xref: /freebsd/sys/netinet/tcp_var.h (revision 1e413cf93298b5b97441a21d9a50fdcd0ee9945e)
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 	TAILQ_ENTRY(sackhole) scblink;	/* scoreboard linkage */
64 };
65 
66 struct sackhint {
67 	struct sackhole	*nexthole;
68 	int		sack_bytes_rexmit;
69 };
70 
71 struct tcptemp {
72 	u_char	tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
73 	struct	tcphdr tt_t;
74 };
75 
76 #define tcp6cb		tcpcb  /* for KAME src sync over BSD*'s */
77 
78 /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
79 #ifdef INET6
80 #define ND6_HINT(tp)						\
81 do {								\
82 	if ((tp) && (tp)->t_inpcb &&				\
83 	    ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0)		\
84 		nd6_nud_hint(NULL, NULL, 0);			\
85 } while (0)
86 #else
87 #define ND6_HINT(tp)
88 #endif
89 
90 /*
91  * Tcp control block, one per tcp; fields:
92  * Organized for 16 byte cacheline efficiency.
93  */
94 struct tcpcb {
95 	struct	tsegqe_head t_segq;	/* segment reassembly queue */
96 	int	t_segqlen;		/* segment reassembly queue length */
97 	int	t_dupacks;		/* consecutive dup acks recd */
98 
99 	struct tcp_timer *t_timers;	/* All the TCP timers in one struct */
100 
101 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
102 	int	t_state;		/* state of this connection */
103 	u_int	t_flags;
104 #define	TF_ACKNOW	0x000001	/* ack peer immediately */
105 #define	TF_DELACK	0x000002	/* ack, but try to delay it */
106 #define	TF_NODELAY	0x000004	/* don't delay packets to coalesce */
107 #define	TF_NOOPT	0x000008	/* don't use tcp options */
108 #define	TF_SENTFIN	0x000010	/* have sent FIN */
109 #define	TF_REQ_SCALE	0x000020	/* have/will request window scaling */
110 #define	TF_RCVD_SCALE	0x000040	/* other side has requested scaling */
111 #define	TF_REQ_TSTMP	0x000080	/* have/will request timestamps */
112 #define	TF_RCVD_TSTMP	0x000100	/* a timestamp was received in SYN */
113 #define	TF_SACK_PERMIT	0x000200	/* other side said I could SACK */
114 #define	TF_NEEDSYN	0x000400	/* send SYN (implicit state) */
115 #define	TF_NEEDFIN	0x000800	/* send FIN (implicit state) */
116 #define	TF_NOPUSH	0x001000	/* don't push */
117 #define	TF_MORETOCOME	0x010000	/* More data to be appended to sock */
118 #define	TF_LQ_OVERFLOW	0x020000	/* listen queue overflow */
119 #define	TF_LASTIDLE	0x040000	/* connection was previously idle */
120 #define	TF_RXWIN0SENT	0x080000	/* sent a receiver win 0 in response */
121 #define	TF_FASTRECOVERY	0x100000	/* in NewReno Fast Recovery */
122 #define	TF_WASFRECOVERY	0x200000	/* was in NewReno Fast Recovery */
123 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
124 #define	TF_FORCEDATA	0x800000	/* force out a byte */
125 #define	TF_TSO		0x1000000	/* TSO enabled on this connection */
126 #define	TF_TOE		0x2000000	/* this connection is offloaded */
127 
128 	tcp_seq	snd_una;		/* send unacknowledged */
129 	tcp_seq	snd_max;		/* highest sequence number sent;
130 					 * used to recognize retransmits
131 					 */
132 	tcp_seq	snd_nxt;		/* send next */
133 	tcp_seq	snd_up;			/* send urgent pointer */
134 
135 	tcp_seq	snd_wl1;		/* window update seg seq number */
136 	tcp_seq	snd_wl2;		/* window update seg ack number */
137 	tcp_seq	iss;			/* initial send sequence number */
138 	tcp_seq	irs;			/* initial receive sequence number */
139 
140 	tcp_seq	rcv_nxt;		/* receive next */
141 	tcp_seq	rcv_adv;		/* advertised window */
142 	u_long	rcv_wnd;		/* receive window */
143 	tcp_seq	rcv_up;			/* receive urgent pointer */
144 
145 	u_long	snd_wnd;		/* send window */
