Expand PMF_FN_* macros.
[netbsd-mini2440.git] / sys / netinet / tcp_var.h
blobd62265b6b8fbd6748be31d2fe8c0d9ccaa481a20
1 /* $NetBSD: tcp_var.h,v 1.161 2009/09/09 22:41:28 darran Exp $ */
3 /*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
10 * 1. Redistributions of source code must retain the above copyright
11 * notice, this list of conditions and the following disclaimer.
12 * 2. Redistributions in binary form must reproduce the above copyright
13 * notice, this list of conditions and the following disclaimer in the
14 * documentation and/or other materials provided with the distribution.
15 * 3. Neither the name of the project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
33 * @(#)COPYRIGHT 1.1 (NRL) 17 January 1995
35 * NRL grants permission for redistribution and use in source and binary
36 * forms, with or without modification, of the software and documentation
37 * created at NRL provided that the following conditions are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 3. All advertising materials mentioning features or use of this software
45 * must display the following acknowledgements:
46 * This product includes software developed by the University of
47 * California, Berkeley and its contributors.
48 * This product includes software developed at the Information
49 * Technology Division, US Naval Research Laboratory.
50 * 4. Neither the name of the NRL nor the names of its contributors
51 * may be used to endorse or promote products derived from this software
52 * without specific prior written permission.
54 * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
55 * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
56 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
57 * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NRL OR
58 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
59 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
60 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
61 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
62 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
63 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
64 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
66 * The views and conclusions contained in the software and documentation
67 * are those of the authors and should not be interpreted as representing
68 * official policies, either expressed or implied, of the US Naval
69 * Research Laboratory (NRL).
72 /*-
73 * Copyright (c) 1997, 1998, 1999, 2001, 2005 The NetBSD Foundation, Inc.
74 * All rights reserved.
76 * This code is derived from software contributed to The NetBSD Foundation
77 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility,
78 * NASA Ames Research Center.
79 * This code is derived from software contributed to The NetBSD Foundation
80 * by Charles M. Hannum.
82 * Redistribution and use in source and binary forms, with or without
83 * modification, are permitted provided that the following conditions
84 * are met:
85 * 1. Redistributions of source code must retain the above copyright
86 * notice, this list of conditions and the following disclaimer.
87 * 2. Redistributions in binary form must reproduce the above copyright
88 * notice, this list of conditions and the following disclaimer in the
89 * documentation and/or other materials provided with the distribution.
91 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
92 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
93 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
94 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
95 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
96 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
97 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
98 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
99 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
100 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
101 * POSSIBILITY OF SUCH DAMAGE.
105 * Copyright (c) 1982, 1986, 1993, 1994, 1995
106 * The Regents of the University of California. All rights reserved.
108 * Redistribution and use in source and binary forms, with or without
109 * modification, are permitted provided that the following conditions
110 * are met:
111 * 1. Redistributions of source code must retain the above copyright
112 * notice, this list of conditions and the following disclaimer.
113 * 2. Redistributions in binary form must reproduce the above copyright
114 * notice, this list of conditions and the following disclaimer in the
115 * documentation and/or other materials provided with the distribution.
116 * 3. Neither the name of the University nor the names of its contributors
117 * may be used to endorse or promote products derived from this software
118 * without specific prior written permission.
120 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
121 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
122 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
123 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
124 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
125 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
126 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
127 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
128 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
129 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
130 * SUCH DAMAGE.
132 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
135 #ifndef _NETINET_TCP_VAR_H_
136 #define _NETINET_TCP_VAR_H_
138 #if defined(_KERNEL_OPT)
139 #include "opt_inet.h"
140 #include "opt_mbuftrace.h"
141 #include "rnd.h"
142 #endif
145 * Kernel variables for tcp.
148 #include <sys/callout.h>
150 #ifdef TCP_SIGNATURE
152 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
153 * for SADB verification and lookup.
155 #define TCP_SIGLEN 16 /* length of computed digest in bytes */
156 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
157 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
159 * Only a single SA per host may be specified at this time. An SPI is
160 * needed in order for the KEY_ALLOCSA() lookup to work.
162 #define TCP_SIG_SPI 0x1000
163 #endif /* TCP_SIGNATURE */
166 * SACK option block.
