1 /* $NetBSD: tcp_var.h,v 1.177 2015/02/14 22:09:53 he Exp $ */
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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
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).
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
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
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
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"
145 * Kernel variables for tcp.
148 #include <sys/callout.h>
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 */
169 tcp_seq left
; /* Left edge of sack block. */
170 tcp_seq right
; /* Right edge of sack block. */
173 TAILQ_HEAD(sackhead
, sackhole
);
179 TAILQ_ENTRY(sackhole
) sackhole_q
;
183 * Tcp control block, one per tcp; fields:
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 */
196 * <0 not in fast recovery.
197 * ==0 in fast recovery. has not received partial acks
198 * >0 in fast recovery. has received partial acks
200 short t_partialacks
; /* partials acks during fast rexmit */
201 u_short t_peermss
; /* peer's maximum segment size */
202 u_short t_ourmss
; /* our's maximum segment size */
203 u_short t_segsz
; /* current segment size in use */
204 char t_force
; /* 1 if forcing out a byte */
206 #define TF_ACKNOW 0x0001 /* ack peer immediately */
207 #define TF_DELACK 0x0002 /* ack, but try to delay it */
208 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */
209 #define TF_NOOPT 0x0008 /* don't use tcp options */
210 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */
211 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */
212 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */
213 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */
214 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */
215 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */
216 #define TF_WILL_SACK 0x0800 /* try to use SACK */
217 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */
218 #define TF_DEAD 0x2000 /* dead and to-be-released */
219 #define TF_PMTUD_PEND 0x4000 /* Path MTU Discovery pending */
220 #define TF_ECN_PERMIT 0x10000 /* other side said is ECN-ready */
221 #define TF_ECN_SND_CWR 0x20000 /* ECN CWR in queue */
222 #define TF_ECN_SND_ECE 0x40000 /* ECN ECE in queue */
223 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
226 struct mbuf
*t_template
; /* skeletal packet for transmit */
227 struct inpcb
*t_inpcb
; /* back pointer to internet pcb */
228 struct in6pcb
*t_in6pcb
; /* back pointer to internet pcb */
229 callout_t t_delack_ch
; /* delayed ACK callout */
231 * The following fields are used as in the protocol specification.
232 * See RFC793, Dec. 1981, page 21.
234 /* send sequence variables */
235 tcp_seq snd_una
; /* send unacknowledged */
236 tcp_seq snd_nxt
; /* send next */
237 tcp_seq snd_up
; /* send urgent pointer */
238 tcp_seq snd_wl1
; /* window update seg seq number */
239 tcp_seq snd_wl2
; /* window update seg ack number */
240 tcp_seq iss
; /* initial send sequence number */
241 u_long snd_wnd
; /* send window */
244 * it's basically same as the "recover" variable in RFC 2852 (NewReno).
245 * when entering fast retransmit, it's set to snd_max.
246 * newreno uses this to detect partial ack.
248 * it's basically same as the "send_high" variable in RFC 2852 (NewReno).
249 * on each RTO, it's set to snd_max.
250 * newreno uses this to avoid false fast retransmits.
254 /* receive sequence variables */
255 u_long rcv_wnd
; /* receive window */
256 tcp_seq rcv_nxt
; /* receive next */
257 tcp_seq rcv_up
; /* receive urgent pointer */
258 tcp_seq irs
; /* initial receive sequence number */
260 * Additional variables for this implementation.
262 /* receive variables */
263 tcp_seq rcv_adv
; /* advertised window */
266 * retransmit variables
269 * the highest sequence number we've ever sent.
270 * used to recognize retransmits.
274 /* congestion control (for slow start, source quench, retransmit after loss) */
275 u_long snd_cwnd
; /* congestion-controlled window */
276 u_long snd_ssthresh
; /* snd_cwnd size threshhold for
277 * for slow start exponential to
280 /* auto-sizing variables */
281 u_int rfbuf_cnt
; /* recv buffer autoscaling byte count */
282 uint32_t rfbuf_ts
; /* recv buffer autoscaling timestamp */
285 * transmit timing stuff. See below for scale of srtt and rttvar.
286 * "Variance" is actually smoothed difference.
