xref: /netbsd-src/sys/netinet/tcp_var.h (revision d9158b13b5dfe46201430699a3f7a235ecf28df3)
1 /*
2  * Copyright (c) 1982, 1986, 1993, 1994
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  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	from: @(#)tcp_var.h	8.3 (Berkeley) 4/10/94
34  *	$Id: tcp_var.h,v 1.8 1994/05/13 06:06:51 mycroft Exp $
35  */
36 
37 /*
38  * Kernel variables for tcp.
39  */
40 
41 /*
42  * Tcp control block, one per tcp; fields:
43  */
44 struct tcpcb {
45 	struct	tcpiphdr *seg_next;	/* sequencing queue */
46 	struct	tcpiphdr *seg_prev;
47 	short	t_state;		/* state of this connection */
48 	short	t_timer[TCPT_NTIMERS];	/* tcp timers */
49 	short	t_rxtshift;		/* log(2) of rexmt exp. backoff */
50 	short	t_rxtcur;		/* current retransmit value */
51 	short	t_dupacks;		/* consecutive dup acks recd */
52 	u_short	t_maxseg;		/* maximum segment size */
53 	char	t_force;		/* 1 if forcing out a byte */
54 	u_short	t_flags;
55 #define	TF_ACKNOW	0x0001		/* ack peer immediately */
56 #define	TF_DELACK	0x0002		/* ack, but try to delay it */
57 #define	TF_NODELAY	0x0004		/* don't delay packets to coalesce */
58 #define	TF_NOOPT	0x0008		/* don't use tcp options */
59 #define	TF_SENTFIN	0x0010		/* have sent FIN */
60 #define	TF_REQ_SCALE	0x0020		/* have/will request window scaling */
61 #define	TF_RCVD_SCALE	0x0040		/* other side has requested scaling */
62 #define	TF_REQ_TSTMP	0x0080		/* have/will request timestamps */
63 #define	TF_RCVD_TSTMP	0x0100		/* a timestamp was received in SYN */
64 #define	TF_SACK_PERMIT	0x0200		/* other side said I could SACK */
65 
66 	struct	tcpiphdr *t_template;	/* skeletal packet for transmit */
67 	struct	inpcb *t_inpcb;		/* back pointer to internet pcb */
68 /*
69  * The following fields are used as in the protocol specification.
70  * See RFC783, Dec. 1981, page 21.
71  */
72 /* send sequence variables */
73 	tcp_seq	snd_una;		/* send unacknowledged */
74 	tcp_seq	snd_nxt;		/* send next */
75 	tcp_seq	snd_up;			/* send urgent pointer */
76 	tcp_seq	snd_wl1;		/* window update seg seq number */
77 	tcp_seq	snd_wl2;		/* window update seg ack number */
78 	tcp_seq	iss;			/* initial send sequence number */
79 	u_long	snd_wnd;		/* send window */
80 /* receive sequence variables */
81 	u_long	rcv_wnd;		/* receive window */
82 	tcp_seq	rcv_nxt;		/* receive next */
83 	tcp_seq	rcv_up;			/* receive urgent pointer */
84 	tcp_seq	irs;			/* initial receive sequence number */
85 /*
86  * Additional variables for this implementation.
87  */
88 /* receive variables */
89 	tcp_seq	rcv_adv;		/* advertised window */
90 /* retransmit variables */
91 	tcp_seq	snd_max;		/* highest sequence number sent;
92 					 * used to recognize retransmits
93 					 */
94 /* congestion control (for slow start, source quench, retransmit after loss) */
95 	u_long	snd_cwnd;		/* congestion-controlled window */
96 	u_long	snd_ssthresh;		/* snd_cwnd size threshhold for
97 					 * for slow start exponential to
98 					 * linear switch
99 					 */
100 /*
101  * transmit timing stuff.  See below for scale of srtt and rttvar.
102  * "Variance" is actually smoothed difference.
