xref: /openbsd-src/sys/netinet/tcp_var.h (revision d13be5d47e4149db2549a9828e244d59dbc43f15)
1 /*	$OpenBSD: tcp_var.h,v 1.99 2011/07/06 23:44:20 sthen Exp $	*/
2 /*	$NetBSD: tcp_var.h,v 1.17 1996/02/13 23:44:24 christos Exp $	*/
3 
4 /*
5  * Copyright (c) 1982, 1986, 1993, 1994
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  *
32  *	@(#)tcp_var.h	8.3 (Berkeley) 4/10/94
33  */
34 
35 #ifndef _NETINET_TCP_VAR_H_
36 #define _NETINET_TCP_VAR_H_
37 
38 /*
39  * Kernel variables for tcp.
40  */
41 
42 struct sackblk {
43 	tcp_seq start;		/* start seq no. of sack block */
44 	tcp_seq end; 		/* end seq no. */
45 };
46 
47 struct sackhole {
48 	tcp_seq start;		/* start seq no. of hole */
49 	tcp_seq end;		/* end seq no. */
50 	int	dups;		/* number of dup(s)acks for this hole */
51 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
52 	struct sackhole *next;	/* next in list */
53 };
54 
55 /*
56  * TCP sequence queue structures.
57  */
58 TAILQ_HEAD(tcpqehead, tcpqent);
59 struct tcpqent {
60 	TAILQ_ENTRY(tcpqent) tcpqe_q;
61 	struct tcphdr	*tcpqe_tcp;
62 	struct mbuf	*tcpqe_m;	/* mbuf contains packet */
63 };
64 
65 /*
66  * Tcp control block, one per tcp; fields:
67  */
68 struct tcpcb {
69 	struct tcpqehead t_segq;		/* sequencing queue */
70 	struct timeout t_timer[TCPT_NTIMERS];	/* tcp timers */
71 	short	t_state;		/* state of this connection */
72 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
73 	short	t_rxtcur;		/* current retransmit value */
74 	short	t_dupacks;		/* consecutive dup acks recd */
75 	u_short	t_maxseg;		/* maximum segment size */
76 	char	t_force;		/* 1 if forcing out a byte */
77 	u_int	t_flags;
78 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
79 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
80 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
81 #define	TF_NOOPT	0x0008		/* don't use tcp options */
82 #define	TF_SENTFIN	0x0010		/* have sent FIN */
83 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
84 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
85 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
86 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
87 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
88 #define	TF_SIGNATURE	0x0400		/* require TCP MD5 signature */
89 #ifdef TCP_ECN
90 #define TF_ECN_PERMIT	0x00008000	/* other side said I could ECN */
91 #define TF_RCVD_CE	0x00010000	/* send ECE in subsequent segs */
92 #define TF_SEND_CWR	0x00020000	/* send CWR in next seg */
93 #define TF_DISABLE_ECN	0x00040000	/* disable ECN for this connection */
94 #endif
95 #define TF_LASTIDLE	0x00100000	/* no outstanding ACK on last send */
96 #define TF_DEAD		0x00200000	/* dead and to-be-released */
97 #define TF_PMTUD_PEND	0x00400000	/* Path MTU Discovery pending */
98 #define TF_NEEDOUTPUT	0x00800000	/* call tcp_output after tcp_input */
99 #define TF_BLOCKOUTPUT	0x01000000	/* avert tcp_output during tcp_input */
100 
101 	struct	mbuf *t_template;	/* skeletal packet for transmit */
102 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
103 	struct	timeout t_delack_to;	/* delayed ACK callback */
104 /*
105  * The following fields are used as in the protocol specification.
106  * See RFC793, Dec. 1981, page 21.