288 uint32_t t_rcvtime
; /* time last segment received */
289 uint32_t t_rtttime
; /* time we started measuring rtt */
290 tcp_seq t_rtseq
; /* sequence number being timed */
291 int32_t t_srtt
; /* smoothed round-trip time */
292 int32_t t_rttvar
; /* variance in round-trip time */
293 uint32_t t_rttmin
; /* minimum rtt allowed */
294 u_long max_sndwnd
; /* largest window peer has offered */
296 /* out-of-band data */
297 char t_oobflags
; /* have some */
298 char t_iobc
; /* input character */
299 #define TCPOOB_HAVEDATA 0x01
300 #define TCPOOB_HADDATA 0x02
301 short t_softerror
; /* possible error not yet reported */
303 /* RFC 1323 variables */
304 u_char snd_scale
; /* window scaling for send window */
305 u_char rcv_scale
; /* window scaling for recv window */
306 u_char request_r_scale
; /* pending window scaling */
307 u_char requested_s_scale
;
308 u_int32_t ts_recent
; /* timestamp echo data */
309 u_int32_t ts_recent_age
; /* when last updated */
310 u_int32_t ts_timebase
; /* our timebase */
311 tcp_seq last_ack_sent
;
313 /* RFC 3465 variables */
314 u_long t_bytes_acked
; /* ABC "bytes_acked" parameter */
317 #define TCP_SACK_MAX 3
318 #define TCPSACK_NONE 0
319 #define TCPSACK_HAVED 1
320 u_char rcv_sack_flags
; /* SACK flags. */
321 struct sackblk rcv_dsack_block
; /* RX D-SACK block. */
322 struct ipqehead timeq
; /* time sequenced queue. */
323 struct sackhead snd_holes
; /* TX SACK holes. */
324 int snd_numholes
; /* Number of TX SACK holes. */
325 tcp_seq rcv_lastsack
; /* last seq number(+1) sack'd by rcv'r*/
326 tcp_seq sack_newdata
; /* New data xmitted in this recovery
327 episode starts at this seq number*/
328 tcp_seq snd_fack
; /* FACK TCP. Forward-most data held by
331 /* CUBIC variables */
332 ulong snd_cubic_wmax
; /* W_max */
333 ulong snd_cubic_wmax_last
; /* Used for fast convergence */
334 ulong snd_cubic_ctime
; /* Last congestion time */
336 /* pointer for syn cache entries*/
337 LIST_HEAD(, syn_cache
) t_sc
; /* list of entries by this tcb */
339 /* prediction of next mbuf when using large window sizes */
340 struct mbuf
*t_lastm
; /* last mbuf that data was sent from */
341 int t_inoff
; /* data offset in previous mbuf */
342 int t_lastoff
; /* last data address in mbuf chain */
343 int t_lastlen
; /* last length read from mbuf chain */
345 /* Path-MTU discovery blackhole detection */
346 int t_mtudisc
; /* perform mtudisc for this tcb */
347 /* Path-MTU Discovery Information */
348 u_int t_pmtud_mss_acked
; /* MSS acked, lower bound for MTU */
349 u_int t_pmtud_mtu_sent
; /* MTU used, upper bound for MTU */
350 tcp_seq t_pmtud_th_seq
; /* TCP SEQ from ICMP payload */
351 u_int t_pmtud_nextmtu
; /* Advertised Next-Hop MTU from ICMP */
352 u_short t_pmtud_ip_len
; /* IP length from ICMP payload */
353 u_short t_pmtud_ip_hl
; /* IP header length from ICMP payload */
355 uint8_t t_ecn_retries
; /* # of ECN setup retries */
357 const struct tcp_congctl
*t_congctl
; /* per TCB congctl algorithm */
359 /* Keepalive per socket */
364 u_int t_maxidle
; /* t_keepcnt * t_keepintvl */
366 u_int t_msl
; /* MSL to use for this connexion */
368 /* maintain a few stats per connection: */
369 uint32_t t_rcvoopack
; /* out-of-order packets received */
370 uint32_t t_sndrexmitpack
; /* retransmit packets sent */
371 uint32_t t_sndzerowin
; /* zero-window updates sent */
375 * Macros to aid ECN TCP.
377 #define TCP_ECN_ALLOWED(tp) (tp->t_flags & TF_ECN_PERMIT)
380 * Macros to aid SACK/FACK TCP.
382 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK)
383 #define TCP_FACK_FASTRECOV(tp) \
384 (TCP_SACK_ENABLED(tp) && \
385 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz)))
389 * TCP reassembly queue locks.