103  */
104 	short	t_idle;			/* inactivity time */
105 	short	t_rtt;			/* round trip time */
106 	tcp_seq	t_rtseq;		/* sequence number being timed */
107 	short	t_srtt;			/* smoothed round-trip time */
108 	short	t_rttvar;		/* variance in round-trip time */
109 	u_short	t_rttmin;		/* minimum rtt allowed */
110 	u_long	max_sndwnd;		/* largest window peer has offered */
111 
112 /* out-of-band data */
113 	char	t_oobflags;		/* have some */
114 	char	t_iobc;			/* input character */
115 #define	TCPOOB_HAVEDATA	0x01
116 #define	TCPOOB_HADDATA	0x02
117 	short	t_softerror;		/* possible error not yet reported */
118 
119 /* RFC 1323 variables */
120 	u_char	snd_scale;		/* window scaling for send window */
121 	u_char	rcv_scale;		/* window scaling for recv window */
122 	u_char	request_r_scale;	/* pending window scaling */
123 	u_char	requested_s_scale;
124 	u_long	ts_recent;		/* timestamp echo data */
125 	u_long	ts_recent_age;		/* when last updated */
126 	tcp_seq	last_ack_sent;
127 
128 /* TUBA stuff */
129 	caddr_t	t_tuba_pcb;		/* next level down pcb for TCP over z */
130 };
131 
132 #define	intotcpcb(ip)	((struct tcpcb *)(ip)->inp_ppcb)
133 #define	sototcpcb(so)	(intotcpcb(sotoinpcb(so)))
134 
135 /*
136  * The smoothed round-trip time and estimated variance
137  * are stored as fixed point numbers scaled by the values below.
138  * For convenience, these scales are also used in smoothing the average
139  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
140  * With these scales, srtt has 3 bits to the right of the binary point,
141  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
142  * binary point, and is smoothed with an ALPHA of 0.75.
143  */
144 #define	TCP_RTT_SCALE		8	/* multiplier for srtt; 3 bits frac. */
145 #define	TCP_RTT_SHIFT		3	/* shift for srtt; 3 bits frac. */
146 #define	TCP_RTTVAR_SCALE	4	/* multiplier for rttvar; 2 bits */
147 #define	TCP_RTTVAR_SHIFT	2	/* multiplier for rttvar; 2 bits */
148 
149 /*
150  * The initial retransmission should happen at rtt + 4 * rttvar.
151  * Because of the way we do the smoothing, srtt and rttvar
152  * will each average +1/2 tick of bias.  When we compute
153  * the retransmit timer, we want 1/2 tick of rounding and
154  * 1 extra tick because of +-1/2 tick uncertainty in the
155  * firing of the timer.  The bias will give us exactly the
156  * 1.5 tick we need.  But, because the bias is
157  * statistical, we have to test that we don't drop below
158  * the minimum feasible timer (which is 2 ticks).
159  * This macro assumes that the value of TCP_RTTVAR_SCALE
160  * is the same as the multiplier for rttvar.
161  */
162 #define	TCP_REXMTVAL(tp) \
163 	(((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar)
164 
165 /* XXX
166  * We want to avoid doing m_pullup on incoming packets but that
167  * means avoiding dtom on the tcp reassembly code.  That in turn means
168  * keeping an mbuf pointer in the reassembly queue (since we might
169  * have a cluster).  As a quick hack, the source & destination
170  * port numbers (which are no longer needed once we've located the
171  * tcpcb) are overlayed with an mbuf pointer.
172  */
173 #define REASS_MBUF(ti) (*(struct mbuf **)&((ti)->ti_t))
174 
175 /*
176  * TCP statistics.
177  * Many of these should be kept per connection,
178  * but that's inconvenient at the moment.
179  */
180 struct	tcpstat {
181 	u_long	tcps_connattempt;	/* connections initiated */
182 	u_long	tcps_accepts;		/* connections accepted */
183 	u_long	tcps_connects;		/* connections established */
184 	u_long	tcps_drops;		/* connections dropped */
185 	u_long	tcps_conndrops;		/* embryonic connections dropped */
186 	u_long	tcps_closed;		/* conn. closed (includes drops) */
187 	u_long	tcps_segstimed;		/* segs where we tried to get rtt */
188 	u_long	tcps_rttupdated;	/* times we succeeded */
189 	u_long	tcps_delack;		/* delayed acks sent */
190 	u_long	tcps_timeoutdrop;	/* conn. dropped in rxmt timeout */
191 	u_long	tcps_rexmttimeo;	/* retransmit timeouts */
192 	u_long	tcps_persisttimeo;	/* persist timeouts */
193 	u_long	tcps_keeptimeo;		/* keepalive timeouts */
194 	u_long	tcps_keepprobe;		/* keepalive probes sent */
