xref: /openbsd-src/sys/netinet/tcp_var.h (revision 897fc685943471cf985a0fe38ba076ea6fe74fa5)
1 /*	$OpenBSD: tcp_var.h,v 1.131 2018/02/07 00:31:10 bluhm 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_DELACK	0x0002		/* ack, but try to delay it */
82 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
83 #define	TF_NOOPT	0x0008		/* don't use tcp options */
84 #define	TF_SENTFIN	0x0010		/* have sent FIN */
85 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
86 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
87 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
88 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
89 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
90 #define	TF_SIGNATURE	0x0400		/* require TCP MD5 signature */
91 #ifdef TCP_ECN
92 #define TF_ECN_PERMIT	0x00008000	/* other side said I could ECN */
93 #define TF_RCVD_CE	0x00010000	/* send ECE in subsequent segs */
94 #define TF_SEND_CWR	0x00020000	/* send CWR in next seg */
95 #define TF_DISABLE_ECN	0x00040000	/* disable ECN for this connection */
96 #endif
97 #define TF_LASTIDLE	0x00100000	/* no outstanding ACK on last send */
98 #define TF_DEAD		0x00200000	/* dead and to-be-released */
99 #define TF_PMTUD_PEND	0x00400000	/* Path MTU Discovery pending */
100 #define TF_NEEDOUTPUT	0x00800000	/* call tcp_output after tcp_input */
101 #define TF_BLOCKOUTPUT	0x01000000	/* avert tcp_output during tcp_input */
102 #define TF_NOPUSH	0x02000000	/* don't push */
103 #define TF_TMR_REXMT	0x04000000	/* retransmit timer armed */
104 #define TF_TMR_PERSIST	0x08000000	/* retransmit persistence timer armed */
105 #define TF_TMR_KEEP	0x10000000	/* keep alive timer armed */
106 #define TF_TMR_2MSL	0x20000000	/* 2*msl quiet time timer armed */
107 #define TF_TMR_REAPER	0x40000000	/* delayed cleanup timer armed, dead */
108 #define TF_TIMER	TF_TMR_REXMT	/* used to shift with TCPT values */
109 
110 	struct	mbuf *t_template;	/* skeletal packet for transmit */
111 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
112 	struct	timeout t_delack_to;	/* delayed ACK callback */
113 /*
114  * The following fields are used as in the protocol specification.
115  * See RFC793, Dec. 1981, page 21.
116  */
117 /* send sequence variables */
118 	tcp_seq	snd_una;		/* send unacknowledged */
119 	tcp_seq	snd_nxt;		/* send next */
120 	tcp_seq	snd_up;			/* send urgent pointer */
121 	tcp_seq	snd_wl1;		/* window update seg seq number */
122 	tcp_seq	snd_wl2;		/* window update seg ack number */
123 	tcp_seq	iss;			/* initial send sequence number */
124 	u_long	snd_wnd;		/* send window */
125 	int	sack_enable;		/* enable SACK for this connection */
126 	int	snd_numholes;		/* number of holes seen by sender */
127 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
128 	tcp_seq snd_last;		/* for use in fast recovery */
129 /* receive sequence variables */
130 	u_long	rcv_wnd;		/* receive window */
131 	tcp_seq	rcv_nxt;		/* receive next */
132 	tcp_seq	rcv_up;			/* receive urgent pointer */
133 	tcp_seq	irs;			/* initial receive sequence number */
134 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
135 	int	rcv_numsacks;		/* # distinct sack blks present */
136 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
137 
138 /*
139  * Additional variables for this implementation.
140  */
141 /* receive variables */
142 	tcp_seq	rcv_adv;		/* advertised window */
143 /* retransmit variables */
144 	tcp_seq	snd_max;		/* highest sequence number sent;
145 					 * used to recognize retransmits
146 					 */
147 /* congestion control (for slow start, source quench, retransmit after loss) */
148 	u_long	snd_cwnd;		/* congestion-controlled window */
149 	u_long	snd_ssthresh;		/* snd_cwnd size threshold for
150 					 * for slow start exponential to
151 					 * linear switch
152 					 */
153 
154 /* auto-sizing variables */
155 	u_int	rfbuf_cnt;	/* recv buffer autoscaling byte count */
156 	u_int32_t rfbuf_ts;	/* recv buffer autoscaling time stamp */
157 
158 	u_short	t_maxopd;		/* mss plus options */
159 	u_short	t_peermss;		/* peer's maximum segment size */
160 
161 /*
162  * transmit timing stuff.  See below for scale of srtt and rttvar.
