xref: /freebsd/sys/netinet/tcp_var.h (revision ceaec73d406831b1251babb61675df0a1aa54a31)
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_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
191 	int	rcv_numsacks;		/* # distinct sack blks present */
192 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
193 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
194 					   episode starts at this seq number */
195 };
196 
197 #define IN_FASTRECOVERY(tp)	(tp->t_flags & TF_FASTRECOVERY)
198 #define ENTER_FASTRECOVERY(tp)	tp->t_flags |= TF_FASTRECOVERY
199 #define EXIT_FASTRECOVERY(tp)	tp->t_flags &= ~TF_FASTRECOVERY
200 
201 #ifdef TCP_SIGNATURE
202 /*
203  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
204  * for SADB verification and lookup.
205  */
206 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
207 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
208 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
209 /*
210  * Only a single SA per host may be specified at this time. An SPI is
211  * needed in order for the KEY_ALLOCSA() lookup to work.
212  */
213 #define	TCP_SIG_SPI	0x1000
214 #endif /* TCP_SIGNATURE */
215 
216 /*
217  * Structure to hold TCP options that are only used during segment
218  * processing (in tcp_input), but not held in the tcpcb.
219  * It's basically used to reduce the number of parameters
220  * to tcp_dooptions.
221  */
222 struct tcpopt {
223 	u_long		to_flags;	/* which options are present */
224 #define TOF_TS		0x0001		/* timestamp */
225 #define	TOF_MSS		0x0010
226 #define	TOF_SCALE	0x0020
227 #define	TOF_SIGNATURE	0x0040		/* signature option present */
228 #define	TOF_SIGLEN	0x0080		/* signature length valid (RFC2385) */
229 #define	TOF_SACK	0x0100		/* Peer sent SACK option */
230 	u_int32_t	to_tsval;
231 	u_int32_t	to_tsecr;
232 	u_int16_t	to_mss;
233 	u_int8_t	to_requested_s_scale;
234 	u_int8_t	to_pad;
235 };
236 
237 #ifdef _NETINET_IN_PCB_H_
238 struct syncache {
239 	inp_gen_t	sc_inp_gencnt;		/* pointer check */
240 	struct		tcpcb *sc_tp;		/* tcb for listening socket */
241 	struct		mbuf *sc_ipopts;	/* source route */
242 	struct		in_conninfo sc_inc;	/* addresses */
243 	u_int32_t	sc_tsrecent;
244 	u_int32_t	sc_flowlabel;		/* IPv6 flowlabel */
245 	tcp_seq		sc_irs;			/* seq from peer */
246 	tcp_seq		sc_iss;			/* our ISS */
247 	u_long		sc_rxttime;		/* retransmit time */
248 	u_int16_t	sc_rxtslot;		/* retransmit counter */
249 	u_int16_t	sc_peer_mss;		/* peer's MSS */
250 	u_int16_t	sc_wnd;			/* advertised window */
251 	u_int8_t	sc_requested_s_scale:4,
252 			sc_request_r_scale:4;
253 	u_int8_t	sc_flags;
254 #define SCF_NOOPT	0x01			/* no TCP options */
255 #define SCF_WINSCALE	0x02			/* negotiated window scaling */
256 #define SCF_TIMESTAMP	0x04			/* negotiated timestamps */
257 #define SCF_UNREACH	0x10			/* icmp unreachable received */
258 #define SCF_SIGNATURE	0x20			/* send MD5 digests */
259 #define SCF_SACK	0x80			/* send SACK option */
260 	TAILQ_ENTRY(syncache)	sc_hash;
261 	TAILQ_ENTRY(syncache)	sc_timerq;
262 };
263 
264 struct syncache_head {
265 	TAILQ_HEAD(, syncache)	sch_bucket;
266 	u_int		sch_length;
267 };
