1 /**	$MirOS: src/sys/netinet/tcp_var.h,v 1.5 2008/02/22 22:50:46 tg Exp $ */
2 /*	$OpenBSD: tcp_var.h,v 1.68 2004/11/25 15:32:08 markus Exp $	*/
3 /*	$NetBSD: tcp_var.h,v 1.17 1996/02/13 23:44:24 christos Exp $	*/
4 
5 /*
6  * Copyright (c) 1982, 1986, 1993, 1994
7  *	The Regents of the University of California.  All rights reserved.
8  *
9  * Redistribution and use in source and binary forms, with or without
10  * modification, are permitted provided that the following conditions
11  * are met:
12  * 1. Redistributions of source code must retain the above copyright
13  *    notice, this list of conditions and the following disclaimer.
14  * 2. Redistributions in binary form must reproduce the above copyright
15  *    notice, this list of conditions and the following disclaimer in the
16  *    documentation and/or other materials provided with the distribution.
17  * 3. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)tcp_var.h	8.3 (Berkeley) 4/10/94
34  */
35 
36 #ifndef _NETINET_TCP_VAR_H_
37 #define _NETINET_TCP_VAR_H_
38 
39 struct sackblk {
40 	tcp_seq start;		/* start seq no. of sack block */
41 	tcp_seq end; 		/* end seq no. */
42 };
43 
44 struct sackhole {
45 	tcp_seq start;		/* start seq no. of hole */
46 	tcp_seq end;		/* end seq no. */
47 	int	dups;		/* number of dup(s)acks for this hole */
48 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
49 	struct sackhole *next;	/* next in list */
50 };
51 
52 /*
53  * Kernel variables for tcp.
54  */
55 
56 /*
57  * Tcp control block, one per tcp; fields:
58  */
59 struct tcpcb {
60 	struct ipqehead segq;		/* sequencing queue */
61 	struct timeout t_timer[TCPT_NTIMERS];	/* tcp timers */
62 	short	t_state;		/* state of this connection */
63 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
64 	short	t_rxtcur;		/* current retransmit value */
65 	short	t_dupacks;		/* consecutive dup acks recd */
66 	u_short	t_maxseg;		/* maximum segment size */
67 	char	t_force;		/* 1 if forcing out a byte */
68 	u_int	t_flags;
69 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
70 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
71 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
72 #define	TF_NOOPT	0x0008		/* don't use tcp options */
73 #define	TF_SENTFIN	0x0010		/* have sent FIN */
74 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
75 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
76 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
77 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
78 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
79 #define	TF_SIGNATURE	0x0400		/* require TCP MD5 signature */
80 #ifdef TCP_ECN
81 #define TF_ECN_PERMIT	0x00008000	/* other side said I could ECN */
82 #define TF_RCVD_CE	0x00010000	/* send ECE in subsequent segs */
83 #define TF_SEND_CWR	0x00020000	/* send CWR in next seg */
84 #define TF_DISABLE_ECN	0x00040000	/* disable ECN for this connection */
85 #endif
86 #define TF_REASSLOCK	0x00080000	/* reassembling or draining */
87 #define TF_LASTIDLE	0x00100000	/* no outstanding ACK on last send */
88 #define TF_DEAD		0x00200000	/* dead and to-be-released */
89 #define TF_PMTUD_PEND	0x00400000	/* Path MTU Discovery pending */
90 
91 	struct	mbuf *t_template;	/* skeletal packet for transmit */
92 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
93 	struct	timeout t_delack_to;	/* delayed ACK callback */
94 /*
95  * The following fields are used as in the protocol specification.
96  * See RFC783, Dec. 1981, page 21.
