xref: /openbsd/sys/netinet/tcp_var.h (revision 905646f0)
1 /*	$OpenBSD: tcp_var.h,v 1.135 2020/08/18 05:21:21 gnezdo Exp $	*/
2 /*	$NetBSD: tcp_var.h,v 1.17 1996/02/13 23:44:24 christos Exp $	*/
3 
4 /*
5  * Copyright (c) 1982, 1986, 1993, 1994
6  *	The Regents of the University of California.  All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the University nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  *
32  *	@(#)tcp_var.h	8.3 (Berkeley) 4/10/94
33  */
34 
35 #ifndef _NETINET_TCP_VAR_H_
36 #define _NETINET_TCP_VAR_H_
37 
38 #include <sys/timeout.h>
39 
40 /*
41  * Kernel variables for tcp.
42  */
43 
44 struct sackblk {
45 	tcp_seq start;		/* start seq no. of sack block */
46 	tcp_seq end; 		/* end seq no. */
47 };
48 
49 struct sackhole {
50 	tcp_seq start;		/* start seq no. of hole */
51 	tcp_seq end;		/* end seq no. */
52 	int	dups;		/* number of dup(s)acks for this hole */
53 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
54 	struct sackhole *next;	/* next in list */
55 };
56 
57 /*
58  * TCP sequence queue structures.
59  */
60 TAILQ_HEAD(tcpqehead, tcpqent);
61 struct tcpqent {
62 	TAILQ_ENTRY(tcpqent) tcpqe_q;
63 	struct tcphdr	*tcpqe_tcp;
64 	struct mbuf	*tcpqe_m;	/* mbuf contains packet */
65 };
66 
67 /*
68  * Tcp control block, one per tcp; fields:
69  */
70 struct tcpcb {
71 	struct tcpqehead t_segq;		/* sequencing queue */
72 	struct timeout t_timer[TCPT_NTIMERS];	/* tcp timers */
73 	short	t_state;		/* state of this connection */
74 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
75 	short	t_rxtcur;		/* current retransmit value */
76 	short	t_dupacks;		/* consecutive dup acks recd */
77 	u_short	t_maxseg;		/* maximum segment size */
78 	char	t_force;		/* 1 if forcing out a byte */
79 	u_int	t_flags;
80 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
81 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
82 #define	TF_NOOPT	0x0008		/* don't use tcp options */
83 #define	TF_SENTFIN	0x0010		/* have sent FIN */
84 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
85 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
86 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
87 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
88 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
89 #define	TF_SIGNATURE	0x0400		/* require TCP MD5 signature */
90 #ifdef TCP_ECN
91 #define TF_ECN_PERMIT	0x00008000	/* other side said I could ECN */
92 #define TF_RCVD_CE	0x00010000	/* send ECE in subsequent segs */
93 #define TF_SEND_CWR	0x00020000	/* send CWR in next seg */
94 #define TF_DISABLE_ECN	0x00040000	/* disable ECN for this connection */
95 #endif
96 #define TF_LASTIDLE	0x00100000	/* no outstanding ACK on last send */
97 #define TF_PMTUD_PEND	0x00400000	/* Path MTU Discovery pending */
98 #define TF_NEEDOUTPUT	0x00800000	/* call tcp_output after tcp_input */
99 #define TF_BLOCKOUTPUT	0x01000000	/* avert tcp_output during tcp_input */
100 #define TF_NOPUSH	0x02000000	/* don't push */
101 #define TF_TMR_REXMT	0x04000000	/* retransmit timer armed */
102 #define TF_TMR_PERSIST	0x08000000	/* retransmit persistence timer armed */
103 #define TF_TMR_KEEP	0x10000000	/* keep alive timer armed */
104 #define TF_TMR_2MSL	0x20000000	/* 2*msl quiet time timer armed */
105 #define TF_TMR_REAPER	0x40000000	/* delayed cleanup timer armed, dead */
106 #define TF_TMR_DELACK	0x80000000	/* delayed ack timer armed */
107 #define TF_TIMER	TF_TMR_REXMT	/* used to shift with TCPT values */
108 
109 	struct	mbuf *t_template;	/* skeletal packet for transmit */
110 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
111 /*
112  * The following fields are used as in the protocol specification.
113  * See RFC793, Dec. 1981, page 21.