268 #else
269 struct in_conninfo;
270 #endif /* _NETINET_IN_PCB_H_ */
271 
272 struct hc_metrics_lite {	/* must stay in sync with hc_metrics */
273 	u_long	rmx_mtu;	/* MTU for this path */
274 	u_long	rmx_ssthresh;	/* outbound gateway buffer limit */
275 	u_long	rmx_rtt;	/* estimated round trip time */
276 	u_long	rmx_rttvar;	/* estimated rtt variance */
277 	u_long	rmx_bandwidth;	/* estimated bandwidth */
278 	u_long	rmx_cwnd;	/* congestion window */
279 	u_long	rmx_sendpipe;   /* outbound delay-bandwidth product */
280 	u_long	rmx_recvpipe;   /* inbound delay-bandwidth product */
281 };
282 
283 struct tcptw {
284 	struct inpcb	*tw_inpcb;	/* XXX back pointer to internet pcb */
285 	tcp_seq		snd_nxt;
286 	tcp_seq		rcv_nxt;
287 	tcp_seq		iss;
288 	tcp_seq		irs;
289 	u_short		last_win;	/* cached window value */
290 	u_short		tw_so_options;	/* copy of so_options */
291 	struct ucred	*tw_cred;	/* user credentials */
292 	u_long		t_recent;
293 	u_long		t_starttime;
294 	int		tw_time;
295 	LIST_ENTRY(tcptw) tw_2msl;
296 };
297 
298 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
299 #define	intotw(ip)	((struct tcptw *)(ip)->inp_ppcb)
300 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
301 
302 /*
303  * The smoothed round-trip time and estimated variance
304  * are stored as fixed point numbers scaled by the values below.
305  * For convenience, these scales are also used in smoothing the average
306  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
307  * With these scales, srtt has 3 bits to the right of the binary point,
308  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
309  * binary point, and is smoothed with an ALPHA of 0.75.
310  */
311 #define	TCP_RTT_SCALE		32	/* multiplier for srtt; 3 bits frac. */
312 #define	TCP_RTT_SHIFT		5	/* shift for srtt; 3 bits frac. */
313 #define	TCP_RTTVAR_SCALE	16	/* multiplier for rttvar; 2 bits */
314 #define	TCP_RTTVAR_SHIFT	4	/* shift for rttvar; 2 bits */
315 #define	TCP_DELTA_SHIFT		2	/* see tcp_input.c */
316 
317 /*
318  * The initial retransmission should happen at rtt + 4 * rttvar.
319  * Because of the way we do the smoothing, srtt and rttvar
320  * will each average +1/2 tick of bias.  When we compute
321  * the retransmit timer, we want 1/2 tick of rounding and
322  * 1 extra tick because of +-1/2 tick uncertainty in the
323  * firing of the timer.  The bias will give us exactly the
324  * 1.5 tick we need.  But, because the bias is
325  * statistical, we have to test that we don't drop below
326  * the minimum feasible timer (which is 2 ticks).
327  * This version of the macro adapted from a paper by Lawrence
328  * Brakmo and Larry Peterson which outlines a problem caused
329  * by insufficient precision in the original implementation,
330  * which results in inappropriately large RTO values for very
331  * fast networks.
332  */
333 #define	TCP_REXMTVAL(tp) \
334 	max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
335 	  + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
336 
337 /*
338  * TCP statistics.
339  * Many of these should be kept per connection,
340  * but that's inconvenient at the moment.
