xref: /openbsd-src/sys/netinet/tcp_var.h (revision b2ea75c1b17e1a9a339660e7ed45cd24946b230e)
1 /*	$OpenBSD: tcp_var.h,v 1.37 2001/06/23 06:03:13 angelos 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. All advertising materials mentioning features or use of this software
17  *    must display the following acknowledgement:
18  *	This product includes software developed by the University of
19  *	California, Berkeley and its contributors.
20  * 4. Neither the name of the University nor the names of its contributors
21  *    may be used to endorse or promote products derived from this software
22  *    without specific prior written permission.
23  *
24  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
25  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
26  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
27  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
28  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
29  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
30  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
31  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
32  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
33  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34  * SUCH DAMAGE.
35  *
36  *	@(#)tcp_var.h	8.3 (Berkeley) 4/10/94
37  */
38 
39 #ifndef _NETINET_TCP_VAR_H_
40 #define _NETINET_TCP_VAR_H_
41 
42 struct sackblk {
43 	tcp_seq start;		/* start seq no. of sack block */
44 	tcp_seq end; 		/* end seq no. */
45 };
46 
47 struct sackhole {
48 	tcp_seq start;		/* start seq no. of hole */
49 	tcp_seq end;		/* end seq no. */
50 	int	dups;		/* number of dup(s)acks for this hole */
51 	tcp_seq rxmit;		/* next seq. no in hole to be retransmitted */
52 	struct sackhole *next;	/* next in list */
53 };
54 
55 /*
56  * Kernel variables for tcp.
57  */
58 
59 /*
60  * Tcp control block, one per tcp; fields:
61  */
62 struct tcpcb {
63 	struct ipqehead segq;		/* sequencing queue */
64 	short	t_state;		/* state of this connection */
65 	short	t_timer[TCPT_NTIMERS];	/* tcp timers */
66 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
67 	short	t_rxtcur;		/* current retransmit value */
68 	short	t_dupacks;		/* consecutive dup acks recd */
69 	u_short	t_maxseg;		/* maximum segment size */
70 	char	t_force;		/* 1 if forcing out a byte */
71 	u_short	t_flags;
72 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
73 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
74 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
75 #define	TF_NOOPT	0x0008		/* don't use tcp options */
76 #define	TF_SENTFIN	0x0010		/* have sent FIN */
77 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
78 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
79 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
80 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
81 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
82 #define	TF_SIGNATURE	0x0400		/* require TCP MD5 signature */
83 
84 	struct	mbuf *t_template;	/* skeletal packet for transmit */
85 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
86 /*
87  * The following fields are used as in the protocol specification.
88  * See RFC783, Dec. 1981, page 21.
89  */
90 /* send sequence variables */
91 	tcp_seq	snd_una;		/* send unacknowledged */
92 	tcp_seq	snd_nxt;		/* send next */
93 	tcp_seq	snd_up;			/* send urgent pointer */
94 	tcp_seq	snd_wl1;		/* window update seg seq number */
95 	tcp_seq	snd_wl2;		/* window update seg ack number */
96 	tcp_seq	iss;			/* initial send sequence number */
97 	u_long	snd_wnd;		/* send window */
98 #ifdef TCP_SACK
99 	int	sack_disable;		/* disable SACK for this connection */
100 	int	snd_numholes;		/* number of holes seen by sender */
101 	struct sackhole *snd_holes;	/* linked list of holes (sorted) */
102 #if defined(TCP_SACK) && defined(TCP_FACK)
103 	tcp_seq snd_fack;		/* for FACK congestion control */
104 	u_long	snd_awnd;		/* snd_nxt - snd_fack + */
105 					/* retransmitted data */
106 	int retran_data;		/* amount of outstanding retx. data  */
107 #endif /* TCP_FACK */
108 #endif /* TCP_SACK */
109 #if defined(TCP_SACK)
110 	tcp_seq snd_last;		/* for use in fast recovery */
111 #endif
112 /* receive sequence variables */
113 	u_long	rcv_wnd;		/* receive window */
114 	tcp_seq	rcv_nxt;		/* receive next */
115 	tcp_seq	rcv_up;			/* receive urgent pointer */
116 	tcp_seq	irs;			/* initial receive sequence number */
117 #ifdef TCP_SACK
118 	tcp_seq rcv_laststart;		/* start of last segment recd. */
119 	tcp_seq rcv_lastend;		/* end of ... */
120 	tcp_seq rcv_lastsack;		/* last seq number(+1) sack'd by rcv'r*/
121 	int	rcv_numsacks;		/* # distinct sack blks present */
122 	struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
123 #endif
124 
125 /*
126  * Additional variables for this implementation.
