xref: /freebsd/sys/netinet/tcp_var.h (revision 6af83ee0d2941d18880b6aaa2b4facd1d30c6106)
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 
43 /* TCP segment queue entry */
44 struct tseg_qent {
45 	LIST_ENTRY(tseg_qent) tqe_q;
46 	int	tqe_len;		/* TCP segment data length */
47 	struct	tcphdr *tqe_th;		/* a pointer to tcp header */
48 	struct	mbuf	*tqe_m;		/* mbuf contains packet */
49 };
50 LIST_HEAD(tsegqe_head, tseg_qent);
51 extern int	tcp_reass_qsize;
52 extern struct uma_zone *tcp_reass_zone;
53 
54 struct sackblk {
55 	tcp_seq start;		/* start seq no. of sack block */
56 	tcp_seq end;		/* end seq no. */
57 };
58 
59 struct sackhole {
60 	tcp_seq start;		/* start seq no. of hole */
61 	tcp_seq end;		/* end seq no. */
62 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
63 	struct sackhole *next;	/* next in list */
64 };
65 
66 struct tcptemp {
67 	u_char	tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
68 	struct	tcphdr tt_t;
69 };
70 
71 #define tcp6cb		tcpcb  /* for KAME src sync over BSD*'s */
72 
73 /*
74  * Tcp control block, one per tcp; fields:
75  * Organized for 16 byte cacheline efficiency.
76  */
77 struct tcpcb {
78 	struct	tsegqe_head t_segq;	/* segment reassembly queue */
79 	int	t_segqlen;		/* segment reassembly queue length */
80 	int	t_dupacks;		/* consecutive dup acks recd */
81 
82 	struct	callout *tt_rexmt;	/* retransmit timer */
83 	struct	callout *tt_persist;	/* retransmit persistence */
84 	struct	callout *tt_keep;	/* keepalive */
85 	struct	callout *tt_2msl;	/* 2*msl TIME_WAIT timer */
86 	struct	callout *tt_delack;	/* delayed ACK timer */
87 
88 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
89 	int	t_state;		/* state of this connection */
90 	u_int	t_flags;
91 #define	TF_ACKNOW	0x000001	/* ack peer immediately */
92 #define	TF_DELACK	0x000002	/* ack, but try to delay it */
93 #define	TF_NODELAY	0x000004	/* don't delay packets to coalesce */
94 #define	TF_NOOPT	0x000008	/* don't use tcp options */
95 #define	TF_SENTFIN	0x000010	/* have sent FIN */
96 #define	TF_REQ_SCALE	0x000020	/* have/will request window scaling */
97 #define	TF_RCVD_SCALE	0x000040	/* other side has requested scaling */
98 #define	TF_REQ_TSTMP	0x000080	/* have/will request timestamps */
99 #define	TF_RCVD_TSTMP	0x000100	/* a timestamp was received in SYN */
100 #define	TF_SACK_PERMIT	0x000200	/* other side said I could SACK */
101 #define	TF_NEEDSYN	0x000400	/* send SYN (implicit state) */
102 #define	TF_NEEDFIN	0x000800	/* send FIN (implicit state) */
103 #define	TF_NOPUSH	0x001000	/* don't push */
104 #define	TF_MORETOCOME	0x010000	/* More data to be appended to sock */
105 #define	TF_LQ_OVERFLOW	0x020000	/* listen queue overflow */
106 #define	TF_LASTIDLE	0x040000	/* connection was previously idle */
107 #define	TF_RXWIN0SENT	0x080000	/* sent a receiver win 0 in response */
108 #define	TF_FASTRECOVERY	0x100000	/* in NewReno Fast Recovery */
109 #define	TF_WASFRECOVERY	0x200000	/* was in NewReno Fast Recovery */
110 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
111 	int	t_force;		/* 1 if forcing out a byte */
112 
113 	tcp_seq	snd_una;		/* send unacknowledged */
114 	tcp_seq	snd_max;		/* highest sequence number sent;
115 					 * used to recognize retransmits
116 					 */
117 	tcp_seq	snd_nxt;		/* send next */
118 	tcp_seq	snd_up;			/* send urgent pointer */
119 
120 	tcp_seq	snd_wl1;		/* window update seg seq number */
121 	tcp_seq	snd_wl2;		/* window update seg ack number */
122 	tcp_seq	iss;			/* initial send sequence number */
123 	tcp_seq	irs;			/* initial receive sequence number */
124 