168 struct sackblk {
169 tcp_seq left; /* Left edge of sack block. */
170 tcp_seq right; /* Right edge of sack block. */
173 TAILQ_HEAD(sackhead, sackhole);
174 struct sackhole {
175 tcp_seq start;
176 tcp_seq end;
177 tcp_seq rxmit;
179 TAILQ_ENTRY(sackhole) sackhole_q;
183 * Tcp control block, one per tcp; fields:
185 struct tcpcb {
186 int t_family; /* address family on the wire */
187 struct ipqehead segq; /* sequencing queue */
188 int t_segqlen; /* length of the above */
189 callout_t t_timer[TCPT_NTIMERS];/* tcp timers */
190 short t_state; /* state of this connection */
191 short t_rxtshift; /* log(2) of rexmt exp. backoff */
192 uint32_t t_rxtcur; /* current retransmit value */
193 short t_dupacks; /* consecutive dup acks recd */
194 short t_partialacks; /* partials acks during fast rexmit */
195 u_short t_peermss; /* peer's maximum segment size */
196 u_short t_ourmss; /* our's maximum segment size */
197 u_short t_segsz; /* current segment size in use */
198 char t_force; /* 1 if forcing out a byte */
199 u_int t_flags;
200 #define TF_ACKNOW 0x0001 /* ack peer immediately */
201 #define TF_DELACK 0x0002 /* ack, but try to delay it */
202 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */
203 #define TF_NOOPT 0x0008 /* don't use tcp options */
204 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */
205 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */
206 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */
207 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */
208 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */
209 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */
210 #define TF_WILL_SACK 0x0800 /* try to use SACK */
211 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */
212 #define TF_DEAD 0x2000 /* dead and to-be-released */
213 #define TF_PMTUD_PEND 0x4000 /* Path MTU Discovery pending */
214 #define TF_ECN_PERMIT 0x10000 /* other side said is ECN-ready */
215 #define TF_ECN_SND_CWR 0x20000 /* ECN CWR in queue */
216 #define TF_ECN_SND_ECE 0x40000 /* ECN ECE in queue */
217 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
220 struct mbuf *t_template; /* skeletal packet for transmit */
221 struct inpcb *t_inpcb; /* back pointer to internet pcb */
222 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */
223 callout_t t_delack_ch; /* delayed ACK callout */
225 * The following fields are used as in the protocol specification.
226 * See RFC793, Dec. 1981, page 21.
228 /* send sequence variables */
229 tcp_seq snd_una; /* send unacknowledged */
230 tcp_seq snd_nxt; /* send next */
231 tcp_seq snd_up; /* send urgent pointer */
232 tcp_seq snd_wl1; /* window update seg seq number */
233 tcp_seq snd_wl2; /* window update seg ack number */
234 tcp_seq iss; /* initial send sequence number */
235 u_long snd_wnd; /* send window */
236 tcp_seq snd_recover; /* for use in fast recovery */
237 tcp_seq snd_high; /* NewReno false fast rexmit seq */
238 /* receive sequence variables */
239 u_long rcv_wnd; /* receive window */
240 tcp_seq rcv_nxt; /* receive next */
241 tcp_seq rcv_up; /* receive urgent pointer */
242 tcp_seq irs; /* initial receive sequence number */
244 * Additional variables for this implementation.
246 /* receive variables */
247 tcp_seq rcv_adv; /* advertised window */
248 /* retransmit variables */
249 tcp_seq snd_max; /* highest sequence number sent;
250 * used to recognize retransmits
252 /* congestion control (for slow start, source quench, retransmit after loss) */
253 u_long snd_cwnd; /* congestion-controlled window */
254 u_long snd_ssthresh; /* snd_cwnd size threshhold for
255 * for slow start exponential to
256 * linear switch
258 /* auto-sizing variables */
259 u_int rfbuf_cnt; /* recv buffer autoscaling byte count */
260 uint32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
263 * transmit timing stuff. See below for scale of srtt and rttvar.
264 * "Variance" is actually smoothed difference.