97  */
98 /* send sequence variables */
99 	tcp_seq	snd_una;		/* send unacknowledged */
100 	tcp_seq	snd_nxt;		/* send next */
101 	tcp_seq	snd_up;			/* send urgent pointer */
102 	tcp_seq	snd_wl1;		/* window update seg seq number */
103 	tcp_seq	snd_wl2;		/* window update seg ack number */
104 	tcp_seq	iss;			/* initial send sequence number */
105 	u_long	snd_wnd;		/* send window */
106 #if 1 /*def TCP_SACK*/
107 	int	sack_enable;		/* enable SACK for this connection */
108 	int	snd_numholes;		/* number of holes seen by sender */
109 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
110 #if 1 /*defined(TCP_SACK) && defined(TCP_FACK)*/
111 	tcp_seq snd_fack;		/* for FACK congestion control */
112 	u_long	snd_awnd;		/* snd_nxt - snd_fack + */
113 					/* retransmitted data */
114 	int retran_data;		/* amount of outstanding retx. data  */
115 #endif /* TCP_FACK */
116 #endif /* TCP_SACK */
117 #if 1 /*defined(TCP_SACK) || defined(TCP_ECN)*/
118 	tcp_seq snd_last;		/* for use in fast recovery */
119 #endif
120 /* receive sequence variables */
121 	u_long	rcv_wnd;		/* receive window */
122 	tcp_seq	rcv_nxt;		/* receive next */
123 	tcp_seq	rcv_up;			/* receive urgent pointer */
124 	tcp_seq	irs;			/* initial receive sequence number */
125 #if 1 /*def TCP_SACK*/
126 	tcp_seq rcv_laststart;		/* start of last segment recd. */
127 	tcp_seq rcv_lastend;		/* end of ... */
128 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
129 	int	rcv_numsacks;		/* # distinct sack blks present */
130 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
131 #endif
132 
133 /*
134  * Additional variables for this implementation.
135  */
136 /* receive variables */
137 	tcp_seq	rcv_adv;		/* advertised window */
138 /* retransmit variables */
139 	tcp_seq	snd_max;		/* highest sequence number sent;
140 					 * used to recognize retransmits
141 					 */
142 /* congestion control (for slow start, source quench, retransmit after loss) */
143 	u_long	snd_cwnd;		/* congestion-controlled window */
144 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
145 					 * for slow start exponential to
146 					 * linear switch
147 					 */
148 	u_short	t_maxopd;		/* mss plus options */
149 	u_short	t_peermss;		/* peer's maximum segment size */
150 
151 /*
152  * transmit timing stuff.  See below for scale of srtt and rttvar.
153  * "Variance" is actually smoothed difference.
154  */
155 	uint32_t t_rcvtime;		/* time last segment received */
156 	uint32_t t_rtttime;		/* time we started measuring rtt */
157 	tcp_seq	t_rtseq;		/* sequence number being timed */
158 	short	t_srtt;			/* smoothed round-trip time */
159 	short	t_rttvar;		/* variance in round-trip time */
160 	u_short	t_rttmin;		/* minimum rtt allowed */
161 	u_long	max_sndwnd;		/* largest window peer has offered */
162 
163 /* out-of-band data */
164 	char	t_oobflags;		/* have some */
165 	char	t_iobc;			/* input character */
166 #define	TCPOOB_HAVEDATA	0x01
167 #define	TCPOOB_HADDATA	0x02
168 	short	t_softerror;		/* possible error not yet reported */
169 
170 /* RFC 1323 variables */
171 	u_char	snd_scale;		/* window scaling for send window */
172 	u_char	rcv_scale;		/* window scaling for recv window */