125 	tcp_seq	rcv_nxt;		/* receive next */
126 	tcp_seq	rcv_adv;		/* advertised window */
127 	u_long	rcv_wnd;		/* receive window */
128 	tcp_seq	rcv_up;			/* receive urgent pointer */
129 
130 	u_long	snd_wnd;		/* send window */
131 	u_long	snd_cwnd;		/* congestion-controlled window */
132 	u_long	snd_bwnd;		/* bandwidth-controlled window */
133 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
134 					 * for slow start exponential to
135 					 * linear switch
136 					 */
137 	u_long	snd_bandwidth;		/* calculated bandwidth or 0 */
138 	tcp_seq	snd_recover;		/* for use in NewReno Fast Recovery */
139 
140 	u_int	t_maxopd;		/* mss plus options */
141 
142 	u_long	t_rcvtime;		/* inactivity time */
143 	u_long	t_starttime;		/* time connection was established */
144 	int	t_rtttime;		/* round trip time */
145 	tcp_seq	t_rtseq;		/* sequence number being timed */
146 
147 	int	t_bw_rtttime;		/* used for bandwidth calculation */
148 	tcp_seq	t_bw_rtseq;		/* used for bandwidth calculation */
149 
150 	int	t_rxtcur;		/* current retransmit value (ticks) */
151 	u_int	t_maxseg;		/* maximum segment size */
152 	int	t_srtt;			/* smoothed round-trip time */
153 	int	t_rttvar;		/* variance in round-trip time */
154 
155 	int	t_rxtshift;		/* log(2) of rexmt exp. backoff */
156 	u_int	t_rttmin;		/* minimum rtt allowed */
157 	u_int	t_rttbest;		/* best rtt we've seen */
158 	u_long	t_rttupdated;		/* number of times rtt sampled */
159 	u_long	max_sndwnd;		/* largest window peer has offered */
160 
161 	int	t_softerror;		/* possible error not yet reported */
162 /* out-of-band data */
163 	char	t_oobflags;		/* have some */
164 	char	t_iobc;			/* input character */
165 #define	TCPOOB_HAVEDATA	0x01
166 #define	TCPOOB_HADDATA	0x02
167 /* RFC 1323 variables */
168 	u_char	snd_scale;		/* window scaling for send window */
169 	u_char	rcv_scale;		/* window scaling for recv window */
170 	u_char	request_r_scale;	/* pending window scaling */
171 	u_char	requested_s_scale;
172 	u_long	ts_recent;		/* timestamp echo data */
173 
174 	u_long	ts_recent_age;		/* when last updated */
175 	tcp_seq	last_ack_sent;
176 /* experimental */
177 	u_long	snd_cwnd_prev;		/* cwnd prior to retransmit */
178 	u_long	snd_ssthresh_prev;	/* ssthresh prior to retransmit */
179 	tcp_seq	snd_recover_prev;	/* snd_recover prior to retransmit */
180 	u_long	t_badrxtwin;		/* window for retransmit recovery */
181 	u_char	snd_limited;		/* segments limited transmitted */
182 /* anti DoS counters */
183 	u_long	rcv_second;		/* start of interval second */
184 	u_long	rcv_pps;		/* received packets per second */
185 	u_long	rcv_byps;		/* received bytes per second */
186 /* SACK related state */
187 	int	sack_enable;		/* enable SACK for this connection */
188 	int	snd_numholes;		/* number of holes seen by sender */
189 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
190 	tcp_seq	rcv_laststart;		/* start of last segment recd. */
191 	tcp_seq	rcv_lastend;		/* end of ... */
192 	tcp_seq	rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
193 	int	rcv_numsacks;		/* # distinct sack blks present */
194 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
195 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
196 					   episode starts at this seq number */
197 };
198 
199 #define IN_FASTRECOVERY(tp)	(tp->t_flags & TF_FASTRECOVERY)
200 #define ENTER_FASTRECOVERY(tp)	tp->t_flags |= TF_FASTRECOVERY
201 #define EXIT_FASTRECOVERY(tp)	tp->t_flags &= ~TF_FASTRECOVERY
202 
203 #ifdef TCP_SIGNATURE
204 /*
205  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
206  * for SADB verification and lookup.