163  * "Variance" is actually smoothed difference.
164  */
165 	uint32_t t_rcvtime;		/* time last segment received */
166 	uint32_t t_rtttime;		/* time we started measuring rtt */
167 	tcp_seq	t_rtseq;		/* sequence number being timed */
168 	short	t_srtt;			/* smoothed round-trip time */
169 	short	t_rttvar;		/* variance in round-trip time */
170 	u_short	t_rttmin;		/* minimum rtt allowed */
171 	u_long	max_sndwnd;		/* largest window peer has offered */
172 
173 /* out-of-band data */
174 	char	t_oobflags;		/* have some */
175 	char	t_iobc;			/* input character */
176 #define	TCPOOB_HAVEDATA	0x01
177 #define	TCPOOB_HADDATA	0x02
178 	short	t_softerror;		/* possible error not yet reported */
179 
180 /* RFC 1323 variables */
181 	u_char	snd_scale;		/* window scaling for send window */
182 	u_char	rcv_scale;		/* window scaling for recv window */
183 	u_char	request_r_scale;	/* pending window scaling */
184 	u_char	requested_s_scale;
185 	u_int32_t ts_recent;		/* timestamp echo data */
186 	u_int32_t ts_modulate;		/* modulation on timestamp */
187 	u_int32_t ts_recent_age;		/* when last updated */
188 	tcp_seq	last_ack_sent;
189 
190 /* pointer for syn cache entries*/
191 	LIST_HEAD(, syn_cache) t_sc;	/* list of entries by this tcb */
192 
193 /* Path-MTU Discovery Information */
194 	u_int	t_pmtud_mss_acked;	/* MSS acked, lower bound for MTU */
195 	u_int	t_pmtud_mtu_sent;	/* MTU used, upper bound for MTU */
196 	tcp_seq	t_pmtud_th_seq;		/* TCP SEQ from ICMP payload */
197 	u_int	t_pmtud_nextmtu;	/* Advertised Next-Hop MTU from ICMP */
198 	u_short	t_pmtud_ip_len;		/* IP length from ICMP payload */
199 	u_short	t_pmtud_ip_hl;		/* IP header length from ICMP payload */
200 
201 	int pf;
202 };
203 
204 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
205 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
206 
207 #ifdef _KERNEL
208 void	tcp_delack(void *);
209 
210 #define TCP_INIT_DELACK(tp)						\
211 	timeout_set_proc(&(tp)->t_delack_to, tcp_delack, tp)
212 
213 #define TCP_RESTART_DELACK(tp)						\
214 	timeout_add_msec(&(tp)->t_delack_to, tcp_delack_msecs)
215 
216 #define	TCP_SET_DELACK(tp)						\
217 do {									\
218 	if (((tp)->t_flags & TF_DELACK) == 0) {				\
219 		(tp)->t_flags |= TF_DELACK;				\
220 		TCP_RESTART_DELACK(tp);					\
221 	}								\
222 } while (/* CONSTCOND */ 0)
223 
224 #define	TCP_CLEAR_DELACK(tp)						\
225 do {									\
226 	if ((tp)->t_flags & TF_DELACK) {				\
227 		(tp)->t_flags &= ~TF_DELACK;				\
228 		timeout_del(&(tp)->t_delack_to);			\
229 	}								\
230 } while (/* CONSTCOND */ 0)
231 
232 /*
233  * Handy way of passing around TCP option info.
234  */
235 struct tcp_opt_info {
236 	int		ts_present;
237 	u_int32_t	ts_val;
238 	u_int32_t	ts_ecr;
239 	u_int16_t	maxseg;
240 };
241 
242 /*
243  * Data for the TCP compressed state engine.