266 uint32_t t_rcvtime; /* time last segment received */
267 uint32_t t_rtttime; /* time we started measuring rtt */
268 tcp_seq t_rtseq; /* sequence number being timed */
269 int32_t t_srtt; /* smoothed round-trip time */
270 int32_t t_rttvar; /* variance in round-trip time */
271 uint32_t t_rttmin; /* minimum rtt allowed */
272 u_long max_sndwnd; /* largest window peer has offered */
274 /* out-of-band data */
275 char t_oobflags; /* have some */
276 char t_iobc; /* input character */
277 #define TCPOOB_HAVEDATA 0x01
278 #define TCPOOB_HADDATA 0x02
279 short t_softerror; /* possible error not yet reported */
281 /* RFC 1323 variables */
282 u_char snd_scale; /* window scaling for send window */
283 u_char rcv_scale; /* window scaling for recv window */
284 u_char request_r_scale; /* pending window scaling */
285 u_char requested_s_scale;
286 u_int32_t ts_recent; /* timestamp echo data */
287 u_int32_t ts_recent_age; /* when last updated */
288 u_int32_t ts_timebase; /* our timebase */
289 tcp_seq last_ack_sent;
291 /* RFC 3465 variables */
292 u_long t_bytes_acked; /* ABC "bytes_acked" parameter */
294 /* SACK stuff */
295 #define TCP_SACK_MAX 3
296 #define TCPSACK_NONE 0
297 #define TCPSACK_HAVED 1
298 u_char rcv_sack_flags; /* SACK flags. */
299 struct sackblk rcv_dsack_block; /* RX D-SACK block. */
300 struct ipqehead timeq; /* time sequenced queue. */
301 struct sackhead snd_holes; /* TX SACK holes. */
302 int snd_numholes; /* Number of TX SACK holes. */
303 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/
304 tcp_seq sack_newdata; /* New data xmitted in this recovery
305 episode starts at this seq number*/
306 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by
307 peer. */
309 /* pointer for syn cache entries*/
310 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */
312 /* prediction of next mbuf when using large window sizes */
313 struct mbuf *t_lastm; /* last mbuf that data was sent from */
314 int t_inoff; /* data offset in previous mbuf */
315 int t_lastoff; /* last data address in mbuf chain */
316 int t_lastlen; /* last length read from mbuf chain */
318 /* Path-MTU discovery blackhole detection */
319 int t_mtudisc; /* perform mtudisc for this tcb */
320 /* Path-MTU Discovery Information */
321 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */
322 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */
323 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */
324 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */
325 u_short t_pmtud_ip_len; /* IP length from ICMP payload */
326 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */
328 uint8_t t_ecn_retries; /* # of ECN setup retries */
330 const struct tcp_congctl *t_congctl; /* per TCB congctl algorithm */
332 /* Keepalive per socket */
333 u_int t_keepinit;
334 u_int t_keepidle;
335 u_int t_keepintvl;
336 u_int t_keepcnt;
337 u_int t_maxidle; /* t_keepcnt * t_keepintvl */
342 * Macros to aid ECN TCP.
344 #define TCP_ECN_ALLOWED(tp) (tp->t_flags & TF_ECN_PERMIT)
347 * Macros to aid SACK/FACK TCP.
349 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK)
350 #define TCP_FACK_FASTRECOV(tp) \
351 (TCP_SACK_ENABLED(tp) && \
352 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz)))
354 #ifdef _KERNEL
356 * TCP reassembly queue locks.
358 static __inline int tcp_reass_lock_try (struct tcpcb *)
359 __unused;
360 static __inline void tcp_reass_unlock (struct tcpcb *)
361 __unused;
363 static __inline int
364 tcp_reass_lock_try(struct tcpcb *tp)
366 int s;
369 * Use splvm() -- we're blocking things that would cause
370 * mbuf allocation.
372 s = splvm();
373 if (tp->t_flags & TF_REASSEMBLING) {
374 splx(s);
375 return (0);
377 tp->t_flags |= TF_REASSEMBLING;
378 splx(s);
379 return (1);
382 static __inline void
383 tcp_reass_unlock(struct tcpcb *tp)
385 int s;
387 s = splvm();
388 tp->t_flags &= ~TF_REASSEMBLING;
389 splx(s);
392 #ifdef DIAGNOSTIC
393 #define TCP_REASS_LOCK(tp) \
394 do { \
395 if (tcp_reass_lock_try(tp) == 0) { \
396 printf("%s:%d: tcpcb %p reass already locked\n", \
397 __FILE__, __LINE__, tp); \
398 panic("tcp_reass_lock"); \
400 } while (/*CONSTCOND*/ 0)
401 #define TCP_REASS_LOCK_CHECK(tp) \
402 do { \
403 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \
404 printf("%s:%d: tcpcb %p reass lock not held\n", \
405 __FILE__, __LINE__, tp); \
406 panic("tcp reass lock check"); \
408 } while (/*CONSTCOND*/ 0)
409 #else
410 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp))
411 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */
412 #endif
414 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp))
415 #endif /* _KERNEL */
418 * Queue for delayed ACK processing.
420 #ifdef _KERNEL
421 extern int tcp_delack_ticks;
422 void tcp_delack(void *);
424 #define TCP_RESTART_DELACK(tp) \
425 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \
426 tcp_delack, tp)
428 #define TCP_SET_DELACK(tp) \
429 do { \
430 if (((tp)->t_flags & TF_DELACK) == 0) { \
431 (tp)->t_flags |= TF_DELACK; \
432 TCP_RESTART_DELACK(tp); \
434 } while (/*CONSTCOND*/0)
436 #define TCP_CLEAR_DELACK(tp) \
437 do { \
438 if ((tp)->t_flags & TF_DELACK) { \
439 (tp)->t_flags &= ~TF_DELACK; \
440 callout_stop(&(tp)->t_delack_ch); \
442 } while (/*CONSTCOND*/0)
443 #endif /* _KERNEL */
446 * Compute the current timestamp for a connection.