341  */
342 struct	tcpstat {
343 	u_long	tcps_connattempt;	/* connections initiated */
344 	u_long	tcps_accepts;		/* connections accepted */
345 	u_long	tcps_connects;		/* connections established */
346 	u_long	tcps_drops;		/* connections dropped */
347 	u_long	tcps_conndrops;		/* embryonic connections dropped */
348 	u_long	tcps_minmssdrops;	/* average minmss too low drops */
349 	u_long	tcps_closed;		/* conn. closed (includes drops) */
350 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
351 	u_long	tcps_rttupdated;	/* times we succeeded */
352 	u_long	tcps_delack;		/* delayed acks sent */
353 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
354 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
355 	u_long	tcps_persisttimeo;	/* persist timeouts */
356 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
357 	u_long	tcps_keepprobe;		/* keepalive probes sent */
358 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
359 
360 	u_long	tcps_sndtotal;		/* total packets sent */
361 	u_long	tcps_sndpack;		/* data packets sent */
362 	u_long	tcps_sndbyte;		/* data bytes sent */
363 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
364 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
365 	u_long	tcps_sndrexmitbad;	/* unnecessary packet retransmissions */
366 	u_long	tcps_sndacks;		/* ack-only packets sent */
367 	u_long	tcps_sndprobe;		/* window probes sent */
368 	u_long	tcps_sndurg;		/* packets sent with URG only */
369 	u_long	tcps_sndwinup;		/* window update-only packets sent */
370 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
371 
372 	u_long	tcps_rcvtotal;		/* total packets received */
373 	u_long	tcps_rcvpack;		/* packets received in sequence */
374 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
375 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
376 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
377 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
378 	u_long	tcps_rcvshort;		/* packets received too short */
379 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
380 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
381 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
382 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
383 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
384 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
385 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
386 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
387 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
388 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
389 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
390 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
391 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
392 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
393 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
394 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
395 	u_long	tcps_predack;		/* times hdr predict ok for acks */
396 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
397 	u_long	tcps_pcbcachemiss;
398 	u_long	tcps_cachedrtt;		/* times cached RTT in route updated */
399 	u_long	tcps_cachedrttvar;	/* times cached rttvar updated */
400 	u_long	tcps_cachedssthresh;	/* times cached ssthresh updated */
401 	u_long	tcps_usedrtt;		/* times RTT initialized from route */
402 	u_long	tcps_usedrttvar;	/* times RTTVAR initialized from rt */
403 	u_long	tcps_usedssthresh;	/* times ssthresh initialized from rt*/
404 	u_long	tcps_persistdrop;	/* timeout in persist state */
405 	u_long	tcps_badsyn;		/* bogus SYN, e.g. premature ACK */
406 	u_long	tcps_mturesent;		/* resends due to MTU discovery */
407 	u_long	tcps_listendrop;	/* listen queue overflows */
408 	u_long	tcps_badrst;		/* ignored RSTs in the window */
409 
410 	u_long	tcps_sc_added;		/* entry added to syncache */
411 	u_long	tcps_sc_retransmitted;	/* syncache entry was retransmitted */
412 	u_long	tcps_sc_dupsyn;		/* duplicate SYN packet */
413 	u_long	tcps_sc_dropped;	/* could not reply to packet */
414 	u_long	tcps_sc_completed;	/* successful extraction of entry */
415 	u_long	tcps_sc_bucketoverflow;	/* syncache per-bucket limit hit */
416 	u_long	tcps_sc_cacheoverflow;	/* syncache cache limit hit */
417 	u_long	tcps_sc_reset;		/* RST removed entry from syncache */
418 	u_long	tcps_sc_stale;		/* timed out or listen socket gone */
419 	u_long	tcps_sc_aborted;	/* syncache entry aborted */
420 	u_long	tcps_sc_badack;		/* removed due to bad ACK */
421 	u_long	tcps_sc_unreach;	/* ICMP unreachable received */
422 	u_long	tcps_sc_zonefail;	/* zalloc() failed */
423 	u_long	tcps_sc_sendcookie;	/* SYN cookie sent */
424 	u_long	tcps_sc_recvcookie;	/* SYN cookie received */
425 
426 	u_long	tcps_hc_added;		/* entry added to hostcache */
427 	u_long	tcps_hc_bucketoverflow;	/* hostcache per bucket limit hit */
428 
429 	/* SACK related stats */
430 	u_long	tcps_sack_recovery_episode; /* SACK recovery episodes */
431 	u_long  tcps_sack_rexmits;	    /* SACK rexmit segments   */
432 	u_long  tcps_sack_rexmit_bytes;	    /* SACK rexmit bytes      */
433 	u_long  tcps_sack_rcv_blocks;	    /* SACK blocks (options) received */
434 	u_long  tcps_sack_send_blocks;	    /* SACK blocks (options) sent     */
435 	u_long  tcps_sack_sboverflow; 	    /* times scoreboard overflowed */
436 };
437 
438 /*
439  * TCB structure exported to user-land via sysctl(3).