146 	u_long	snd_cwnd;		/* congestion-controlled window */
147 	u_long	snd_bwnd;		/* bandwidth-controlled window */
148 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
149 					 * for slow start exponential to
150 					 * linear switch
151 					 */
152 	u_long	snd_bandwidth;		/* calculated bandwidth or 0 */
153 	tcp_seq	snd_recover;		/* for use in NewReno Fast Recovery */
154 
155 	u_int	t_maxopd;		/* mss plus options */
156 
157 	u_long	t_rcvtime;		/* inactivity time */
158 	u_long	t_starttime;		/* time connection was established */
159 	int	t_rtttime;		/* round trip time */
160 	tcp_seq	t_rtseq;		/* sequence number being timed */
161 
162 	int	t_bw_rtttime;		/* used for bandwidth calculation */
163 	tcp_seq	t_bw_rtseq;		/* used for bandwidth calculation */
164 
165 	int	t_rxtcur;		/* current retransmit value (ticks) */
166 	u_int	t_maxseg;		/* maximum segment size */
167 	int	t_srtt;			/* smoothed round-trip time */
168 	int	t_rttvar;		/* variance in round-trip time */
169 
170 	int	t_rxtshift;		/* log(2) of rexmt exp. backoff */
171 	u_int	t_rttmin;		/* minimum rtt allowed */
172 	u_int	t_rttbest;		/* best rtt we've seen */
173 	u_long	t_rttupdated;		/* number of times rtt sampled */
174 	u_long	max_sndwnd;		/* largest window peer has offered */
175 
176 	int	t_softerror;		/* possible error not yet reported */
177 /* out-of-band data */
178 	char	t_oobflags;		/* have some */
179 	char	t_iobc;			/* input character */
180 #define	TCPOOB_HAVEDATA	0x01
181 #define	TCPOOB_HADDATA	0x02
182 /* RFC 1323 variables */
183 	u_char	snd_scale;		/* window scaling for send window */
184 	u_char	rcv_scale;		/* window scaling for recv window */
185 	u_char	request_r_scale;	/* pending window scaling */
186 	u_int32_t  ts_recent;		/* timestamp echo data */
187 	u_long	ts_recent_age;		/* when last updated */
188 	u_int32_t  ts_offset;		/* our timestamp offset */
189 
190 	tcp_seq	last_ack_sent;
191 /* experimental */
192 	u_long	snd_cwnd_prev;		/* cwnd prior to retransmit */
193 	u_long	snd_ssthresh_prev;	/* ssthresh prior to retransmit */
194 	tcp_seq	snd_recover_prev;	/* snd_recover prior to retransmit */
195 	u_long	t_badrxtwin;		/* window for retransmit recovery */
196 	u_char	snd_limited;		/* segments limited transmitted */
197 /* SACK related state */
198 	int	snd_numholes;		/* number of holes seen by sender */
199 	TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
200 					/* SACK scoreboard (sorted) */
201 	tcp_seq	snd_fack;		/* last seq number(+1) sack'd by rcv'r*/
202 	int	rcv_numsacks;		/* # distinct sack blks present */
203 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
204 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
205 					   episode starts at this seq number */
206 	struct sackhint	sackhint;	/* SACK scoreboard hint */
207 	int	t_rttlow;		/* smallest observerved RTT */
208 	u_int32_t	rfbuf_ts;	/* recv buffer autoscaling timestamp */
209 	int	rfbuf_cnt;		/* recv buffer autoscaling byte count */
210 	void	*t_pspare[3];		/* toe usrreqs / toepcb * / congestion algo / vimage / 1 general use */
211 	struct toe_usrreqs *t_tu;       /* offload operations vector */
212 	void	*t_toe;			/* TOE pcb pointer */
213 };
214 
215 #define IN_FASTRECOVERY(tp)	(tp->t_flags & TF_FASTRECOVERY)
216 #define ENTER_FASTRECOVERY(tp)	tp->t_flags |= TF_FASTRECOVERY
217 #define EXIT_FASTRECOVERY(tp)	tp->t_flags &= ~TF_FASTRECOVERY
218 
219 #ifdef TCP_SIGNATURE
220 /*
221  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
222  * for SADB verification and lookup.