207  */
208 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
209 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
210 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
211 /*
212  * Only a single SA per host may be specified at this time. An SPI is
213  * needed in order for the KEY_ALLOCSA() lookup to work.
214  */
215 #define	TCP_SIG_SPI	0x1000
216 #endif /* TCP_SIGNATURE */
217 
218 /*
219  * Structure to hold TCP options that are only used during segment
220  * processing (in tcp_input), but not held in the tcpcb.
221  * It's basically used to reduce the number of parameters
222  * to tcp_dooptions.
223  */
224 struct tcpopt {
225 	u_long		to_flags;	/* which options are present */
226 #define TOF_TS		0x0001		/* timestamp */
227 #define	TOF_MSS		0x0010
228 #define	TOF_SCALE	0x0020
229 #define	TOF_SIGNATURE	0x0040		/* signature option present */
230 #define	TOF_SIGLEN	0x0080		/* signature length valid (RFC2385) */
231 #define	TOF_SACK	0x0100		/* Peer sent SACK option */
232 	u_int32_t	to_tsval;
233 	u_int32_t	to_tsecr;
234 	u_int16_t	to_mss;
235 	u_int8_t	to_requested_s_scale;
236 	u_int8_t	to_pad;
237 };
238 
239 #ifdef _NETINET_IN_PCB_H_
240 struct syncache {
241 	inp_gen_t	sc_inp_gencnt;		/* pointer check */
242 	struct		tcpcb *sc_tp;		/* tcb for listening socket */
243 	struct		mbuf *sc_ipopts;	/* source route */
244 	struct		in_conninfo sc_inc;	/* addresses */
245 	u_int32_t	sc_tsrecent;
246 	u_int32_t	sc_flowlabel;		/* IPv6 flowlabel */
247 	tcp_seq		sc_irs;			/* seq from peer */
248 	tcp_seq		sc_iss;			/* our ISS */
249 	u_long		sc_rxttime;		/* retransmit time */
250 	u_int16_t	sc_rxtslot;		/* retransmit counter */
251 	u_int16_t	sc_peer_mss;		/* peer's MSS */
252 	u_int16_t	sc_wnd;			/* advertised window */
253 	u_int8_t	sc_requested_s_scale:4,
254 			sc_request_r_scale:4;
255 	u_int8_t	sc_flags;
256 #define SCF_NOOPT	0x01			/* no TCP options */
257 #define SCF_WINSCALE	0x02			/* negotiated window scaling */
258 #define SCF_TIMESTAMP	0x04			/* negotiated timestamps */
259 #define SCF_UNREACH	0x10			/* icmp unreachable received */
260 #define SCF_SIGNATURE	0x20			/* send MD5 digests */
261 #define SCF_SACK	0x80			/* send SACK option */
262 	TAILQ_ENTRY(syncache)	sc_hash;
263 	TAILQ_ENTRY(syncache)	sc_timerq;
264 };
265 
266 struct syncache_head {
267 	TAILQ_HEAD(, syncache)	sch_bucket;
268 	u_int		sch_length;
269 };
270 #else
271 struct in_conninfo;
272 #endif /* _NETINET_IN_PCB_H_ */
273 
274 struct hc_metrics_lite {	/* must stay in sync with hc_metrics */
275 	u_long	rmx_mtu;	/* MTU for this path */
276 	u_long	rmx_ssthresh;	/* outbound gateway buffer limit */
277 	u_long	rmx_rtt;	/* estimated round trip time */
278 	u_long	rmx_rttvar;	/* estimated rtt variance */
