xref: /freebsd/sys/netinet/tcp_var.h (revision 2546665afcaf0d53dc2c7058fee96354b3680f5a)
1 /*
2  * Copyright (c) 1982, 1986, 1993, 1994, 1995
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 4. Neither the name of the University nor the names of its contributors
14  *    may be used to endorse or promote products derived from this software
15  *    without specific prior written permission.
16  *
17  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27  * SUCH DAMAGE.
28  *
29  *	@(#)tcp_var.h	8.4 (Berkeley) 5/24/95
30  * $FreeBSD$
31  */
32 
33 #ifndef _NETINET_TCP_VAR_H_
34 #define _NETINET_TCP_VAR_H_
35 
36 #include <netinet/tcp.h>
37 
38 /*
39  * Kernel variables for tcp.
40  */
41 extern int	tcp_do_rfc1323;
42 extern int	tcp_do_rfc1644;
43 
44 /* TCP segment queue entry */
45 struct tseg_qent {
46 	LIST_ENTRY(tseg_qent) tqe_q;
47 	int	tqe_len;		/* TCP segment data length */
48 	struct	tcphdr *tqe_th;		/* a pointer to tcp header */
49 	struct	mbuf	*tqe_m;		/* mbuf contains packet */
50 };
51 LIST_HEAD(tsegqe_head, tseg_qent);
52 extern int	tcp_reass_qsize;
53 extern struct uma_zone *tcp_reass_zone;
54 
55 struct sackblk {
56 	tcp_seq start;		/* start seq no. of sack block */
57 	tcp_seq end;		/* end seq no. */
58 };
59 
60 struct sackhole {
61 	tcp_seq start;		/* start seq no. of hole */
62 	tcp_seq end;		/* end seq no. */
63 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
64 	struct sackhole *next;	/* next in list */
65 };
66 
67 struct tcptemp {
68 	u_char	tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
69 	struct	tcphdr tt_t;
70 };
71 
72 #define tcp6cb		tcpcb  /* for KAME src sync over BSD*'s */
73 
74 /*
75  * Tcp control block, one per tcp; fields:
76  * Organized for 16 byte cacheline efficiency.
77  */
78 struct tcpcb {
79 	struct	tsegqe_head t_segq;	/* segment reassembly queue */
80 	int	t_segqlen;		/* segment reassembly queue length */
81 	int	t_dupacks;		/* consecutive dup acks recd */
82 	struct	tcptemp	*unused;	/* unused */
83 
84 	struct	callout *tt_rexmt;	/* retransmit timer */
85 	struct	callout *tt_persist;	/* retransmit persistence */
86 	struct	callout *tt_keep;	/* keepalive */
87 	struct	callout *tt_2msl;	/* 2*msl TIME_WAIT timer */
88 	struct	callout *tt_delack;	/* delayed ACK timer */
89 
90 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
91 	int	t_state;		/* state of this connection */
92 	u_int	t_flags;
93 #define	TF_ACKNOW	0x000001	/* ack peer immediately */
94 #define	TF_DELACK	0x000002	/* ack, but try to delay it */
95 #define	TF_NODELAY	0x000004	/* don't delay packets to coalesce */
96 #define	TF_NOOPT	0x000008	/* don't use tcp options */
97 #define	TF_SENTFIN	0x000010	/* have sent FIN */
98 #define	TF_REQ_SCALE	0x000020	/* have/will request window scaling */
99 #define	TF_RCVD_SCALE	0x000040	/* other side has requested scaling */
100 #define	TF_REQ_TSTMP	0x000080	/* have/will request timestamps */
101 #define	TF_RCVD_TSTMP	0x000100	/* a timestamp was received in SYN */
102 #define	TF_SACK_PERMIT	0x000200	/* other side said I could SACK */
103 #define	TF_NEEDSYN	0x000400	/* send SYN (implicit state) */
104 #define	TF_NEEDFIN	0x000800	/* send FIN (implicit state) */
105 #define	TF_NOPUSH	0x001000	/* don't push */
106 #define	TF_REQ_CC	0x002000	/* have/will request CC */
107 #define	TF_RCVD_CC	0x004000	/* a CC was received in SYN */
108 #define	TF_SENDCCNEW	0x008000	/* send CCnew instead of CC in SYN */
109 #define	TF_MORETOCOME	0x010000	/* More data to be appended to sock */
110 #define	TF_LQ_OVERFLOW	0x020000	/* listen queue overflow */
111 #define	TF_LASTIDLE	0x040000	/* connection was previously idle */
112 #define	TF_RXWIN0SENT	0x080000	/* sent a receiver win 0 in response */
113 #define	TF_FASTRECOVERY	0x100000	/* in NewReno Fast Recovery */
114 #define	TF_WASFRECOVERY	0x200000	/* was in NewReno Fast Recovery */
115 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
116 	int	t_force;		/* 1 if forcing out a byte */
117 
118 	tcp_seq	snd_una;		/* send unacknowledged */
119 	tcp_seq	snd_max;		/* highest sequence number sent;