223  */
224 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
225 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
226 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
227 /*
228  * Only a single SA per host may be specified at this time. An SPI is
229  * needed in order for the KEY_ALLOCSA() lookup to work.
230  */
231 #define	TCP_SIG_SPI	0x1000
232 #endif /* TCP_SIGNATURE */
233 
234 /*
235  * Structure to hold TCP options that are only used during segment
236  * processing (in tcp_input), but not held in the tcpcb.
237  * It's basically used to reduce the number of parameters
238  * to tcp_dooptions and tcp_addoptions.
239  * The binary order of the to_flags is relevant for packing of the
240  * options in tcp_addoptions.
241  */
242 struct tcpopt {
243 	u_long		to_flags;	/* which options are present */
244 #define	TOF_MSS		0x0001		/* maximum segment size */
245 #define	TOF_SCALE	0x0002		/* window scaling */
246 #define	TOF_SACKPERM	0x0004		/* SACK permitted */
247 #define	TOF_TS		0x0010		/* timestamp */
248 #define	TOF_SIGNATURE	0x0040		/* TCP-MD5 signature option (RFC2385) */
249 #define	TOF_SACK	0x0080		/* Peer sent SACK option */
250 #define	TOF_MAXOPT	0x0100
251 	u_int32_t	to_tsval;	/* new timestamp */
252 	u_int32_t	to_tsecr;	/* reflected timestamp */
253 	u_int16_t	to_mss;		/* maximum segment size */
254 	u_int8_t	to_wscale;	/* window scaling */
255 	u_int8_t	to_nsacks;	/* number of SACK blocks */
256 	u_char		*to_sacks;	/* pointer to the first SACK blocks */
257 	u_char		*to_signature;	/* pointer to the TCP-MD5 signature */
258 };
259 
260 /*
261  * Flags for tcp_dooptions.
262  */
263 #define	TO_SYN		0x01		/* parse SYN-only options */
264 
265 struct hc_metrics_lite {	/* must stay in sync with hc_metrics */
266 	u_long	rmx_mtu;	/* MTU for this path */
267 	u_long	rmx_ssthresh;	/* outbound gateway buffer limit */
268 	u_long	rmx_rtt;	/* estimated round trip time */
269 	u_long	rmx_rttvar;	/* estimated rtt variance */
270 	u_long	rmx_bandwidth;	/* estimated bandwidth */
271 	u_long	rmx_cwnd;	/* congestion window */
272 	u_long	rmx_sendpipe;   /* outbound delay-bandwidth product */
273 	u_long	rmx_recvpipe;   /* inbound delay-bandwidth product */
274 };
275 
276 #ifndef _NETINET_IN_PCB_H_
277 struct in_conninfo;
278 #endif /* _NETINET_IN_PCB_H_ */
279 
280 struct tcptw {
281 	struct inpcb	*tw_inpcb;	/* XXX back pointer to internet pcb */
282 	tcp_seq		snd_nxt;
283 	tcp_seq		rcv_nxt;
284 	tcp_seq		iss;
285 	tcp_seq		irs;
286 	u_short		last_win;	/* cached window value */
287 	u_short		tw_so_options;	/* copy of so_options */
288 	struct ucred	*tw_cred;	/* user credentials */
289 	u_long		t_recent;
290 	u_int32_t	ts_offset;	/* our timestamp offset */
291 	u_long		t_starttime;
292 	int		tw_time;
293 	TAILQ_ENTRY(tcptw) tw_2msl;
294 };
295 
296 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
297 #define	intotw(ip)	((struct tcptw *)(ip)->inp_ppcb)
298 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
299 
300 /*
301  * The smoothed round-trip time and estimated variance
302  * are stored as fixed point numbers scaled by the values below.
