xref: /freebsd/sys/netinet/tcp_var.h (revision 0e1497aefd602cea581d2380d22e67dfdcac6b4e)
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 #ifdef _KERNEL
39 #include <net/vnet.h>
40 
41 /*
42  * Kernel variables for tcp.
43  */
44 VNET_DECLARE(int, tcp_do_rfc1323);
45 #define	V_tcp_do_rfc1323	VNET(tcp_do_rfc1323)
46 
47 #endif /* _KERNEL */
48 
49 /* TCP segment queue entry */
50 struct tseg_qent {
51 	LIST_ENTRY(tseg_qent) tqe_q;
52 	int	tqe_len;		/* TCP segment data length */
53 	struct	tcphdr *tqe_th;		/* a pointer to tcp header */
54 	struct	mbuf	*tqe_m;		/* mbuf contains packet */
55 };
56 LIST_HEAD(tsegqe_head, tseg_qent);
57 
58 struct sackblk {
59 	tcp_seq start;		/* start seq no. of sack block */
60 	tcp_seq end;		/* end seq no. */
61 };
62 
63 struct sackhole {
64 	tcp_seq start;		/* start seq no. of hole */
65 	tcp_seq end;		/* end seq no. */
66 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
67 	TAILQ_ENTRY(sackhole) scblink;	/* scoreboard linkage */
68 };
69 
70 struct sackhint {
71 	struct sackhole	*nexthole;
72 	int		sack_bytes_rexmit;
73 
74 	int		ispare;		/* explicit pad for 64bit alignment */
75 	uint64_t	_pad[2];	/* 1 sacked_bytes, 1 TBD */
76 };
77 
78 struct tcptemp {
79 	u_char	tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
80 	struct	tcphdr tt_t;
81 };
82 
83 #define tcp6cb		tcpcb  /* for KAME src sync over BSD*'s */
84 
85 /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
86 #ifdef INET6
87 #define ND6_HINT(tp)						\
88 do {								\
89 	if ((tp) && (tp)->t_inpcb &&				\
90 	    ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0)		\
91 		nd6_nud_hint(NULL, NULL, 0);			\
92 } while (0)
93 #else
94 #define ND6_HINT(tp)
95 #endif
96 
97 /*
98  * Tcp control block, one per tcp; fields:
99  * Organized for 16 byte cacheline efficiency.
100  */
101 struct tcpcb {
102 	struct	tsegqe_head t_segq;	/* segment reassembly queue */
103 	void	*t_pspare[2];		/* new reassembly queue */
104 	int	t_segqlen;		/* segment reassembly queue length */
105 	int	t_dupacks;		/* consecutive dup acks recd */
106 
107 	struct tcp_timer *t_timers;	/* All the TCP timers in one struct */
108 
109 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
110 	int	t_state;		/* state of this connection */
111 	u_int	t_flags;
112 
113 	struct	vnet *t_vnet;		/* back pointer to parent vnet */
114 
115 	tcp_seq	snd_una;		/* send unacknowledged */
116 	tcp_seq	snd_max;		/* highest sequence number sent;
117 					 * used to recognize retransmits
118 					 */
119 	tcp_seq	snd_nxt;		/* send next */
120 	tcp_seq	snd_up;			/* send urgent pointer */
121 
122 	tcp_seq	snd_wl1;		/* window update seg seq number */
123 	tcp_seq	snd_wl2;		/* window update seg ack number */
124 	tcp_seq	iss;			/* initial send sequence number */
125 	tcp_seq	irs;			/* initial receive sequence number */
126 
127 	tcp_seq	rcv_nxt;		/* receive next */
128 	tcp_seq	rcv_adv;		/* advertised window */
129 	u_long	rcv_wnd;		/* receive window */
130 	tcp_seq	rcv_up;			/* receive urgent pointer */
131 
132 	u_long	snd_wnd;		/* send window */
133 	u_long	snd_cwnd;		/* congestion-controlled window */