114  */
115 /* send sequence variables */
116 	tcp_seq	snd_una;		/* send unacknowledged */
117 	tcp_seq	snd_nxt;		/* send next */
118 	tcp_seq	snd_up;			/* send urgent pointer */
119 	tcp_seq	snd_wl1;		/* window update seg seq number */
120 	tcp_seq	snd_wl2;		/* window update seg ack number */
121 	tcp_seq	iss;			/* initial send sequence number */
122 	u_long	snd_wnd;		/* send window */
123 	int	sack_enable;		/* enable SACK for this connection */
124 	int	snd_numholes;		/* number of holes seen by sender */
125 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
126 	tcp_seq snd_last;		/* for use in fast recovery */
127 /* receive sequence variables */
128 	u_long	rcv_wnd;		/* receive window */
129 	tcp_seq	rcv_nxt;		/* receive next */
130 	tcp_seq	rcv_up;			/* receive urgent pointer */
131 	tcp_seq	irs;			/* initial receive sequence number */
132 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
133 	int	rcv_numsacks;		/* # distinct sack blks present */
134 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
135 
136 /*
137  * Additional variables for this implementation.
138  */
139 /* receive variables */
140 	tcp_seq	rcv_adv;		/* advertised window */
141 /* retransmit variables */
142 	tcp_seq	snd_max;		/* highest sequence number sent;
143 					 * used to recognize retransmits
144 					 */
145 /* congestion control (for slow start, source quench, retransmit after loss) */
146 	u_long	snd_cwnd;		/* congestion-controlled window */
147 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
148 					 * for slow start exponential to
149 					 * linear switch
150 					 */
151 
152 /* auto-sizing variables */
153 	u_int	rfbuf_cnt;	/* recv buffer autoscaling byte count */
154 	u_int32_t rfbuf_ts;	/* recv buffer autoscaling time stamp */
155 
156 	u_short	t_maxopd;		/* mss plus options */
157 	u_short	t_peermss;		/* peer's maximum segment size */
158 
159 /*
160  * transmit timing stuff.  See below for scale of srtt and rttvar.
161  * "Variance" is actually smoothed difference.
162  */
163 	uint32_t t_rcvtime;		/* time last segment received */
164 	uint32_t t_rtttime;		/* time we started measuring rtt */
165 	tcp_seq	t_rtseq;		/* sequence number being timed */
166 	short	t_srtt;			/* smoothed round-trip time */
167 	short	t_rttvar;		/* variance in round-trip time */
168 	u_short	t_rttmin;		/* minimum rtt allowed */
169 	u_long	max_sndwnd;		/* largest window peer has offered */
170 
171 /* out-of-band data */
172 	char	t_oobflags;		/* have some */
173 	char	t_iobc;			/* input character */
174 #define	TCPOOB_HAVEDATA	0x01
175 #define	TCPOOB_HADDATA	0x02
176 	short	t_softerror;		/* possible error not yet reported */
177 
178 /* RFC 1323 variables */
179 	u_char	snd_scale;		/* window scaling for send window */
180 	u_char	rcv_scale;		/* window scaling for recv window */
181 	u_char	request_r_scale;	/* pending window scaling */
182 	u_char	requested_s_scale;
183 	u_int32_t ts_recent;		/* timestamp echo data */
184 	u_int32_t ts_modulate;		/* modulation on timestamp */
185 	u_int32_t ts_recent_age;		/* when last updated */
186 	tcp_seq	last_ack_sent;
187 
188 /* pointer for syn cache entries*/
189 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
190 
191 /* Path-MTU Discovery Information */
192 	u_int	t_pmtud_mss_acked;	/* MSS acked, lower bound for MTU */
193 	u_int	t_pmtud_mtu_sent;	/* MTU used, upper bound for MTU */
194 	tcp_seq	t_pmtud_th_seq;		/* TCP SEQ from ICMP payload */
195 	u_int	t_pmtud_nextmtu;	/* Advertised Next-Hop MTU from ICMP */
196 	u_short	t_pmtud_ip_len;		/* IP length from ICMP payload */
197 	u_short	t_pmtud_ip_hl;		/* IP header length from ICMP payload */
198 
199 	int pf;
200 };
201 
202 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
203 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
204 
205 #ifdef _KERNEL
206 /*
207  * Handy way of passing around TCP option info.
208  */
209 struct tcp_opt_info {
210 	int		ts_present;
211 	u_int32_t	ts_val;
212 	u_int32_t	ts_ecr;
213 	u_int16_t	maxseg;
214 };
215 
216 /*
217  * Data for the TCP compressed state engine.