440  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
441  * included.  Not all of our clients do.
442  */
443 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
444 struct	xtcpcb {
445 	size_t	xt_len;
446 	struct	inpcb	xt_inp;
447 	struct	tcpcb	xt_tp;
448 	struct	xsocket	xt_socket;
449 	u_quad_t	xt_alignment_hack;
450 };
451 #endif
452 
453 /*
454  * Names for TCP sysctl objects
455  */
456 #define	TCPCTL_DO_RFC1323	1	/* use RFC-1323 extensions */
457 #define	TCPCTL_MSSDFLT		3	/* MSS default */
458 #define TCPCTL_STATS		4	/* statistics (read-only) */
459 #define	TCPCTL_RTTDFLT		5	/* default RTT estimate */
460 #define	TCPCTL_KEEPIDLE		6	/* keepalive idle timer */
461 #define	TCPCTL_KEEPINTVL	7	/* interval to send keepalives */
462 #define	TCPCTL_SENDSPACE	8	/* send buffer space */
463 #define	TCPCTL_RECVSPACE	9	/* receive buffer space */
464 #define	TCPCTL_KEEPINIT		10	/* timeout for establishing syn */
465 #define	TCPCTL_PCBLIST		11	/* list of all outstanding PCBs */
466 #define	TCPCTL_DELACKTIME	12	/* time before sending delayed ACK */
467 #define	TCPCTL_V6MSSDFLT	13	/* MSS default for IPv6 */
468 #define	TCPCTL_SACK		14	/* Selective Acknowledgement,rfc 2018 */
469 #define	TCPCTL_DROP		15	/* drop tcp connection */
470 #define	TCPCTL_MAXID		16
471 
472 #define TCPCTL_NAMES { \
473 	{ 0, 0 }, \
474 	{ "rfc1323", CTLTYPE_INT }, \
475 	{ "mssdflt", CTLTYPE_INT }, \
476 	{ "stats", CTLTYPE_STRUCT }, \
477 	{ "rttdflt", CTLTYPE_INT }, \
478 	{ "keepidle", CTLTYPE_INT }, \
479 	{ "keepintvl", CTLTYPE_INT }, \
480 	{ "sendspace", CTLTYPE_INT }, \
481 	{ "recvspace", CTLTYPE_INT }, \
482 	{ "keepinit", CTLTYPE_INT }, \
483 	{ "pcblist", CTLTYPE_STRUCT }, \
484 	{ "delacktime", CTLTYPE_INT }, \
485 	{ "v6mssdflt", CTLTYPE_INT }, \
486 	{ "maxid", CTLTYPE_INT }, \
487 }
488 
489 
490 #ifdef _KERNEL
491 #ifdef SYSCTL_DECL
492 SYSCTL_DECL(_net_inet_tcp);
493 SYSCTL_DECL(_net_inet_tcp_sack);
494 #endif
495 
496 extern	struct inpcbhead tcb;		/* head of queue of active tcpcb's */
497 extern	struct inpcbinfo tcbinfo;
498 extern	struct tcpstat tcpstat;	/* tcp statistics */
499 extern	int tcp_mssdflt;	/* XXX */
500 extern	int tcp_minmss;
501 extern	int tcp_minmssoverload;
502 extern	int tcp_delack_enabled;
503 extern	int tcp_do_newreno;
504 extern	int path_mtu_discovery;
505 extern	int ss_fltsz;
506 extern	int ss_fltsz_local;
507 
508 extern	int tcp_do_sack;	/* SACK enabled/disabled */
509 
510 struct tcpcb *
511 	 tcp_close(struct tcpcb *);
512 void	 tcp_twstart(struct tcpcb *);
513 int	 tcp_twrecycleable(struct tcptw *tw);
514 struct tcptw *
515 	 tcp_twclose(struct tcptw *_tw, int _reuse);
516 void	 tcp_ctlinput(int, struct sockaddr *, void *);
517 int	 tcp_ctloutput(struct socket *, struct sockopt *);
518 struct tcpcb *
519 	 tcp_drop(struct tcpcb *, int);
520 void	 tcp_drain(void);
521 void	 tcp_fasttimo(void);
522 void	 tcp_init(void);
523 void	 tcp_fini(void *);