279 	u_long	rmx_bandwidth;	/* estimated bandwidth */
280 	u_long	rmx_cwnd;	/* congestion window */
281 	u_long	rmx_sendpipe;   /* outbound delay-bandwidth product */
282 	u_long	rmx_recvpipe;   /* inbound delay-bandwidth product */
283 };
284 
285 struct tcptw {
286 	struct inpcb	*tw_inpcb;	/* XXX back pointer to internet pcb */
287 	tcp_seq		snd_nxt;
288 	tcp_seq		rcv_nxt;
289 	tcp_seq		iss;
290 	tcp_seq		irs;
291 	u_short		last_win;	/* cached window value */
292 	u_short		tw_so_options;	/* copy of so_options */
293 	struct ucred	*tw_cred;	/* user credentials */
294 	u_long		t_recent;
295 	u_long		t_starttime;
296 	int		tw_time;
297 	LIST_ENTRY(tcptw) tw_2msl;
298 };
299 
300 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
301 #define	intotw(ip)	((struct tcptw *)(ip)->inp_ppcb)
302 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
303 
304 /*
305  * The smoothed round-trip time and estimated variance
306  * are stored as fixed point numbers scaled by the values below.
307  * For convenience, these scales are also used in smoothing the average
308  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
309  * With these scales, srtt has 3 bits to the right of the binary point,
310  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
311  * binary point, and is smoothed with an ALPHA of 0.75.
312  */
313 #define	TCP_RTT_SCALE		32	/* multiplier for srtt; 3 bits frac. */
314 #define	TCP_RTT_SHIFT		5	/* shift for srtt; 3 bits frac. */
315 #define	TCP_RTTVAR_SCALE	16	/* multiplier for rttvar; 2 bits */
316 #define	TCP_RTTVAR_SHIFT	4	/* shift for rttvar; 2 bits */
317 #define	TCP_DELTA_SHIFT		2	/* see tcp_input.c */
318 
319 /*
320  * The initial retransmission should happen at rtt + 4 * rttvar.
321  * Because of the way we do the smoothing, srtt and rttvar
322  * will each average +1/2 tick of bias.  When we compute
323  * the retransmit timer, we want 1/2 tick of rounding and
324  * 1 extra tick because of +-1/2 tick uncertainty in the
325  * firing of the timer.  The bias will give us exactly the
326  * 1.5 tick we need.  But, because the bias is
327  * statistical, we have to test that we don't drop below
328  * the minimum feasible timer (which is 2 ticks).
329  * This version of the macro adapted from a paper by Lawrence
330  * Brakmo and Larry Peterson which outlines a problem caused
331  * by insufficient precision in the original implementation,
332  * which results in inappropriately large RTO values for very
333  * fast networks.
334  */
335 #define	TCP_REXMTVAL(tp) \
336 	max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
337 	  + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
338 
339 /*
340  * TCP statistics.
341  * Many of these should be kept per connection,
342  * but that's inconvenient at the moment.