134 	u_long	snd_spare1;		/* unused */
135 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
136 					 * for slow start exponential to
137 					 * linear switch
138 					 */
139 	u_long	snd_spare2;		/* unused */
140 	tcp_seq	snd_recover;		/* for use in NewReno Fast Recovery */
141 
142 	u_int	t_maxopd;		/* mss plus options */
143 
144 	u_int	t_rcvtime;		/* inactivity time */
145 	u_int	t_starttime;		/* time connection was established */
146 	u_int	t_rtttime;		/* RTT measurement start time */
147 	tcp_seq	t_rtseq;		/* sequence number being timed */
148 
149 	u_int	t_bw_spare1;		/* unused */
150 	tcp_seq	t_bw_spare2;		/* unused */
151 
152 	int	t_rxtcur;		/* current retransmit value (ticks) */
153 	u_int	t_maxseg;		/* maximum segment size */
154 	int	t_srtt;			/* smoothed round-trip time */
155 	int	t_rttvar;		/* variance in round-trip time */
156 
157 	int	t_rxtshift;		/* log(2) of rexmt exp. backoff */
158 	u_int	t_rttmin;		/* minimum rtt allowed */
159 	u_int	t_rttbest;		/* best rtt we've seen */
160 	u_long	t_rttupdated;		/* number of times rtt sampled */
161 	u_long	max_sndwnd;		/* largest window peer has offered */
162 
163 	int	t_softerror;		/* possible error not yet reported */
164 /* out-of-band data */
165 	char	t_oobflags;		/* have some */
166 	char	t_iobc;			/* input character */
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_int32_t  ts_recent;		/* timestamp echo data */
172 	u_int	ts_recent_age;		/* when last updated */
173 	u_int32_t  ts_offset;		/* our timestamp offset */
174 
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 	int	t_sndzerowin;		/* zero-window updates sent */
181 	u_int	t_badrxtwin;		/* window for retransmit recovery */
182 	u_char	snd_limited;		/* segments limited transmitted */
183 /* SACK related state */
184 	int	snd_numholes;		/* number of holes seen by sender */
185 	TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
186 					/* SACK scoreboard (sorted) */
187 	tcp_seq	snd_fack;		/* last seq number(+1) sack'd by rcv'r*/
188 	int	rcv_numsacks;		/* # distinct sack blks present */
189 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
190 	tcp_seq sack_newdata;		/* New data xmitted in this recovery
191 					   episode starts at this seq number */
192 	struct sackhint	sackhint;	/* SACK scoreboard hint */
193 	int	t_rttlow;		/* smallest observerved RTT */
194 	u_int32_t	rfbuf_ts;	/* recv buffer autoscaling timestamp */
195 	int	rfbuf_cnt;		/* recv buffer autoscaling byte count */
196 	struct toe_usrreqs *t_tu;	/* offload operations vector */
197 	int	t_sndrexmitpack;	/* retransmit packets sent */
198 	int	t_rcvoopack;		/* out-of-order packets received */
199 	void	*t_toe;			/* TOE pcb pointer */
200 	int	t_bytes_acked;		/* # bytes acked during current RTT */
201 	struct cc_algo	*cc_algo;	/* congestion control algorithm */
202 	struct cc_var	*ccv;
203 
204 	int	t_ispare;		/* explicit pad for 64bit alignment */
205 	void	*t_pspare2[4];		/* 4 TBD */
206 	uint64_t _pad[12];		/* 7 UTO, 5 TBD (1-2 CC/RTT?) */
207 };
208 
209 /*
210  * Flags and utility macros for the t_flags field.