448 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase)
451 * Handy way of passing around TCP option info.
453 struct tcp_opt_info {
454 int ts_present;
455 u_int32_t ts_val;
456 u_int32_t ts_ecr;
457 u_int16_t maxseg;
460 #define TOF_SIGNATURE 0x0040 /* signature option present */
461 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */
464 * Data for the TCP compressed state engine.
466 union syn_cache_sa {
467 struct sockaddr sa;
468 struct sockaddr_in sin;
469 #if 1 /*def INET6*/
470 struct sockaddr_in6 sin6;
471 #endif
474 struct syn_cache {
475 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */
476 callout_t sc_timer; /* rexmt timer */
477 struct route sc_route;
478 long sc_win; /* advertised window */
479 int sc_bucketidx; /* our bucket index */
480 u_int32_t sc_hash;
481 u_int32_t sc_timestamp; /* timestamp from SYN */
482 u_int32_t sc_timebase; /* our local timebase */
483 union syn_cache_sa sc_src;
484 union syn_cache_sa sc_dst;
485 tcp_seq sc_irs;
486 tcp_seq sc_iss;
487 u_int sc_rxtcur; /* current rxt timeout */
488 u_int sc_rxttot; /* total time spend on queues */
489 u_short sc_rxtshift; /* for computing backoff */
490 u_short sc_flags;
492 #define SCF_UNREACH 0x0001 /* we've had an unreach error */
493 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */
494 #define SCF_DEAD 0x0004 /* this entry to be released */
495 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */
496 #define SCF_ECN_PERMIT 0x0010 /* peer will do ECN */
497 #define SCF_SIGNATURE 0x40 /* send MD5 digests */
499 struct mbuf *sc_ipopts; /* IP options */
500 u_int16_t sc_peermaxseg;
501 u_int16_t sc_ourmaxseg;
502 u_int8_t sc_request_r_scale : 4,
503 sc_requested_s_scale : 4;
505 struct tcpcb *sc_tp; /* tcb for listening socket */
506 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */
509 struct syn_cache_head {
510 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */
511 u_short sch_length; /* # entries in bucket */
514 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
515 #ifdef INET6
516 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb)
517 #endif
518 #ifndef INET6
519 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
520 #else
521 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \
522 ? intotcpcb(sotoinpcb(so)) \
523 : in6totcpcb(sotoin6pcb(so)))
524 #endif
527 * The smoothed round-trip time and estimated variance
528 * are stored as fixed point numbers scaled by the values below.
529 * For convenience, these scales are also used in smoothing the average
530 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
531 * With these scales, srtt has 3 bits to the right of the binary point,
532 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
533 * binary point, and is smoothed with an ALPHA of 0.75.
535 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */
536 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */
539 * The initial retransmission should happen at rtt + 4 * rttvar.
540 * Because of the way we do the smoothing, srtt and rttvar
541 * will each average +1/2 tick of bias. When we compute
542 * the retransmit timer, we want 1/2 tick of rounding and
543 * 1 extra tick because of +-1/2 tick uncertainty in the
544 * firing of the timer. The bias will give us exactly the
545 * 1.5 tick we need. But, because the bias is
546 * statistical, we have to test that we don't drop below
547 * the minimum feasible timer (which is 2 ticks).
548 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
549 * is the same as the multiplier for rttvar.
551 #define TCP_REXMTVAL(tp) \
552 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
555 * Compute the initial window for slow start.
557 #define TCP_INITIAL_WINDOW(iw, segsz) \
558 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
559 ((segsz) * (iw)))
562 * TCP statistics.
563 * Each counter is an unsigned 64-bit value.
565 * Many of these should be kept per connection, but that's inconvenient
566 * at the moment.