343  */
344 struct	tcpstat {
345 	u_long	tcps_connattempt;	/* connections initiated */
346 	u_long	tcps_accepts;		/* connections accepted */
347 	u_long	tcps_connects;		/* connections established */
348 	u_long	tcps_drops;		/* connections dropped */
349 	u_long	tcps_conndrops;		/* embryonic connections dropped */
350 	u_long	tcps_minmssdrops;	/* average minmss too low drops */
351 	u_long	tcps_closed;		/* conn. closed (includes drops) */
352 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
353 	u_long	tcps_rttupdated;	/* times we succeeded */
354 	u_long	tcps_delack;		/* delayed acks sent */
355 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
356 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
357 	u_long	tcps_persisttimeo;	/* persist timeouts */
358 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
359 	u_long	tcps_keepprobe;		/* keepalive probes sent */
360 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
361 
362 	u_long	tcps_sndtotal;		/* total packets sent */
363 	u_long	tcps_sndpack;		/* data packets sent */
364 	u_long	tcps_sndbyte;		/* data bytes sent */
365 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
366 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
367 	u_long	tcps_sndrexmitbad;	/* unnecessary packet retransmissions */
368 	u_long	tcps_sndacks;		/* ack-only packets sent */
369 	u_long	tcps_sndprobe;		/* window probes sent */
370 	u_long	tcps_sndurg;		/* packets sent with URG only */
371 	u_long	tcps_sndwinup;		/* window update-only packets sent */
372 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
373 
374 	u_long	tcps_rcvtotal;		/* total packets received */
375 	u_long	tcps_rcvpack;		/* packets received in sequence */
376 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
377 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
378 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
379 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
380 	u_long	tcps_rcvshort;		/* packets received too short */
381 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
382 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
383 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
384 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
385 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
386 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
387 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
388 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
389 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
390 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
391 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
392 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
393 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
394 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
395 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
396 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
397 	u_long	tcps_predack;		/* times hdr predict ok for acks */
398 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
399 	u_long	tcps_pcbcachemiss;
400 	u_long	tcps_cachedrtt;		/* times cached RTT in route updated */
401 	u_long	tcps_cachedrttvar;	/* times cached rttvar updated */
402 	u_long	tcps_cachedssthresh;	/* times cached ssthresh updated */
403 	u_long	tcps_usedrtt;		/* times RTT initialized from route */
404 	u_long	tcps_usedrttvar;	/* times RTTVAR initialized from rt */
405 	u_long	tcps_usedssthresh;	/* times ssthresh initialized from rt*/
406 	u_long	tcps_persistdrop;	/* timeout in persist state */
407 	u_long	tcps_badsyn;		/* bogus SYN, e.g. premature ACK */
408 	u_long	tcps_mturesent;		/* resends due to MTU discovery */
409 	u_long	tcps_listendrop;	/* listen queue overflows */
410 	u_long	tcps_badrst;		/* ignored RSTs in the window */
411 
412 	u_long	tcps_sc_added;		/* entry added to syncache */
413 	u_long	tcps_sc_retransmitted;	/* syncache entry was retransmitted */
414 	u_long	tcps_sc_dupsyn;		/* duplicate SYN packet */
415 	u_long	tcps_sc_dropped;	/* could not reply to packet */
416 	u_long	tcps_sc_completed;	/* successful extraction of entry */
417 	u_long	tcps_sc_bucketoverflow;	/* syncache per-bucket limit hit */
418 	u_long	tcps_sc_cacheoverflow;	/* syncache cache limit hit */
419 	u_long	tcps_sc_reset;		/* RST removed entry from syncache */
420 	u_long	tcps_sc_stale;		/* timed out or listen socket gone */
421 	u_long	tcps_sc_aborted;	/* syncache entry aborted */
422 	u_long	tcps_sc_badack;		/* removed due to bad ACK */
423 	u_long	tcps_sc_unreach;	/* ICMP unreachable received */
424 	u_long	tcps_sc_zonefail;	/* zalloc() failed */
425 	u_long	tcps_sc_sendcookie;	/* SYN cookie sent */
426 	u_long	tcps_sc_recvcookie;	/* SYN cookie received */
427 
428 	u_long	tcps_hc_added;		/* entry added to hostcache */
429 	u_long	tcps_hc_bucketoverflow;	/* hostcache per bucket limit hit */
430 
431 	/* SACK related stats */
432 	u_long	tcps_sack_recovery_episode; /* SACK recovery episodes */
433 	u_long  tcps_sack_rexmits;	    /* SACK rexmit segments   */
434 	u_long  tcps_sack_rexmit_bytes;	    /* SACK rexmit bytes      */
435 	u_long  tcps_sack_rcv_blocks;	    /* SACK blocks (options) received */
436 	u_long  tcps_sack_send_blocks;	    /* SACK blocks (options) sent     */
437 };
438 
439 /*
440  * TCB structure exported to user-land via sysctl(3).