173 	u_char	request_r_scale;	/* pending window scaling */
174 	u_char	requested_s_scale;
175 	u_int32_t ts_recent;		/* timestamp echo data */
176 	u_int32_t ts_recent_age;		/* when last updated */
177 	tcp_seq	last_ack_sent;
178 
179 /* pointer for syn cache entries*/
180 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
181 
182 /* TUBA stuff */
183 	caddr_t	t_tuba_pcb;		/* next level down pcb for TCP over z */
184 
185 /* Path-MTU Discovery Information */
186 	u_int	t_pmtud_mss_acked;	/* MSS acked, lower bound for MTU */
187 	u_int	t_pmtud_mtu_sent;	/* MTU used, upper bound for MTU */
188 	tcp_seq	t_pmtud_th_seq;		/* TCP SEQ from ICMP payload */
189 	u_int	t_pmtud_nextmtu;	/* Advertised Next-Hop MTU from ICMP */
190 	u_short	t_pmtud_ip_len;		/* IP length from ICMP payload */
191 	u_short	t_pmtud_ip_hl;		/* IP header length from ICMP payload */
192 
193 	int pf;
194 
195 	struct	timeout t_reap_to;	/* delayed cleanup timeout */
196 };
197 
198 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
199 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
200 
201 #ifdef _KERNEL
202 extern int tcp_delack_ticks;
203 void	tcp_delack(void *);
204 
205 #define TCP_INIT_DELACK(tp)						\
206 	timeout_set(&(tp)->t_delack_to, tcp_delack, tp)
207 
208 #define TCP_RESTART_DELACK(tp)						\
209 	timeout_add(&(tp)->t_delack_to, tcp_delack_ticks)
210 
211 #define	TCP_SET_DELACK(tp)						\
212 do {									\
213 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
214 		(tp)->t_flags |= TF_DELACK;				\
215 		TCP_RESTART_DELACK(tp);					\
216 	}								\
217 } while (/*CONSTCOND*/0)
218 
219 #define	TCP_CLEAR_DELACK(tp)						\
220 do {									\
221 	if ((tp)->t_flags & TF_DELACK) {				\
222 		(tp)->t_flags &= ~TF_DELACK;				\
223 		timeout_del(&(tp)->t_delack_to);			\
224 	}								\
225 } while (/*CONSTCOND*/0)
226 
227 /*
228  * Handy way of passing around TCP option info.
229  */
230 struct tcp_opt_info {
231 	int		ts_present;
232 	u_int32_t	ts_val;
233 	u_int32_t	ts_ecr;
234 	u_int16_t	maxseg;
235 };
236 
237 /*
238  * Data for the TCP compressed state engine.
239  */
240 union syn_cache_sa {
241 	struct sockaddr sa;
242 	struct sockaddr_in sin;
243 #if 1 /*def INET6*/
244 	struct sockaddr_in6 sin6;
245 #endif
246 };
247 
248 struct syn_cache {
249 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
250 	struct timeout sc_timer;		/* rexmt timer */
251 	union {					/* cached route */
252 		struct route route4;
253 #ifdef INET6
254 		struct route_in6 route6;
255 #endif
256 	} sc_route_u;
257 #define sc_route4	sc_route_u.route4
258 #ifdef INET6
259 #define sc_route6	sc_route_u.route6
260 #endif
261 	long sc_win;				/* advertised window */
262 	int sc_bucketidx;			/* our bucket index */
263 	u_int32_t sc_hash;
264 	u_int32_t sc_timestamp;			/* timestamp from SYN */
265 #if 0
266 	u_int32_t sc_timebase;			/* our local timebase */
267 #endif
268 	union syn_cache_sa sc_src;
269 	union syn_cache_sa sc_dst;
270 	tcp_seq sc_irs;
271 	tcp_seq sc_iss;
272 	u_int sc_rxtcur;			/* current rxt timeout */
273 	u_int sc_rxttot;			/* total time spend on queues */
274 	u_short sc_rxtshift;			/* for computing backoff */
275 	u_short sc_flags;
276 