218  */
219 
220 #define	TCP_SYN_HASH_SIZE	293
221 #define	TCP_SYN_BUCKET_SIZE	35
222 
223 union syn_cache_sa {
224 	struct sockaddr sa;
225 	struct sockaddr_in sin;
226 	struct sockaddr_in6 sin6;
227 };
228 
229 struct syn_cache {
230 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
231 	struct timeout sc_timer;		/* rexmt timer */
232 	union {					/* cached route */
233 		struct route route4;
234 #ifdef INET6
235 		struct route_in6 route6;
236 #endif
237 	} sc_route_u;
238 #define sc_route4	sc_route_u.route4
239 #ifdef INET6
240 #define sc_route6	sc_route_u.route6
241 #endif
242 	long sc_win;				/* advertised window */
243 	struct syn_cache_head *sc_buckethead;	/* our bucket index */
244 	struct syn_cache_set *sc_set;		/* our syn cache set */
245 	u_int32_t sc_hash;
246 	u_int32_t sc_timestamp;			/* timestamp from SYN */
247 	u_int32_t sc_modulate;			/* our timestamp modulator */
248 #if 0
249 	u_int32_t sc_timebase;			/* our local timebase */
250 #endif
251 	union syn_cache_sa sc_src;
252 	union syn_cache_sa sc_dst;
253 	tcp_seq sc_irs;
254 	tcp_seq sc_iss;
255 	u_int sc_rtableid;
256 	u_int sc_rxtcur;			/* current rxt timeout */
257 	u_int sc_rxttot;			/* total time spend on queues */
258 	u_short sc_rxtshift;			/* for computing backoff */
259 	u_short sc_flags;
260 
261 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
262 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
263 #define	SCF_DEAD		0x0004		/* this entry to be released */
264 #define	SCF_SACK_PERMIT		0x0008		/* permit sack */
265 #define	SCF_ECN_PERMIT		0x0010		/* permit ecn */
266 #define	SCF_SIGNATURE		0x0020		/* enforce tcp signatures */
267 
268 	struct mbuf *sc_ipopts;			/* IP options */
269 	u_int16_t sc_peermaxseg;
270 	u_int16_t sc_ourmaxseg;
271 	u_int     sc_request_r_scale	: 4,
272 		  sc_requested_s_scale	: 4;
273 
274 	struct tcpcb *sc_tp;			/* tcb for listening socket */
275 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
276 };
277 
278 struct syn_cache_head {
279 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
280 	u_short sch_length;			/* # entries in bucket */
281 };
282 
283 struct syn_cache_set {
284 	struct		syn_cache_head *scs_buckethead;
285 	int		scs_size;
286 	int		scs_count;
287 	int		scs_use;
288 	u_int32_t	scs_random[5];
289 };
290 
291 #endif /* _KERNEL */
292 
293 /*
294  * The smoothed round-trip time and estimated variance
295  * are stored as fixed point numbers scaled by the values below.
296  * For convenience, these scales are also used in smoothing the average
297  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
298  * With these scales, srtt has 5 bits to the right of the binary point,
299  * and thus an "ALPHA" of 0.875.  rttvar has 4 bits to the right of the
300  * binary point, and is smoothed with an ALPHA of 0.75.
301  */
302 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 5 bits frac. */
303 #define	TCP_RTTVAR_SHIFT	2	/* shift for rttvar; 4 bits */
304 #define	TCP_RTT_BASE_SHIFT	2	/* remaining 2 bit shift */
305 #define	TCP_RTT_MAX		(1<<9)	/* maximum rtt */
306 
307 /*
308  * The initial retransmission should happen at rtt + 4 * rttvar.
309  * Because of the way we do the smoothing, srtt and rttvar
310  * will each average +1/2 tick of bias.  When we compute
311  * the retransmit timer, we want 1/2 tick of rounding and
312  * 1 extra tick because of +-1/2 tick uncertainty in the
313  * firing of the timer.  The bias will give us exactly the
314  * 1.5 tick we need.  But, because the bias is
315  * statistical, we have to test that we don't drop below
316  * the minimum feasible timer (which is 2 ticks).
317  * This macro assumes that the value of (1 << TCP_RTTVAR_SHIFT)
318  * is the same as the multiplier for rttvar.