441  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
442  * included.  Not all of our clients do.
443  */
444 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
445 struct	xtcpcb {
446 	size_t	xt_len;
447 	struct	inpcb	xt_inp;
448 	struct	tcpcb	xt_tp;
449 	struct	xsocket	xt_socket;
450 	u_quad_t	xt_alignment_hack;
451 };
452 #endif
453 
454 struct tcp_ident_mapping {
455 	struct sockaddr_storage faddr, laddr;
456 	uid_t euid, ruid;
457 };
458 
459 /*
460  * Names for TCP sysctl objects
461  */
462 #define	TCPCTL_DO_RFC1323	1	/* use RFC-1323 extensions */
463 #define	TCPCTL_MSSDFLT		3	/* MSS default */
464 #define TCPCTL_STATS		4	/* statistics (read-only) */
465 #define	TCPCTL_RTTDFLT		5	/* default RTT estimate */
466 #define	TCPCTL_KEEPIDLE		6	/* keepalive idle timer */
467 #define	TCPCTL_KEEPINTVL	7	/* interval to send keepalives */
468 #define	TCPCTL_SENDSPACE	8	/* send buffer space */
469 #define	TCPCTL_RECVSPACE	9	/* receive buffer space */
470 #define	TCPCTL_KEEPINIT		10	/* timeout for establishing syn */
471 #define	TCPCTL_PCBLIST		11	/* list of all outstanding PCBs */
472 #define	TCPCTL_DELACKTIME	12	/* time before sending delayed ACK */
473 #define	TCPCTL_V6MSSDFLT	13	/* MSS default for IPv6 */
474 #define	TCPCTL_SACK		14	/* Selective Acknowledgement,rfc 2018 */
475 #define	TCPCTL_DROP		15	/* drop tcp connection */
476 #define	TCPCTL_MAXID		16
477 
478 #define TCPCTL_NAMES { \
479 	{ 0, 0 }, \
480 	{ "rfc1323", CTLTYPE_INT }, \
481 	{ "mssdflt", CTLTYPE_INT }, \
482 	{ "stats", CTLTYPE_STRUCT }, \
483 	{ "rttdflt", CTLTYPE_INT }, \
484 	{ "keepidle", CTLTYPE_INT }, \
485 	{ "keepintvl", CTLTYPE_INT }, \
486 	{ "sendspace", CTLTYPE_INT }, \
487 	{ "recvspace", CTLTYPE_INT }, \
488 	{ "keepinit", CTLTYPE_INT }, \
489 	{ "pcblist", CTLTYPE_STRUCT }, \
490 	{ "delacktime", CTLTYPE_INT }, \
491 	{ "v6mssdflt", CTLTYPE_INT }, \
492 	{ "maxid", CTLTYPE_INT }, \
493 }
494 
495 
496 #ifdef _KERNEL
497 #ifdef SYSCTL_DECL
498 SYSCTL_DECL(_net_inet_tcp);
499 SYSCTL_DECL(_net_inet_tcp_sack);
500 #endif
501 
502 extern	struct inpcbhead tcb;		/* head of queue of active tcpcb's */
503 extern	struct inpcbinfo tcbinfo;
504 extern	struct tcpstat tcpstat;	/* tcp statistics */
505 extern	int tcp_mssdflt;	/* XXX */
506 extern	int tcp_minmss;
507 extern	int tcp_minmssoverload;
508 extern	int tcp_delack_enabled;
509 extern	int tcp_do_newreno;
510 extern	int path_mtu_discovery;
511 extern	int ss_fltsz;
512 extern	int ss_fltsz_local;
513 
514 extern	int tcp_do_sack;	/* SACK enabled/disabled */
515 
516 struct tcpcb *
517 	 tcp_close(struct tcpcb *);
518 void	 tcp_twstart(struct tcpcb *);
519 int	 tcp_twrecycleable(struct tcptw *tw);
520 struct tcptw *
521 	 tcp_twclose(struct tcptw *_tw, int _reuse);
522 void	 tcp_ctlinput(int, struct sockaddr *, void *);
523 int	 tcp_ctloutput(struct socket *, struct sockopt *);
524 struct tcpcb *
525 	 tcp_drop(struct tcpcb *, int);