277 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
278 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
279 #define	SCF_DEAD		0x0004		/* this entry to be released */
280 #define	SCF_SACK_PERMIT		0x0008		/* permit sack */
281 #define	SCF_ECN_PERMIT		0x0010		/* permit ecn */
282 #define	SCF_SIGNATURE		0x0020		/* enforce tcp signatures */
283 
284 	struct mbuf *sc_ipopts;			/* IP options */
285 	u_int16_t sc_peermaxseg;
286 	u_int16_t sc_ourmaxseg;
287 	u_int8_t sc_request_r_scale	: 4,
288 		 sc_requested_s_scale	: 4;
289 
290 	struct tcpcb *sc_tp;			/* tcb for listening socket */
291 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
292 };
293 
294 struct syn_cache_head {
295 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
296 	u_short sch_length;			/* # entries in bucket */
297 };
298 
299 static __inline int tcp_reass_lock_try(struct tcpcb *);
300 static __inline void tcp_reass_unlock(struct tcpcb *);
301 #define tcp_reass_lock(tp) tcp_reass_lock_try(tp)
302 
303 static __inline int
tcp_reass_lock_try(struct tcpcb * tp)304 tcp_reass_lock_try(struct tcpcb *tp)
305 {
306 	int s;
307 
308 	s = splimp();
309 	if (tp->t_flags & TF_REASSLOCK) {
310 		splx(s);
311 		return (0);
312 	}
313 	tp->t_flags |= TF_REASSLOCK;
314 	splx(s);
315 	return (1);
316 }
317 
318 static __inline void
tcp_reass_unlock(struct tcpcb * tp)319 tcp_reass_unlock(struct tcpcb *tp)
320 {
321 	int s;
322 
323 	s = splimp();
324 	tp->t_flags &= ~TF_REASSLOCK;
325 	splx(s);
326 }
327 #endif /* _KERNEL */
328 
329 /*
330  * The smoothed round-trip time and estimated variance
331  * are stored as fixed point numbers scaled by the values below.
332  * For convenience, these scales are also used in smoothing the average
333  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
334  * With these scales, srtt has 3 bits to the right of the binary point,
335  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
336  * binary point, and is smoothed with an ALPHA of 0.75.
337  */
338 #define	TCP_RTT_SCALE		8	/* multiplier for srtt; 3 bits frac. */
339 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
340 #define	TCP_RTTVAR_SCALE	4	/* multiplier for rttvar; 2 bits */
341 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
342 #define TCP_RTT_MAX		(1<<9)	/* maximum rtt */
343 
344 /*
345  * The initial retransmission should happen at rtt + 4 * rttvar.
346  * Because of the way we do the smoothing, srtt and rttvar
347  * will each average +1/2 tick of bias.  When we compute
348  * the retransmit timer, we want 1/2 tick of rounding and
349  * 1 extra tick because of +-1/2 tick uncertainty in the
350  * firing of the timer.  The bias will give us exactly the
351  * 1.5 tick we need.  But, because the bias is
352  * statistical, we have to test that we don't drop below
353  * the minimum feasible timer (which is 2 ticks).
354  * This macro assumes that the value of TCP_RTTVAR_SCALE
355  * is the same as the multiplier for rttvar.
356  */
357 #define	TCP_REXMTVAL(tp) \
358 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
359 
360 /*
361  * TCP statistics.
362  * Many of these should be kept per connection,
363  * but that's inconvenient at the moment.