211  */
212 #define	TF_ACKNOW	0x000001	/* ack peer immediately */
213 #define	TF_DELACK	0x000002	/* ack, but try to delay it */
214 #define	TF_NODELAY	0x000004	/* don't delay packets to coalesce */
215 #define	TF_NOOPT	0x000008	/* don't use tcp options */
216 #define	TF_SENTFIN	0x000010	/* have sent FIN */
217 #define	TF_REQ_SCALE	0x000020	/* have/will request window scaling */
218 #define	TF_RCVD_SCALE	0x000040	/* other side has requested scaling */
219 #define	TF_REQ_TSTMP	0x000080	/* have/will request timestamps */
220 #define	TF_RCVD_TSTMP	0x000100	/* a timestamp was received in SYN */
221 #define	TF_SACK_PERMIT	0x000200	/* other side said I could SACK */
222 #define	TF_NEEDSYN	0x000400	/* send SYN (implicit state) */
223 #define	TF_NEEDFIN	0x000800	/* send FIN (implicit state) */
224 #define	TF_NOPUSH	0x001000	/* don't push */
225 #define	TF_MORETOCOME	0x010000	/* More data to be appended to sock */
226 #define	TF_LQ_OVERFLOW	0x020000	/* listen queue overflow */
227 #define	TF_LASTIDLE	0x040000	/* connection was previously idle */
228 #define	TF_RXWIN0SENT	0x080000	/* sent a receiver win 0 in response */
229 #define	TF_FASTRECOVERY	0x100000	/* in NewReno Fast Recovery */
230 #define	TF_WASFRECOVERY	0x200000	/* was in NewReno Fast Recovery */
231 #define	TF_SIGNATURE	0x400000	/* require MD5 digests (RFC2385) */
232 #define	TF_FORCEDATA	0x800000	/* force out a byte */
233 #define	TF_TSO		0x1000000	/* TSO enabled on this connection */
234 #define	TF_TOE		0x2000000	/* this connection is offloaded */
235 #define	TF_ECN_PERMIT	0x4000000	/* connection ECN-ready */
236 #define	TF_ECN_SND_CWR	0x8000000	/* ECN CWR in queue */
237 #define	TF_ECN_SND_ECE	0x10000000	/* ECN ECE in queue */
238 #define	TF_CONGRECOVERY	0x20000000	/* congestion recovery mode */
239 #define	TF_WASCRECOVERY	0x40000000	/* was in congestion recovery */
240 
241 #define	IN_FASTRECOVERY(t_flags)	(t_flags & TF_FASTRECOVERY)
242 #define	ENTER_FASTRECOVERY(t_flags)	t_flags |= TF_FASTRECOVERY
243 #define	EXIT_FASTRECOVERY(t_flags)	t_flags &= ~TF_FASTRECOVERY
244 
245 #define	IN_CONGRECOVERY(t_flags)	(t_flags & TF_CONGRECOVERY)
246 #define	ENTER_CONGRECOVERY(t_flags)	t_flags |= TF_CONGRECOVERY
247 #define	EXIT_CONGRECOVERY(t_flags)	t_flags &= ~TF_CONGRECOVERY
248 
249 #define	IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY))
250 #define	ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY)
251 #define	EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY)
252 
253 #define	BYTES_THIS_ACK(tp, th)	(th->th_ack - tp->snd_una)
254 
255 /*
256  * Flags for the t_oobflags field.
257  */
258 #define	TCPOOB_HAVEDATA	0x01
259 #define	TCPOOB_HADDATA	0x02
260 
261 #ifdef TCP_SIGNATURE
262 /*
263  * Defines which are needed by the xform_tcp module and tcp_[in|out]put
264  * for SADB verification and lookup.
265  */
266 #define	TCP_SIGLEN	16	/* length of computed digest in bytes */
267 #define	TCP_KEYLEN_MIN	1	/* minimum length of TCP-MD5 key */
268 #define	TCP_KEYLEN_MAX	80	/* maximum length of TCP-MD5 key */
269 /*
270  * Only a single SA per host may be specified at this time. An SPI is
271  * needed in order for the KEY_ALLOCSA() lookup to work.
272  */
273 #define	TCP_SIG_SPI	0x1000
274 #endif /* TCP_SIGNATURE */
275 
276 /*
277  * Structure to hold TCP options that are only used during segment
278  * processing (in tcp_input), but not held in the tcpcb.
279  * It's basically used to reduce the number of parameters
280  * to tcp_dooptions and tcp_addoptions.
281  * The binary order of the to_flags is relevant for packing of the
282  * options in tcp_addoptions.