391 static __inline
int tcp_reass_lock_try (struct tcpcb
*)
393 static __inline
void tcp_reass_unlock (struct tcpcb
*)
397 tcp_reass_lock_try(struct tcpcb
*tp
)
402 * Use splvm() -- we're blocking things that would cause
406 if (tp
->t_flags
& TF_REASSEMBLING
) {
410 tp
->t_flags
|= TF_REASSEMBLING
;
416 tcp_reass_unlock(struct tcpcb
*tp
)
421 KASSERT((tp
->t_flags
& TF_REASSEMBLING
) != 0);
422 tp
->t_flags
&= ~TF_REASSEMBLING
;
427 #define TCP_REASS_LOCK(tp) \
429 if (tcp_reass_lock_try(tp) == 0) { \
430 printf("%s:%d: tcpcb %p reass already locked\n", \
431 __FILE__, __LINE__, tp); \
432 panic("tcp_reass_lock"); \
434 } while (/*CONSTCOND*/ 0)
435 #define TCP_REASS_LOCK_CHECK(tp) \
437 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \
438 printf("%s:%d: tcpcb %p reass lock not held\n", \
439 __FILE__, __LINE__, tp); \
440 panic("tcp reass lock check"); \
442 } while (/*CONSTCOND*/ 0)
444 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp))
445 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */
448 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp))
452 * Queue for delayed ACK processing.
455 extern int tcp_delack_ticks
;
456 void tcp_delack(void *);
458 #define TCP_RESTART_DELACK(tp) \
459 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \
462 #define TCP_SET_DELACK(tp) \
464 if (((tp)->t_flags & TF_DELACK) == 0) { \
465 (tp)->t_flags |= TF_DELACK; \
466 TCP_RESTART_DELACK(tp); \
468 } while (/*CONSTCOND*/0)
470 #define TCP_CLEAR_DELACK(tp) \
472 if ((tp)->t_flags & TF_DELACK) { \
473 (tp)->t_flags &= ~TF_DELACK; \
474 callout_stop(&(tp)->t_delack_ch); \
476 } while (/*CONSTCOND*/0)
480 * Compute the current timestamp for a connection.
482 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase)
485 * Handy way of passing around TCP option info.
487 struct tcp_opt_info
{
494 #define TOF_SIGNATURE 0x0040 /* signature option present */
495 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */
498 * Data for the TCP compressed state engine.
502 struct sockaddr_in sin
;
504 struct sockaddr_in6 sin6
;
509 TAILQ_ENTRY(syn_cache
) sc_bucketq
; /* link on bucket list */
510 callout_t sc_timer
; /* rexmt timer */
511 struct route sc_route
;
512 long sc_win
; /* advertised window */
513 int sc_bucketidx
; /* our bucket index */
515 u_int32_t sc_timestamp
; /* timestamp from SYN */
516 u_int32_t sc_timebase
; /* our local timebase */
517 union syn_cache_sa sc_src
;
518 union syn_cache_sa sc_dst
;
521 u_int sc_rxtcur
; /* current rxt timeout */
522 u_int sc_rxttot
; /* total time spend on queues */
523 u_short sc_rxtshift
; /* for computing backoff */
526 #define SCF_UNREACH 0x0001 /* we've had an unreach error */
527 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */
528 #define SCF_DEAD 0x0004 /* this entry to be released */
529 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */
530 #define SCF_ECN_PERMIT 0x0010 /* peer will do ECN */
531 #define SCF_SIGNATURE 0x40 /* send MD5 digests */
533 struct mbuf
*sc_ipopts
; /* IP options */
534 u_int16_t sc_peermaxseg
;
535 u_int16_t sc_ourmaxseg
;
536 u_int8_t sc_request_r_scale
: 4,
537 sc_requested_s_scale
: 4;
539 struct tcpcb
*sc_tp
; /* tcb for listening socket */
540 LIST_ENTRY(syn_cache
) sc_tpq
; /* list of entries by same tp */
543 struct syn_cache_head
{
544 TAILQ_HEAD(, syn_cache
) sch_bucket
; /* bucket entries */
545 u_short sch_length
; /* # entries in bucket */
548 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
550 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb)
553 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
555 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \
556 ? intotcpcb(sotoinpcb(so)) \
557 : in6totcpcb(sotoin6pcb(so)))
561 * See RFC2988 for a discussion of RTO calculation; comments assume
562 * familiarity with that document.