364  */
365 struct	tcpstat {
366 	u_int32_t tcps_connattempt;	/* connections initiated */
367 	u_int32_t tcps_accepts;		/* connections accepted */
368 	u_int32_t tcps_connects;	/* connections established */
369 	u_int32_t tcps_drops;		/* connections dropped */
370 	u_int32_t tcps_conndrops;	/* embryonic connections dropped */
371 	u_int32_t tcps_closed;		/* conn. closed (includes drops) */
372 	u_int32_t tcps_segstimed;	/* segs where we tried to get rtt */
373 	u_int32_t tcps_rttupdated;	/* times we succeeded */
374 	u_int32_t tcps_delack;		/* delayed acks sent */
375 	u_int32_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
376 	u_int32_t tcps_rexmttimeo;	/* retransmit timeouts */
377 	u_int32_t tcps_persisttimeo;	/* persist timeouts */
378 	u_int32_t tcps_persistdrop;	/* connections dropped in persist */
379 	u_int32_t tcps_keeptimeo;	/* keepalive timeouts */
380 	u_int32_t tcps_keepprobe;	/* keepalive probes sent */
381 	u_int32_t tcps_keepdrops;	/* connections dropped in keepalive */
382 
383 	u_int32_t tcps_sndtotal;		/* total packets sent */
384 	u_int32_t tcps_sndpack;		/* data packets sent */
385 	u_int64_t tcps_sndbyte;		/* data bytes sent */
386 	u_int32_t tcps_sndrexmitpack;	/* data packets retransmitted */
387 	u_int64_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
388 	u_int64_t tcps_sndrexmitfast;	/* Fast retransmits */
389 	u_int32_t tcps_sndacks;		/* ack-only packets sent */
390 	u_int32_t tcps_sndprobe;	/* window probes sent */
391 	u_int32_t tcps_sndurg;		/* packets sent with URG only */
392 	u_int32_t tcps_sndwinup;	/* window update-only packets sent */
393 	u_int32_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
394 
395 	u_int32_t tcps_rcvtotal;	/* total packets received */
396 	u_int32_t tcps_rcvpack;		/* packets received in sequence */
397 	u_int64_t tcps_rcvbyte;		/* bytes received in sequence */
398 	u_int32_t tcps_rcvbadsum;	/* packets received with ccksum errs */
399 	u_int32_t tcps_rcvbadoff;	/* packets received with bad offset */
400 	u_int32_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
401 	u_int32_t tcps_rcvnosec;	/* packets dropped for lack of ipsec */
402 	u_int32_t tcps_rcvshort;	/* packets received too short */
403 	u_int32_t tcps_rcvduppack;	/* duplicate-only packets received */
404 	u_int64_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
405 	u_int32_t tcps_rcvpartduppack;	/* packets with some duplicate data */
406 	u_int64_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
407 	u_int32_t tcps_rcvoopack;	/* out-of-order packets received */
408 	u_int64_t tcps_rcvoobyte;	/* out-of-order bytes received */
409 	u_int32_t tcps_rcvpackafterwin;	/* packets with data after window */
410 	u_int64_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
411 	u_int32_t tcps_rcvafterclose;	/* packets rcvd after "close" */
412 	u_int32_t tcps_rcvwinprobe;	/* rcvd window probe packets */
413 	u_int32_t tcps_rcvdupack;	/* rcvd duplicate acks */
414 	u_int32_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
415 	u_int32_t tcps_rcvacktooold;	/* rcvd acks for old data */
416 	u_int32_t tcps_rcvackpack;	/* rcvd ack packets */
417 	u_int64_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
418 	u_int32_t tcps_rcvwinupd;	/* rcvd window update packets */
419 	u_int32_t tcps_pawsdrop;	/* segments dropped due to PAWS */
420 	u_int32_t tcps_predack;		/* times hdr predict ok for acks */
421 	u_int32_t tcps_preddat;		/* times hdr predict ok for data pkts */
422 
423 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
424 	u_int32_t tcps_noport;		/* no socket on port */
425 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
426 
427 	u_int32_t tcps_rcvbadsig;	/* rcvd bad/missing TCP signatures */
428 	u_int64_t tcps_rcvgoodsig;	/* rcvd good TCP signatures */
429 	u_int32_t tcps_inhwcsum;	/* input hardware-checksummed packets */
430 	u_int32_t tcps_outhwcsum;	/* output hardware-checksummed packets */
431 
432 	/* ECN stats */
433 	u_int32_t tcps_ecn_accepts;	/* ecn connections accepted */
434 	u_int32_t tcps_ecn_rcvece;	/* # of rcvd ece */
435 	u_int32_t tcps_ecn_rcvcwr;	/* # of rcvd cwr */
436 	u_int32_t tcps_ecn_rcvce;	/* # of rcvd ce in ip header */
437 	u_int32_t tcps_ecn_sndect;	/* # of cwr sent */
438 	u_int32_t tcps_ecn_sndece;	/* # of ece sent */
439 	u_int32_t tcps_ecn_sndcwr;	/* # of cwr sent */
440 	u_int32_t tcps_cwr_ecn;		/* # of cwnd reduced by ecn */
441 	u_int32_t tcps_cwr_frecovery;	/* # of cwnd reduced by fastrecovery */
442 	u_int32_t tcps_cwr_timeout;	/* # of cwnd reduced by timeout */
443 
444 	/* These statistics deal with the SYN cache. */
445 	u_int64_t tcps_sc_added;	/* # of entries added */
446 	u_int64_t tcps_sc_completed;	/* # of connections completed */
447 	u_int64_t tcps_sc_timed_out;	/* # of entries timed out */
448 	u_int64_t tcps_sc_overflowed;	/* # dropped due to overflow */
449 	u_int64_t tcps_sc_reset;	/* # dropped due to RST */
450 	u_int64_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
451 	u_int64_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
452 	u_int64_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
453 	u_int64_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
454 	u_int64_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
455 	u_int64_t tcps_sc_collisions;	/* # of hash collisions */
456 	u_int64_t tcps_sc_retransmitted;/* # of retransmissions */
457 
458 	u_int64_t tcps_conndrained;	/* # of connections drained */
459 };
460 
461 /*
462  * Names for TCP sysctl objects.