244  */
245 
246 #define	TCP_SYN_HASH_SIZE	293
247 #define	TCP_SYN_BUCKET_SIZE	35
248 
249 union syn_cache_sa {
250 	struct sockaddr sa;
251 	struct sockaddr_in sin;
252 	struct sockaddr_in6 sin6;
253 };
254 
255 struct syn_cache {
256 	TAILQ_ENTRY(syn_cache) sc_bucketq;	/* link on bucket list */
257 	struct timeout sc_timer;		/* rexmt timer */
258 	union {					/* cached route */
259 		struct route route4;
260 #ifdef INET6
261 		struct route_in6 route6;
262 #endif
263 	} sc_route_u;
264 #define sc_route4	sc_route_u.route4
265 #ifdef INET6
266 #define sc_route6	sc_route_u.route6
267 #endif
268 	long sc_win;				/* advertised window */
269 	struct syn_cache_head *sc_buckethead;	/* our bucket index */
270 	struct syn_cache_set *sc_set;		/* our syn cache set */
271 	u_int32_t sc_hash;
272 	u_int32_t sc_timestamp;			/* timestamp from SYN */
273 	u_int32_t sc_modulate;			/* our timestamp modulator */
274 #if 0
275 	u_int32_t sc_timebase;			/* our local timebase */
276 #endif
277 	union syn_cache_sa sc_src;
278 	union syn_cache_sa sc_dst;
279 	tcp_seq sc_irs;
280 	tcp_seq sc_iss;
281 	u_int sc_rtableid;
282 	u_int sc_rxtcur;			/* current rxt timeout */
283 	u_int sc_rxttot;			/* total time spend on queues */
284 	u_short sc_rxtshift;			/* for computing backoff */
285 	u_short sc_flags;
286 
287 #define	SCF_UNREACH		0x0001		/* we've had an unreach error */
288 #define	SCF_TIMESTAMP		0x0002		/* peer will do timestamps */
289 #define	SCF_DEAD		0x0004		/* this entry to be released */
290 #define	SCF_SACK_PERMIT		0x0008		/* permit sack */
291 #define	SCF_ECN_PERMIT		0x0010		/* permit ecn */
292 #define	SCF_SIGNATURE		0x0020		/* enforce tcp signatures */
293 
294 	struct mbuf *sc_ipopts;			/* IP options */
295 	u_int16_t sc_peermaxseg;
296 	u_int16_t sc_ourmaxseg;
297 	u_int     sc_request_r_scale	: 4,
298 		  sc_requested_s_scale	: 4;
299 
300 	struct tcpcb *sc_tp;			/* tcb for listening socket */
301 	LIST_ENTRY(syn_cache) sc_tpq;		/* list of entries by same tp */
302 };
303 
304 struct syn_cache_head {
305 	TAILQ_HEAD(, syn_cache) sch_bucket;	/* bucket entries */
306 	u_short sch_length;			/* # entries in bucket */
307 };
308 
309 struct syn_cache_set {
310 	struct		syn_cache_head *scs_buckethead;
311 	int		scs_size;
312 	int		scs_count;
313 	int		scs_use;
314 	u_int32_t	scs_random[5];
315 };
316 
317 #endif /* _KERNEL */
318 
319 /*
320  * The smoothed round-trip time and estimated variance
321  * are stored as fixed point numbers scaled by the values below.
322  * For convenience, these scales are also used in smoothing the average
323  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
324  * With these scales, srtt has 5 bits to the right of the binary point,
325  * and thus an "ALPHA" of 0.875.  rttvar has 4 bits to the right of the
326  * binary point, and is smoothed with an ALPHA of 0.75.
327  */
328 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 5 bits frac. */
329 #define	TCP_RTTVAR_SHIFT	2	/* shift for rttvar; 4 bits */
330 #define	TCP_RTT_BASE_SHIFT	2	/* remaining 2 bit shift */
331 #define	TCP_RTT_MAX		(1<<9)	/* maximum rtt */
332 
333 /*
334  * The initial retransmission should happen at rtt + 4 * rttvar.
335  * Because of the way we do the smoothing, srtt and rttvar
336  * will each average +1/2 tick of bias.  When we compute
337  * the retransmit timer, we want 1/2 tick of rounding and
338  * 1 extra tick because of +-1/2 tick uncertainty in the
339  * firing of the timer.  The bias will give us exactly the
340  * 1.5 tick we need.  But, because the bias is
341  * statistical, we have to test that we don't drop below
342  * the minimum feasible timer (which is 2 ticks).
343  * This macro assumes that the value of (1 << TCP_RTTVAR_SHIFT)
344  * is the same as the multiplier for rttvar.
345  */
346 #define	TCP_REXMTVAL(tp) \
347 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> TCP_RTT_BASE_SHIFT)
348 
349 /*
350  * TCP statistics.
351  * Many of these should be kept per connection,
352  * but that's inconvenient at the moment.