283  */
284 struct tcpopt {
285 	u_int64_t	to_flags;	/* which options are present */
286 #define	TOF_MSS		0x0001		/* maximum segment size */
287 #define	TOF_SCALE	0x0002		/* window scaling */
288 #define	TOF_SACKPERM	0x0004		/* SACK permitted */
289 #define	TOF_TS		0x0010		/* timestamp */
290 #define	TOF_SIGNATURE	0x0040		/* TCP-MD5 signature option (RFC2385) */
291 #define	TOF_SACK	0x0080		/* Peer sent SACK option */
292 #define	TOF_MAXOPT	0x0100
293 	u_int32_t	to_tsval;	/* new timestamp */
294 	u_int32_t	to_tsecr;	/* reflected timestamp */
295 	u_char		*to_sacks;	/* pointer to the first SACK blocks */
296 	u_char		*to_signature;	/* pointer to the TCP-MD5 signature */
297 	u_int16_t	to_mss;		/* maximum segment size */
298 	u_int8_t	to_wscale;	/* window scaling */
299 	u_int8_t	to_nsacks;	/* number of SACK blocks */
300 };
301 
302 /*
303  * Flags for tcp_dooptions.
304  */
305 #define	TO_SYN		0x01		/* parse SYN-only options */
306 
307 struct hc_metrics_lite {	/* must stay in sync with hc_metrics */
308 	u_long	rmx_mtu;	/* MTU for this path */
309 	u_long	rmx_ssthresh;	/* outbound gateway buffer limit */
310 	u_long	rmx_rtt;	/* estimated round trip time */
311 	u_long	rmx_rttvar;	/* estimated rtt variance */
312 	u_long	rmx_bandwidth;	/* estimated bandwidth */
313 	u_long	rmx_cwnd;	/* congestion window */
314 	u_long	rmx_sendpipe;   /* outbound delay-bandwidth product */
315 	u_long	rmx_recvpipe;   /* inbound delay-bandwidth product */
316 };
317 
318 #ifndef _NETINET_IN_PCB_H_
319 struct in_conninfo;
320 #endif /* _NETINET_IN_PCB_H_ */
321 
322 struct tcptw {
323 	struct inpcb	*tw_inpcb;	/* XXX back pointer to internet pcb */
324 	tcp_seq		snd_nxt;
325 	tcp_seq		rcv_nxt;
326 	tcp_seq		iss;
327 	tcp_seq		irs;
328 	u_short		last_win;	/* cached window value */
329 	u_short		tw_so_options;	/* copy of so_options */
330 	struct ucred	*tw_cred;	/* user credentials */
331 	u_int32_t	t_recent;
332 	u_int32_t	ts_offset;	/* our timestamp offset */
333 	u_int		t_starttime;
334 	int		tw_time;
335 	TAILQ_ENTRY(tcptw) tw_2msl;
336 };
337 
338 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
339 #define	intotw(ip)	((struct tcptw *)(ip)->inp_ppcb)
340 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
341 
342 /*
343  * The smoothed round-trip time and estimated variance
344  * are stored as fixed point numbers scaled by the values below.
345  * For convenience, these scales are also used in smoothing the average
346  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
347  * With these scales, srtt has 3 bits to the right of the binary point,
348  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
349  * binary point, and is smoothed with an ALPHA of 0.75.
350  */
351 #define	TCP_RTT_SCALE		32	/* multiplier for srtt; 3 bits frac. */
352 #define	TCP_RTT_SHIFT		5	/* shift for srtt; 3 bits frac. */
353 #define	TCP_RTTVAR_SCALE	16	/* multiplier for rttvar; 2 bits */
354 #define	TCP_RTTVAR_SHIFT	4	/* shift for rttvar; 2 bits */
355 #define	TCP_DELTA_SHIFT		2	/* see tcp_input.c */
356 
357 /*
358  * The initial retransmission should happen at rtt + 4 * rttvar.
359  * Because of the way we do the smoothing, srtt and rttvar
360  * will each average +1/2 tick of bias.  When we compute
361  * the retransmit timer, we want 1/2 tick of rounding and
362  * 1 extra tick because of +-1/2 tick uncertainty in the
363  * firing of the timer.  The bias will give us exactly the
364  * 1.5 tick we need.  But, because the bias is
365  * statistical, we have to test that we don't drop below
366  * the minimum feasible timer (which is 2 ticks).
367  * This version of the macro adapted from a paper by Lawrence
368  * Brakmo and Larry Peterson which outlines a problem caused
369  * by insufficient precision in the original implementation,
370  * which results in inappropriately large RTO values for very
371  * fast networks.