107  */
108 /* send sequence variables */
109 	tcp_seq	snd_una;		/* send unacknowledged */
110 	tcp_seq	snd_nxt;		/* send next */
111 	tcp_seq	snd_up;			/* send urgent pointer */
112 	tcp_seq	snd_wl1;		/* window update seg seq number */
113 	tcp_seq	snd_wl2;		/* window update seg ack number */
114 	tcp_seq	iss;			/* initial send sequence number */
115 	u_long	snd_wnd;		/* send window */
116 #if 1 /*def TCP_SACK*/
117 	int	sack_enable;		/* enable SACK for this connection */
118 	int	snd_numholes;		/* number of holes seen by sender */
119 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
120 #if 1 /*defined(TCP_SACK) && defined(TCP_FACK)*/
121 	tcp_seq snd_fack;		/* for FACK congestion control */
122 	u_long	snd_awnd;		/* snd_nxt - snd_fack + */
123 					/* retransmitted data */
124 	int retran_data;		/* amount of outstanding retx. data  */
125 #endif /* TCP_FACK */
126 #endif /* TCP_SACK */
127 #if 1 /*defined(TCP_SACK) || defined(TCP_ECN)*/
128 	tcp_seq snd_last;		/* for use in fast recovery */
129 #endif
130 /* receive sequence variables */
131 	u_long	rcv_wnd;		/* receive window */
132 	tcp_seq	rcv_nxt;		/* receive next */
133 	tcp_seq	rcv_up;			/* receive urgent pointer */
134 	tcp_seq	irs;			/* initial receive sequence number */
135 #if 1 /*def TCP_SACK*/
136 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
137 	int	rcv_numsacks;		/* # distinct sack blks present */
138 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
139 #endif
140 
141 /*
142  * Additional variables for this implementation.
143  */
144 /* receive variables */
145 	tcp_seq	rcv_adv;		/* advertised window */
146 /* retransmit variables */
147 	tcp_seq	snd_max;		/* highest sequence number sent;
148 					 * used to recognize retransmits
149 					 */
150 /* congestion control (for slow start, source quench, retransmit after loss) */
151 	u_long	snd_cwnd;		/* congestion-controlled window */
152 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
153 					 * for slow start exponential to
154 					 * linear switch
155 					 */
156 
157 /* auto-sizing variables */
158 	u_int	rfbuf_cnt;	/* recv buffer autoscaling byte count */
159 	u_int32_t rfbuf_ts;	/* recv buffer autoscaling time stamp */
160 
161 	u_short	t_maxopd;		/* mss plus options */
162 	u_short	t_peermss;		/* peer's maximum segment size */
163 
164 /*
165  * transmit timing stuff.  See below for scale of srtt and rttvar.
166  * "Variance" is actually smoothed difference.
167  */
168 	uint32_t t_rcvtime;		/* time last segment received */
169 	uint32_t t_rtttime;		/* time we started measuring rtt */
170 	tcp_seq	t_rtseq;		/* sequence number being timed */
171 	short	t_srtt;			/* smoothed round-trip time */
172 	short	t_rttvar;		/* variance in round-trip time */
173 	u_short	t_rttmin;		/* minimum rtt allowed */
174 	u_long	max_sndwnd;		/* largest window peer has offered */
175 
176 /* out-of-band data */
177 	char	t_oobflags;		/* have some */
178 	char	t_iobc;			/* input character */
179 #define	TCPOOB_HAVEDATA	0x01
180 #define	TCPOOB_HADDATA	0x02
181 	short	t_softerror;		/* possible error not yet reported */
182 
183 /* RFC 1323 variables */
184 	u_char	snd_scale;		/* window scaling for send window */
185 	u_char	rcv_scale;		/* window scaling for recv window */
186 	u_char	request_r_scale;	/* pending window scaling */
187 	u_char	requested_s_scale;
188 	u_int32_t ts_recent;		/* timestamp echo data */
189 	u_int32_t ts_modulate;		/* modulation on timestamp */
190 	u_int32_t ts_recent_age;		/* when last updated */
191 	tcp_seq	last_ack_sent;
192 
193 /* pointer for syn cache entries*/
194 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
195 
196 /* Path-MTU Discovery Information */
197 	u_int	t_pmtud_mss_acked;	/* MSS acked, lower bound for MTU */