120 					 * used to recognize retransmits
121 					 */
122 	tcp_seq	snd_nxt;		/* send next */
123 	tcp_seq	snd_up;			/* send urgent pointer */
124 
125 	tcp_seq	snd_wl1;		/* window update seg seq number */
126 	tcp_seq	snd_wl2;		/* window update seg ack number */
127 	tcp_seq	iss;			/* initial send sequence number */
128 	tcp_seq	irs;			/* initial receive sequence number */
129 
130 	tcp_seq	rcv_nxt;		/* receive next */
131 	tcp_seq	rcv_adv;		/* advertised window */
132 	u_long	rcv_wnd;		/* receive window */
133 	tcp_seq	rcv_up;			/* receive urgent pointer */
134 
135 	u_long	snd_wnd;		/* send window */
136 	u_long	snd_cwnd;		/* congestion-controlled window */
137 	u_long	snd_bwnd;		/* bandwidth-controlled window */
138 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
139 					 * for slow start exponential to
140 					 * linear switch
141 					 */
142 	u_long	snd_bandwidth;		/* calculated bandwidth or 0 */
143 	tcp_seq	snd_recover;		/* for use in NewReno Fast Recovery */
144 
145 	u_int	t_maxopd;		/* mss plus options */
146 
147 	u_long	t_rcvtime;		/* inactivity time */
148 	u_long	t_starttime;		/* time connection was established */
149 	int	t_rtttime;		/* round trip time */
150 	tcp_seq	t_rtseq;		/* sequence number being timed */
151 
152 	int	t_bw_rtttime;		/* used for bandwidth calculation */
153 	tcp_seq	t_bw_rtseq;		/* used for bandwidth calculation */
154 
155 	int	t_rxtcur;		/* current retransmit value (ticks) */
156 	u_int	t_maxseg;		/* maximum segment size */
157 	int	t_srtt;			/* smoothed round-trip time */
158 	int	t_rttvar;		/* variance in round-trip time */
159 
160 	int	t_rxtshift;		/* log(2) of rexmt exp. backoff */
161 	u_int	t_rttmin;		/* minimum rtt allowed */
162 	u_int	t_rttbest;		/* best rtt we've seen */
163 	u_long	t_rttupdated;		/* number of times rtt sampled */
164 	u_long	max_sndwnd;		/* largest window peer has offered */
165 
166 	int	t_softerror;		/* possible error not yet reported */
167 /* out-of-band data */
168 	char	t_oobflags;		/* have some */
169 	char	t_iobc;			/* input character */
170 #define	TCPOOB_HAVEDATA	0x01
171 #define	TCPOOB_HADDATA	0x02
172 /* RFC 1323 variables */
173 	u_char	snd_scale;		/* window scaling for send window */
174 	u_char	rcv_scale;		/* window scaling for recv window */
175 	u_char	request_r_scale;	/* pending window scaling */
176 	u_char	requested_s_scale;
177 	u_long	ts_recent;		/* timestamp echo data */
178 
179 	u_long	ts_recent_age;		/* when last updated */
180 	tcp_seq	last_ack_sent;
181 /* RFC 1644 variables */
182 	tcp_cc	cc_send;		/* send connection count */
183 	tcp_cc	cc_recv;		/* receive connection count */
184 /* experimental */
185 	u_long	snd_cwnd_prev;		/* cwnd prior to retransmit */
186 	u_long	snd_ssthresh_prev;	/* ssthresh prior to retransmit */
187 	tcp_seq	snd_recover_prev;	/* snd_recover prior to retransmit */
188 	u_long	t_badrxtwin;		/* window for retransmit recovery */
189 	u_char	snd_limited;		/* segments limited transmitted */
190 /* anti DoS counters */
191 	u_long	rcv_second;		/* start of interval second */
192 	u_long	rcv_pps;		/* received packets per second */
193 	u_long	rcv_byps;		/* received bytes per second */
194 /* SACK related state */
195 	int	sack_enable;		/* enable SACK for this connection */
196 	int	snd_numholes;		/* number of holes seen by sender */
197 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
198 	tcp_seq	rcv_laststart;		/* start of last segment recd. */
199 	tcp_seq	rcv_lastend;		/* end of ... */
200 	tcp_seq	rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
201 	int	rcv_numsacks;		/* # distinct sack blks present */
202 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
203 };
204 
205 #define IN_FASTRECOVERY(tp)	(tp->t_flags & TF_FASTRECOVERY)
206 #define ENTER_FASTRECOVERY(tp)	tp->t_flags |= TF_FASTRECOVERY
207 #define EXIT_FASTRECOVERY(tp)	tp->t_flags &= ~TF_FASTRECOVERY
208 
209 #ifdef TCP_SIGNATURE
210 /*
211  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
212  * for SADB verification and lookup.