353  */
354 struct	tcpstat {
355 	u_int32_t tcps_connattempt;	/* connections initiated */
356 	u_int32_t tcps_accepts;		/* connections accepted */
357 	u_int32_t tcps_connects;	/* connections established */
358 	u_int32_t tcps_drops;		/* connections dropped */
359 	u_int32_t tcps_conndrops;	/* embryonic connections dropped */
360 	u_int32_t tcps_closed;		/* conn. closed (includes drops) */
361 	u_int32_t tcps_segstimed;	/* segs where we tried to get rtt */
362 	u_int32_t tcps_rttupdated;	/* times we succeeded */
363 	u_int32_t tcps_delack;		/* delayed acks sent */
364 	u_int32_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
365 	u_int32_t tcps_rexmttimeo;	/* retransmit timeouts */
366 	u_int32_t tcps_persisttimeo;	/* persist timeouts */
367 	u_int32_t tcps_persistdrop;	/* connections dropped in persist */
368 	u_int32_t tcps_keeptimeo;	/* keepalive timeouts */
369 	u_int32_t tcps_keepprobe;	/* keepalive probes sent */
370 	u_int32_t tcps_keepdrops;	/* connections dropped in keepalive */
371 
372 	u_int32_t tcps_sndtotal;		/* total packets sent */
373 	u_int32_t tcps_sndpack;		/* data packets sent */
374 	u_int64_t tcps_sndbyte;		/* data bytes sent */
375 	u_int32_t tcps_sndrexmitpack;	/* data packets retransmitted */
376 	u_int64_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
377 	u_int64_t tcps_sndrexmitfast;	/* Fast retransmits */
378 	u_int32_t tcps_sndacks;		/* ack-only packets sent */
379 	u_int32_t tcps_sndprobe;	/* window probes sent */
380 	u_int32_t tcps_sndurg;		/* packets sent with URG only */
381 	u_int32_t tcps_sndwinup;	/* window update-only packets sent */
382 	u_int32_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
383 
384 	u_int32_t tcps_rcvtotal;	/* total packets received */
385 	u_int32_t tcps_rcvpack;		/* packets received in sequence */
386 	u_int64_t tcps_rcvbyte;		/* bytes received in sequence */
387 	u_int32_t tcps_rcvbadsum;	/* packets received with ccksum errs */
388 	u_int32_t tcps_rcvbadoff;	/* packets received with bad offset */
389 	u_int32_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
390 	u_int32_t tcps_rcvnosec;	/* packets dropped for lack of ipsec */
391 	u_int32_t tcps_rcvshort;	/* packets received too short */
392 	u_int32_t tcps_rcvduppack;	/* duplicate-only packets received */
393 	u_int64_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
394 	u_int32_t tcps_rcvpartduppack;	/* packets with some duplicate data */
395 	u_int64_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
396 	u_int32_t tcps_rcvoopack;	/* out-of-order packets received */
397 	u_int64_t tcps_rcvoobyte;	/* out-of-order bytes received */
398 	u_int32_t tcps_rcvpackafterwin;	/* packets with data after window */
399 	u_int64_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
400 	u_int32_t tcps_rcvafterclose;	/* packets rcvd after "close" */
401 	u_int32_t tcps_rcvwinprobe;	/* rcvd window probe packets */
402 	u_int32_t tcps_rcvdupack;	/* rcvd duplicate acks */
403 	u_int32_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
404 	u_int32_t tcps_rcvacktooold;	/* rcvd acks for old data */
405 	u_int32_t tcps_rcvackpack;	/* rcvd ack packets */
406 	u_int64_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
407 	u_int32_t tcps_rcvwinupd;	/* rcvd window update packets */
408 	u_int32_t tcps_pawsdrop;	/* segments dropped due to PAWS */
409 	u_int32_t tcps_predack;		/* times hdr predict ok for acks */