319  */
320 #define	TCP_REXMTVAL(tp) \
321 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> TCP_RTT_BASE_SHIFT)
322 
323 /*
324  * TCP statistics.
325  * Many of these should be kept per connection,
326  * but that's inconvenient at the moment.
327  */
328 struct	tcpstat {
329 	u_int32_t tcps_connattempt;	/* connections initiated */
330 	u_int32_t tcps_accepts;		/* connections accepted */
331 	u_int32_t tcps_connects;	/* connections established */
332 	u_int32_t tcps_drops;		/* connections dropped */
333 	u_int32_t tcps_conndrops;	/* embryonic connections dropped */
334 	u_int32_t tcps_closed;		/* conn. closed (includes drops) */
335 	u_int32_t tcps_segstimed;	/* segs where we tried to get rtt */
336 	u_int32_t tcps_rttupdated;	/* times we succeeded */
337 	u_int32_t tcps_delack;		/* delayed acks sent */
338 	u_int32_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
339 	u_int32_t tcps_rexmttimeo;	/* retransmit timeouts */
340 	u_int32_t tcps_persisttimeo;	/* persist timeouts */
341 	u_int32_t tcps_persistdrop;	/* connections dropped in persist */
342 	u_int32_t tcps_keeptimeo;	/* keepalive timeouts */
343 	u_int32_t tcps_keepprobe;	/* keepalive probes sent */
344 	u_int32_t tcps_keepdrops;	/* connections dropped in keepalive */
345 
346 	u_int32_t tcps_sndtotal;		/* total packets sent */
347 	u_int32_t tcps_sndpack;		/* data packets sent */
348 	u_int64_t tcps_sndbyte;		/* data bytes sent */
349 	u_int32_t tcps_sndrexmitpack;	/* data packets retransmitted */
350 	u_int64_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
351 	u_int64_t tcps_sndrexmitfast;	/* Fast retransmits */
352 	u_int32_t tcps_sndacks;		/* ack-only packets sent */
353 	u_int32_t tcps_sndprobe;	/* window probes sent */
354 	u_int32_t tcps_sndurg;		/* packets sent with URG only */
355 	u_int32_t tcps_sndwinup;	/* window update-only packets sent */
356 	u_int32_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
357 
358 	u_int32_t tcps_rcvtotal;	/* total packets received */
359 	u_int32_t tcps_rcvpack;		/* packets received in sequence */
360 	u_int64_t tcps_rcvbyte;		/* bytes received in sequence */
361 	u_int32_t tcps_rcvbadsum;	/* packets received with ccksum errs */
362 	u_int32_t tcps_rcvbadoff;	/* packets received with bad offset */
363 	u_int32_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
364 	u_int32_t tcps_rcvnosec;	/* packets dropped for lack of ipsec */
365 	u_int32_t tcps_rcvshort;	/* packets received too short */
366 	u_int32_t tcps_rcvduppack;	/* duplicate-only packets received */
367 	u_int64_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
368 	u_int32_t tcps_rcvpartduppack;	/* packets with some duplicate data */
369 	u_int64_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
370 	u_int32_t tcps_rcvoopack;	/* out-of-order packets received */
371 	u_int64_t tcps_rcvoobyte;	/* out-of-order bytes received */
372 	u_int32_t tcps_rcvpackafterwin;	/* packets with data after window */
373 	u_int64_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
374 	u_int32_t tcps_rcvafterclose;	/* packets rcvd after "close" */
375 	u_int32_t tcps_rcvwinprobe;	/* rcvd window probe packets */
376 	u_int32_t tcps_rcvdupack;	/* rcvd duplicate acks */
377 	u_int32_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
378 	u_int32_t tcps_rcvacktooold;	/* rcvd acks for old data */