198 	u_int	t_pmtud_mtu_sent;	/* MTU used, upper bound for MTU */
199 	tcp_seq	t_pmtud_th_seq;		/* TCP SEQ from ICMP payload */
200 	u_int	t_pmtud_nextmtu;	/* Advertised Next-Hop MTU from ICMP */
201 	u_short	t_pmtud_ip_len;		/* IP length from ICMP payload */
202 	u_short	t_pmtud_ip_hl;		/* IP header length from ICMP payload */
203 
204 	int pf;
205 
206 	struct	timeout t_reap_to;	/* delayed cleanup timeout */
207 };
208 
209 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
210 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
211 
212 #ifdef _KERNEL
213 extern int tcp_delack_ticks;
214 void	tcp_delack(void *);
215 
216 #define TCP_INIT_DELACK(tp)						\
217 	timeout_set(&(tp)->t_delack_to, tcp_delack, tp)
218 
219 #define TCP_RESTART_DELACK(tp)						\
220 	timeout_add(&(tp)->t_delack_to, tcp_delack_ticks)
221 
222 #define	TCP_SET_DELACK(tp)						\
223 do {									\
224 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
225 		(tp)->t_flags |= TF_DELACK;				\
226 		TCP_RESTART_DELACK(tp);					\
227 	}								\
228 } while (/* CONSTCOND */ 0)
229 
230 #define	TCP_CLEAR_DELACK(tp)						\
231 do {									\
232 	if ((tp)->t_flags & TF_DELACK) {				\
233 		(tp)->t_flags &= ~TF_DELACK;				\
234 		timeout_del(&(tp)->t_delack_to);			\
235 	}								\
236 } while (/* CONSTCOND */ 0)
237 
238 /*
239  * Handy way of passing around TCP option info.
240  */
241 struct tcp_opt_info {
242 	int		ts_present;
243 	u_int32_t	ts_val;
244 	u_int32_t	ts_ecr;
245 	u_int16_t	maxseg;
246 };
247 
248 /*
249  * Data for the TCP compressed state engine.
250  */
251 union syn_cache_sa {
252 	struct sockaddr sa;
253 	struct sockaddr_in sin;
254 #if 1 /*def INET6*/
255 	struct sockaddr_in6 sin6;
256 #endif
257 };
258 
259 struct syn_cache {
260 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
261 	struct timeout sc_timer;		/* rexmt timer */
262 	union {					/* cached route */
263 		struct route route4;
264 #ifdef INET6
265 		struct route_in6 route6;
266 #endif
267 	} sc_route_u;
268 #define sc_route4	sc_route_u.route4
269 #ifdef INET6
270 #define sc_route6	sc_route_u.route6
271 #endif
272 	long sc_win;				/* advertised window */
273 	int sc_bucketidx;			/* our bucket index */
274 	u_int32_t sc_hash;
275 	u_int32_t sc_timestamp;			/* timestamp from SYN */
276 	u_int32_t sc_modulate;			/* our timestamp modulator */
277 #if 0
278 	u_int32_t sc_timebase;			/* our local timebase */
279 #endif
280 	union syn_cache_sa sc_src;
281 	union syn_cache_sa sc_dst;
282 	tcp_seq sc_irs;
283 	tcp_seq sc_iss;
284 	u_int sc_rtableid;
285 	u_int sc_rxtcur;			/* current rxt timeout */
286 	u_int sc_rxttot;			/* total time spend on queues */
287 	u_short sc_rxtshift;			/* for computing backoff */
288 	u_short sc_flags;
289 
290 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
291 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
292 #define	SCF_DEAD		0x0004		/* this entry to be released */
293 #define	SCF_SACK_PERMIT		0x0008		/* permit sack */
294 #define	SCF_ECN_PERMIT		0x0010		/* permit ecn */
295 #define	SCF_SIGNATURE		0x0020		/* enforce tcp signatures */
296 
297 	struct mbuf *sc_ipopts;			/* IP options */
298 	u_int16_t sc_peermaxseg;
299 	u_int16_t sc_ourmaxseg;
300 	u_int     sc_request_r_scale	: 4,
301 		  sc_requested_s_scale	: 4;
302 
303 	struct tcpcb *sc_tp;			/* tcb for listening socket */
304 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
305 };
306 
307 struct syn_cache_head {
308 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
309 	u_short sch_length;			/* # entries in bucket */
310 };
311 
312 #endif /* _KERNEL */
313 
314 /*
315  * The smoothed round-trip time and estimated variance
316  * are stored as fixed point numbers scaled by the values below.