463  */
464 
465 #define	TCPCTL_RFC1323		1 /* enable/disable RFC1323 timestamps/scaling */
466 #define	TCPCTL_KEEPINITTIME	2 /* TCPT_KEEP value */
467 #define TCPCTL_KEEPIDLE		3 /* allow tcp_keepidle to be changed */
468 #define TCPCTL_KEEPINTVL	4 /* allow tcp_keepintvl to be changed */
469 #define TCPCTL_SLOWHZ		5 /* return kernel idea of PR_SLOWHZ */
470 #define TCPCTL_BADDYNAMIC	6 /* return bad dynamic port bitmap */
471 #define	TCPCTL_RECVSPACE	7 /* receive buffer space */
472 #define	TCPCTL_SENDSPACE	8 /* send buffer space */
473 #define	TCPCTL_IDENT		9 /* get connection owner */
474 #define	TCPCTL_SACK	       10 /* selective acknowledgement, rfc 2018 */
475 #define TCPCTL_MSSDFLT	       11 /* Default maximum segment size */
476 #define	TCPCTL_RSTPPSLIMIT     12 /* RST pps limit */
477 #define	TCPCTL_ACK_ON_PUSH     13 /* ACK immediately on PUSH */
478 #define	TCPCTL_ECN	       14 /* RFC3168 ECN */
479 #define	TCPCTL_SYN_CACHE_LIMIT 15 /* max size of comp. state engine */
480 #define	TCPCTL_SYN_BUCKET_LIMIT	16 /* max size of hash bucket */
481 #define	TCPCTL_RFC3390	       17 /* enable/disable RFC3390 increased cwnd */
482 #define	TCPCTL_REASS_LIMIT     18 /* max entries for tcp reass queues */
483 #define	TCPCTL_DROP	       19 /* drop tcp connection */
484 #define	TCPCTL_SACKHOLE_LIMIT  20 /* max entries for tcp sack queues */
485 #define	TCPCTL_MAXID	       21
486 
487 #define	TCPCTL_NAMES { \
488 	{ 0, 0 }, \
489 	{ "rfc1323",	CTLTYPE_INT }, \
490 	{ "keepinittime",	CTLTYPE_INT }, \
491 	{ "keepidle",	CTLTYPE_INT }, \
492 	{ "keepintvl",	CTLTYPE_INT }, \
493 	{ "slowhz",	CTLTYPE_INT }, \
494 	{ "baddynamic", CTLTYPE_STRUCT }, \
495 	{ "recvspace",	CTLTYPE_INT }, \
496 	{ "sendspace",	CTLTYPE_INT }, \
497 	{ "ident", 	CTLTYPE_STRUCT }, \
498 	{ "sack",	CTLTYPE_INT }, \
499 	{ "mssdflt",	CTLTYPE_INT }, \
500 	{ "rstppslimit",	CTLTYPE_INT }, \
501 	{ "ackonpush",	CTLTYPE_INT }, \
502 	{ "ecn", 	CTLTYPE_INT }, \
503 	{ "syncachelimit", 	CTLTYPE_INT }, \
504 	{ "synbucketlimit", 	CTLTYPE_INT }, \
505 	{ "rfc3390", 	CTLTYPE_INT }, \
506 	{ "reasslimit", 	CTLTYPE_INT }, \
507 	{ "drop", 	CTLTYPE_STRUCT }, \
508 	{ "sackholelimit", 	CTLTYPE_INT }, \
509 }
510 
511 #define	TCPCTL_VARS { \
512 	NULL, \
513 	&tcp_do_rfc1323, \
514 	&tcptv_keep_init, \
515 	&tcp_keepidle, \
516 	&tcp_keepintvl, \
517 	NULL, \
518 	NULL, \
519 	&tcp_recvspace, \
520 	&tcp_sendspace, \
521 	NULL, \
522 	NULL, \
523 	&tcp_mssdflt, \
524 	&tcp_rst_ppslim, \
525 	&tcp_ack_on_push, \
526 	NULL, \
527 	&tcp_syn_cache_limit, \
528 	&tcp_syn_bucket_limit, \
529 	&tcp_do_rfc3390, \
530 	NULL, \
531 	NULL, \
532 	NULL \
533 }
534 
535 struct tcp_ident_mapping {
536 	struct sockaddr_storage faddr, laddr;