213  */
214 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
215 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
216 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
217 /*
218  * Only a single SA per host may be specified at this time. An SPI is
219  * needed in order for the KEY_ALLOCSA() lookup to work.
220  */
221 #define	TCP_SIG_SPI	0x1000
222 #endif /* TCP_SIGNATURE */
223 
224 /*
225  * Structure to hold TCP options that are only used during segment
226  * processing (in tcp_input), but not held in the tcpcb.
227  * It's basically used to reduce the number of parameters
228  * to tcp_dooptions.
229  */
230 struct tcpopt {
231 	u_long		to_flags;	/* which options are present */
232 #define TOF_TS		0x0001		/* timestamp */
233 #define TOF_CC		0x0002		/* CC and CCnew are exclusive */
234 #define TOF_CCNEW	0x0004
235 #define	TOF_CCECHO	0x0008
236 #define	TOF_MSS		0x0010
237 #define	TOF_SCALE	0x0020
238 #define	TOF_SIGNATURE	0x0040		/* signature option present */
239 #define	TOF_SIGLEN	0x0080		/* signature length valid (RFC2385) */
240 #define	TOF_SACK	0x0100		/* Peer sent SACK option */
241 	u_int32_t	to_tsval;
242 	u_int32_t	to_tsecr;
243 	tcp_cc		to_cc;		/* holds CC or CCnew */
244 	tcp_cc		to_ccecho;
245 	u_int16_t	to_mss;
246 	u_int8_t 	to_requested_s_scale;
247 	u_int8_t 	to_pad;
248 };
249 
250 #ifdef _NETINET_IN_PCB_H_
251 struct syncache {
252 	inp_gen_t	sc_inp_gencnt;		/* pointer check */
253 	struct 		tcpcb *sc_tp;		/* tcb for listening socket */
254 	struct		mbuf *sc_ipopts;	/* source route */
255 	struct 		in_conninfo sc_inc;	/* addresses */
256 	u_int32_t	sc_tsrecent;
257 	u_int32_t	sc_flowlabel;		/* IPv6 flowlabel */
258 	tcp_cc		sc_cc_send;		/* holds CC or CCnew */
259 	tcp_cc		sc_cc_recv;
260 	tcp_seq 	sc_irs;			/* seq from peer */
261 	tcp_seq 	sc_iss;			/* our ISS */
262 	u_long		sc_rxttime;		/* retransmit time */
263 	u_int16_t	sc_rxtslot; 		/* retransmit counter */
264 	u_int16_t	sc_peer_mss;		/* peer's MSS */
265 	u_int16_t	sc_wnd;			/* advertised window */
266 	u_int8_t 	sc_requested_s_scale:4,
267 			sc_request_r_scale:4;
268 	u_int8_t	sc_flags;
269 #define SCF_NOOPT	0x01			/* no TCP options */
270 #define SCF_WINSCALE	0x02			/* negotiated window scaling */
271 #define SCF_TIMESTAMP	0x04			/* negotiated timestamps */
272 #define SCF_CC		0x08			/* negotiated CC */
273 #define SCF_UNREACH	0x10			/* icmp unreachable received */
274 #define SCF_SIGNATURE	0x20			/* send MD5 digests */
275 #define SCF_SACK	0x80			/* send SACK option */
276 	TAILQ_ENTRY(syncache)	sc_hash;
277 	TAILQ_ENTRY(syncache)	sc_timerq;
278 };
279 
280 struct syncache_head {
281 	TAILQ_HEAD(, syncache)	sch_bucket;
282 	u_int		sch_length;
283 };
284 #else
285 struct in_conninfo;
286 #endif /* _NETINET_IN_PCB_H_ */
287 
288 struct hc_metrics_lite {	/* must stay in sync with hc_metrics */
289 	u_long	rmx_mtu;	/* MTU for this path */
290 	u_long	rmx_ssthresh;	/* outbound gateway buffer limit */
291 	u_long	rmx_rtt;	/* estimated round trip time */
292 	u_long	rmx_rttvar;	/* estimated rtt variance */
