1 //==========================================================================
3 // include/netinet/tcp_var.h
5 //==========================================================================
6 //####BSDCOPYRIGHTBEGIN####
8 // -------------------------------------------
10 // Portions of this software may have been derived from OpenBSD,
11 // FreeBSD or other sources, and are covered by the appropriate
12 // copyright disclaimers included herein.
14 // Portions created by Red Hat are
15 // Copyright (C) 2002 Red Hat, Inc. All Rights Reserved.
17 // -------------------------------------------
19 //####BSDCOPYRIGHTEND####
20 //==========================================================================
23 * Copyright (c) 1982, 1986, 1993, 1994, 1995
24 * The Regents of the University of California. All rights reserved.
26 * Redistribution and use in source and binary forms, with or without
27 * modification, are permitted provided that the following conditions
29 * 1. Redistributions of source code must retain the above copyright
30 * notice, this list of conditions and the following disclaimer.
31 * 2. Redistributions in binary form must reproduce the above copyright
32 * notice, this list of conditions and the following disclaimer in the
33 * documentation and/or other materials provided with the distribution.
34 * 3. All advertising materials mentioning features or use of this software
35 * must display the following acknowledgement:
36 * This product includes software developed by the University of
37 * California, Berkeley and its contributors.
38 * 4. Neither the name of the University nor the names of its contributors
39 * may be used to endorse or promote products derived from this software
40 * without specific prior written permission.
42 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
43 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
44 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
45 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
46 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
47 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
48 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
49 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
50 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
51 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
54 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
55 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.8 2001/08/22 00:59:13 silby Exp $
58 #ifndef _NETINET_TCP_VAR_H_
59 #define _NETINET_TCP_VAR_H_
61 * Kernel variables for tcp.
64 /* TCP segment queue entry */
66 LIST_ENTRY(tseg_qent) tqe_q;
67 int tqe_len; /* TCP segment data length */
68 struct tcphdr *tqe_th; /* a pointer to tcp header */
69 struct mbuf *tqe_m; /* mbuf contains packet */
71 LIST_HEAD(tsegqe_head, tseg_qent);
73 MALLOC_DECLARE(M_TSEGQ);
77 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
81 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
84 * Tcp control block, one per tcp; fields:
85 * Organized for 16 byte cacheline efficiency.
88 struct tsegqe_head t_segq;
89 int t_dupacks; /* consecutive dup acks recd */
90 struct tcptemp *unused; /* unused */
92 struct callout *tt_rexmt; /* retransmit timer */
93 struct callout *tt_persist; /* retransmit persistence */
94 struct callout *tt_keep; /* keepalive */
95 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */
96 struct callout *tt_delack; /* delayed ACK timer */
98 struct inpcb *t_inpcb; /* back pointer to internet pcb */
99 int t_state; /* state of this connection */
101 #define TF_ACKNOW 0x00001 /* ack peer immediately */
102 #define TF_DELACK 0x00002 /* ack, but try to delay it */
103 #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */
104 #define TF_NOOPT 0x00008 /* don't use tcp options */
105 #define TF_SENTFIN 0x00010 /* have sent FIN */
106 #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */
107 #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */
108 #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */
109 #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */
110 #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */
111 #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */
112 #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */
113 #define TF_NOPUSH 0x01000 /* don't push */
114 #define TF_REQ_CC 0x02000 /* have/will request CC */
115 #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */
116 #define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */
117 #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */
118 #define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */
119 int t_force; /* 1 if forcing out a byte */
121 tcp_seq snd_una; /* send unacknowledged */
122 tcp_seq snd_max; /* highest sequence number sent;
123 * used to recognize retransmits
125 tcp_seq snd_nxt; /* send next */
126 tcp_seq snd_up; /* send urgent pointer */
128 tcp_seq snd_wl1; /* window update seg seq number */
129 tcp_seq snd_wl2; /* window update seg ack number */
130 tcp_seq iss; /* initial send sequence number */
131 tcp_seq irs; /* initial receive sequence number */
133 tcp_seq rcv_nxt; /* receive next */
134 tcp_seq rcv_adv; /* advertised window */
135 u_long rcv_wnd; /* receive window */
136 tcp_seq rcv_up; /* receive urgent pointer */
138 u_long snd_wnd; /* send window */
139 u_long snd_cwnd; /* congestion-controlled window */
140 u_long snd_ssthresh; /* snd_cwnd size threshold for
141 * for slow start exponential to
144 tcp_seq snd_recover; /* for use in fast recovery */
146 u_int t_maxopd; /* mss plus options */
148 u_long t_rcvtime; /* inactivity time */
149 u_long t_starttime; /* time connection was established */
150 int t_rtttime; /* round trip time */
151 tcp_seq t_rtseq; /* sequence number being timed */
153 int t_rxtcur; /* current retransmit value (ticks) */
154 u_int t_maxseg; /* maximum segment size */
155 int t_srtt; /* smoothed round-trip time */
156 int t_rttvar; /* variance in round-trip time */
158 int t_rxtshift; /* log(2) of rexmt exp. backoff */
159 u_int t_rttmin; /* minimum rtt allowed */
160 u_long t_rttupdated; /* number of times rtt sampled */
161 u_long max_sndwnd; /* largest window peer has offered */
163 int t_softerror; /* possible error not yet reported */
164 /* out-of-band data */
165 char t_oobflags; /* have some */
166 char t_iobc; /* input character */
167 #define TCPOOB_HAVEDATA 0x01
168 #define TCPOOB_HADDATA 0x02
169 /* RFC 1323 variables */
170 u_char snd_scale; /* window scaling for send window */
171 u_char rcv_scale; /* window scaling for recv window */
172 u_char request_r_scale; /* pending window scaling */
173 u_char requested_s_scale;
174 u_long ts_recent; /* timestamp echo data */
176 u_long ts_recent_age; /* when last updated */
177 tcp_seq last_ack_sent;
178 /* RFC 1644 variables */
179 tcp_cc cc_send; /* send connection count */
180 tcp_cc cc_recv; /* receive connection count */
182 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
183 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
184 u_long t_badrxtwin; /* window for retransmit recovery */
188 * Structure to hold TCP options that are only used during segment
189 * processing (in tcp_input), but not held in the tcpcb.