303  * For convenience, these scales are also used in smoothing the average
304  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
305  * With these scales, srtt has 3 bits to the right of the binary point,
306  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
307  * binary point, and is smoothed with an ALPHA of 0.75.
308  */
309 #define	TCP_RTT_SCALE		32	/* multiplier for srtt; 3 bits frac. */
310 #define	TCP_RTT_SHIFT		5	/* shift for srtt; 3 bits frac. */
311 #define	TCP_RTTVAR_SCALE	16	/* multiplier for rttvar; 2 bits */
312 #define	TCP_RTTVAR_SHIFT	4	/* shift for rttvar; 2 bits */
313 #define	TCP_DELTA_SHIFT		2	/* see tcp_input.c */
314 
315 /*
316  * The initial retransmission should happen at rtt + 4 * rttvar.
317  * Because of the way we do the smoothing, srtt and rttvar
318  * will each average +1/2 tick of bias.  When we compute
319  * the retransmit timer, we want 1/2 tick of rounding and
320  * 1 extra tick because of +-1/2 tick uncertainty in the
321  * firing of the timer.  The bias will give us exactly the
322  * 1.5 tick we need.  But, because the bias is
323  * statistical, we have to test that we don't drop below
324  * the minimum feasible timer (which is 2 ticks).
325  * This version of the macro adapted from a paper by Lawrence
326  * Brakmo and Larry Peterson which outlines a problem caused
327  * by insufficient precision in the original implementation,
328  * which results in inappropriately large RTO values for very
329  * fast networks.
330  */
331 #define	TCP_REXMTVAL(tp) \
332 	max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
333 	  + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
334 
335 /*
336  * TCP statistics.
337  * Many of these should be kept per connection,
338  * but that's inconvenient at the moment.
339  */
340 struct	tcpstat {
341 	u_long	tcps_connattempt;	/* connections initiated */
342 	u_long	tcps_accepts;		/* connections accepted */
343 	u_long	tcps_connects;		/* connections established */
344 	u_long	tcps_drops;		/* connections dropped */
345 	u_long	tcps_conndrops;		/* embryonic connections dropped */
346 	u_long	tcps_minmssdrops;	/* average minmss too low drops */
347 	u_long	tcps_closed;		/* conn. closed (includes drops) */
348 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
349 	u_long	tcps_rttupdated;	/* times we succeeded */
350 	u_long	tcps_delack;		/* delayed acks sent */
351 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
352 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
353 	u_long	tcps_persisttimeo;	/* persist timeouts */
354 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
355 	u_long	tcps_keepprobe;		/* keepalive probes sent */
356 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
357 
358 	u_long	tcps_sndtotal;		/* total packets sent */
359 	u_long	tcps_sndpack;		/* data packets sent */
360 	u_long	tcps_sndbyte;		/* data bytes sent */
361 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
362 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
363 	u_long	tcps_sndrexmitbad;	/* unnecessary packet retransmissions */
364 	u_long	tcps_sndacks;		/* ack-only packets sent */
365 	u_long	tcps_sndprobe;		/* window probes sent */
366 	u_long	tcps_sndurg;		/* packets sent with URG only */
367 	u_long	tcps_sndwinup;		/* window update-only packets sent */
368 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
369 
370 	u_long	tcps_rcvtotal;		/* total packets received */
371 	u_long	tcps_rcvpack;		/* packets received in sequence */
372 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
373 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
374 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