293 	u_long	rmx_bandwidth;	/* estimated bandwidth */
294 	u_long	rmx_cwnd;	/* congestion window */
295 	u_long	rmx_sendpipe;   /* outbound delay-bandwidth product */
296 	u_long	rmx_recvpipe;   /* inbound delay-bandwidth product */
297 };
298 
299 struct tcptw {
300 	struct inpcb	*tw_inpcb;	/* XXX back pointer to internet pcb */
301 	tcp_seq		snd_nxt;
302 	tcp_seq		rcv_nxt;
303 	tcp_seq		iss;
304 	tcp_seq		irs;
305 	tcp_cc		cc_recv;
306 	tcp_cc		cc_send;
307 	u_short		last_win;	/* cached window value */
308 	u_short		tw_so_options;	/* copy of so_options */
309 	struct ucred	*tw_cred;	/* user credentials */
310 	u_long 		t_recent;
311 	u_long		t_starttime;
312 	int		tw_time;
313 	LIST_ENTRY(tcptw) tw_2msl;
314 };
315 
316 /*
317  * The TAO cache entry which is stored in the tcp hostcache.
318  */
319 struct rmxp_tao {
320 	tcp_cc	tao_cc;			/* latest CC in valid SYN */
321 	tcp_cc	tao_ccsent;		/* latest CC sent to peer */
322 	u_short	tao_mssopt;		/* peer's cached MSS */
323 #ifdef notyet
324 	u_short	tao_flags;		/* cache status flags */
325 #define	TAOF_DONT	0x0001		/* peer doesn't understand rfc1644 */
326 #define	TAOF_OK		0x0002		/* peer does understand rfc1644 */
327 #define	TAOF_UNDEF	0		/* we don't know yet */
328 #endif /* notyet */
329 };
330 
331 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
332 #define	intotw(ip)	((struct tcptw *)(ip)->inp_ppcb)
333 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
334 
335 /*
336  * The smoothed round-trip time and estimated variance
337  * are stored as fixed point numbers scaled by the values below.
338  * For convenience, these scales are also used in smoothing the average
339  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
340  * With these scales, srtt has 3 bits to the right of the binary point,
341  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
342  * binary point, and is smoothed with an ALPHA of 0.75.
343  */
344 #define	TCP_RTT_SCALE		32	/* multiplier for srtt; 3 bits frac. */
345 #define	TCP_RTT_SHIFT		5	/* shift for srtt; 3 bits frac. */
346 #define	TCP_RTTVAR_SCALE	16	/* multiplier for rttvar; 2 bits */
347 #define	TCP_RTTVAR_SHIFT	4	/* shift for rttvar; 2 bits */
348 #define	TCP_DELTA_SHIFT		2	/* see tcp_input.c */
349 
350 /*
351  * The initial retransmission should happen at rtt + 4 * rttvar.
352  * Because of the way we do the smoothing, srtt and rttvar
353  * will each average +1/2 tick of bias.  When we compute
354  * the retransmit timer, we want 1/2 tick of rounding and
355  * 1 extra tick because of +-1/2 tick uncertainty in the
356  * firing of the timer.  The bias will give us exactly the
357  * 1.5 tick we need.  But, because the bias is
358  * statistical, we have to test that we don't drop below
359  * the minimum feasible timer (which is 2 ticks).
360  * This version of the macro adapted from a paper by Lawrence
361  * Brakmo and Larry Peterson which outlines a problem caused
362  * by insufficient precision in the original implementation,
363  * which results in inappropriately large RTO values for very
364  * fast networks.