537 	int euid, ruid;
538 };
539 
540 #ifdef _KERNEL
541 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
542 extern	struct tcpstat tcpstat;	/* tcp statistics */
543 extern	u_int32_t tcp_now;		/* for RFC 1323 timestamps */
544 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
545 extern	int tcp_mssdflt;	/* default maximum segment size */
546 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
547 #ifdef TCP_SACK
548 extern	int tcp_do_sack;	/* SACK enabled/disabled */
549 extern	struct pool sackhl_pool;
550 extern	int tcp_sackhole_limit;	/* max entries for tcp sack queues */
551 #endif
552 extern	int tcp_do_ecn;		/* RFC3168 ECN enabled/disabled? */
553 extern	int tcp_do_rfc3390;	/* RFC3390 Increasing TCP's Initial Window */
554 
555 extern	struct pool tcpqe_pool;
556 extern	int tcp_reass_limit;	/* max entries for tcp reass queues */
557 
558 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
559 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
560 
561 extern	int tcp_syn_cache_size;
562 extern	struct syn_cache_head tcp_syn_cache[];
563 extern	u_long syn_cache_count;
564 
565 extern	struct pool tcpqe_pool;
566 extern	int tcp_reass_limit;	/* max entries for tcp reass queues */
567 
568 int	 tcp_attach(struct socket *);
569 void	 tcp_canceltimers(struct tcpcb *);
570 struct tcpcb *
571 	 tcp_close(struct tcpcb *);
572 void	 tcp_reaper(void *);
573 int	 tcp_freeq(struct tcpcb *);
574 #if defined(INET6) && !defined(TCP6)
575 void	 tcp6_ctlinput(int, struct sockaddr *, void *);
576 #endif
577 void	 *tcp_ctlinput(int, struct sockaddr *, void *);
578 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf **);
579 struct tcpcb *
580 	 tcp_disconnect(struct tcpcb *);
581 struct tcpcb *
582 	 tcp_drop(struct tcpcb *, int);
583 int	 tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
584 		struct mbuf *, int, struct tcp_opt_info *);
585 void	 tcp_drain(void);
586 void	 tcp_init(void);
587 #if defined(INET6) && !defined(TCP6)
588 int	 tcp6_input(struct mbuf **, int *, int);
589 #endif
590 void	 tcp_input(struct mbuf *, ...);
591 int	 tcp_mss(struct tcpcb *, int);
592 void	 tcp_mss_update(struct tcpcb *);
593 u_int	 tcp_hdrsz(struct tcpcb *);
594 void	 tcp_mtudisc(struct inpcb *, int);
595 void	 tcp_mtudisc_increase(struct inpcb *, int);
596 #ifdef INET6
597 void	tcp6_mtudisc(struct inpcb *, int);
598 void	tcp6_mtudisc_callback(struct in6_addr *);
599 #endif
600 struct tcpcb *
601 	 tcp_newtcpcb(struct inpcb *);
602 void	 tcp_notify(struct inpcb *, int);
603 int	 tcp_output(struct tcpcb *);
604 void	 tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
605 void	 tcp_quench(struct inpcb *, int);
606 int	 tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