379 	u_int32_t tcps_rcvackpack;	/* rcvd ack packets */
380 	u_int64_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
381 	u_int32_t tcps_rcvwinupd;	/* rcvd window update packets */
382 	u_int32_t tcps_pawsdrop;	/* segments dropped due to PAWS */
383 	u_int32_t tcps_predack;		/* times hdr predict ok for acks */
384 	u_int32_t tcps_preddat;		/* times hdr predict ok for data pkts */
385 
386 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
387 	u_int32_t tcps_noport;		/* no socket on port */
388 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
389 	u_int32_t tcps_dropsyn;		/* SYN packet dropped */
390 
391 	u_int32_t tcps_rcvbadsig;	/* rcvd bad/missing TCP signatures */
392 	u_int64_t tcps_rcvgoodsig;	/* rcvd good TCP signatures */
393 	u_int32_t tcps_inswcsum;	/* input software-checksummed packets */
394 	u_int32_t tcps_outswcsum;	/* output software-checksummed packets */
395 
396 	/* ECN stats */
397 	u_int32_t tcps_ecn_accepts;	/* ecn connections accepted */
398 	u_int32_t tcps_ecn_rcvece;	/* # of rcvd ece */
399 	u_int32_t tcps_ecn_rcvcwr;	/* # of rcvd cwr */
400 	u_int32_t tcps_ecn_rcvce;	/* # of rcvd ce in ip header */
401 	u_int32_t tcps_ecn_sndect;	/* # of cwr sent */
402 	u_int32_t tcps_ecn_sndece;	/* # of ece sent */
403 	u_int32_t tcps_ecn_sndcwr;	/* # of cwr sent */
404 	u_int32_t tcps_cwr_ecn;		/* # of cwnd reduced by ecn */
405 	u_int32_t tcps_cwr_frecovery;	/* # of cwnd reduced by fastrecovery */
406 	u_int32_t tcps_cwr_timeout;	/* # of cwnd reduced by timeout */
407 
408 	/* These statistics deal with the SYN cache. */
409 	u_int64_t tcps_sc_added;	/* # of entries added */
410 	u_int64_t tcps_sc_completed;	/* # of connections completed */
411 	u_int64_t tcps_sc_timed_out;	/* # of entries timed out */
412 	u_int64_t tcps_sc_overflowed;	/* # dropped due to overflow */
413 	u_int64_t tcps_sc_reset;	/* # dropped due to RST */
414 	u_int64_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
415 	u_int64_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
416 	u_int64_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
417 	u_int64_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
418 	u_int64_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
419 	u_int64_t tcps_sc_collisions;	/* # of hash collisions */
420 	u_int64_t tcps_sc_retransmitted;/* # of retransmissions */
421 	u_int64_t tcps_sc_seedrandom;	/* # of syn cache seeds with random */
422 	u_int64_t tcps_sc_hash_size;	/* hash buckets in current syn cache */
423 	u_int64_t tcps_sc_entry_count;	/* # of entries in current syn cache */
424 	u_int64_t tcps_sc_entry_limit;	/* limit of syn cache entries */
425 	u_int64_t tcps_sc_bucket_maxlen;/* maximum # of entries in any bucket */
426 	u_int64_t tcps_sc_bucket_limit;	/* limit of syn cache bucket list */
427 	u_int64_t tcps_sc_uses_left;	/* use counter of current syn cache */
428 
429 	u_int64_t tcps_conndrained;	/* # of connections drained */
430 
431 	u_int64_t tcps_sack_recovery_episode;	/* SACK recovery episodes */
432 	u_int64_t tcps_sack_rexmits;		/* SACK rexmit segments */
433 	u_int64_t tcps_sack_rexmit_bytes;	/* SACK rexmit bytes */
434 	u_int64_t tcps_sack_rcv_opts;		/* SACK options received */
435 	u_int64_t tcps_sack_snd_opts;		/* SACK options sent */
436 	u_int64_t tcps_sack_drop_opts;		/* SACK options dropped */
437 };
438 
439 /*
440  * Names for TCP sysctl objects.