372  */
373 #define	TCP_REXMTVAL(tp) \
374 	max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
375 	  + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
376 
377 /*
378  * TCP statistics.
379  * Many of these should be kept per connection,
380  * but that's inconvenient at the moment.
381  */
382 struct	tcpstat {
383 	u_long	tcps_connattempt;	/* connections initiated */
384 	u_long	tcps_accepts;		/* connections accepted */
385 	u_long	tcps_connects;		/* connections established */
386 	u_long	tcps_drops;		/* connections dropped */
387 	u_long	tcps_conndrops;		/* embryonic connections dropped */
388 	u_long	tcps_minmssdrops;	/* average minmss too low drops */
389 	u_long	tcps_closed;		/* conn. closed (includes drops) */
390 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
391 	u_long	tcps_rttupdated;	/* times we succeeded */
392 	u_long	tcps_delack;		/* delayed acks sent */
393 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
394 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
395 	u_long	tcps_persisttimeo;	/* persist timeouts */
396 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
397 	u_long	tcps_keepprobe;		/* keepalive probes sent */
398 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
399 
400 	u_long	tcps_sndtotal;		/* total packets sent */
401 	u_long	tcps_sndpack;		/* data packets sent */
402 	u_long	tcps_sndbyte;		/* data bytes sent */
403 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
404 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
405 	u_long	tcps_sndrexmitbad;	/* unnecessary packet retransmissions */
406 	u_long	tcps_sndacks;		/* ack-only packets sent */
407 	u_long	tcps_sndprobe;		/* window probes sent */
408 	u_long	tcps_sndurg;		/* packets sent with URG only */
409 	u_long	tcps_sndwinup;		/* window update-only packets sent */
410 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
411 
412 	u_long	tcps_rcvtotal;		/* total packets received */
413 	u_long	tcps_rcvpack;		/* packets received in sequence */
414 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
415 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
416 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
417 	u_long	tcps_rcvmemdrop;	/* packets dropped for lack of memory */
418 	u_long	tcps_rcvshort;		/* packets received too short */
419 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
420 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
421 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
422 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
423 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
424 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
425 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
426 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
427 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
428 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
429 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
430 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
431 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
432 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
433 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
434 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
435 	u_long	tcps_predack;		/* times hdr predict ok for acks */
436 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
437 	u_long	tcps_pcbcachemiss;
438 	u_long	tcps_cachedrtt;		/* times cached RTT in route updated */
439 	u_long	tcps_cachedrttvar;	/* times cached rttvar updated */
440 	u_long	tcps_cachedssthresh;	/* times cached ssthresh updated */