317  * For convenience, these scales are also used in smoothing the average
318  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
319  * With these scales, srtt has 5 bits to the right of the binary point,
320  * and thus an "ALPHA" of 0.875.  rttvar has 4 bits to the right of the
321  * binary point, and is smoothed with an ALPHA of 0.75.
322  */
323 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 5 bits frac. */
324 #define	TCP_RTTVAR_SHIFT	2	/* shift for rttvar; 4 bits */
325 #define	TCP_RTT_BASE_SHIFT	2	/* remaining 2 bit shift */
326 #define	TCP_RTT_MAX		(1<<9)	/* maximum rtt */
327 
328 /*
329  * The initial retransmission should happen at rtt + 4 * rttvar.
330  * Because of the way we do the smoothing, srtt and rttvar
331  * will each average +1/2 tick of bias.  When we compute
332  * the retransmit timer, we want 1/2 tick of rounding and
333  * 1 extra tick because of +-1/2 tick uncertainty in the
334  * firing of the timer.  The bias will give us exactly the
335  * 1.5 tick we need.  But, because the bias is
336  * statistical, we have to test that we don't drop below
337  * the minimum feasible timer (which is 2 ticks).
338  * This macro assumes that the value of (1 << TCP_RTTVAR_SHIFT)
339  * is the same as the multiplier for rttvar.
340  */
341 #define	TCP_REXMTVAL(tp) \
342 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> TCP_RTT_BASE_SHIFT)
343 
344 /*
345  * TCP statistics.
346  * Many of these should be kept per connection,
347  * but that's inconvenient at the moment.
348  */
349 struct	tcpstat {
350 	u_int32_t tcps_connattempt;	/* connections initiated */
351 	u_int32_t tcps_accepts;		/* connections accepted */
352 	u_int32_t tcps_connects;	/* connections established */
353 	u_int32_t tcps_drops;		/* connections dropped */
354 	u_int32_t tcps_conndrops;	/* embryonic connections dropped */
355 	u_int32_t tcps_closed;		/* conn. closed (includes drops) */
356 	u_int32_t tcps_segstimed;	/* segs where we tried to get rtt */
357 	u_int32_t tcps_rttupdated;	/* times we succeeded */
358 	u_int32_t tcps_delack;		/* delayed acks sent */
359 	u_int32_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
360 	u_int32_t tcps_rexmttimeo;	/* retransmit timeouts */
361 	u_int32_t tcps_persisttimeo;	/* persist timeouts */
362 	u_int32_t tcps_persistdrop;	/* connections dropped in persist */
363 	u_int32_t tcps_keeptimeo;	/* keepalive timeouts */
364 	u_int32_t tcps_keepprobe;	/* keepalive probes sent */
365 	u_int32_t tcps_keepdrops;	/* connections dropped in keepalive */
366 
367 	u_int32_t tcps_sndtotal;		/* total packets sent */
368 	u_int32_t tcps_sndpack;		/* data packets sent */
369 	u_int64_t tcps_sndbyte;		/* data bytes sent */
370 	u_int32_t tcps_sndrexmitpack;	/* data packets retransmitted */
371 	u_int64_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
372 	u_int64_t tcps_sndrexmitfast;	/* Fast retransmits */
373 	u_int32_t tcps_sndacks;		/* ack-only packets sent */
374 	u_int32_t tcps_sndprobe;	/* window probes sent */
375 	u_int32_t tcps_sndurg;		/* packets sent with URG only */
376 	u_int32_t tcps_sndwinup;	/* window update-only packets sent */
377 	u_int32_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
378 
379 	u_int32_t tcps_rcvtotal;	/* total packets received */
380 	u_int32_t tcps_rcvpack;		/* packets received in sequence */
381 	u_int64_t tcps_rcvbyte;		/* bytes received in sequence */
382 	u_int32_t tcps_rcvbadsum;	/* packets received with ccksum errs */
383 	u_int32_t tcps_rcvbadoff;	/* packets received with bad offset */
384 	u_int32_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
385 	u_int32_t tcps_rcvnosec;	/* packets dropped for lack of ipsec */