190 * It's basically used to reduce the number of parameters
194 u_long to_flag; /* which options are present */
195 #define TOF_TS 0x0001 /* timestamp */
196 #define TOF_CC 0x0002 /* CC and CCnew are exclusive */
197 #define TOF_CCNEW 0x0004
198 #define TOF_CCECHO 0x0008
201 tcp_cc to_cc; /* holds CC or CCnew */
206 * The TAO cache entry which is stored in the protocol family specific
207 * portion of the route metrics.
210 tcp_cc tao_cc; /* latest CC in valid SYN */
211 tcp_cc tao_ccsent; /* latest CC sent to peer */
212 u_short tao_mssopt; /* peer's cached MSS */
214 u_short tao_flags; /* cache status flags */
215 #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */
216 #define TAOF_OK 0x0002 /* peer does understand rfc1644 */
217 #define TAOF_UNDEF 0 /* we don't know yet */
220 #define rmx_taop(r) ((struct rmxp_tao *)(r).rmx_filler)
222 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
223 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
226 * The smoothed round-trip time and estimated variance
227 * are stored as fixed point numbers scaled by the values below.
228 * For convenience, these scales are also used in smoothing the average
229 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
230 * With these scales, srtt has 3 bits to the right of the binary point,
231 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
232 * binary point, and is smoothed with an ALPHA of 0.75.
234 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
235 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
236 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
237 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
238 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
241 * The initial retransmission should happen at rtt + 4 * rttvar.
242 * Because of the way we do the smoothing, srtt and rttvar
243 * will each average +1/2 tick of bias. When we compute
244 * the retransmit timer, we want 1/2 tick of rounding and
245 * 1 extra tick because of +-1/2 tick uncertainty in the
246 * firing of the timer. The bias will give us exactly the
247 * 1.5 tick we need. But, because the bias is
248 * statistical, we have to test that we don't drop below
249 * the minimum feasible timer (which is 2 ticks).
250 * This version of the macro adapted from a paper by Lawrence
251 * Brakmo and Larry Peterson which outlines a problem caused
252 * by insufficient precision in the original implementation,
253 * which results in inappropriately large RTO values for very
256 #define TCP_REXMTVAL(tp) \
257 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
258 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
262 * Many of these should be kept per connection,
263 * but that's inconvenient at the moment.
266 u_long tcps_connattempt; /* connections initiated */
267 u_long tcps_accepts; /* connections accepted */
268 u_long tcps_connects; /* connections established */
269 u_long tcps_drops; /* connections dropped */
270 u_long tcps_conndrops; /* embryonic connections dropped */
271 u_long tcps_closed; /* conn. closed (includes drops) */
272 u_long tcps_segstimed; /* segs where we tried to get rtt */
273 u_long tcps_rttupdated; /* times we succeeded */
274 u_long tcps_delack; /* delayed acks sent */
275 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
276 u_long tcps_rexmttimeo; /* retransmit timeouts */
277 u_long tcps_persisttimeo; /* persist timeouts */
278 u_long tcps_keeptimeo; /* keepalive timeouts */
279 u_long tcps_keepprobe; /* keepalive probes sent */
280 u_long tcps_keepdrops; /* connections dropped in keepalive */
282 u_long tcps_sndtotal; /* total packets sent */
283 u_long tcps_sndpack; /* data packets sent */
284 u_long tcps_sndbyte; /* data bytes sent */
285 u_long tcps_sndrexmitpack; /* data packets retransmitted */
286 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
287 u_long tcps_sndacks; /* ack-only packets sent */
288 u_long tcps_sndprobe; /* window probes sent */
289 u_long tcps_sndurg; /* packets sent with URG only */
290 u_long tcps_sndwinup; /* window update-only packets sent */
291 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
293 u_long tcps_rcvtotal; /* total packets received */
294 u_long tcps_rcvpack; /* packets received in sequence */
295 u_long tcps_rcvbyte; /* bytes received in sequence */
296 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
297 u_long tcps_rcvbadoff; /* packets received with bad offset */
298 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
299 u_long tcps_rcvshort; /* packets received too short */
300 u_long tcps_rcvduppack; /* duplicate-only packets received */
301 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
302 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
303 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
304 u_long tcps_rcvoopack; /* out-of-order packets received */
305 u_long tcps_rcvoobyte; /* out-of-order bytes received */
306 u_long tcps_rcvpackafterwin; /* packets with data after window */
307 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
308 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
309 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
310 u_long tcps_rcvdupack; /* rcvd duplicate acks */
311 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
312 u_long tcps_rcvackpack; /* rcvd ack packets */
313 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
314 u_long tcps_rcvwinupd; /* rcvd window update packets */
315 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
316 u_long tcps_predack; /* times hdr predict ok for acks */
317 u_long tcps_preddat; /* times hdr predict ok for data pkts */
318 u_long tcps_pcbcachemiss;
319 u_long tcps_cachedrtt; /* times cached RTT in route updated */
320 u_long tcps_cachedrttvar; /* times cached rttvar updated */
321 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
322 u_long tcps_usedrtt; /* times RTT initialized from route */
323 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
324 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
325 u_long tcps_persistdrop; /* timeout in persist state */
326 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
327 u_long tcps_mturesent; /* resends due to MTU discovery */
328 u_long tcps_listendrop; /* listen queue overflows */
332 * TCB structure exported to user-land via sysctl(3).