441 	u_long	tcps_usedrtt;		/* times RTT initialized from route */
442 	u_long	tcps_usedrttvar;	/* times RTTVAR initialized from rt */
443 	u_long	tcps_usedssthresh;	/* times ssthresh initialized from rt*/
444 	u_long	tcps_persistdrop;	/* timeout in persist state */
445 	u_long	tcps_badsyn;		/* bogus SYN, e.g. premature ACK */
446 	u_long	tcps_mturesent;		/* resends due to MTU discovery */
447 	u_long	tcps_listendrop;	/* listen queue overflows */
448 	u_long	tcps_badrst;		/* ignored RSTs in the window */
449 
450 	u_long	tcps_sc_added;		/* entry added to syncache */
451 	u_long	tcps_sc_retransmitted;	/* syncache entry was retransmitted */
452 	u_long	tcps_sc_dupsyn;		/* duplicate SYN packet */
453 	u_long	tcps_sc_dropped;	/* could not reply to packet */
454 	u_long	tcps_sc_completed;	/* successful extraction of entry */
455 	u_long	tcps_sc_bucketoverflow;	/* syncache per-bucket limit hit */
456 	u_long	tcps_sc_cacheoverflow;	/* syncache cache limit hit */
457 	u_long	tcps_sc_reset;		/* RST removed entry from syncache */
458 	u_long	tcps_sc_stale;		/* timed out or listen socket gone */
459 	u_long	tcps_sc_aborted;	/* syncache entry aborted */
460 	u_long	tcps_sc_badack;		/* removed due to bad ACK */
461 	u_long	tcps_sc_unreach;	/* ICMP unreachable received */
462 	u_long	tcps_sc_zonefail;	/* zalloc() failed */
463 	u_long	tcps_sc_sendcookie;	/* SYN cookie sent */
464 	u_long	tcps_sc_recvcookie;	/* SYN cookie received */
465 
466 	u_long	tcps_hc_added;		/* entry added to hostcache */
467 	u_long	tcps_hc_bucketoverflow;	/* hostcache per bucket limit hit */
468 
469 	u_long  tcps_finwait2_drops;    /* Drop FIN_WAIT_2 connection after time limit */
470 
471 	/* SACK related stats */
472 	u_long	tcps_sack_recovery_episode; /* SACK recovery episodes */
473 	u_long  tcps_sack_rexmits;	    /* SACK rexmit segments   */
474 	u_long  tcps_sack_rexmit_bytes;	    /* SACK rexmit bytes      */
475 	u_long  tcps_sack_rcv_blocks;	    /* SACK blocks (options) received */
476 	u_long  tcps_sack_send_blocks;	    /* SACK blocks (options) sent     */
477 	u_long  tcps_sack_sboverflow; 	    /* times scoreboard overflowed */
478 
479 	/* ECN related stats */
480 	u_long	tcps_ecn_ce;		/* ECN Congestion Experienced */
481 	u_long	tcps_ecn_ect0;		/* ECN Capable Transport */
482 	u_long	tcps_ecn_ect1;		/* ECN Capable Transport */
483 	u_long	tcps_ecn_shs;		/* ECN successful handshakes */
484 	u_long	tcps_ecn_rcwnd;		/* # times ECN reduced the cwnd */
485 
486 	u_long	_pad[12];		/* 6 UTO, 6 TBD */
487 };
488 
489 #ifdef _KERNEL
490 /*
491  * In-kernel consumers can use these accessor macros directly to update
492  * stats.
493  */
494 #define	TCPSTAT_ADD(name, val)	V_tcpstat.name += (val)
495 #define	TCPSTAT_INC(name)	TCPSTAT_ADD(name, 1)
496 
497 /*
498  * Kernel module consumers must use this accessor macro.
499  */
500 void	kmod_tcpstat_inc(int statnum);
501 #define	KMOD_TCPSTAT_INC(name)						\
502 	kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(u_long))
503 #endif
504 
505 /*
506  * TCB structure exported to user-land via sysctl(3).
507  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
508  * included.  Not all of our clients do.
509  */
510 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
511 struct xtcp_timer {
512 	int tt_rexmt;	/* retransmit timer */
513 	int tt_persist;	/* retransmit persistence */
514 	int tt_keep;	/* keepalive */
515 	int tt_2msl;	/* 2*msl TIME_WAIT timer */
516 	int tt_delack;	/* delayed ACK timer */
517 	int t_rcvtime; 	/* Time since last packet received */