386 	u_int32_t tcps_rcvshort;	/* packets received too short */
387 	u_int32_t tcps_rcvduppack;	/* duplicate-only packets received */
388 	u_int64_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
389 	u_int32_t tcps_rcvpartduppack;	/* packets with some duplicate data */
390 	u_int64_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
391 	u_int32_t tcps_rcvoopack;	/* out-of-order packets received */
392 	u_int64_t tcps_rcvoobyte;	/* out-of-order bytes received */
393 	u_int32_t tcps_rcvpackafterwin;	/* packets with data after window */
394 	u_int64_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
395 	u_int32_t tcps_rcvafterclose;	/* packets rcvd after "close" */
396 	u_int32_t tcps_rcvwinprobe;	/* rcvd window probe packets */
397 	u_int32_t tcps_rcvdupack;	/* rcvd duplicate acks */
398 	u_int32_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
399 	u_int32_t tcps_rcvacktooold;	/* rcvd acks for old data */
400 	u_int32_t tcps_rcvackpack;	/* rcvd ack packets */
401 	u_int64_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
402 	u_int32_t tcps_rcvwinupd;	/* rcvd window update packets */
403 	u_int32_t tcps_pawsdrop;	/* segments dropped due to PAWS */
404 	u_int32_t tcps_predack;		/* times hdr predict ok for acks */
405 	u_int32_t tcps_preddat;		/* times hdr predict ok for data pkts */
406 
407 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
408 	u_int32_t tcps_noport;		/* no socket on port */
409 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
410 
411 	u_int32_t tcps_rcvbadsig;	/* rcvd bad/missing TCP signatures */
412 	u_int64_t tcps_rcvgoodsig;	/* rcvd good TCP signatures */
413 	u_int32_t tcps_inhwcsum;	/* input hardware-checksummed packets */
414 	u_int32_t tcps_outhwcsum;	/* output hardware-checksummed packets */
415 
416 	/* ECN stats */
417 	u_int32_t tcps_ecn_accepts;	/* ecn connections accepted */
418 	u_int32_t tcps_ecn_rcvece;	/* # of rcvd ece */
419 	u_int32_t tcps_ecn_rcvcwr;	/* # of rcvd cwr */
420 	u_int32_t tcps_ecn_rcvce;	/* # of rcvd ce in ip header */
421 	u_int32_t tcps_ecn_sndect;	/* # of cwr sent */
422 	u_int32_t tcps_ecn_sndece;	/* # of ece sent */
423 	u_int32_t tcps_ecn_sndcwr;	/* # of cwr sent */
424 	u_int32_t tcps_cwr_ecn;		/* # of cwnd reduced by ecn */
425 	u_int32_t tcps_cwr_frecovery;	/* # of cwnd reduced by fastrecovery */
426 	u_int32_t tcps_cwr_timeout;	/* # of cwnd reduced by timeout */
427 
428 	/* These statistics deal with the SYN cache. */
429 	u_int64_t tcps_sc_added;	/* # of entries added */
430 	u_int64_t tcps_sc_completed;	/* # of connections completed */
431 	u_int64_t tcps_sc_timed_out;	/* # of entries timed out */
432 	u_int64_t tcps_sc_overflowed;	/* # dropped due to overflow */
433 	u_int64_t tcps_sc_reset;	/* # dropped due to RST */
434 	u_int64_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
435 	u_int64_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
436 	u_int64_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
437 	u_int64_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
438 	u_int64_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
439 	u_int64_t tcps_sc_collisions;	/* # of hash collisions */
440 	u_int64_t tcps_sc_retransmitted;/* # of retransmissions */
441 
442 	u_int64_t tcps_conndrained;	/* # of connections drained */
443 
444 	u_int64_t tcps_sack_recovery_episode;	/* SACK recovery episodes */
445 	u_int64_t tcps_sack_rexmits;		/* SACK rexmit segments */
446 	u_int64_t tcps_sack_rexmit_bytes;	/* SACK rexmit bytes */
447 	u_int64_t tcps_sack_rcv_opts;		/* SACK options received */
448 	u_int64_t tcps_sack_snd_opts;		/* SACK options sent */
449 };
450 
451 /*
452  * Names for TCP sysctl objects.