564 * The smoothed round-trip time and estimated variance are stored as
565 * fixed point numbers. Historically, srtt was scaled by
566 * TCP_RTT_SHIFT bits, and rttvar by TCP_RTTVAR_SHIFT bits. Because
567 * the values coincide with the alpha and beta parameters suggested
568 * for RTO calculation (1/8 for srtt, 1/4 for rttvar), the combination
569 * of computing 1/8 of the new value and transforming it to the
570 * fixed-point representation required zero instructions. However,
571 * the storage representations no longer coincide with the alpha/beta
572 * shifts; instead, more fractional bits are present.
574 * The storage representation of srtt is 1/32 slow ticks, or 1/64 s.
575 * (The assumption that a slow tick is 500 ms should not be present in
578 * The storage representation of rttvar is 1/16 slow ticks, or 1/32 s.
579 * There may be some confusion about this in the code.
581 * For historical reasons, these scales are also used in smoothing the
582 * average (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
583 * This results in alpha of 0.125 and beta of 0.25, following RFC2988
586 * XXX Change SHIFT values to LGWEIGHT and REP_SHIFT, and adjust
587 * the code to use the correct ones.
589 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */
590 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */
593 * Compute TCP retransmission timer, following RFC2988.
594 * This macro returns a value in slow timeout ticks.
596 * Section 2.2 requires that the RTO value be
597 * srtt + max(G, 4*RTTVAR)
598 * where G is the clock granularity.
600 * This comment has not necessarily been updated for the new storage
603 * Because of the way we do the smoothing, srtt and rttvar
604 * will each average +1/2 tick of bias. When we compute
605 * the retransmit timer, we want 1/2 tick of rounding and
606 * 1 extra tick because of +-1/2 tick uncertainty in the
607 * firing of the timer. The bias will give us exactly the
608 * 1.5 tick we need. But, because the bias is
609 * statistical, we have to test that we don't drop below
610 * the minimum feasible timer (which is 2 ticks).
611 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
612 * is the same as the multiplier for rttvar.
614 * This macro appears to be wrong; it should be checking rttvar*4 in
615 * ticks and making sure we use 1 instead if rttvar*4 rounds to 0. It
616 * appears to be treating srtt as being in the old storage
617 * representation, resulting in a factor of 4 extra.
619 #define TCP_REXMTVAL(tp) \
620 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
623 * Compute the initial window for slow start.
625 #define TCP_INITIAL_WINDOW(iw, segsz) \
626 min((iw) * (segsz), max(2 * (segsz), tcp_init_win_max[(iw)]))
630 * Each counter is an unsigned 64-bit value.
632 * Many of these should be kept per connection, but that's inconvenient
635 #define TCP_STAT_CONNATTEMPT 0 /* connections initiated */
636 #define TCP_STAT_ACCEPTS 1 /* connections accepted */
637 #define TCP_STAT_CONNECTS 2 /* connections established */
638 #define TCP_STAT_DROPS 3 /* connections dropped */
639 #define TCP_STAT_CONNDROPS 4 /* embryonic connections dropped */
640 #define TCP_STAT_CLOSED 5 /* conn. closed (includes drops) */
641 #define TCP_STAT_SEGSTIMED 6 /* segs where we tried to get rtt */
642 #define TCP_STAT_RTTUPDATED 7 /* times we succeeded */
643 #define TCP_STAT_DELACK 8 /* delayed ACKs sent */
644 #define TCP_STAT_TIMEOUTDROP 9 /* conn. dropped in rxmt timeout */
645 #define TCP_STAT_REXMTTIMEO 10 /* retransmit timeouts */
646 #define TCP_STAT_PERSISTTIMEO 11 /* persist timeouts */
647 #define TCP_STAT_KEEPTIMEO 12 /* keepalive timeouts */