375 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
376 	u_long	tcps_rcvshort;		/* packets received too short */
377 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
378 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
379 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
380 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
381 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
382 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
383 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
384 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
385 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
386 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
387 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
388 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
389 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
390 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
391 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
392 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
393 	u_long	tcps_predack;		/* times hdr predict ok for acks */
394 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
395 	u_long	tcps_pcbcachemiss;
396 	u_long	tcps_cachedrtt;		/* times cached RTT in route updated */
397 	u_long	tcps_cachedrttvar;	/* times cached rttvar updated */
398 	u_long	tcps_cachedssthresh;	/* times cached ssthresh updated */
399 	u_long	tcps_usedrtt;		/* times RTT initialized from route */
400 	u_long	tcps_usedrttvar;	/* times RTTVAR initialized from rt */
401 	u_long	tcps_usedssthresh;	/* times ssthresh initialized from rt*/
402 	u_long	tcps_persistdrop;	/* timeout in persist state */
403 	u_long	tcps_badsyn;		/* bogus SYN, e.g. premature ACK */
404 	u_long	tcps_mturesent;		/* resends due to MTU discovery */
405 	u_long	tcps_listendrop;	/* listen queue overflows */
406 	u_long	tcps_badrst;		/* ignored RSTs in the window */
407 
408 	u_long	tcps_sc_added;		/* entry added to syncache */
409 	u_long	tcps_sc_retransmitted;	/* syncache entry was retransmitted */
410 	u_long	tcps_sc_dupsyn;		/* duplicate SYN packet */
411 	u_long	tcps_sc_dropped;	/* could not reply to packet */
412 	u_long	tcps_sc_completed;	/* successful extraction of entry */
413 	u_long	tcps_sc_bucketoverflow;	/* syncache per-bucket limit hit */
414 	u_long	tcps_sc_cacheoverflow;	/* syncache cache limit hit */
415 	u_long	tcps_sc_reset;		/* RST removed entry from syncache */
416 	u_long	tcps_sc_stale;		/* timed out or listen socket gone */
417 	u_long	tcps_sc_aborted;	/* syncache entry aborted */
418 	u_long	tcps_sc_badack;		/* removed due to bad ACK */
419 	u_long	tcps_sc_unreach;	/* ICMP unreachable received */
420 	u_long	tcps_sc_zonefail;	/* zalloc() failed */
421 	u_long	tcps_sc_sendcookie;	/* SYN cookie sent */
422 	u_long	tcps_sc_recvcookie;	/* SYN cookie received */
423 
424 	u_long	tcps_hc_added;		/* entry added to hostcache */
425 	u_long	tcps_hc_bucketoverflow;	/* hostcache per bucket limit hit */
426 
427 	u_long  tcps_finwait2_drops;    /* Drop FIN_WAIT_2 connection after time limit */
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 #define TCPCTL_FINWAIT2_TIMEOUT        17
472 
473 #define TCPCTL_NAMES { \
474 	{ 0, 0 }, \
475 	{ "rfc1323", CTLTYPE_INT }, \
476 	{ "mssdflt", CTLTYPE_INT }, \
477 	{ "stats", CTLTYPE_STRUCT }, \
478 	{ "rttdflt", CTLTYPE_INT }, \
479 	{ "keepidle", CTLTYPE_INT }, \
480 	{ "keepintvl", CTLTYPE_INT }, \
481 	{ "sendspace", CTLTYPE_INT }, \
482 	{ "recvspace", CTLTYPE_INT }, \
483 	{ "keepinit", CTLTYPE_INT }, \
484 	{ "pcblist", CTLTYPE_STRUCT }, \
485 	{ "delacktime", CTLTYPE_INT }, \