518 };
519 struct	xtcpcb {
520 	size_t	xt_len;
521 	struct	inpcb	xt_inp;
522 	struct	tcpcb	xt_tp;
523 	struct	xsocket	xt_socket;
524 	struct	xtcp_timer xt_timer;
525 	u_quad_t	xt_alignment_hack;
526 };
527 #endif
528 
529 /*
530  * Names for TCP sysctl objects
531  */
532 #define	TCPCTL_DO_RFC1323	1	/* use RFC-1323 extensions */
533 #define	TCPCTL_MSSDFLT		3	/* MSS default */
534 #define TCPCTL_STATS		4	/* statistics (read-only) */
535 #define	TCPCTL_RTTDFLT		5	/* default RTT estimate */
536 #define	TCPCTL_KEEPIDLE		6	/* keepalive idle timer */
537 #define	TCPCTL_KEEPINTVL	7	/* interval to send keepalives */
538 #define	TCPCTL_SENDSPACE	8	/* send buffer space */
539 #define	TCPCTL_RECVSPACE	9	/* receive buffer space */
540 #define	TCPCTL_KEEPINIT		10	/* timeout for establishing syn */
541 #define	TCPCTL_PCBLIST		11	/* list of all outstanding PCBs */
542 #define	TCPCTL_DELACKTIME	12	/* time before sending delayed ACK */
543 #define	TCPCTL_V6MSSDFLT	13	/* MSS default for IPv6 */
544 #define	TCPCTL_SACK		14	/* Selective Acknowledgement,rfc 2018 */
545 #define	TCPCTL_DROP		15	/* drop tcp connection */
546 #define	TCPCTL_MAXID		16
547 #define TCPCTL_FINWAIT2_TIMEOUT        17
548 
549 #define TCPCTL_NAMES { \
550 	{ 0, 0 }, \
551 	{ "rfc1323", CTLTYPE_INT }, \
552 	{ "mssdflt", CTLTYPE_INT }, \
553 	{ "stats", CTLTYPE_STRUCT }, \
554 	{ "rttdflt", CTLTYPE_INT }, \
555 	{ "keepidle", CTLTYPE_INT }, \
556 	{ "keepintvl", CTLTYPE_INT }, \
557 	{ "sendspace", CTLTYPE_INT }, \
558 	{ "recvspace", CTLTYPE_INT }, \
559 	{ "keepinit", CTLTYPE_INT }, \
560 	{ "pcblist", CTLTYPE_STRUCT }, \
561 	{ "delacktime", CTLTYPE_INT }, \
562 	{ "v6mssdflt", CTLTYPE_INT }, \
563 	{ "maxid", CTLTYPE_INT }, \
564 }
565 
566 
567 #ifdef _KERNEL
568 #ifdef SYSCTL_DECL
569 SYSCTL_DECL(_net_inet_tcp);
570 SYSCTL_DECL(_net_inet_tcp_sack);
571 MALLOC_DECLARE(M_TCPLOG);
572 #endif
573 
574 VNET_DECLARE(struct inpcbhead, tcb);		/* queue of active tcpcb's */
575 VNET_DECLARE(struct inpcbinfo, tcbinfo);
576 VNET_DECLARE(struct tcpstat, tcpstat);		/* tcp statistics */
577 extern	int tcp_log_in_vain;
578 VNET_DECLARE(int, tcp_mssdflt);	/* XXX */
579 VNET_DECLARE(int, tcp_minmss);
580 VNET_DECLARE(int, tcp_delack_enabled);
581 VNET_DECLARE(int, tcp_do_rfc3390);
582 VNET_DECLARE(int, path_mtu_discovery);
583 VNET_DECLARE(int, ss_fltsz);
584 VNET_DECLARE(int, ss_fltsz_local);
585 VNET_DECLARE(int, tcp_do_rfc3465);
586 VNET_DECLARE(int, tcp_abc_l_var);
587 #define	V_tcb			VNET(tcb)
588 #define	V_tcbinfo		VNET(tcbinfo)
589 #define	V_tcpstat		VNET(tcpstat)
590 #define	V_tcp_mssdflt		VNET(tcp_mssdflt)
591 #define	V_tcp_minmss		VNET(tcp_minmss)
592 #define	V_tcp_delack_enabled	VNET(tcp_delack_enabled)
593 #define	V_tcp_do_rfc3390	VNET(tcp_do_rfc3390)
594 #define	V_path_mtu_discovery	VNET(path_mtu_discovery)
595 #define	V_ss_fltsz		VNET(ss_fltsz)
596 #define	V_ss_fltsz_local	VNET(ss_fltsz_local)
597 #define	V_tcp_do_rfc3465	VNET(tcp_do_rfc3465)
598 #define	V_tcp_abc_l_var		VNET(tcp_abc_l_var)
599 
600 VNET_DECLARE(int, tcp_do_sack);			/* SACK enabled/disabled */
601 VNET_DECLARE(int, tcp_sc_rst_sock_fail);	/* RST on sock alloc failure */
602 #define	V_tcp_do_sack		VNET(tcp_do_sack)
603 #define	V_tcp_sc_rst_sock_fail	VNET(tcp_sc_rst_sock_fail)
604 
605 VNET_DECLARE(int, tcp_do_ecn);			/* TCP ECN enabled/disabled */
606 VNET_DECLARE(int, tcp_ecn_maxretries);
607 #define	V_tcp_do_ecn		VNET(tcp_do_ecn)
608 #define	V_tcp_ecn_maxretries	VNET(tcp_ecn_maxretries)
609 
610 int	 tcp_addoptions(struct tcpopt *, u_char *);