127  */
128 /* receive variables */
129 	tcp_seq	rcv_adv;		/* advertised window */
130 /* retransmit variables */
131 	tcp_seq	snd_max;		/* highest sequence number sent;
132 					 * used to recognize retransmits
133 					 */
134 /* congestion control (for slow start, source quench, retransmit after loss) */
135 	u_long	snd_cwnd;		/* congestion-controlled window */
136 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
137 					 * for slow start exponential to
138 					 * linear switch
139 					 */
140 	u_short	t_maxopd;		/* mss plus options */
141 	u_short	t_peermss;		/* peer's maximum segment size */
142 
143 /*
144  * transmit timing stuff.  See below for scale of srtt and rttvar.
145  * "Variance" is actually smoothed difference.
146  */
147 	short	t_idle;			/* inactivity time */
148 	short	t_rtt;			/* round trip time */
149 	tcp_seq	t_rtseq;		/* sequence number being timed */
150 	short	t_srtt;			/* smoothed round-trip time */
151 	short	t_rttvar;		/* variance in round-trip time */
152 	u_short	t_rttmin;		/* minimum rtt allowed */
153 	u_long	max_sndwnd;		/* largest window peer has offered */
154 
155 /* out-of-band data */
156 	char	t_oobflags;		/* have some */
157 	char	t_iobc;			/* input character */
158 #define	TCPOOB_HAVEDATA	0x01
159 #define	TCPOOB_HADDATA	0x02
160 	short	t_softerror;		/* possible error not yet reported */
161 
162 /* RFC 1323 variables */
163 	u_char	snd_scale;		/* window scaling for send window */
164 	u_char	rcv_scale;		/* window scaling for recv window */
165 	u_char	request_r_scale;	/* pending window scaling */
166 	u_char	requested_s_scale;
167 	u_int32_t ts_recent;		/* timestamp echo data */
168 	u_int32_t ts_recent_age;		/* when last updated */
169 	tcp_seq	last_ack_sent;
170 
171 /* TUBA stuff */
172 	caddr_t	t_tuba_pcb;		/* next level down pcb for TCP over z */
173 
174 	int pf;
175 };
176 
177 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
178 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
179 
180 /*
181  * The smoothed round-trip time and estimated variance
182  * are stored as fixed point numbers scaled by the values below.
183  * For convenience, these scales are also used in smoothing the average
184  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
185  * With these scales, srtt has 3 bits to the right of the binary point,
186  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
187  * binary point, and is smoothed with an ALPHA of 0.75.
188  */
189 #define	TCP_RTT_SCALE		8	/* multiplier for srtt; 3 bits frac. */
190 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
191 #define	TCP_RTTVAR_SCALE	4	/* multiplier for rttvar; 2 bits */
192 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
193 
194 /*
195  * The initial retransmission should happen at rtt + 4 * rttvar.
196  * Because of the way we do the smoothing, srtt and rttvar
197  * will each average +1/2 tick of bias.  When we compute
198  * the retransmit timer, we want 1/2 tick of rounding and
199  * 1 extra tick because of +-1/2 tick uncertainty in the
200  * firing of the timer.  The bias will give us exactly the
201  * 1.5 tick we need.  But, because the bias is
202  * statistical, we have to test that we don't drop below
203  * the minimum feasible timer (which is 2 ticks).
204  * This macro assumes that the value of TCP_RTTVAR_SCALE
205  * is the same as the multiplier for rttvar.
206  */
207 #define	TCP_REXMTVAL(tp) \
208 	((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
209 
210 /*
211  * TCP statistics.
212  * Many of these should be kept per connection,
213  * but that's inconvenient at the moment.