365  */
366 #define	TCP_REXMTVAL(tp) \
367 	max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
368 	  + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
369 
370 /*
371  * TCP statistics.
372  * Many of these should be kept per connection,
373  * but that's inconvenient at the moment.
374  */
375 struct	tcpstat {
376 	u_long	tcps_connattempt;	/* connections initiated */
377 	u_long	tcps_accepts;		/* connections accepted */
378 	u_long	tcps_connects;		/* connections established */
379 	u_long	tcps_drops;		/* connections dropped */
380 	u_long	tcps_conndrops;		/* embryonic connections dropped */
381 	u_long	tcps_minmssdrops;	/* average minmss too low drops */
382 	u_long	tcps_closed;		/* conn. closed (includes drops) */
383 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
384 	u_long	tcps_rttupdated;	/* times we succeeded */
385 	u_long	tcps_delack;		/* delayed acks sent */
386 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
387 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
388 	u_long	tcps_persisttimeo;	/* persist timeouts */
389 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
390 	u_long	tcps_keepprobe;		/* keepalive probes sent */
391 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
392 
393 	u_long	tcps_sndtotal;		/* total packets sent */
394 	u_long	tcps_sndpack;		/* data packets sent */
395 	u_long	tcps_sndbyte;		/* data bytes sent */
396 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
397 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
398 	u_long	tcps_sndrexmitbad;	/* unnecessary packet retransmissions */
399 	u_long	tcps_sndacks;		/* ack-only packets sent */
400 	u_long	tcps_sndprobe;		/* window probes sent */
401 	u_long	tcps_sndurg;		/* packets sent with URG only */
402 	u_long	tcps_sndwinup;		/* window update-only packets sent */
403 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
404 
405 	u_long	tcps_rcvtotal;		/* total packets received */
406 	u_long	tcps_rcvpack;		/* packets received in sequence */
407 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
408 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
409 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
410 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
411 	u_long	tcps_rcvshort;		/* packets received too short */
412 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
413 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
414 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
415 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
416 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
417 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
418 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
419 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
420 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
421 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
422 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
423 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
424 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
425 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
426 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
427 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
428 	u_long	tcps_predack;		/* times hdr predict ok for acks */
429 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
430 	u_long	tcps_pcbcachemiss;
431 	u_long	tcps_cachedrtt;		/* times cached RTT in route updated */
432 	u_long	tcps_cachedrttvar;	/* times cached rttvar updated */
433 	u_long	tcps_cachedssthresh;	/* times cached ssthresh updated */
434 	u_long	tcps_usedrtt;		/* times RTT initialized from route */
435 	u_long	tcps_usedrttvar;	/* times RTTVAR initialized from rt */
436 	u_long	tcps_usedssthresh;	/* times ssthresh initialized from rt*/
437 	u_long	tcps_persistdrop;	/* timeout in persist state */
438 	u_long	tcps_badsyn;		/* bogus SYN, e.g. premature ACK */
439 	u_long	tcps_mturesent;		/* resends due to MTU discovery */
440 	u_long	tcps_listendrop;	/* listen queue overflows */
441 	u_long	tcps_badrst;		/* ignored RSTs in the window */
442 
443 	u_long	tcps_sc_added;		/* entry added to syncache */
444 	u_long	tcps_sc_retransmitted;	/* syncache entry was retransmitted */
445 	u_long	tcps_sc_dupsyn;		/* duplicate SYN packet */
446 	u_long	tcps_sc_dropped;	/* could not reply to packet */
447 	u_long	tcps_sc_completed;	/* successful extraction of entry */
448 	u_long	tcps_sc_bucketoverflow;	/* syncache per-bucket limit hit */
449 	u_long	tcps_sc_cacheoverflow;	/* syncache cache limit hit */
450 	u_long	tcps_sc_reset;		/* RST removed entry from syncache */
451 	u_long	tcps_sc_stale;		/* timed out or listen socket gone */
452 	u_long	tcps_sc_aborted;	/* syncache entry aborted */
453 	u_long	tcps_sc_badack;		/* removed due to bad ACK */
454 	u_long	tcps_sc_unreach;	/* ICMP unreachable received */
455 	u_long	tcps_sc_zonefail;	/* zalloc() failed */
456 	u_long	tcps_sc_sendcookie;	/* SYN cookie sent */
457 	u_long	tcps_sc_recvcookie;	/* SYN cookie received */
458 
459 	u_long	tcps_hc_added;		/* entry added to hostcache */
460 	u_long	tcps_hc_bucketoverflow;	/* hostcache per bucket limit hit */
461 
462 	/* SACK related stats */
463 	u_long	tcps_sack_recovery_episode; /* SACK recovery episodes */
464 	u_long  tcps_sack_rexmits;	    /* SACK rexmit segments   */
465 	u_long  tcps_sack_rexmit_bytes;	    /* SACK rexmit bytes      */
466 	u_long  tcps_sack_rcv_blocks;	    /* SACK blocks (options) received */
467 	u_long  tcps_sack_send_blocks;	    /* SACK blocks (options) sent     */
468 };
469 
470 /*
471  * TCB structure exported to user-land via sysctl(3).