568 #define TCP_STAT_CONNATTEMPT 0 /* connections initiated */
569 #define TCP_STAT_ACCEPTS 1 /* connections accepted */
570 #define TCP_STAT_CONNECTS 2 /* connections established */
571 #define TCP_STAT_DROPS 3 /* connections dropped */
572 #define TCP_STAT_CONNDROPS 4 /* embryonic connections dropped */
573 #define TCP_STAT_CLOSED 5 /* conn. closed (includes drops) */
574 #define TCP_STAT_SEGSTIMED 6 /* segs where we tried to get rtt */
575 #define TCP_STAT_RTTUPDATED 7 /* times we succeeded */
576 #define TCP_STAT_DELACK 8 /* delayed ACKs sent */
577 #define TCP_STAT_TIMEOUTDROP 9 /* conn. dropped in rxmt timeout */
578 #define TCP_STAT_REXMTTIMEO 10 /* retransmit timeouts */
579 #define TCP_STAT_PERSISTTIMEO 11 /* persist timeouts */
580 #define TCP_STAT_KEEPTIMEO 12 /* keepalive timeouts */
581 #define TCP_STAT_KEEPPROBE 13 /* keepalive probes sent */
582 #define TCP_STAT_KEEPDROPS 14 /* connections dropped in keepalive */
583 #define TCP_STAT_PERSISTDROPS 15 /* connections dropped in persist */
584 #define TCP_STAT_CONNSDRAINED 16 /* connections drained due to memory
585 shortage */
586 #define TCP_STAT_PMTUBLACKHOLE 17 /* PMTUD blackhole detected */
587 #define TCP_STAT_SNDTOTAL 18 /* total packets sent */
588 #define TCP_STAT_SNDPACK 19 /* data packlets sent */
589 #define TCP_STAT_SNDBYTE 20 /* data bytes sent */
590 #define TCP_STAT_SNDREXMITPACK 21 /* data packets retransmitted */
591 #define TCP_STAT_SNDREXMITBYTE 22 /* data bytes retransmitted */
592 #define TCP_STAT_SNDACKS 23 /* ACK-only packets sent */
593 #define TCP_STAT_SNDPROBE 24 /* window probes sent */
594 #define TCP_STAT_SNDURG 25 /* packets sent with URG only */
595 #define TCP_STAT_SNDWINUP 26 /* window update-only packets sent */
596 #define TCP_STAT_SNDCTRL 27 /* control (SYN|FIN|RST) packets sent */
597 #define TCP_STAT_RCVTOTAL 28 /* total packets received */
598 #define TCP_STAT_RCVPACK 29 /* packets received in sequence */
599 #define TCP_STAT_RCVBYTE 30 /* bytes received in sequence */
600 #define TCP_STAT_RCVBADSUM 31 /* packets received with cksum errs */
601 #define TCP_STAT_RCVBADOFF 32 /* packets received with bad offset */
602 #define TCP_STAT_RCVMEMDROP 33 /* packets dropped for lack of memory */
603 #define TCP_STAT_RCVSHORT 34 /* packets received too short */
604 #define TCP_STAT_RCVDUPPACK 35 /* duplicate-only packets received */
605 #define TCP_STAT_RCVDUPBYTE 36 /* duplicate-only bytes received */
606 #define TCP_STAT_RCVPARTDUPPACK 37 /* packets with some duplicate data */
607 #define TCP_STAT_RCVPARTDUPBYTE 38 /* dup. bytes in part-dup. packets */
608 #define TCP_STAT_RCVOOPACK 39 /* out-of-order packets received */
609 #define TCP_STAT_RCVOOBYTE 40 /* out-of-order bytes received */
610 #define TCP_STAT_RCVPACKAFTERWIN 41 /* packets with data after window */
611 #define TCP_STAT_RCVBYTEAFTERWIN 42 /* bytes received after window */
612 #define TCP_STAT_RCVAFTERCLOSE 43 /* packets received after "close" */
613 #define TCP_STAT_RCVWINPROBE 44 /* rcvd window probe packets */
614 #define TCP_STAT_RCVDUPACK 45 /* rcvd duplicate ACKs */
615 #define TCP_STAT_RCVACKTOOMUCH 46 /* rcvd ACKs for unsent data */
616 #define TCP_STAT_RCVACKPACK 47 /* rcvd ACK packets */
617 #define TCP_STAT_RCVACKBYTE 48 /* bytes ACKed by rcvd ACKs */
618 #define TCP_STAT_RCVWINUPD 49 /* rcvd window update packets */
619 #define TCP_STAT_PAWSDROP 50 /* segments dropped due to PAWS */
620 #define TCP_STAT_PREDACK 51 /* times hdr predict OK for ACKs */
621 #define TCP_STAT_PREDDAT 52 /* times hdr predict OK for data pkts */
622 #define TCP_STAT_PCBHASHMISS 53 /* input packets missing PCB hash */
623 #define TCP_STAT_NOPORT 54 /* no socket on port */
624 #define TCP_STAT_BADSYN 55 /* received ACK for which we have
625 no SYN in compressed state */
626 #define TCP_STAT_DELAYED_FREE 56 /* delayed pool_put() of tcpcb */
627 #define TCP_STAT_SC_ADDED 57 /* # of sc entries added */
628 #define TCP_STAT_SC_COMPLETED 58 /* # of sc connections completed */