441  */
442 
443 #define	TCPCTL_RFC1323		1 /* enable/disable RFC1323 timestamps/scaling */
444 #define	TCPCTL_KEEPINITTIME	2 /* TCPT_KEEP value */
445 #define TCPCTL_KEEPIDLE		3 /* allow tcp_keepidle to be changed */
446 #define TCPCTL_KEEPINTVL	4 /* allow tcp_keepintvl to be changed */
447 #define TCPCTL_SLOWHZ		5 /* return kernel idea of PR_SLOWHZ */
448 #define TCPCTL_BADDYNAMIC	6 /* return bad dynamic port bitmap */
449 #define	TCPCTL_RECVSPACE	7 /* receive buffer space */
450 #define	TCPCTL_SENDSPACE	8 /* send buffer space */
451 #define	TCPCTL_IDENT		9 /* get connection owner */
452 #define	TCPCTL_SACK	       10 /* selective acknowledgement, rfc 2018 */
453 #define TCPCTL_MSSDFLT	       11 /* Default maximum segment size */
454 #define	TCPCTL_RSTPPSLIMIT     12 /* RST pps limit */
455 #define	TCPCTL_ACK_ON_PUSH     13 /* ACK immediately on PUSH */
456 #define	TCPCTL_ECN	       14 /* RFC3168 ECN */
457 #define	TCPCTL_SYN_CACHE_LIMIT 15 /* max size of comp. state engine */
458 #define	TCPCTL_SYN_BUCKET_LIMIT	16 /* max size of hash bucket */
459 #define	TCPCTL_RFC3390	       17 /* enable/disable RFC3390 increased cwnd */
460 #define	TCPCTL_REASS_LIMIT     18 /* max entries for tcp reass queues */
461 #define	TCPCTL_DROP	       19 /* drop tcp connection */
462 #define	TCPCTL_SACKHOLE_LIMIT  20 /* max entries for tcp sack queues */
463 #define	TCPCTL_STATS	       21 /* TCP statistics */
464 #define	TCPCTL_ALWAYS_KEEPALIVE 22 /* assume SO_KEEPALIVE is always set */
465 #define	TCPCTL_SYN_USE_LIMIT   23 /* number of uses before reseeding hash */
466 #define TCPCTL_ROOTONLY	       24 /* return root only port bitmap */
467 #define	TCPCTL_SYN_HASH_SIZE   25 /* number of buckets in the hash */
468 #define	TCPCTL_MAXID	       26
469 
470 #define	TCPCTL_NAMES { \
471 	{ 0, 0 }, \
472 	{ "rfc1323",	CTLTYPE_INT }, \
473 	{ "keepinittime",	CTLTYPE_INT }, \
474 	{ "keepidle",	CTLTYPE_INT }, \
475 	{ "keepintvl",	CTLTYPE_INT }, \
476 	{ "slowhz",	CTLTYPE_INT }, \
477 	{ "baddynamic", CTLTYPE_STRUCT }, \
478 	{ NULL,	0 }, \
479 	{ NULL,	0 }, \
480 	{ "ident", 	CTLTYPE_STRUCT }, \
481 	{ "sack",	CTLTYPE_INT }, \
482 	{ "mssdflt",	CTLTYPE_INT }, \
483 	{ "rstppslimit",	CTLTYPE_INT }, \
484 	{ "ackonpush",	CTLTYPE_INT }, \
485 	{ "ecn", 	CTLTYPE_INT }, \
486 	{ "syncachelimit", 	CTLTYPE_INT }, \
487 	{ "synbucketlimit", 	CTLTYPE_INT }, \
488 	{ "rfc3390", 	CTLTYPE_INT }, \
489 	{ "reasslimit", 	CTLTYPE_INT }, \
490 	{ "drop", 	CTLTYPE_STRUCT }, \
491 	{ "sackholelimit", 	CTLTYPE_INT }, \
492 	{ "stats",	CTLTYPE_STRUCT }, \
493 	{ "always_keepalive",	CTLTYPE_INT }, \
494 	{ "synuselimit", 	CTLTYPE_INT }, \
495 	{ "rootonly", CTLTYPE_STRUCT }, \
496 	{ "synhashsize", 	CTLTYPE_INT }, \
497 }
498 
499 struct tcp_ident_mapping {
500 	struct sockaddr_storage faddr, laddr;
501 	int euid, ruid;
502 	u_int rdomain;
503 };
504 
505 #ifdef _KERNEL
506 
507 #include <sys/percpu.h>
508 
509 enum tcpstat_counters {
510 	tcps_connattempt,
511 	tcps_accepts,
512 	tcps_connects,
513 	tcps_drops,
514 	tcps_conndrops,
515 	tcps_closed,
516 	tcps_segstimed,
517 	tcps_rttupdated,
518 	tcps_delack,
519 	tcps_timeoutdrop,
520 	tcps_rexmttimeo,
521 	tcps_persisttimeo,
522 	tcps_persistdrop,
523 	tcps_keeptimeo,
524 	tcps_keepprobe,
525 	tcps_keepdrops,
526 	tcps_sndtotal,
527 	tcps_sndpack,
528 	tcps_sndbyte,
529 	tcps_sndrexmitpack,