410 	u_int32_t tcps_preddat;		/* times hdr predict ok for data pkts */
411 
412 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
413 	u_int32_t tcps_noport;		/* no socket on port */
414 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
415 	u_int32_t tcps_dropsyn;		/* SYN packet dropped */
416 
417 	u_int32_t tcps_rcvbadsig;	/* rcvd bad/missing TCP signatures */
418 	u_int64_t tcps_rcvgoodsig;	/* rcvd good TCP signatures */
419 	u_int32_t tcps_inswcsum;	/* input software-checksummed packets */
420 	u_int32_t tcps_outswcsum;	/* output software-checksummed packets */
421 
422 	/* ECN stats */
423 	u_int32_t tcps_ecn_accepts;	/* ecn connections accepted */
424 	u_int32_t tcps_ecn_rcvece;	/* # of rcvd ece */
425 	u_int32_t tcps_ecn_rcvcwr;	/* # of rcvd cwr */
426 	u_int32_t tcps_ecn_rcvce;	/* # of rcvd ce in ip header */
427 	u_int32_t tcps_ecn_sndect;	/* # of cwr sent */
428 	u_int32_t tcps_ecn_sndece;	/* # of ece sent */
429 	u_int32_t tcps_ecn_sndcwr;	/* # of cwr sent */
430 	u_int32_t tcps_cwr_ecn;		/* # of cwnd reduced by ecn */
431 	u_int32_t tcps_cwr_frecovery;	/* # of cwnd reduced by fastrecovery */
432 	u_int32_t tcps_cwr_timeout;	/* # of cwnd reduced by timeout */
433 
434 	/* These statistics deal with the SYN cache. */
435 	u_int64_t tcps_sc_added;	/* # of entries added */
436 	u_int64_t tcps_sc_completed;	/* # of connections completed */
437 	u_int64_t tcps_sc_timed_out;	/* # of entries timed out */
438 	u_int64_t tcps_sc_overflowed;	/* # dropped due to overflow */
439 	u_int64_t tcps_sc_reset;	/* # dropped due to RST */
440 	u_int64_t tcps_sc_unreach;	/* # dropped due to ICMP unreach */
441 	u_int64_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
442 	u_int64_t tcps_sc_aborted;	/* # of entries aborted (no mem) */
443 	u_int64_t tcps_sc_dupesyn;	/* # of duplicate SYNs received */
444 	u_int64_t tcps_sc_dropped;	/* # of SYNs dropped (no route/mem) */
445 	u_int64_t tcps_sc_collisions;	/* # of hash collisions */
446 	u_int64_t tcps_sc_retransmitted;/* # of retransmissions */
447 	u_int64_t tcps_sc_seedrandom;	/* # of syn cache seeds with random */
448 	u_int64_t tcps_sc_hash_size;	/* hash buckets in current syn cache */
449 	u_int64_t tcps_sc_entry_count;	/* # of entries in current syn cache */
450 	u_int64_t tcps_sc_entry_limit;	/* limit of syn cache entries */
451 	u_int64_t tcps_sc_bucket_maxlen;/* maximum # of entries in any bucket */
452 	u_int64_t tcps_sc_bucket_limit;	/* limit of syn cache bucket list */
453 	u_int64_t tcps_sc_uses_left;	/* use counter of current syn cache */
454 
455 	u_int64_t tcps_conndrained;	/* # of connections drained */
456 
457 	u_int64_t tcps_sack_recovery_episode;	/* SACK recovery episodes */
458 	u_int64_t tcps_sack_rexmits;		/* SACK rexmit segments */
459 	u_int64_t tcps_sack_rexmit_bytes;	/* SACK rexmit bytes */
460 	u_int64_t tcps_sack_rcv_opts;		/* SACK options received */
461 	u_int64_t tcps_sack_snd_opts;		/* SACK options sent */
462 };
463 
464 /*
465  * Names for TCP sysctl objects.
466  */
467 
468 #define	TCPCTL_RFC1323		1 /* enable/disable RFC1323 timestamps/scaling */
469 #define	TCPCTL_KEEPINITTIME	2 /* TCPT_KEEP value */
470 #define TCPCTL_KEEPIDLE		3 /* allow tcp_keepidle to be changed */
471 #define TCPCTL_KEEPINTVL	4 /* allow tcp_keepintvl to be changed */
472 #define TCPCTL_SLOWHZ		5 /* return kernel idea of PR_SLOWHZ */
473 #define TCPCTL_BADDYNAMIC	6 /* return bad dynamic port bitmap */