453  */
454 
455 #define	TCPCTL_RFC1323		1 /* enable/disable RFC1323 timestamps/scaling */
456 #define	TCPCTL_KEEPINITTIME	2 /* TCPT_KEEP value */
457 #define TCPCTL_KEEPIDLE		3 /* allow tcp_keepidle to be changed */
458 #define TCPCTL_KEEPINTVL	4 /* allow tcp_keepintvl to be changed */
459 #define TCPCTL_SLOWHZ		5 /* return kernel idea of PR_SLOWHZ */
460 #define TCPCTL_BADDYNAMIC	6 /* return bad dynamic port bitmap */
461 #define	TCPCTL_RECVSPACE	7 /* receive buffer space */
462 #define	TCPCTL_SENDSPACE	8 /* send buffer space */
463 #define	TCPCTL_IDENT		9 /* get connection owner */
464 #define	TCPCTL_SACK	       10 /* selective acknowledgement, rfc 2018 */
465 #define TCPCTL_MSSDFLT	       11 /* Default maximum segment size */
466 #define	TCPCTL_RSTPPSLIMIT     12 /* RST pps limit */
467 #define	TCPCTL_ACK_ON_PUSH     13 /* ACK immediately on PUSH */
468 #define	TCPCTL_ECN	       14 /* RFC3168 ECN */
469 #define	TCPCTL_SYN_CACHE_LIMIT 15 /* max size of comp. state engine */
470 #define	TCPCTL_SYN_BUCKET_LIMIT	16 /* max size of hash bucket */
471 #define	TCPCTL_RFC3390	       17 /* enable/disable RFC3390 increased cwnd */
472 #define	TCPCTL_REASS_LIMIT     18 /* max entries for tcp reass queues */
473 #define	TCPCTL_DROP	       19 /* drop tcp connection */
474 #define	TCPCTL_SACKHOLE_LIMIT  20 /* max entries for tcp sack queues */
475 #define	TCPCTL_STATS	       21 /* TCP statistics */
476 #define	TCPCTL_ALWAYS_KEEPALIVE 22 /* assume SO_KEEPALIVE is always set */
477 #define	TCPCTL_MAXID	       23
478 
479 #define	TCPCTL_NAMES { \
480 	{ 0, 0 }, \
481 	{ "rfc1323",	CTLTYPE_INT }, \
482 	{ "keepinittime",	CTLTYPE_INT }, \
483 	{ "keepidle",	CTLTYPE_INT }, \
484 	{ "keepintvl",	CTLTYPE_INT }, \
485 	{ "slowhz",	CTLTYPE_INT }, \
486 	{ "baddynamic", CTLTYPE_STRUCT }, \
487 	{ NULL,	0 }, \
488 	{ NULL,	0 }, \
489 	{ "ident", 	CTLTYPE_STRUCT }, \
490 	{ "sack",	CTLTYPE_INT }, \
491 	{ "mssdflt",	CTLTYPE_INT }, \
492 	{ "rstppslimit",	CTLTYPE_INT }, \
493 	{ "ackonpush",	CTLTYPE_INT }, \
494 	{ "ecn", 	CTLTYPE_INT }, \
495 	{ "syncachelimit", 	CTLTYPE_INT }, \
496 	{ "synbucketlimit", 	CTLTYPE_INT }, \
497 	{ "rfc3390", 	CTLTYPE_INT }, \
498 	{ "reasslimit", 	CTLTYPE_INT }, \
499 	{ "drop", 	CTLTYPE_STRUCT }, \
500 	{ "sackholelimit", 	CTLTYPE_INT }, \
501 	{ "stats",	CTLTYPE_STRUCT }, \
502 	{ "always_keepalive",	CTLTYPE_INT } \
503 }
504 
505 #define	TCPCTL_VARS { \
506 	NULL, \
507 	&tcp_do_rfc1323, \
508 	&tcptv_keep_init, \
509 	&tcp_keepidle, \
510 	&tcp_keepintvl, \
511 	NULL, \
512 	NULL, \
513 	NULL, \
514 	NULL, \
515 	NULL, \
516 	NULL, \
517 	&tcp_mssdflt, \
518 	&tcp_rst_ppslim, \
519 	&tcp_ack_on_push, \
520 	NULL, \
521 	&tcp_syn_cache_limit, \