530 	tcps_sndrexmitbyte,
531 	tcps_sndrexmitfast,
532 	tcps_sndacks,
533 	tcps_sndprobe,
534 	tcps_sndurg,
535 	tcps_sndwinup,
536 	tcps_sndctrl,
537 	tcps_rcvtotal,
538 	tcps_rcvpack,
539 	tcps_rcvbyte,
540 	tcps_rcvbadsum,
541 	tcps_rcvbadoff,
542 	tcps_rcvmemdrop,
543 	tcps_rcvnosec,
544 	tcps_rcvshort,
545 	tcps_rcvduppack,
546 	tcps_rcvdupbyte,
547 	tcps_rcvpartduppack,
548 	tcps_rcvpartdupbyte,
549 	tcps_rcvoopack,
550 	tcps_rcvoobyte,
551 	tcps_rcvpackafterwin,
552 	tcps_rcvbyteafterwin,
553 	tcps_rcvafterclose,
554 	tcps_rcvwinprobe,
555 	tcps_rcvdupack,
556 	tcps_rcvacktoomuch,
557 	tcps_rcvacktooold,
558 	tcps_rcvackpack,
559 	tcps_rcvackbyte,
560 	tcps_rcvwinupd,
561 	tcps_pawsdrop,
562 	tcps_predack,
563 	tcps_preddat,
564 	tcps_pcbhashmiss,
565 	tcps_noport,
566 	tcps_badsyn,
567 	tcps_dropsyn,
568 	tcps_rcvbadsig,
569 	tcps_rcvgoodsig,
570 	tcps_inswcsum,
571 	tcps_outswcsum,
572 	tcps_ecn_accepts,
573 	tcps_ecn_rcvece,
574 	tcps_ecn_rcvcwr,
575 	tcps_ecn_rcvce,
576 	tcps_ecn_sndect,
577 	tcps_ecn_sndece,
578 	tcps_ecn_sndcwr,
579 	tcps_cwr_ecn,
580 	tcps_cwr_frecovery,
581 	tcps_cwr_timeout,
582 	tcps_sc_added,
583 	tcps_sc_completed,
584 	tcps_sc_timed_out,
585 	tcps_sc_overflowed,
586 	tcps_sc_reset,
587 	tcps_sc_unreach,
588 	tcps_sc_bucketoverflow,
589 	tcps_sc_aborted,
590 	tcps_sc_dupesyn,
591 	tcps_sc_dropped,
592 	tcps_sc_collisions,
593 	tcps_sc_retransmitted,
594 	tcps_sc_seedrandom,
595 	tcps_sc_hash_size,
596 	tcps_sc_entry_count,
597 	tcps_sc_entry_limit,
598 	tcps_sc_bucket_maxlen,
599 	tcps_sc_bucket_limit,
600 	tcps_sc_uses_left,
601 	tcps_conndrained,
602 	tcps_sack_recovery_episode,
603 	tcps_sack_rexmits,
604 	tcps_sack_rexmit_bytes,
605 	tcps_sack_rcv_opts,
606 	tcps_sack_snd_opts,
607 	tcps_sack_drop_opts,
608 	tcps_ncounters,
609 };
610 
611 extern struct cpumem *tcpcounters;
612 
613 static inline void
614 tcpstat_inc(enum tcpstat_counters c)
615 {
616 	counters_inc(tcpcounters, c);
617 }
618 
619 static inline void
620 tcpstat_add(enum tcpstat_counters c, uint64_t v)
621 {
622 	counters_add(tcpcounters, c, v);
623 }
624 
625 static inline void
626 tcpstat_pkt(enum tcpstat_counters pcounter, enum tcpstat_counters bcounter,
627     uint64_t v)
628 {
629 	counters_pkt(tcpcounters, pcounter, bcounter, v);
630 }
631 
632 extern	struct pool tcpcb_pool;
633 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
634 extern	u_int32_t tcp_now;		/* for RFC 1323 timestamps */
635 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
636 extern	int tcptv_keep_init;	/* time to keep alive the initial SYN packet */
637 extern	int tcp_mssdflt;	/* default maximum segment size */
638 extern	int tcp_rst_ppslim;	/* maximum outgoing RST packet per second */
639 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
640 extern	int tcp_do_sack;	/* SACK enabled/disabled */
641 extern	struct pool sackhl_pool;
642 extern	int tcp_sackhole_limit;	/* max entries for tcp sack queues */
643 extern	int tcp_do_ecn;		/* RFC3168 ECN enabled/disabled? */
644 extern	int tcp_do_rfc3390;	/* RFC3390 Increasing TCP's Initial Window */
645 
646 extern	struct pool tcpqe_pool;
647 extern	int tcp_reass_limit;	/* max entries for tcp reass queues */
648 
649 extern	int tcp_syn_hash_size;  /* adjustable size of the hash array */
650 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