472  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
473  * included.  Not all of our clients do.
474  */
475 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
476 struct	xtcpcb {
477 	size_t	xt_len;
478 	struct	inpcb	xt_inp;
479 	struct	tcpcb	xt_tp;
480 	struct	xsocket	xt_socket;
481 	u_quad_t	xt_alignment_hack;
482 };
483 #endif
484 
485 /*
486  * Names for TCP sysctl objects
487  */
488 #define	TCPCTL_DO_RFC1323	1	/* use RFC-1323 extensions */
489 #define	TCPCTL_DO_RFC1644	2	/* use RFC-1644 extensions */
490 #define	TCPCTL_MSSDFLT		3	/* MSS default */
491 #define TCPCTL_STATS		4	/* statistics (read-only) */
492 #define	TCPCTL_RTTDFLT		5	/* default RTT estimate */
493 #define	TCPCTL_KEEPIDLE		6	/* keepalive idle timer */
494 #define	TCPCTL_KEEPINTVL	7	/* interval to send keepalives */
495 #define	TCPCTL_SENDSPACE	8	/* send buffer space */
496 #define	TCPCTL_RECVSPACE	9	/* receive buffer space */
497 #define	TCPCTL_KEEPINIT		10	/* timeout for establishing syn */
498 #define	TCPCTL_PCBLIST		11	/* list of all outstanding PCBs */
499 #define	TCPCTL_DELACKTIME	12	/* time before sending delayed ACK */
500 #define	TCPCTL_V6MSSDFLT	13	/* MSS default for IPv6 */
501 #define	TCPCTL_SACK		14	/* Selective Acknowledgement,rfc 2018 */
502 #define	TCPCTL_MAXID		15
503 
504 #define TCPCTL_NAMES { \
505 	{ 0, 0 }, \
506 	{ "rfc1323", CTLTYPE_INT }, \
507 	{ "rfc1644", CTLTYPE_INT }, \
508 	{ "mssdflt", CTLTYPE_INT }, \
509 	{ "stats", CTLTYPE_STRUCT }, \
510 	{ "rttdflt", CTLTYPE_INT }, \
511 	{ "keepidle", CTLTYPE_INT }, \
512 	{ "keepintvl", CTLTYPE_INT }, \
513 	{ "sendspace", CTLTYPE_INT }, \
514 	{ "recvspace", CTLTYPE_INT }, \
515 	{ "keepinit", CTLTYPE_INT }, \
516 	{ "pcblist", CTLTYPE_STRUCT }, \
517 	{ "delacktime", CTLTYPE_INT }, \
518 	{ "v6mssdflt", CTLTYPE_INT }, \
519 	{ "maxid", CTLTYPE_INT }, \
520 }
521 
522 
523 #ifdef _KERNEL
524 #ifdef SYSCTL_DECL
525 SYSCTL_DECL(_net_inet_tcp);
526 #endif
527 
528 extern	struct inpcbhead tcb;		/* head of queue of active tcpcb's */
529 extern	struct inpcbinfo tcbinfo;
530 extern	struct tcpstat tcpstat;	/* tcp statistics */
531 extern	int tcp_mssdflt;	/* XXX */
532 extern	int tcp_minmss;
533 extern	int tcp_minmssoverload;
534 extern	int tcp_delack_enabled;
535 extern	int tcp_do_newreno;
536 extern	int path_mtu_discovery;
537 extern	int ss_fltsz;
538 extern	int ss_fltsz_local;
539 
540 extern	int tcp_do_sack;	/* SACK enabled/disabled */
541 
542 void	 tcp_canceltimers(struct tcpcb *);
543 struct tcpcb *
544 	 tcp_close(struct tcpcb *);
545 void	 tcp_twstart(struct tcpcb *);
546 int	 tcp_twrecycleable(struct tcptw *tw);
547 struct tcptw *
548 	 tcp_twclose(struct tcptw *_tw, int _reuse);
549 void	 tcp_ctlinput(int, struct sockaddr *, void *);
550 int	 tcp_ctloutput(struct socket *, struct sockopt *);
551 struct tcpcb *
552 	 tcp_drop(struct tcpcb *, int);
553 void	 tcp_drain(void);
554 void	 tcp_fasttimo(void);
555 void	 tcp_init(void);
556 void	 tcp_fini(void *);
557 void	 tcp_reass_init(void);
558 void	 tcp_input(struct mbuf *, int);