629 #define TCP_STAT_SC_TIMED_OUT 59 /* # of sc entries timed out */
630 #define TCP_STAT_SC_OVERFLOWED 60 /* # of sc drops due to overflow */
631 #define TCP_STAT_SC_RESET 61 /* # of sc drops due to RST */
632 #define TCP_STAT_SC_UNREACH 62 /* # of sc drops due to ICMP unreach */
633 #define TCP_STAT_SC_BUCKETOVERFLOW 63 /* # of sc drops due to bucket ovflow */
634 #define TCP_STAT_SC_ABORTED 64 /* # of sc entries aborted (no mem) */
635 #define TCP_STAT_SC_DUPESYN 65 /* # of duplicate SYNs received */
636 #define TCP_STAT_SC_DROPPED 66 /* # of SYNs dropped (no route/mem) */
637 #define TCP_STAT_SC_COLLISIONS 67 /* # of sc hash collisions */
638 #define TCP_STAT_SC_RETRANSMITTED 68 /* # of sc retransmissions */
639 #define TCP_STAT_SC_DELAYED_FREE 69 /* # of delayed pool_put()s */
640 #define TCP_STAT_SELFQUENCH 70 /* # of ENOBUFS we get on output */
641 #define TCP_STAT_BADSIG 71 /* # of drops due to bad signature */
642 #define TCP_STAT_GOODSIG 72 /* # of packets with good signature */
643 #define TCP_STAT_ECN_SHS 73 /* # of successful ECN handshakes */
644 #define TCP_STAT_ECN_CE 74 /* # of packets with CE bit */
645 #define TCP_STAT_ECN_ECT 75 /* # of packets with ECT(0) bit */
647 #define TCP_NSTATS 76
650 * Names for TCP sysctl objects.
652 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */
653 #define TCPCTL_SENDSPACE 2 /* default send buffer */
654 #define TCPCTL_RECVSPACE 3 /* default recv buffer */
655 #define TCPCTL_MSSDFLT 4 /* default seg size */
656 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */
657 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */
658 #if 0 /*obsoleted*/
659 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */
660 #endif
661 #define TCPCTL_INIT_WIN 8 /* initial window */
662 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */
663 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */
664 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */
665 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */
666 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */
667 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */
668 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */
669 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */
670 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */
671 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */
672 #define TCPCTL_KEEPCNT 19 /* keepalive count */
673 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */
674 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */
675 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */
676 #if 0 /*obsoleted*/
677 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */
678 #endif
679 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */
680 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */
681 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */
682 #define TCPCTL_IDENT 27 /* rfc 931 identd */
683 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */
684 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */
685 #define TCPCTL_STATS 30 /* TCP statistics */
686 #define TCPCTL_DEBUG 31 /* TCP debug sockets */
687 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */
688 #define TCPCTL_DROP 33 /* drop tcp connection */
689 #define TCPCTL_MSL 34 /* Max Segment Life */
690 #define TCPCTL_MAXID 35
692 #define TCPCTL_NAMES { \
693 { 0, 0 }, \
694 { "rfc1323", CTLTYPE_INT }, \
695 { "sendspace", CTLTYPE_INT }, \
696 { "recvspace", CTLTYPE_INT }, \
697 { "mssdflt", CTLTYPE_INT }, \
698 { "syn_cache_limit", CTLTYPE_INT }, \
699 { "syn_bucket_limit", CTLTYPE_INT }, \
700 { 0, 0 },\
701 { "init_win", CTLTYPE_INT }, \
702 { "mss_ifmtu", CTLTYPE_INT }, \
703 { "sack", CTLTYPE_INT }, \
704 { "win_scale", CTLTYPE_INT }, \
705 { "timestamps", CTLTYPE_INT }, \
706 { "compat_42", CTLTYPE_INT }, \
707 { "cwm", CTLTYPE_INT }, \
708 { "cwm_burstsize", CTLTYPE_INT }, \
709 { "ack_on_push", CTLTYPE_INT }, \
710 { "keepidle", CTLTYPE_INT }, \