474 #define	TCPCTL_RECVSPACE	7 /* receive buffer space */
475 #define	TCPCTL_SENDSPACE	8 /* send buffer space */
476 #define	TCPCTL_IDENT		9 /* get connection owner */
477 #define	TCPCTL_SACK	       10 /* selective acknowledgement, rfc 2018 */
478 #define TCPCTL_MSSDFLT	       11 /* Default maximum segment size */
479 #define	TCPCTL_RSTPPSLIMIT     12 /* RST pps limit */
480 #define	TCPCTL_ACK_ON_PUSH     13 /* ACK immediately on PUSH */
481 #define	TCPCTL_ECN	       14 /* RFC3168 ECN */
482 #define	TCPCTL_SYN_CACHE_LIMIT 15 /* max size of comp. state engine */
483 #define	TCPCTL_SYN_BUCKET_LIMIT	16 /* max size of hash bucket */
484 #define	TCPCTL_RFC3390	       17 /* enable/disable RFC3390 increased cwnd */
485 #define	TCPCTL_REASS_LIMIT     18 /* max entries for tcp reass queues */
486 #define	TCPCTL_DROP	       19 /* drop tcp connection */
487 #define	TCPCTL_SACKHOLE_LIMIT  20 /* max entries for tcp sack queues */
488 #define	TCPCTL_STATS	       21 /* TCP statistics */
489 #define	TCPCTL_ALWAYS_KEEPALIVE 22 /* assume SO_KEEPALIVE is always set */
490 #define	TCPCTL_SYN_USE_LIMIT   23 /* number of uses before reseeding hash */
491 #define TCPCTL_ROOTONLY	       24 /* return root only port bitmap */
492 #define	TCPCTL_SYN_HASH_SIZE   25 /* number of buckets in the hash */
493 #define	TCPCTL_MAXID	       26
494 
495 #define	TCPCTL_NAMES { \
496 	{ 0, 0 }, \
497 	{ "rfc1323",	CTLTYPE_INT }, \
498 	{ "keepinittime",	CTLTYPE_INT }, \
499 	{ "keepidle",	CTLTYPE_INT }, \
500 	{ "keepintvl",	CTLTYPE_INT }, \
501 	{ "slowhz",	CTLTYPE_INT }, \
502 	{ "baddynamic", CTLTYPE_STRUCT }, \
503 	{ NULL,	0 }, \
504 	{ NULL,	0 }, \
505 	{ "ident", 	CTLTYPE_STRUCT }, \
506 	{ "sack",	CTLTYPE_INT }, \
507 	{ "mssdflt",	CTLTYPE_INT }, \
508 	{ "rstppslimit",	CTLTYPE_INT }, \
509 	{ "ackonpush",	CTLTYPE_INT }, \
510 	{ "ecn", 	CTLTYPE_INT }, \
511 	{ "syncachelimit", 	CTLTYPE_INT }, \
512 	{ "synbucketlimit", 	CTLTYPE_INT }, \
513 	{ "rfc3390", 	CTLTYPE_INT }, \
514 	{ "reasslimit", 	CTLTYPE_INT }, \
515 	{ "drop", 	CTLTYPE_STRUCT }, \
516 	{ "sackholelimit", 	CTLTYPE_INT }, \
517 	{ "stats",	CTLTYPE_STRUCT }, \
518 	{ "always_keepalive",	CTLTYPE_INT }, \
519 	{ "synuselimit", 	CTLTYPE_INT }, \
520 	{ "rootonly", CTLTYPE_STRUCT }, \
521 	{ "synhashsize", 	CTLTYPE_INT }, \
522 }
523 
524 #define	TCPCTL_VARS { \
525 	NULL, \
526 	&tcp_do_rfc1323, \
527 	&tcptv_keep_init, \
528 	&tcp_keepidle, \
529 	&tcp_keepintvl, \
530 	NULL, \
531 	NULL, \
532 	NULL, \
533 	NULL, \
534 	NULL, \
535 	NULL, \
536 	&tcp_mssdflt, \
537 	&tcp_rst_ppslim, \
538 	&tcp_ack_on_push, \
539 	NULL, \
540 	&tcp_syn_cache_limit, \
541 	&tcp_syn_bucket_limit, \
542 	&tcp_do_rfc3390, \
543 	NULL, \
544 	NULL, \
545 	NULL, \
546 	NULL, \
547 	NULL, \
548 	NULL, \
549 	NULL, \
550 	NULL \
551 }
552 
553 struct tcp_ident_mapping {
554 	struct sockaddr_storage faddr, laddr;
555 	int euid, ruid;
556 	u_int rdomain;
557 };
558 
559 #ifdef _KERNEL
560 
561 #include <sys/percpu.h>
562 
563 enum tcpstat_counters {
564 	tcps_connattempt,
565 	tcps_accepts,
566 	tcps_connects,
567 	tcps_drops,
568 	tcps_conndrops,
569 	tcps_closed,
570 	tcps_segstimed,
571 	tcps_rttupdated,
572 	tcps_delack,
573 	tcps_timeoutdrop,
574 	tcps_rexmttimeo,
575 	tcps_persisttimeo,
576 	tcps_persistdrop,
577 	tcps_keeptimeo,
578 	tcps_keepprobe,
579 	tcps_keepdrops,
580 	tcps_sndtotal,
581 	tcps_sndpack,