559 u_long	 tcp_maxmtu(struct in_conninfo *);
560 u_long	 tcp_maxmtu6(struct in_conninfo *);
561 void	 tcp_mss(struct tcpcb *, int);
562 int	 tcp_mssopt(struct in_conninfo *);
563 struct inpcb *
564 	 tcp_drop_syn_sent(struct inpcb *, int);
565 struct inpcb *
566 	 tcp_mtudisc(struct inpcb *, int);
567 struct tcpcb *
568 	 tcp_newtcpcb(struct inpcb *);
569 int	 tcp_output(struct tcpcb *);
570 struct inpcb *
571 	 tcp_quench(struct inpcb *, int);
572 void	 tcp_respond(struct tcpcb *, void *,
573 	    struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
574 int	 tcp_twrespond(struct tcptw *, int);
575 void	 tcp_setpersist(struct tcpcb *);
576 #ifdef TCP_SIGNATURE
577 int	 tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
578 #endif
579 void	 tcp_slowtimo(void);
580 struct tcptemp *
581 	 tcpip_maketemplate(struct inpcb *);
582 void	 tcpip_fillheaders(struct inpcb *, void *, void *);
583 struct tcpcb *
584 	 tcp_timers(struct tcpcb *, int);
585 void	 tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
586 void	 tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
587 void	 syncache_init(void);
588 void	 syncache_unreach(struct in_conninfo *, struct tcphdr *);
589 int	 syncache_expand(struct in_conninfo *, struct tcphdr *,
590 	     struct socket **, struct mbuf *);
591 int	 syncache_add(struct in_conninfo *, struct tcpopt *,
592 	     struct tcphdr *, struct socket **, struct mbuf *);
593 void	 syncache_chkrst(struct in_conninfo *, struct tcphdr *);
594 void	 syncache_badack(struct in_conninfo *);
595 /*
596  * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
597  */
598 void	 tcp_hc_init(void);
599 void	 tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
600 u_long	 tcp_hc_getmtu(struct in_conninfo *);
601 void	 tcp_hc_gettao(struct in_conninfo *, struct rmxp_tao *);
602 void	 tcp_hc_updatemtu(struct in_conninfo *, u_long);
603 void	 tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
604 void	 tcp_hc_updatetao(struct in_conninfo *, int, tcp_cc, u_short);
605 /* update which tao field */
606 #define	TCP_HC_TAO_CC		0x1
607 #define TCP_HC_TAO_CCSENT	0x2
608 #define TCP_HC_TAO_MSSOPT	0x3
609 
610 extern	struct pr_usrreqs tcp_usrreqs;
611 extern	u_long tcp_sendspace;
612 extern	u_long tcp_recvspace;
613 tcp_seq tcp_new_isn(struct tcpcb *);
614 
615 int	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
616 void	 tcp_update_sack_list(struct tcpcb *tp);
617 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
618 void	 tcp_clean_sackreport(struct tcpcb *tp);
619 void	 tcp_sack_adjust(struct tcpcb *tp);
620 struct sackhole *tcp_sack_output(struct tcpcb *tp);
621 void	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
622 void	 tcp_free_sackholes(struct tcpcb *tp);
623 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
624 u_long	 tcp_seq_subtract(u_long, u_long );
625 #ifdef TCP_SACK_DEBUG
626 void 	 tcp_print_holes(struct tcpcb *tp);
627 #endif /* TCP_SACK_DEBUG */
628 
629 #endif /* _KERNEL */
630 
631 #endif /* _NETINET_TCP_VAR_H_ */
632