522 	&tcp_syn_bucket_limit, \
523 	&tcp_do_rfc3390, \
524 	NULL, \
525 	NULL, \
526 	NULL, \
527 	NULL \
528 }
529 
530 struct tcp_ident_mapping {
531 	struct sockaddr_storage faddr, laddr;
532 	int euid, ruid;
533 	u_int rdomain;
534 };
535 
536 #ifdef _KERNEL
537 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
538 extern	struct tcpstat tcpstat;	/* tcp statistics */
539 extern	u_int32_t tcp_now;		/* for RFC 1323 timestamps */
540 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
541 extern	int tcp_mssdflt;	/* default maximum segment size */
542 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
543 #ifdef TCP_SACK
544 extern	int tcp_do_sack;	/* SACK enabled/disabled */
545 extern	struct pool sackhl_pool;
546 extern	int tcp_sackhole_limit;	/* max entries for tcp sack queues */
547 #endif
548 extern	int tcp_do_ecn;		/* RFC3168 ECN enabled/disabled? */
549 extern	int tcp_do_rfc3390;	/* RFC3390 Increasing TCP's Initial Window */
550 
551 extern	struct pool tcpqe_pool;
552 extern	int tcp_reass_limit;	/* max entries for tcp reass queues */
553 
554 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
555 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
556 
557 extern	int tcp_syn_cache_size;
558 extern	struct syn_cache_head tcp_syn_cache[];
559 extern	u_long syn_cache_count;
560 
561 int	 tcp_attach(struct socket *);
562 void	 tcp_canceltimers(struct tcpcb *);
563 struct tcpcb *
564 	 tcp_close(struct tcpcb *);
565 void	 tcp_reaper(void *);
566 int	 tcp_freeq(struct tcpcb *);
567 #ifdef INET6
568 void	 tcp6_ctlinput(int, struct sockaddr *, void *);
569 #endif
570 void	 *tcp_ctlinput(int, struct sockaddr *, u_int, void *);
571 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf **);
572 struct tcpcb *
573 	 tcp_disconnect(struct tcpcb *);
574 struct tcpcb *
575 	 tcp_drop(struct tcpcb *, int);
576 int	 tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
577 		struct mbuf *, int, struct tcp_opt_info *, u_int);
578 void	 tcp_init(void);
579 #ifdef INET6
580 int	 tcp6_input(struct mbuf **, int *, int);
581 #endif
582 void	 tcp_input(struct mbuf *, ...);
583 int	 tcp_mss(struct tcpcb *, int);
584 void	 tcp_mss_update(struct tcpcb *);
585 u_int	 tcp_hdrsz(struct tcpcb *);
586 void	 tcp_mtudisc(struct inpcb *, int);
587 void	 tcp_mtudisc_increase(struct inpcb *, int);
588 #ifdef INET6
589 void	tcp6_mtudisc(struct inpcb *, int);
590 void	tcp6_mtudisc_callback(struct in6_addr *);
591 #endif
592 struct tcpcb *
593 	 tcp_newtcpcb(struct inpcb *);
594 void	 tcp_notify(struct inpcb *, int);
595 int	 tcp_output(struct tcpcb *);
596 void	 tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
597 int	 tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