582 	tcps_sndbyte,
583 	tcps_sndrexmitpack,
584 	tcps_sndrexmitbyte,
585 	tcps_sndrexmitfast,
586 	tcps_sndacks,
587 	tcps_sndprobe,
588 	tcps_sndurg,
589 	tcps_sndwinup,
590 	tcps_sndctrl,
591 	tcps_rcvtotal,
592 	tcps_rcvpack,
593 	tcps_rcvbyte,
594 	tcps_rcvbadsum,
595 	tcps_rcvbadoff,
596 	tcps_rcvmemdrop,
597 	tcps_rcvnosec,
598 	tcps_rcvshort,
599 	tcps_rcvduppack,
600 	tcps_rcvdupbyte,
601 	tcps_rcvpartduppack,
602 	tcps_rcvpartdupbyte,
603 	tcps_rcvoopack,
604 	tcps_rcvoobyte,
605 	tcps_rcvpackafterwin,
606 	tcps_rcvbyteafterwin,
607 	tcps_rcvafterclose,
608 	tcps_rcvwinprobe,
609 	tcps_rcvdupack,
610 	tcps_rcvacktoomuch,
611 	tcps_rcvacktooold,
612 	tcps_rcvackpack,
613 	tcps_rcvackbyte,
614 	tcps_rcvwinupd,
615 	tcps_pawsdrop,
616 	tcps_predack,
617 	tcps_preddat,
618 	tcps_pcbhashmiss,
619 	tcps_noport,
620 	tcps_badsyn,
621 	tcps_dropsyn,
622 	tcps_rcvbadsig,
623 	tcps_rcvgoodsig,
624 	tcps_inswcsum,
625 	tcps_outswcsum,
626 	tcps_ecn_accepts,
627 	tcps_ecn_rcvece,
628 	tcps_ecn_rcvcwr,
629 	tcps_ecn_rcvce,
630 	tcps_ecn_sndect,
631 	tcps_ecn_sndece,
632 	tcps_ecn_sndcwr,
633 	tcps_cwr_ecn,
634 	tcps_cwr_frecovery,
635 	tcps_cwr_timeout,
636 	tcps_sc_added,
637 	tcps_sc_completed,
638 	tcps_sc_timed_out,
639 	tcps_sc_overflowed,
640 	tcps_sc_reset,
641 	tcps_sc_unreach,
642 	tcps_sc_bucketoverflow,
643 	tcps_sc_aborted,
644 	tcps_sc_dupesyn,
645 	tcps_sc_dropped,
646 	tcps_sc_collisions,
647 	tcps_sc_retransmitted,
648 	tcps_sc_seedrandom,
649 	tcps_sc_hash_size,
650 	tcps_sc_entry_count,
651 	tcps_sc_entry_limit,
652 	tcps_sc_bucket_maxlen,
653 	tcps_sc_bucket_limit,
654 	tcps_sc_uses_left,
655 	tcps_conndrained,
656 	tcps_sack_recovery_episode,
657 	tcps_sack_rexmits,
658 	tcps_sack_rexmit_bytes,
659 	tcps_sack_rcv_opts,
660 	tcps_sack_snd_opts,
661 	tcps_ncounters,
662 };
663 
664 extern struct cpumem *tcpcounters;
665 
666 static inline void
667 tcpstat_inc(enum tcpstat_counters c)
668 {
669 	counters_inc(tcpcounters, c);
670 }
671 
672 static inline void
673 tcpstat_add(enum tcpstat_counters c, uint64_t v)
674 {
675 	counters_add(tcpcounters, c, v);
676 }
677 
678 static inline void
679 tcpstat_pkt(enum tcpstat_counters pcounter, enum tcpstat_counters bcounter,
680     uint64_t v)
681 {
682 	counters_pkt(tcpcounters, pcounter, bcounter, v);
683 }
684 
685 extern	struct pool tcpcb_pool;
686 extern	struct inpcbtable tcbtable;	/* head of queue of active tcpcb's */
687 extern	u_int32_t tcp_now;		/* for RFC 1323 timestamps */
688 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
689 extern	int tcptv_keep_init;	/* time to keep alive the initial SYN packet */
690 extern	int tcp_mssdflt;	/* default maximum segment size */
691 extern	int tcp_rst_ppslim;	/* maximum outgoing RST packet per second */
692 extern	int tcp_ack_on_push;	/* ACK immediately on PUSH */
693 extern	int tcp_do_sack;	/* SACK enabled/disabled */
694 extern	struct pool sackhl_pool;
695 extern	int tcp_sackhole_limit;	/* max entries for tcp sack queues */
696 extern	int tcp_do_ecn;		/* RFC3168 ECN enabled/disabled? */
697 extern	int tcp_do_rfc3390;	/* RFC3390 Increasing TCP's Initial Window */
698 
699 extern	struct pool tcpqe_pool;
700 extern	int tcp_reass_limit;	/* max entries for tcp reass queues */
701 
702 extern	int tcp_syn_hash_size;  /* adjustable size of the hash array */
703 extern	int tcp_syn_cache_limit; /* max entries for compressed state engine */