611 int	 tcp_ccalgounload(struct cc_algo *unload_algo);
612 struct tcpcb *
613 	 tcp_close(struct tcpcb *);
614 void	 tcp_discardcb(struct tcpcb *);
615 void	 tcp_twstart(struct tcpcb *);
616 #if 0
617 int	 tcp_twrecycleable(struct tcptw *tw);
618 #endif
619 void	 tcp_twclose(struct tcptw *_tw, int _reuse);
620 void	 tcp_ctlinput(int, struct sockaddr *, void *);
621 int	 tcp_ctloutput(struct socket *, struct sockopt *);
622 struct tcpcb *
623 	 tcp_drop(struct tcpcb *, int);
624 void	 tcp_drain(void);
625 void	 tcp_init(void);
626 #ifdef VIMAGE
627 void	 tcp_destroy(void);
628 #endif
629 void	 tcp_fini(void *);
630 char 	*tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
631 	    const void *);
632 char	*tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *,
633 	    const void *);
634 int	 tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *);
635 void	 tcp_reass_init(void);
636 void	 tcp_reass_flush(struct tcpcb *);
637 #ifdef VIMAGE
638 void	 tcp_reass_destroy(void);
639 #endif
640 void	 tcp_input(struct mbuf *, int);
641 u_long	 tcp_maxmtu(struct in_conninfo *, int *);
642 u_long	 tcp_maxmtu6(struct in_conninfo *, int *);
643 void	 tcp_mss_update(struct tcpcb *, int, struct hc_metrics_lite *, int *);
644 void	 tcp_mss(struct tcpcb *, int);
645 int	 tcp_mssopt(struct in_conninfo *);
646 struct inpcb *
647 	 tcp_drop_syn_sent(struct inpcb *, int);
648 struct inpcb *
649 	 tcp_mtudisc(struct inpcb *, int);
650 struct tcpcb *
651 	 tcp_newtcpcb(struct inpcb *);
652 int	 tcp_output(struct tcpcb *);
653 void	 tcp_respond(struct tcpcb *, void *,
654 	    struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
655 void	 tcp_tw_init(void);
656 #ifdef VIMAGE
657 void	 tcp_tw_destroy(void);
658 #endif
659 void	 tcp_tw_zone_change(void);
660 int	 tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
661 	    struct mbuf *, int);
662 int	 tcp_twrespond(struct tcptw *, int);
663 void	 tcp_setpersist(struct tcpcb *);
664 #ifdef TCP_SIGNATURE
665 int	 tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
666 #endif
667 void	 tcp_slowtimo(void);
668 struct tcptemp *
669 	 tcpip_maketemplate(struct inpcb *);
670 void	 tcpip_fillheaders(struct inpcb *, void *, void *);
671 void	 tcp_timer_activate(struct tcpcb *, int, u_int);
672 int	 tcp_timer_active(struct tcpcb *, int);
673 void	 tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
674 /*
675  * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
676  */
677 void	 tcp_hc_init(void);
678 #ifdef VIMAGE
679 void	 tcp_hc_destroy(void);
680 #endif
681 void	 tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
682 u_long	 tcp_hc_getmtu(struct in_conninfo *);
683 void	 tcp_hc_updatemtu(struct in_conninfo *, u_long);
684 void	 tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
685 
686 extern	struct pr_usrreqs tcp_usrreqs;
687 extern	u_long tcp_sendspace;
688 extern	u_long tcp_recvspace;
689 tcp_seq tcp_new_isn(struct tcpcb *);
690 
691 void	 tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
692 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
693 void	 tcp_clean_sackreport(struct tcpcb *tp);
694 void	 tcp_sack_adjust(struct tcpcb *tp);
695 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
696 void	 tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
697 void	 tcp_free_sackholes(struct tcpcb *tp);
698 int	 tcp_newreno(struct tcpcb *, struct tcphdr *);
699 u_long	 tcp_seq_subtract(u_long, u_long );
700 
701 void	cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type);
702 
703 #endif /* _KERNEL */
704 
705 #endif /* _NETINET_TCP_VAR_H_ */
706