486 	{ "v6mssdflt", CTLTYPE_INT }, \
487 	{ "maxid", CTLTYPE_INT }, \
488 }
489 
490 
491 #ifdef _KERNEL
492 #ifdef SYSCTL_DECL
493 SYSCTL_DECL(_net_inet_tcp);
494 SYSCTL_DECL(_net_inet_tcp_sack);
495 MALLOC_DECLARE(M_TCPLOG);
496 #endif
497 
498 extern	struct inpcbhead tcb;		/* head of queue of active tcpcb's */
499 extern	struct inpcbinfo tcbinfo;
500 extern	struct tcpstat tcpstat;	/* tcp statistics */
501 extern	int tcp_log_in_vain;
502 extern	int tcp_mssdflt;	/* XXX */
503 extern	int tcp_minmss;
504 extern	int tcp_delack_enabled;
505 extern	int tcp_do_newreno;
506 extern	int path_mtu_discovery;
507 extern	int ss_fltsz;
508 extern	int ss_fltsz_local;
509 
510 extern	int tcp_do_sack;		/* SACK enabled/disabled */
511 extern	int tcp_sc_rst_sock_fail;	/* RST on sock alloc failure */
512 
513 int	 tcp_addoptions(struct tcpopt *, u_char *);
514 struct tcpcb *
515 	 tcp_close(struct tcpcb *);
516 void	 tcp_discardcb(struct tcpcb *);
517 void	 tcp_twstart(struct tcpcb *);
518 #if 0
519 int	 tcp_twrecycleable(struct tcptw *tw);
520 #endif
521 void	 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 char 	*tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
531 	    const void *);
532 int	 tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *);
533 void	 tcp_reass_init(void);
534 void	 tcp_input(struct mbuf *, int);
535 u_long	 tcp_maxmtu(struct in_conninfo *, int *);
536 u_long	 tcp_maxmtu6(struct in_conninfo *, int *);
537 void	 tcp_mss(struct tcpcb *, int);
538 int	 tcp_mssopt(struct in_conninfo *);
539 struct inpcb *
540 	 tcp_drop_syn_sent(struct inpcb *, int);
541 struct inpcb *
542 	 tcp_mtudisc(struct inpcb *, int);
543 struct tcpcb *
544 	 tcp_newtcpcb(struct inpcb *);
545 int	 tcp_output(struct tcpcb *);
546 void	 tcp_respond(struct tcpcb *, void *,
547 	    struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
548 void	 tcp_tw_init(void);
549 void	 tcp_tw_zone_change(void);
550 int	 tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
551 	    struct mbuf *, int);
552 int	 tcp_twrespond(struct tcptw *, int);
553 void	 tcp_setpersist(struct tcpcb *);
554 #ifdef TCP_SIGNATURE
555 int	 tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
556 #endif
557 void	 tcp_slowtimo(void);
558 struct tcptemp *
559 	 tcpip_maketemplate(struct inpcb *);
560 void	 tcpip_fillheaders(struct inpcb *, void *, void *);
561 void	 tcp_timer_activate(struct tcpcb *, int, u_int);
562 int	 tcp_timer_active(struct tcpcb *, int);
563 void	 tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
564 void	 tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
565 /*
566  * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
567  */
568 void	 tcp_hc_init(void);
569 void	 tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
570 u_long	 tcp_hc_getmtu(struct in_conninfo *);
571 void	 tcp_hc_updatemtu(struct in_conninfo *, u_long);
572 void	 tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
573 
574 extern	struct pr_usrreqs tcp_usrreqs;
575 extern	u_long tcp_sendspace;
576 extern	u_long tcp_recvspace;
577 tcp_seq tcp_new_isn(struct tcpcb *);
578 
579 void	 tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
580 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
581 void	 tcp_clean_sackreport(struct tcpcb *tp);
582 void	 tcp_sack_adjust(struct tcpcb *tp);
583 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
584 void	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
585 void	 tcp_free_sackholes(struct tcpcb *tp);
586 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
587 u_long	 tcp_seq_subtract(u_long, u_long );
588 
589 #endif /* _KERNEL */
590 
591 #endif /* _NETINET_TCP_VAR_H_ */
592