651 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
652 extern	int tcp_syn_use_limit;   /* number of uses before reseeding hash */
653 extern	struct syn_cache_set tcp_syn_cache[];
654 extern	int tcp_syn_cache_active; /* active syn cache, may be 0 or 1 */
655 
656 void	 tcp_canceltimers(struct tcpcb *);
657 struct tcpcb *
658 	 tcp_close(struct tcpcb *);
659 int	 tcp_freeq(struct tcpcb *);
660 #ifdef INET6
661 void	 tcp6_ctlinput(int, struct sockaddr *, u_int, void *);
662 #endif
663 void	 tcp_ctlinput(int, struct sockaddr *, u_int, void *);
664 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf *);
665 struct tcpcb *
666 	 tcp_disconnect(struct tcpcb *);
667 struct tcpcb *
668 	 tcp_drop(struct tcpcb *, int);
669 int	 tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
670 		struct mbuf *, int, struct tcp_opt_info *, u_int);
671 void	 tcp_init(void);
672 int	 tcp_input(struct mbuf **, int *, int, int);
673 int	 tcp_mss(struct tcpcb *, int);
674 void	 tcp_mss_update(struct tcpcb *);
675 u_int	 tcp_hdrsz(struct tcpcb *);
676 void	 tcp_mtudisc(struct inpcb *, int);
677 void	 tcp_mtudisc_increase(struct inpcb *, int);
678 #ifdef INET6
679 void	tcp6_mtudisc(struct inpcb *, int);
680 void	tcp6_mtudisc_callback(struct sockaddr_in6 *, u_int);
681 #endif
682 struct tcpcb *
683 	 tcp_newtcpcb(struct inpcb *);
684 void	 tcp_notify(struct inpcb *, int);
685 int	 tcp_output(struct tcpcb *);
686 void	 tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
687 int	 tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
688 void	 tcp_rscale(struct tcpcb *, u_long);
689 void	 tcp_respond(struct tcpcb *, caddr_t, struct tcphdr *, tcp_seq,
690 		tcp_seq, int, u_int);
691 void	 tcp_setpersist(struct tcpcb *);
692 void	 tcp_update_sndspace(struct tcpcb *);
693 void	 tcp_update_rcvspace(struct tcpcb *);
694 void	 tcp_slowtimo(void);
695 struct mbuf *
696 	 tcp_template(struct tcpcb *);
697 void	 tcp_trace(short, short, struct tcpcb *, struct tcpcb *, caddr_t,
698 		int, int);
699 struct tcpcb *
700 	 tcp_usrclosed(struct tcpcb *);
701 int	 tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
702 int	 tcp_usrreq(struct socket *,
703 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *);
704 int	 tcp_attach(struct socket *, int);
705 int	 tcp_detach(struct socket *);
706 void	 tcp_xmit_timer(struct tcpcb *, int);
707 void	 tcpdropoldhalfopen(struct tcpcb *, u_int16_t);
708 void	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
709 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq, tcp_seq);
710 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
711 void	 tcp_clean_sackreport(struct tcpcb *tp);
712 void	 tcp_sack_adjust(struct tcpcb *tp);
713 struct sackhole *
714 	 tcp_sack_output(struct tcpcb *tp);
715 #ifdef DEBUG
716 void	 tcp_print_holes(struct tcpcb *tp);
717 #endif
718 u_long	 tcp_seq_subtract(u_long, u_long );
719 #ifdef TCP_SIGNATURE
720 int	tcp_signature_apply(caddr_t, caddr_t, unsigned int);
721 int	tcp_signature(struct tdb *, int, struct mbuf *, struct tcphdr *,
722 	    int, int, char *);
723 #endif /* TCP_SIGNATURE */
724 void     tcp_set_iss_tsm(struct tcpcb *);
725 
726 void	 syn_cache_unreach(struct sockaddr *, struct sockaddr *,
727 	   struct tcphdr *, u_int);
728 void	 syn_cache_init(void);
729 void	 syn_cache_cleanup(struct tcpcb *);
730 
731 #endif /* _KERNEL */
732 #endif /* _NETINET_TCP_VAR_H_ */
733