711 { "keepintvl", CTLTYPE_INT }, \
712 { "keepcnt", CTLTYPE_INT }, \
713 { "slowhz", CTLTYPE_INT }, \
714 { 0, 0 }, \
715 { "log_refused",CTLTYPE_INT }, \
716 { 0, 0 }, \
717 { "rstppslimit", CTLTYPE_INT }, \
718 { "delack_ticks", CTLTYPE_INT }, \
719 { "init_win_local", CTLTYPE_INT }, \
720 { "ident", CTLTYPE_STRUCT }, \
721 { "ackdropppslimit", CTLTYPE_INT }, \
722 { "do_loopback_cksum", CTLTYPE_INT }, \
723 { "stats", CTLTYPE_STRUCT }, \
724 { "debug", CTLTYPE_STRUCT }, \
725 { "debx", CTLTYPE_INT }, \
726 { "drop", CTLTYPE_STRUCT }, \
727 { "msl", CTLTYPE_INT }, \
730 #ifdef _KERNEL
731 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */
732 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */
733 extern int tcp_do_rfc1323; /* enabled/disabled? */
734 extern int tcp_do_sack; /* SACK enabled/disabled? */
735 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */
736 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */
737 extern int tcp_mssdflt; /* default seg size */
738 extern int tcp_minmss; /* minimal seg size */
739 extern int tcp_msl; /* max segment life */
740 extern int tcp_init_win; /* initial window */
741 extern int tcp_init_win_local; /* initial window for local nets */
742 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */
743 extern int tcp_compat_42; /* work around ancient broken TCP peers */
744 extern int tcp_cwm; /* enable Congestion Window Monitoring */
745 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */
746 extern int tcp_ack_on_push; /* ACK immediately on PUSH */
747 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */
748 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */
749 extern int tcp_log_refused; /* log refused connections */
750 extern int tcp_do_ecn; /* TCP ECN enabled/disabled? */
751 extern int tcp_ecn_maxretries; /* Max ECN setup retries */
752 #if NRND > 0
753 extern int tcp_do_rfc1948; /* ISS by cryptographic hash */
754 #endif
755 extern int tcp_sack_tp_maxholes; /* Max holes per connection. */
756 extern int tcp_sack_globalmaxholes; /* Max holes per system. */
757 extern int tcp_sack_globalholes; /* Number of holes present. */
758 extern int tcp_do_abc; /* RFC3465 ABC enabled/disabled? */
759 extern int tcp_abc_aggressive; /* 1: L=2*SMSS 0: L=1*SMSS */
761 extern int tcp_rst_ppslim;
762 extern int tcp_ackdrop_ppslim;
764 extern int tcp_syn_cache_size;
765 extern struct syn_cache_head tcp_syn_cache[];
766 extern u_long syn_cache_count;
768 #ifdef MBUFTRACE
769 extern struct mowner tcp_rx_mowner;
770 extern struct mowner tcp_tx_mowner;
771 extern struct mowner tcp_reass_mowner;
772 extern struct mowner tcp_sock_mowner;
773 extern struct mowner tcp_sock_rx_mowner;
774 extern struct mowner tcp_sock_tx_mowner;
775 extern struct mowner tcp_mowner;
776 #endif
778 extern int tcp_do_autorcvbuf;
779 extern int tcp_autorcvbuf_inc;
780 extern int tcp_autorcvbuf_max;
781 extern int tcp_do_autosndbuf;
782 extern int tcp_autosndbuf_inc;
783 extern int tcp_autosndbuf_max;
786 #define TCPCTL_VARIABLES { \
787 { 0 }, \
788 { 1, 0, &tcp_do_rfc1323 }, \
789 { 1, 0, &tcp_sendspace }, \
790 { 1, 0, &tcp_recvspace }, \
791 { 1, 0, &tcp_mssdflt }, \
792 { 1, 0, &tcp_syn_cache_limit }, \
793 { 1, 0, &tcp_syn_bucket_limit }, \
794 { 0 }, \
795 { 1, 0, &tcp_init_win }, \
796 { 1, 0, &tcp_mss_ifmtu }, \
797 { 1, 0, &tcp_do_sack }, \
798 { 1, 0, &tcp_do_win_scale }, \
799 { 1, 0, &tcp_do_timestamps }, \
800 { 1, 0, &tcp_compat_42 }, \
801 { 1, 0, &tcp_cwm }, \
802 { 1, 0, &tcp_cwm_burstsize }, \
803 { 1, 0, &tcp_ack_on_push }, \
804 { 1, 0, &tcp_keepidle }, \
805 { 1, 0, &tcp_keepintvl }, \
806 { 1, 0, &tcp_keepcnt }, \
807 { 1, 1, 0, PR_SLOWHZ }, \
808 { 0 }, \
809 { 1, 0, &tcp_log_refused }, \
810 { 0 }, \
811 { 1, 0, &tcp_rst_ppslim }, \
812 { 1, 0, &tcp_delack_ticks }, \
813 { 1, 0, &tcp_init_win_local }, \
814 { 1, 0, &tcp_ackdrop_ppslim }, \
817 struct secasvar;
819 int tcp_attach(struct socket *);
820 void tcp_canceltimers(struct tcpcb *);
821 struct tcpcb *