214  */
215 struct	tcpstat {
216 	u_int32_t tcps_connattempt;	/* connections initiated */
217 	u_int32_t tcps_accepts;		/* connections accepted */
218 	u_int32_t tcps_connects;	/* connections established */
219 	u_int32_t tcps_drops;		/* connections dropped */
220 	u_int32_t tcps_conndrops;	/* embryonic connections dropped */
221 	u_int32_t tcps_closed;		/* conn. closed (includes drops) */
222 	u_int32_t tcps_segstimed;	/* segs where we tried to get rtt */
223 	u_int32_t tcps_rttupdated;	/* times we succeeded */
224 	u_int32_t tcps_delack;		/* delayed acks sent */
225 	u_int32_t tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
226 	u_int32_t tcps_rexmttimeo;	/* retransmit timeouts */
227 	u_int32_t tcps_persisttimeo;	/* persist timeouts */
228 	u_int32_t tcps_persistdrop;	/* connections dropped in persist */
229 	u_int32_t tcps_keeptimeo;	/* keepalive timeouts */
230 	u_int32_t tcps_keepprobe;	/* keepalive probes sent */
231 	u_int32_t tcps_keepdrops;	/* connections dropped in keepalive */
232 
233 	u_int32_t tcps_sndtotal;		/* total packets sent */
234 	u_int32_t tcps_sndpack;		/* data packets sent */
235 	u_int64_t tcps_sndbyte;		/* data bytes sent */
236 	u_int32_t tcps_sndrexmitpack;	/* data packets retransmitted */
237 	u_int64_t tcps_sndrexmitbyte;	/* data bytes retransmitted */
238 	u_int64_t tcps_sndrexmitfast;	/* Fast retransmits */
239 	u_int32_t tcps_sndacks;		/* ack-only packets sent */
240 	u_int32_t tcps_sndprobe;	/* window probes sent */
241 	u_int32_t tcps_sndurg;		/* packets sent with URG only */
242 	u_int32_t tcps_sndwinup;	/* window update-only packets sent */
243 	u_int32_t tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
244 
245 	u_int32_t tcps_rcvtotal;	/* total packets received */
246 	u_int32_t tcps_rcvpack;		/* packets received in sequence */
247 	u_int64_t tcps_rcvbyte;		/* bytes received in sequence */
248 	u_int32_t tcps_rcvbadsum;	/* packets received with ccksum errs */
249 	u_int32_t tcps_rcvbadoff;	/* packets received with bad offset */
250 	u_int32_t tcps_rcvmemdrop;	/* packets dropped for lack of memory */
251 	u_int32_t tcps_rcvnosec;	/* packets dropped for lack of ipsec */
252 	u_int32_t tcps_rcvshort;	/* packets received too short */
253 	u_int32_t tcps_rcvduppack;	/* duplicate-only packets received */
254 	u_int64_t tcps_rcvdupbyte;	/* duplicate-only bytes received */
255 	u_int32_t tcps_rcvpartduppack;	/* packets with some duplicate data */
256 	u_int64_t tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
257 	u_int32_t tcps_rcvoopack;	/* out-of-order packets received */
258 	u_int64_t tcps_rcvoobyte;	/* out-of-order bytes received */
259 	u_int32_t tcps_rcvpackafterwin;	/* packets with data after window */
260 	u_int64_t tcps_rcvbyteafterwin;	/* bytes rcvd after window */
261 	u_int32_t tcps_rcvafterclose;	/* packets rcvd after "close" */
262 	u_int32_t tcps_rcvwinprobe;	/* rcvd window probe packets */
263 	u_int32_t tcps_rcvdupack;	/* rcvd duplicate acks */
264 	u_int32_t tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
265 	u_int32_t tcps_rcvackpack;	/* rcvd ack packets */
266 	u_int64_t tcps_rcvackbyte;	/* bytes acked by rcvd acks */
267 	u_int32_t tcps_rcvwinupd;	/* rcvd window update packets */
268 	u_int32_t tcps_pawsdrop;	/* segments dropped due to PAWS */
269 	u_int32_t tcps_predack;		/* times hdr predict ok for acks */
270 	u_int32_t tcps_preddat;		/* times hdr predict ok for data pkts */
271 
272 	u_int32_t tcps_pcbhashmiss;	/* input packets missing pcb hash */
273 	u_int32_t tcps_noport;		/* no socket on port */
274 	u_int32_t tcps_badsyn;		/* SYN packet with src==dst rcv'ed */
275 
276 	u_int32_t tcps_rcvbadsig;	/* rcvd bad/missing TCP signatures */
277 	u_int64_t tcps_rcvgoodsig;	/* rcvd good TCP signatures */
278 	u_int32_t tcps_inhwcsum;	/* input hardware-checksummed packets */
279 	u_int32_t tcps_outhwcsum;	/* output hardware-checksummed packets */
280 };
281 
282 /*
283  * Names for TCP sysctl objects.