526 void	 tcp_drain(void);
527 void	 tcp_fasttimo(void);
528 void	 tcp_init(void);
529 void	 tcp_fini(void *);
530 void	 tcp_reass_init(void);
531 void	 tcp_input(struct mbuf *, int);
532 u_long	 tcp_maxmtu(struct in_conninfo *);
533 u_long	 tcp_maxmtu6(struct in_conninfo *);
534 void	 tcp_mss(struct tcpcb *, int);
535 int	 tcp_mssopt(struct in_conninfo *);
536 struct inpcb *
537 	 tcp_drop_syn_sent(struct inpcb *, int);
538 struct inpcb *
539 	 tcp_mtudisc(struct inpcb *, int);
540 struct tcpcb *
541 	 tcp_newtcpcb(struct inpcb *);
542 int	 tcp_output(struct tcpcb *);
543 struct inpcb *
544 	 tcp_quench(struct inpcb *, int);
545 void	 tcp_respond(struct tcpcb *, void *,
546 	    struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
547 int	 tcp_twrespond(struct tcptw *, int);
548 void	 tcp_setpersist(struct tcpcb *);
549 #ifdef TCP_SIGNATURE
550 int	 tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
551 #endif
552 void	 tcp_slowtimo(void);
553 struct tcptemp *
554 	 tcpip_maketemplate(struct inpcb *);
555 void	 tcpip_fillheaders(struct inpcb *, void *, void *);
556 struct tcpcb *
557 	 tcp_timers(struct tcpcb *, int);
558 void	 tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
559 void	 tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
560 void	 syncache_init(void);
561 void	 syncache_unreach(struct in_conninfo *, struct tcphdr *);
562 int	 syncache_expand(struct in_conninfo *, struct tcphdr *,
563 	     struct socket **, struct mbuf *);
564 int	 syncache_add(struct in_conninfo *, struct tcpopt *,
565 	     struct tcphdr *, struct socket **, struct mbuf *);
566 void	 syncache_chkrst(struct in_conninfo *, struct tcphdr *);
567 void	 syncache_badack(struct in_conninfo *);
568 /*
569  * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
570  */
571 void	 tcp_hc_init(void);
572 void	 tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
573 u_long	 tcp_hc_getmtu(struct in_conninfo *);
574 void	 tcp_hc_updatemtu(struct in_conninfo *, u_long);
575 void	 tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
576 
577 extern	struct pr_usrreqs tcp_usrreqs;
578 extern	u_long tcp_sendspace;
579 extern	u_long tcp_recvspace;
580 tcp_seq tcp_new_isn(struct tcpcb *);
581 
582 int	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
583 void	 tcp_update_sack_list(struct tcpcb *tp);
584 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
585 void	 tcp_clean_sackreport(struct tcpcb *tp);
586 void	 tcp_sack_adjust(struct tcpcb *tp);
587 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
588 void	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
589 void	 tcp_free_sackholes(struct tcpcb *tp);
590 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
591 u_long	 tcp_seq_subtract(u_long, u_long );
592 #ifdef TCP_SACK_DEBUG
593 void	 tcp_print_holes(struct tcpcb *tp);
594 #endif /* TCP_SACK_DEBUG */
595 
596 #endif /* _KERNEL */
597 
598 #endif /* _NETINET_TCP_VAR_H_ */
599