704 extern	int tcp_syn_bucket_limit;/* max entries per hash bucket */
705 extern	int tcp_syn_use_limit;   /* number of uses before reseeding hash */
706 extern	struct syn_cache_set tcp_syn_cache[];
707 extern	int tcp_syn_cache_active; /* active syn cache, may be 0 or 1 */
708 
709 void	 tcp_canceltimers(struct tcpcb *);
710 struct tcpcb *
711 	 tcp_close(struct tcpcb *);
712 int	 tcp_freeq(struct tcpcb *);
713 #ifdef INET6
714 void	 tcp6_ctlinput(int, struct sockaddr *, u_int, void *);
715 #endif
716 void	 tcp_ctlinput(int, struct sockaddr *, u_int, void *);
717 int	 tcp_ctloutput(int, struct socket *, int, int, struct mbuf *);
718 struct tcpcb *
719 	 tcp_disconnect(struct tcpcb *);
720 struct tcpcb *
721 	 tcp_drop(struct tcpcb *, int);
722 int	 tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *,
723 		struct mbuf *, int, struct tcp_opt_info *, u_int);
724 void	 tcp_init(void);
725 int	 tcp_input(struct mbuf **, int *, int, int);
726 int	 tcp_mss(struct tcpcb *, int);
727 void	 tcp_mss_update(struct tcpcb *);
728 u_int	 tcp_hdrsz(struct tcpcb *);
729 void	 tcp_mtudisc(struct inpcb *, int);
730 void	 tcp_mtudisc_increase(struct inpcb *, int);
731 #ifdef INET6
732 void	tcp6_mtudisc(struct inpcb *, int);
733 void	tcp6_mtudisc_callback(struct sockaddr_in6 *, u_int);
734 #endif
735 struct tcpcb *
736 	 tcp_newtcpcb(struct inpcb *);
737 void	 tcp_notify(struct inpcb *, int);
738 int	 tcp_output(struct tcpcb *);
739 void	 tcp_pulloutofband(struct socket *, u_int, struct mbuf *, int);
740 int	 tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *);
741 void	 tcp_rscale(struct tcpcb *, u_long);
742 void	 tcp_respond(struct tcpcb *, caddr_t, struct tcphdr *, tcp_seq,
743 		tcp_seq, int, u_int);
744 void	 tcp_setpersist(struct tcpcb *);
745 void	 tcp_update_sndspace(struct tcpcb *);
746 void	 tcp_update_rcvspace(struct tcpcb *);
747 void	 tcp_slowtimo(void);
748 struct mbuf *
749 	 tcp_template(struct tcpcb *);
750 void	 tcp_trace(short, short, struct tcpcb *, caddr_t, int, int);
751 struct tcpcb *
752 	 tcp_usrclosed(struct tcpcb *);
753 int	 tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t);
754 int	 tcp_usrreq(struct socket *,
755 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *);
756 int	 tcp_attach(struct socket *, int);
757 int	 tcp_detach(struct socket *);
758 void	 tcp_xmit_timer(struct tcpcb *, int);
759 void	 tcpdropoldhalfopen(struct tcpcb *, u_int16_t);
760 void	 tcp_sack_option(struct tcpcb *,struct tcphdr *,u_char *,int);
761 void	 tcp_update_sack_list(struct tcpcb *tp, tcp_seq, tcp_seq);
762 void	 tcp_del_sackholes(struct tcpcb *, struct tcphdr *);
763 void	 tcp_clean_sackreport(struct tcpcb *tp);
764 void	 tcp_sack_adjust(struct tcpcb *tp);
765 struct sackhole *
766 	 tcp_sack_output(struct tcpcb *tp);
767 #ifdef DEBUG
768 void	 tcp_print_holes(struct tcpcb *tp);
769 #endif
770 u_long	 tcp_seq_subtract(u_long, u_long );
771 #ifdef TCP_SIGNATURE
772 int	tcp_signature_apply(caddr_t, caddr_t, unsigned int);
773 int	tcp_signature(struct tdb *, int, struct mbuf *, struct tcphdr *,
774 	    int, int, char *);
775 #endif /* TCP_SIGNATURE */
776 void     tcp_set_iss_tsm(struct tcpcb *);
777 
778 void	 syn_cache_unreach(struct sockaddr *, struct sockaddr *,
779 	   struct tcphdr *, u_int);
780 void	 syn_cache_init(void);
781 void	 syn_cache_cleanup(struct tcpcb *);
782 
783 #endif /* _KERNEL */
784 #endif /* _NETINET_TCP_VAR_H_ */
785