195 	u_long	tcps_keepdrops;		/* connections dropped in keepalive */
196 
197 	u_long	tcps_sndtotal;		/* total packets sent */
198 	u_long	tcps_sndpack;		/* data packets sent */
199 	u_long	tcps_sndbyte;		/* data bytes sent */
200 	u_long	tcps_sndrexmitpack;	/* data packets retransmitted */
201 	u_long	tcps_sndrexmitbyte;	/* data bytes retransmitted */
202 	u_long	tcps_sndacks;		/* ack-only packets sent */
203 	u_long	tcps_sndprobe;		/* window probes sent */
204 	u_long	tcps_sndurg;		/* packets sent with URG only */
205 	u_long	tcps_sndwinup;		/* window update-only packets sent */
206 	u_long	tcps_sndctrl;		/* control (SYN|FIN|RST) packets sent */
207 
208 	u_long	tcps_rcvtotal;		/* total packets received */
209 	u_long	tcps_rcvpack;		/* packets received in sequence */
210 	u_long	tcps_rcvbyte;		/* bytes received in sequence */
211 	u_long	tcps_rcvbadsum;		/* packets received with ccksum errs */
212 	u_long	tcps_rcvbadoff;		/* packets received with bad offset */
213 	u_long	tcps_rcvshort;		/* packets received too short */
214 	u_long	tcps_rcvduppack;	/* duplicate-only packets received */
215 	u_long	tcps_rcvdupbyte;	/* duplicate-only bytes received */
216 	u_long	tcps_rcvpartduppack;	/* packets with some duplicate data */
217 	u_long	tcps_rcvpartdupbyte;	/* dup. bytes in part-dup. packets */
218 	u_long	tcps_rcvoopack;		/* out-of-order packets received */
219 	u_long	tcps_rcvoobyte;		/* out-of-order bytes received */
220 	u_long	tcps_rcvpackafterwin;	/* packets with data after window */
221 	u_long	tcps_rcvbyteafterwin;	/* bytes rcvd after window */
222 	u_long	tcps_rcvafterclose;	/* packets rcvd after "close" */
223 	u_long	tcps_rcvwinprobe;	/* rcvd window probe packets */
224 	u_long	tcps_rcvdupack;		/* rcvd duplicate acks */
225 	u_long	tcps_rcvacktoomuch;	/* rcvd acks for unsent data */
226 	u_long	tcps_rcvackpack;	/* rcvd ack packets */
227 	u_long	tcps_rcvackbyte;	/* bytes acked by rcvd acks */
228 	u_long	tcps_rcvwinupd;		/* rcvd window update packets */
229 	u_long	tcps_pawsdrop;		/* segments dropped due to PAWS */
230 	u_long	tcps_predack;		/* times hdr predict ok for acks */
231 	u_long	tcps_preddat;		/* times hdr predict ok for data pkts */
232 	u_long	tcps_pcbcachemiss;
233 };
234 
235 #ifdef KERNEL
236 struct	inpcb tcb;		/* head of queue of active tcpcb's */
237 struct	tcpstat tcpstat;	/* tcp statistics */
238 u_long	tcp_now;		/* for RFC 1323 timestamps */
239 
240 int	 tcp_attach __P((struct socket *));
241 void	 tcp_canceltimers __P((struct tcpcb *));
242 struct tcpcb *
243 	 tcp_close __P((struct tcpcb *));
244 void	 tcp_ctlinput __P((int, struct sockaddr *, struct ip *));
245 int	 tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **));
246 struct tcpcb *
247 	 tcp_disconnect __P((struct tcpcb *));
248 struct tcpcb *
249 	 tcp_drop __P((struct tcpcb *, int));
250 void	 tcp_dooptions __P((struct tcpcb *,
251 	    u_char *, int, struct tcpiphdr *, int *, u_long *, u_long *));
252 void	 tcp_drain __P((void));
253 void	 tcp_fasttimo __P((void));
254 void	 tcp_init __P((void));
255 void	 tcp_input __P((struct mbuf *, int));
256 int	 tcp_mss __P((struct tcpcb *, u_int));
257 struct tcpcb *
258 	 tcp_newtcpcb __P((struct inpcb *));
259 void	 tcp_notify __P((struct inpcb *, int));
260 int	 tcp_output __P((struct tcpcb *));
261 void	 tcp_pulloutofband __P((struct socket *,
262 	    struct tcpiphdr *, struct mbuf *));
263 void	 tcp_quench __P((struct inpcb *, int));
264 int	 tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *));
265 void	 tcp_respond __P((struct tcpcb *,
266 	    struct tcpiphdr *, struct mbuf *, u_long, u_long, int));
267 void	 tcp_setpersist __P((struct tcpcb *));
268 void	 tcp_slowtimo __P((void));
269 struct tcpiphdr *
270 	 tcp_template __P((struct tcpcb *));
271 struct tcpcb *
272 	 tcp_timers __P((struct tcpcb *, int));
273 void	 tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int));
274 struct tcpcb *
275 	 tcp_usrclosed __P((struct tcpcb *));
276 int	 tcp_usrreq __P((struct socket *,
277 	    int, struct mbuf *, struct mbuf *, struct mbuf *));
278 void	 tcp_xmit_timer __P((struct tcpcb *, int));
279 #endif
280