524 void	 tcp_reass_init(void);
525 void	 tcp_input(struct mbuf *, int);
526 u_long	 tcp_maxmtu(struct in_conninfo *);
527 u_long	 tcp_maxmtu6(struct in_conninfo *);
528 void	 tcp_mss(struct tcpcb *, int);
529 int	 tcp_mssopt(struct in_conninfo *);
530 struct inpcb *
531 	 tcp_drop_syn_sent(struct inpcb *, int);
532 struct inpcb *
533 	 tcp_mtudisc(struct inpcb *, int);
534 struct tcpcb *
535 	 tcp_newtcpcb(struct inpcb *);
536 int	 tcp_output(struct tcpcb *);
537 struct inpcb *
538 	 tcp_quench(struct inpcb *, int);
539 void	 tcp_respond(struct tcpcb *, void *,
540 	    struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
541 int	 tcp_twrespond(struct tcptw *, int);
542 void	 tcp_setpersist(struct tcpcb *);
543 #ifdef TCP_SIGNATURE
544 int	 tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
545 #endif
546 void	 tcp_slowtimo(void);
547 struct tcptemp *
548 	 tcpip_maketemplate(struct inpcb *);
549 void	 tcpip_fillheaders(struct inpcb *, void *, void *);
550 struct tcpcb *
551 	 tcp_timers(struct tcpcb *, int);
552 void	 tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
553 void	 tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
554 void	 syncache_init(void);
555 void	 syncache_unreach(struct in_conninfo *, struct tcphdr *);
556 int	 syncache_expand(struct in_conninfo *, struct tcphdr *,
557 	     struct socket **, struct mbuf *);
558 int	 syncache_add(struct in_conninfo *, struct tcpopt *,
559 	     struct tcphdr *, struct socket **, struct mbuf *);
560 void	 syncache_chkrst(struct in_conninfo *, struct tcphdr *);
561 void	 syncache_badack(struct in_conninfo *);
562 /*
563  * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
564  */
565 void	 tcp_hc_init(void);
566 void	 tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
567 u_long	 tcp_hc_getmtu(struct in_conninfo *);
568 void	 tcp_hc_updatemtu(struct in_conninfo *, u_long);
569 void	 tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
570 
571 extern	struct pr_usrreqs tcp_usrreqs;
572 extern	u_long tcp_sendspace;
573 extern	u_long tcp_recvspace;
574 tcp_seq tcp_new_isn(struct tcpcb *);
575 
576 int	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
577 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
578 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
579 void	 tcp_clean_sackreport(struct tcpcb *tp);
580 void	 tcp_sack_adjust(struct tcpcb *tp);
581 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
582 int      tcp_sacked_bytes(struct tcpcb *tp, int *lost_not_rexmitted);
583 void	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
584 void	 tcp_free_sackholes(struct tcpcb *tp);
585 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
586 u_long	 tcp_seq_subtract(u_long, u_long );
587 #ifdef TCP_SACK_DEBUG
588 void	 tcp_print_holes(struct tcpcb *tp);
589 #endif /* TCP_SACK_DEBUG */
590 
591 #endif /* _KERNEL */
592 
593 #endif /* _NETINET_TCP_VAR_H_ */
594