598 void	 tcp_rscale(struct tcpcb *, u_long);
599 void	 tcp_respond(struct tcpcb *, caddr_t, struct tcphdr *, tcp_seq,
600 		tcp_seq, int, u_int);
601 void	 tcp_setpersist(struct tcpcb *);
602 void	 tcp_update_sndspace(struct tcpcb *);
603 void	 tcp_update_rcvspace(struct tcpcb *);
604 void	 tcp_slowtimo(void);
605 struct mbuf *
606 	 tcp_template(struct tcpcb *);
607 void	 tcp_trace(short, short, struct tcpcb *, caddr_t, int, int);
608 struct tcpcb *
609 	 tcp_usrclosed(struct tcpcb *);
610 int	 tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
611 int	 tcp_usrreq(struct socket *,
612 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *);
613 void	 tcp_xmit_timer(struct tcpcb *, int);
614 void	 tcpdropoldhalfopen(struct tcpcb *, u_int16_t);
615 #ifdef TCP_SACK
616 void	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
617 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq, tcp_seq);
618 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
619 void	 tcp_clean_sackreport(struct tcpcb *tp);
620 void	 tcp_sack_adjust(struct tcpcb *tp);
621 struct sackhole *
622 	 tcp_sack_output(struct tcpcb *tp);
623 int	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
624 #ifdef DEBUG
625 void	 tcp_print_holes(struct tcpcb *tp);
626 #endif
627 #endif /* TCP_SACK */
628 #if defined(TCP_SACK)
629 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
630 u_long	 tcp_seq_subtract(u_long, u_long );
631 #endif /* TCP_SACK */
632 #ifdef TCP_SIGNATURE
633 int	tcp_signature_apply(caddr_t, caddr_t, unsigned int);
634 int	tcp_signature(struct tdb *, int, struct mbuf *, struct tcphdr *,
635 	    int, int, char *);
636 #endif /* TCP_SIGNATURE */
637 void     tcp_set_iss_tsm(struct tcpcb *);
638 
639 int	 syn_cache_add(struct sockaddr *, struct sockaddr *,
640 		struct tcphdr *, unsigned int, struct socket *,
641 		struct mbuf *, u_char *, int, struct tcp_opt_info *, tcp_seq *);
642 void	 syn_cache_unreach(struct sockaddr *, struct sockaddr *,
643 	   struct tcphdr *, u_int);
644 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
645 		struct tcphdr *, unsigned int, unsigned int,
646 		struct socket *so, struct mbuf *);
647 void	 syn_cache_init(void);
648 void	 syn_cache_insert(struct syn_cache *, struct tcpcb *);
649 struct syn_cache *syn_cache_lookup(struct sockaddr *, struct sockaddr *,
650 		struct syn_cache_head **, u_int);
651 void	 syn_cache_reset(struct sockaddr *, struct sockaddr *,
652 		struct tcphdr *, u_int);
653 int	 syn_cache_respond(struct syn_cache *, struct mbuf *);
654 void	 syn_cache_timer(void *);
655 void	 syn_cache_cleanup(struct tcpcb *);
656 void	 syn_cache_reaper(void *);
657 
658 #endif /* _KERNEL */
659 #endif /* _NETINET_TCP_VAR_H_ */
660