284  */
285 
286 #define	TCPCTL_RFC1323		1 /* enable/disable RFC1323 timestamps/scaling */
287 #define	TCPCTL_KEEPINITTIME	2 /* TCPT_KEEP value */
288 #define TCPCTL_KEEPIDLE		3 /* allow tcp_keepidle to be changed */
289 #define TCPCTL_KEEPINTVL	4 /* allow tcp_keepintvl to be changed */
290 #define TCPCTL_SLOWHZ		5 /* return kernel idea of PR_SLOWHZ */
291 #define TCPCTL_BADDYNAMIC	6 /* return bad dynamic port bitmap */
292 #define	TCPCTL_RECVSPACE	7 /* receive buffer space */
293 #define	TCPCTL_SENDSPACE	8 /* send buffer space */
294 #define	TCPCTL_IDENT		9 /* get connection owner */
295 #define	TCPCTL_SACK	       10 /* selective acknowledgement, rfc 2018 */
296 #define TCPCTL_MSSDFLT	       11 /* Default maximum segment size */
297 #define	TCPCTL_RSTPPSLIMIT     12 /* RST pps limit */
298 #define	TCPCTL_MAXID	       13
299 
300 #define	TCPCTL_NAMES { \
301 	{ 0, 0 }, \
302 	{ "rfc1323",	CTLTYPE_INT }, \
303 	{ "keepinittime",	CTLTYPE_INT }, \
304 	{ "keepidle",	CTLTYPE_INT }, \
305 	{ "keepintvl",	CTLTYPE_INT }, \
306 	{ "slowhz",	CTLTYPE_INT }, \
307 	{ "baddynamic", CTLTYPE_STRUCT }, \
308 	{ "recvspace",	CTLTYPE_INT }, \
309 	{ "sendspace",	CTLTYPE_INT }, \
310 	{ "ident", 	CTLTYPE_STRUCT }, \
311 	{ "sack",	CTLTYPE_INT }, \
312 	{ "mssdflt",	CTLTYPE_INT }, \
313 	{ "rstppslimit",	CTLTYPE_INT }, \
314 }
315 
316 struct tcp_ident_mapping {
317 	struct sockaddr_storage faddr, laddr;
318 	int euid, ruid;
319 };
320 
321 #ifdef _KERNEL
322 struct	inpcbtable tcbtable;	/* head of queue of active tcpcb's */
323 extern	struct tcpstat tcpstat;	/* tcp statistics */
324 u_int32_t tcp_now;		/* for RFC 1323 timestamps */
325 extern	int tcp_do_rfc1323;	/* enabled/disabled? */
326 extern	int tcp_mssdflt;	/* default maximum segment size */
327 #ifdef TCP_SACK
328 extern	int tcp_do_sack;	/* SACK enabled/disabled */
329 #endif
330 
331 int	 tcp_attach __P((struct socket *));
332 void	 tcp_canceltimers __P((struct tcpcb *));
333 struct tcpcb *
334 	 tcp_close __P((struct tcpcb *));
335 #if defined(INET6) && !defined(TCP6)
336 void	 tcp6_ctlinput __P((int, struct sockaddr *, void *));
337 #endif
338 void	 *tcp_ctlinput __P((int, struct sockaddr *, void *));
339 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
340 struct tcpcb *
341 	 tcp_disconnect __P((struct tcpcb *));
342 struct tcpcb *
343 	 tcp_drop __P((struct tcpcb *, int));
344 void	 tcp_dooptions __P((struct tcpcb *, u_char *, int, struct tcphdr *,
345 		int *, u_int32_t *, u_int32_t *));
346 void	 tcp_drain __P((void));
347 void	 tcp_fasttimo __P((void));
348 void	 tcp_init __P((void));
349 #if defined(INET6) && !defined(TCP6)
350 int	 tcp6_input __P((struct mbuf **, int *, int));
351 #endif
352 void	 tcp_input __P((struct mbuf *, ...));