607 void	 tcp_rscale(struct tcpcb *, u_long);
608 void	 tcp_respond(struct tcpcb *, caddr_t, struct tcphdr *, tcp_seq,
609 		tcp_seq, int);
610 void	 tcp_setpersist(struct tcpcb *);
611 void	 tcp_slowtimo(void);
612 struct mbuf *
613 	 tcp_template(struct tcpcb *);
614 void	 tcp_trace(int, int, struct tcpcb *, caddr_t, int, int);
615 struct tcpcb *
616 	 tcp_usrclosed(struct tcpcb *);
617 int	 tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
618 #if defined(INET6) && !defined(TCP6)
619 int	 tcp6_usrreq(struct socket *,
620 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *);
621 #endif
622 int	 tcp_usrreq(struct socket *,
623 	    int, struct mbuf *, struct mbuf *, struct mbuf *);
624 void	 tcp_xmit_timer(struct tcpcb *, int);
625 void	 tcpdropoldhalfopen(struct tcpcb *, u_int16_t);
626 #ifdef TCP_SACK
627 void	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
628 void	 tcp_update_sack_list(struct tcpcb *tp);
629 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
630 void	 tcp_clean_sackreport(struct tcpcb *tp);
631 void	 tcp_sack_adjust(struct tcpcb *tp);
632 struct sackhole *
633 	 tcp_sack_output(struct tcpcb *tp);
634 int	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
635 #ifdef DEBUG
636 void	 tcp_print_holes(struct tcpcb *tp);
637 #endif
638 #endif /* TCP_SACK */
639 #if defined(TCP_SACK)
640 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
641 u_long	 tcp_seq_subtract(u_long, u_long );
642 #endif /* TCP_SACK */
643 #ifdef TCP_SIGNATURE
644 int	tcp_signature_apply(caddr_t, caddr_t, unsigned int);
645 #endif /* TCP_SIGNATURE */
646 void	tcp_rndiss_init(void);
647 tcp_seq	tcp_rndiss_next(void);
648 u_int16_t
649 	tcp_rndiss_encrypt(u_int16_t);
650 
651 int	 syn_cache_add(struct sockaddr *, struct sockaddr *,
652 		struct tcphdr *, unsigned int, struct socket *,
653 		struct mbuf *, u_char *, int, struct tcp_opt_info *);
654 void	 syn_cache_unreach(struct sockaddr *, struct sockaddr *,
655 	   struct tcphdr *);
656 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
657 		struct tcphdr *, unsigned int, unsigned int,
658 		struct socket *so, struct mbuf *);
659 void	 syn_cache_init(void);
660 void	 syn_cache_insert(struct syn_cache *, struct tcpcb *);
661 struct syn_cache *syn_cache_lookup(struct sockaddr *, struct sockaddr *,
662 		struct syn_cache_head **);
663 void	 syn_cache_reset(struct sockaddr *, struct sockaddr *,
664 		struct tcphdr *);
665 int	 syn_cache_respond(struct syn_cache *, struct mbuf *);
666 void	 syn_cache_timer(void *);
667 void	 syn_cache_cleanup(struct tcpcb *);
668 void	 syn_cache_reaper(void *);
669 
670 #endif /* _KERNEL */
671 #endif /* _NETINET_TCP_VAR_H_ */
672