822 tcp_close(struct tcpcb *);
823 int tcp_isdead(struct tcpcb *);
824 #ifdef INET6
825 void *tcp6_ctlinput(int, const struct sockaddr *, void *);
826 #endif
827 void *tcp_ctlinput(int, const struct sockaddr *, void *);
828 int tcp_ctloutput(int, struct socket *, struct sockopt *);
829 struct tcpcb *
830 tcp_disconnect(struct tcpcb *);
831 struct tcpcb *
832 tcp_drop(struct tcpcb *, int);
833 #ifdef TCP_SIGNATURE
834 int tcp_signature_apply(void *, void *, u_int);
835 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *);
836 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *,
837 char *);
838 #endif
839 void tcp_drain(void);
840 void tcp_established(struct tcpcb *);
841 void tcp_init(void);
842 #ifdef INET6
843 int tcp6_input(struct mbuf **, int *, int);
844 #endif
845 void tcp_input(struct mbuf *, ...);
846 u_int tcp_hdrsz(struct tcpcb *);
847 u_long tcp_mss_to_advertise(const struct ifnet *, int);
848 void tcp_mss_from_peer(struct tcpcb *, int);
849 void tcp_tcpcb_template(void);
850 struct tcpcb *
851 tcp_newtcpcb(int, void *);
852 void tcp_notify(struct inpcb *, int);
853 #ifdef INET6
854 void tcp6_notify(struct in6pcb *, int);
855 #endif
856 u_int tcp_optlen(struct tcpcb *);
857 int tcp_output(struct tcpcb *);
858 void tcp_pulloutofband(struct socket *,
859 struct tcphdr *, struct mbuf *, int);
860 void tcp_quench(struct inpcb *, int);
861 #ifdef INET6
862 void tcp6_quench(struct in6pcb *, int);
863 #endif
864 void tcp_mtudisc(struct inpcb *, int);
866 void tcpipqent_init(void);
867 struct ipqent *tcpipqent_alloc(void);
868 void tcpipqent_free(struct ipqent *);
870 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *,
871 struct tcphdr *, tcp_seq, tcp_seq, int);
872 void tcp_rmx_rtt(struct tcpcb *);
873 void tcp_setpersist(struct tcpcb *);
874 #ifdef TCP_SIGNATURE
875 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int,
876 int, u_char *, u_int);
877 #endif
878 void tcp_slowtimo(void);
879 struct mbuf *
880 tcp_template(struct tcpcb *);
881 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int);
882 struct tcpcb *
883 tcp_usrclosed(struct tcpcb *);
884 void tcp_usrreq_init(void);
885 int tcp_usrreq(struct socket *,
886 int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *);
887 void tcp_xmit_timer(struct tcpcb *, uint32_t);
888 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq);
889 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t,
890 tcp_seq);
892 void tcp_sack_init(void);
893 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t);
894 void tcp_sack_option(struct tcpcb *, const struct tcphdr *,
895 const u_char *, int);
896 void tcp_del_sackholes(struct tcpcb *, const struct tcphdr *);
897 void tcp_free_sackholes(struct tcpcb *);
898 void tcp_sack_adjust(struct tcpcb *tp);
899 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
900 void tcp_sack_newack(struct tcpcb *, const struct tcphdr *);
901 int tcp_sack_numblks(const struct tcpcb *);
902 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2)
904 void tcp_statinc(u_int);
905 void tcp_statadd(u_int, uint64_t);
907 int syn_cache_add(struct sockaddr *, struct sockaddr *,
908 struct tcphdr *, unsigned int, struct socket *,
909 struct mbuf *, u_char *, int, struct tcp_opt_info *);
910 void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *,
911 struct tcphdr *);
912 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
913 struct tcphdr *, unsigned int, unsigned int,
914 struct socket *so, struct mbuf *);
915 void syn_cache_init(void);
916 void syn_cache_insert(struct syn_cache *, struct tcpcb *);
917 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *,
918 struct syn_cache_head **);
919 void syn_cache_reset(struct sockaddr *, struct sockaddr *,
920 struct tcphdr *);
921 int syn_cache_respond(struct syn_cache *, struct mbuf *);
922 void syn_cache_timer(void *);
923 void syn_cache_cleanup(struct tcpcb *);
925 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int,
926 int);
927 #endif
929 #endif /* !_NETINET_TCP_VAR_H_ */