE_INT }, \
    462 	{ "syn_cache_interval", CTLTYPE_INT },\
    463 	{ "init_win", CTLTYPE_INT }, \
    464 	{ "mss_ifmtu", CTLTYPE_INT }, \
    465 	{ "sack", CTLTYPE_INT }, \
    466 	{ "win_scale", CTLTYPE_INT }, \
    467 	{ "timestamps", CTLTYPE_INT }, \
    468 	{ "compat_42", CTLTYPE_INT }, \
    469 	{ "cwm", CTLTYPE_INT }, \
    470 	{ "cwm_burstsize", CTLTYPE_INT }, \
    471 	{ "ack_on_push", CTLTYPE_INT }, \
    472 	{ "keepidle",	CTLTYPE_INT }, \
    473 	{ "keepintvl",	CTLTYPE_INT }, \
    474 	{ "keepcnt",	CTLTYPE_INT }, \
    475 	{ "slowhz",	CTLTYPE_INT }, \
    476 	{ "newreno",	CTLTYPE_INT }, \
    477 }
    478 
    479 #ifdef _KERNEL
    480 struct	inpcbtable tcbtable;	/* head of queue of active tcpcb's */
    481 struct	tcpstat tcpstat;	/* tcp statistics */
    482 u_int32_t tcp_now;		/* for RFC 1323 timestamps */
    483 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
    484 extern	int tcp_do_sack;	/* SACK enabled/disabled? */
    485 extern	int tcp_do_win_scale;	/* RFC1323 window scaling enabled/disabled? */
    486 extern	int tcp_do_timestamps;	/* RFC1323 timestamps enabled/disabled? */
    487 extern	int tcp_do_newreno;	/* Use the New Reno algorithms */
    488 extern	int tcp_mssdflt;	/* default seg size */
    489 extern	int tcp_init_win;	/* initial window */
    490 extern	int tcp_mss_ifmtu;	/* take MSS from interface, not in_maxmtu */
    491 extern	int tcp_compat_42;	/* work around ancient broken TCP peers */
    492 extern	int tcp_cwm;		/* enable Congestion Window Monitoring */
    493 extern	int tcp_cwm_burstsize;	/* burst size allowed by CWM */
    494 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
    495 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
    496 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
    497 extern	int tcp_syn_cache_interval; /* compressed state timer */
    498 
    499 extern	int tcp_syn_cache_size;
    500 extern	int tcp_syn_cache_timeo;
    501 extern	struct syn_cache_head tcp_syn_cache[];
    502 extern	u_long syn_cache_count;
    503 
    504 #define	TCPCTL_VARIABLES { \
    505 	{ 0 },					\
    506 	{ 1, 0, &tcp_do_rfc1323 },		\
    507 	{ 1, 0, &tcp_sendspace },		\
    508 	{ 1, 0, &tcp_recvspace },		\
    509 	{ 1, 0, &tcp_mssdflt },			\
    510 	{ 1, 0, &tcp_syn_cache_limit },		\
    511 	{ 1, 0, &tcp_syn_bucket_limit },	\
    512 	{ 1, 0, &tcp_syn_cache_interval },	\
    513 	{ 1, 0, &tcp_init_win },		\
    514 	{ 1, 0, &tcp_mss_ifmtu },		\
    515 	{ 1, 0, &tcp_do_sack },			\
    516 	{ 1, 0, &tcp_do_win_scale },		\
    517 	{ 1, 0, &tcp_do_timestamps },		\
    518 	{ 1, 0, &tcp_compat_42 },		\
    519 	{ 1, 0, &tcp_cwm },			\
    520 	{ 1, 0, &tcp_cwm_burstsize },		\
    521 	{ 1, 0, &tcp_ack_on_push },		\
    522 	{ 1, 0, &tcp_keepidle },		\
    523 	{ 1, 0, &tcp_keepintvl },		\
    524 	{ 1, 0, &tcp_keepcnt },			\
    525 	{ 1, 1, 0, PR_SLOWHZ },			\
    526 	{ 1, 0, &tcp_do_newreno },		\
    527 }
    528 
    529 int	 tcp_attach __P((struct socket *));
    530 void	 tcp_canceltimers __P((struct tcpcb *));
    531 struct tcpcb *
    532 	 tcp_close __P((struct tcpcb *));
    533 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
    534 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
    535 struct tcpcb *
    536 	 tcp_disconnect __P((struct tcpcb *));
    537 struct tcpcb *
    538 	 tcp_drop __P((struct tcpcb *, int));
    539 void	 tcp_dooptions __P((struct tcpcb *,
    540 	    u_char *, int, struct tcpiphdr *, struct tcp_opt_info *));
    541 void	 tcp_drain __P((void));
    542 void	 tcp_established __P((struct tcpcb *));
    543 void	 tcp_fasttimo __P((void));
    544 void	 tcp_init __P((void));
    545 void	 tcp_input __P((struct mbuf *, ...));
    546 u_long	 tcp_mss_to_advertise __P((const struct ifnet *));
    547 void	 tcp_mss_from_peer __P((struct tcpcb *, int));
    548 void	 tcp_mtudisc __P((struct inpcb *, int));
    549 struct tcpcb *
    550 	 tcp_newtcpcb __P((struct inpcb *));
    551 void	 tcp_notify __P((struct inpcb *, int));
    552 u_int	 tcp_optlen __P((struct tcpcb *));
    553 int	 tcp_output __P((struct tcpcb *));
    554 void	 tcp_pulloutofband __P((struct socket *,
    555 	    struct tcpiphdr *, struct mbuf *));
    556 void	 tcp_quench __P((struct inpcb *, int));
    557 int	 tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *));
    558 int	 tcp_respond __P((struct tcpcb *,
    559 	    struct tcpiphdr *, struct mbuf *, tcp_seq, tcp_seq, int));
    560 void	 tcp_rmx_rtt __P((struct tcpcb *));
    561 void	 tcp_setpersist __P((struct tcpcb *));
    562 void	 tcp_slowtimo __P((void));
    563 struct tcpiphdr *
    564 	 tcp_template __P((struct tcpcb *));
    565 struct tcpcb *
    566 	 tcp_timers __P((struct tcpcb *, int));
    567 void	 tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int));
    568 struct tcpcb *
    569 	 tcp_usrclosed __P((struct tcpcb *));
    570 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
    571 int	 tcp_usrreq __P((struct socket *,
    572 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
    573 void	 tcp_xmit_timer __P((struct tcpcb *, int));
    574 tcp_seq  tcp_new_iss __P((void *, u_long, tcp_seq));
    575 
    576 int	 syn_cache_add __P((struct socket *, struct mbuf *, u_char *,
    577 	    int, struct tcp_opt_info *));
    578 void	 syn_cache_unreach __P((struct ip *, struct tcphdr *));
    579 struct socket *
    580 	 syn_cache_get __P((struct socket *so, struct mbuf *));
    581 void	 syn_cache_init __P((void));
    582 void	 syn_cache_insert __P((struct syn_cache *));
    583 struct syn_cache *
    584 	 syn_cache_lookup __P((struct tcpiphdr *, struct syn_cache_head **));
    585 void	 syn_cache_reset __P((struct tcpiphdr *));
    586 int	 syn_cache_respond __P((struct syn_cache *, struct mbuf *,
    587 	    register struct tcpiphdr *, long, u_long));
    588 void	 syn_cache_timer __P((void));
    589 
    590 int	tcp_newreno __P((struct tcpcb *, struct tcpiphdr *));
    591 #endif
    592 
    593 #endif /* _NETINET_TCP_VAR_H_ */
    594