333 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
334 * included. Not all of our clients do.
336 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
341 struct xsocket xt_socket;
342 u_quad_t xt_alignment_hack;
347 * Names for TCP sysctl objects
349 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
350 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */
351 #define TCPCTL_MSSDFLT 3 /* MSS default */
352 #define TCPCTL_STATS 4 /* statistics (read-only) */
353 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */
354 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
355 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
356 #define TCPCTL_SENDSPACE 8 /* send buffer space */
357 #define TCPCTL_RECVSPACE 9 /* receive buffer space */
358 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
359 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
360 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
361 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
362 #define TCPCTL_MAXID 14
364 #define TCPCTL_NAMES { \
366 { "rfc1323", CTLTYPE_INT }, \
367 { "rfc1644", CTLTYPE_INT }, \
368 { "mssdflt", CTLTYPE_INT }, \
369 { "stats", CTLTYPE_STRUCT }, \
370 { "rttdflt", CTLTYPE_INT }, \
371 { "keepidle", CTLTYPE_INT }, \
372 { "keepintvl", CTLTYPE_INT }, \
373 { "sendspace", CTLTYPE_INT }, \
374 { "recvspace", CTLTYPE_INT }, \
375 { "keepinit", CTLTYPE_INT }, \
376 { "pcblist", CTLTYPE_STRUCT }, \
377 { "delacktime", CTLTYPE_INT }, \
378 { "v6mssdflt", CTLTYPE_INT }, \
384 SYSCTL_DECL(_net_inet_tcp);
387 extern struct inpcbhead tcb; /* head of queue of active tcpcb's */
388 extern struct inpcbinfo tcbinfo;
389 extern struct tcpstat tcpstat; /* tcp statistics */
390 extern int tcp_mssdflt; /* XXX */
391 extern int tcp_delack_enabled;
392 extern int tcp_do_newreno;
394 extern int ss_fltsz_local;
396 void tcp_canceltimers __P((struct tcpcb *));
398 tcp_close __P((struct tcpcb *));
399 void tcp_ctlinput __P((int, struct sockaddr *, void *));
400 int tcp_ctloutput __P((struct socket *, struct sockopt *));
402 tcp_drop __P((struct tcpcb *, int));
403 void tcp_drain __P((void));
404 void tcp_fasttimo __P((void));
406 tcp_gettaocache __P((struct inpcb *));
407 void tcp_init __P((void));
408 void tcp_input __P((struct mbuf *, int));
409 void tcp_mss __P((struct tcpcb *, int));
410 int tcp_mssopt __P((struct tcpcb *));
411 void tcp_drop_syn_sent __P((struct inpcb *, int));
412 void tcp_mtudisc __P((struct inpcb *, int));
414 tcp_newtcpcb __P((struct inpcb *));
415 int tcp_output __P((struct tcpcb *));
416 void tcp_quench __P((struct inpcb *, int));
417 void tcp_respond __P((struct tcpcb *, void *,
418 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int));
420 tcp_rtlookup __P((struct inpcb *));
421 void tcp_setpersist __P((struct tcpcb *));
422 void tcp_slowtimo __P((void));
424 tcp_maketemplate __P((struct tcpcb *));
425 void tcp_fillheaders __P((struct tcpcb *, void *, void *));
427 tcp_timers __P((struct tcpcb *, int));
428 void tcp_trace __P((int, int, struct tcpcb *, void *, struct tcphdr *,
431 extern struct pr_usrreqs tcp_usrreqs;
432 extern u_long tcp_sendspace;
433 extern u_long tcp_recvspace;
434 tcp_seq tcp_new_isn __P((struct tcpcb *));
438 #endif /* _NETINET_TCP_VAR_H_ */