353 int	 tcp_mss __P((struct tcpcb *, int));
354 void	 tcp_mss_update __P((struct tcpcb *));
355 void	 tcp_mtudisc __P((struct inpcb *, int));
356 void	 tcp_mtudisc_increase __P((struct inpcb *, int));
357 #ifdef INET6
358 void	tcp6_mtudisc __P((struct inpcb *, int));
359 void	tcp6_mtudisc_callback __P((struct in6_addr *));
360 #endif
361 struct tcpcb *
362 	 tcp_newtcpcb __P((struct inpcb *));
363 void	 tcp_notify __P((struct inpcb *, int));
364 int	 tcp_output __P((struct tcpcb *));
365 void	 tcp_pulloutofband __P((struct socket *, u_int, struct mbuf *, int));
366 void	 tcp_quench __P((struct inpcb *, int));
367 int	 tcp_reass __P((struct tcpcb *, struct tcphdr *, struct mbuf *, int *));
368 void	 tcp_rscale __P((struct tcpcb *, u_long));
369 void	 tcp_respond __P((struct tcpcb *, caddr_t, struct mbuf *, tcp_seq,
370 		tcp_seq, int));
371 void	 tcp_setpersist __P((struct tcpcb *));
372 void	 tcp_slowtimo __P((void));
373 struct mbuf *
374 	 tcp_template __P((struct tcpcb *));
375 struct tcpcb *
376 	 tcp_timers __P((struct tcpcb *, int));
377 void	 tcp_trace __P((int, int, struct tcpcb *, caddr_t, int, int));
378 struct tcpcb *
379 	 tcp_usrclosed __P((struct tcpcb *));
380 int	 tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t));
381 #if defined(INET6) && !defined(TCP6)
382 int	 tcp6_usrreq __P((struct socket *,
383 	    int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *));
384 #endif
385 int	 tcp_usrreq __P((struct socket *,
386 	    int, struct mbuf *, struct mbuf *, struct mbuf *));
387 void	 tcp_xmit_timer __P((struct tcpcb *, int));
388 void	 tcpdropoldhalfopen __P((struct tcpcb *, u_int16_t));
389 #ifdef TCP_SACK
390 int	 tcp_sack_option __P((struct tcpcb *,struct tcphdr *,u_char *,int));
391 void	 tcp_update_sack_list __P((struct tcpcb *tp));
392 void	 tcp_del_sackholes __P((struct tcpcb *, struct tcphdr *));
393 void	 tcp_clean_sackreport __P((struct tcpcb *tp));
394 void	 tcp_sack_adjust __P((struct tcpcb *tp));
395 struct sackhole *
396 	 tcp_sack_output __P((struct tcpcb *tp));
397 int	 tcp_sack_partialack __P((struct tcpcb *, struct tcphdr *));
398 #ifdef DEBUG
399 void	 tcp_print_holes __P((struct tcpcb *tp));
400 #endif
401 #endif /* TCP_SACK */
402 #if defined(TCP_SACK)
403 int	 tcp_newreno __P((struct tcpcb *, struct tcphdr *));
404 u_long	 tcp_seq_subtract  __P((u_long, u_long ));
405 #endif /* TCP_SACK */
406 #ifdef TCP_SIGNATURE
407 int	tcp_signature_apply __P((caddr_t, caddr_t, unsigned int));
408 #endif /* TCP_SIGNATURE */
409 void	tcp_rndiss_init __P((void));
410 tcp_seq	tcp_rndiss_next __P((void));
411 u_int16_t
412 	tcp_rndiss_encrypt __P((u_int16_t));
413 #endif /* _KERNEL */
414 #endif /* _NETINET_TCP_VAR_H_ */
415