648 #define TCP_STAT_KEEPPROBE 13 /* keepalive probes sent */
649 #define TCP_STAT_KEEPDROPS 14 /* connections dropped in keepalive */
650 #define TCP_STAT_PERSISTDROPS 15 /* connections dropped in persist */
651 #define TCP_STAT_CONNSDRAINED 16 /* connections drained due to memory
653 #define TCP_STAT_PMTUBLACKHOLE 17 /* PMTUD blackhole detected */
654 #define TCP_STAT_SNDTOTAL 18 /* total packets sent */
655 #define TCP_STAT_SNDPACK 19 /* data packlets sent */
656 #define TCP_STAT_SNDBYTE 20 /* data bytes sent */
657 #define TCP_STAT_SNDREXMITPACK 21 /* data packets retransmitted */
658 #define TCP_STAT_SNDREXMITBYTE 22 /* data bytes retransmitted */
659 #define TCP_STAT_SNDACKS 23 /* ACK-only packets sent */
660 #define TCP_STAT_SNDPROBE 24 /* window probes sent */
661 #define TCP_STAT_SNDURG 25 /* packets sent with URG only */
662 #define TCP_STAT_SNDWINUP 26 /* window update-only packets sent */
663 #define TCP_STAT_SNDCTRL 27 /* control (SYN|FIN|RST) packets sent */
664 #define TCP_STAT_RCVTOTAL 28 /* total packets received */
665 #define TCP_STAT_RCVPACK 29 /* packets received in sequence */
666 #define TCP_STAT_RCVBYTE 30 /* bytes received in sequence */
667 #define TCP_STAT_RCVBADSUM 31 /* packets received with cksum errs */
668 #define TCP_STAT_RCVBADOFF 32 /* packets received with bad offset */
669 #define TCP_STAT_RCVMEMDROP 33 /* packets dropped for lack of memory */
670 #define TCP_STAT_RCVSHORT 34 /* packets received too short */
671 #define TCP_STAT_RCVDUPPACK 35 /* duplicate-only packets received */
672 #define TCP_STAT_RCVDUPBYTE 36 /* duplicate-only bytes received */
673 #define TCP_STAT_RCVPARTDUPPACK 37 /* packets with some duplicate data */
674 #define TCP_STAT_RCVPARTDUPBYTE 38 /* dup. bytes in part-dup. packets */
675 #define TCP_STAT_RCVOOPACK 39 /* out-of-order packets received */
676 #define TCP_STAT_RCVOOBYTE 40 /* out-of-order bytes received */
677 #define TCP_STAT_RCVPACKAFTERWIN 41 /* packets with data after window */
678 #define TCP_STAT_RCVBYTEAFTERWIN 42 /* bytes received after window */
679 #define TCP_STAT_RCVAFTERCLOSE 43 /* packets received after "close" */
680 #define TCP_STAT_RCVWINPROBE 44 /* rcvd window probe packets */
681 #define TCP_STAT_RCVDUPACK 45 /* rcvd duplicate ACKs */
682 #define TCP_STAT_RCVACKTOOMUCH 46 /* rcvd ACKs for unsent data */
683 #define TCP_STAT_RCVACKPACK 47 /* rcvd ACK packets */
684 #define TCP_STAT_RCVACKBYTE 48 /* bytes ACKed by rcvd ACKs */
685 #define TCP_STAT_RCVWINUPD 49 /* rcvd window update packets */
686 #define TCP_STAT_PAWSDROP 50 /* segments dropped due to PAWS */
687 #define TCP_STAT_PREDACK 51 /* times hdr predict OK for ACKs */
688 #define TCP_STAT_PREDDAT 52 /* times hdr predict OK for data pkts */
689 #define TCP_STAT_PCBHASHMISS 53 /* input packets missing PCB hash */
690 #define TCP_STAT_NOPORT 54 /* no socket on port */
691 #define TCP_STAT_BADSYN 55 /* received ACK for which we have
692 no SYN in compressed state */
693 #define TCP_STAT_DELAYED_FREE 56 /* delayed pool_put() of tcpcb */
694 #define TCP_STAT_SC_ADDED 57 /* # of sc entries added */
695 #define TCP_STAT_SC_COMPLETED 58 /* # of sc connections completed */
696 #define TCP_STAT_SC_TIMED_OUT 59 /* # of sc entries timed out */
697 #define TCP_STAT_SC_OVERFLOWED 60 /* # of sc drops due to overflow */
698 #define TCP_STAT_SC_RESET 61 /* # of sc drops due to RST */
699 #define TCP_STAT_SC_UNREACH 62 /* # of sc drops due to ICMP unreach */
700 #define TCP_STAT_SC_BUCKETOVERFLOW 63 /* # of sc drops due to bucket ovflow */
701 #define TCP_STAT_SC_ABORTED 64 /* # of sc entries aborted (no mem) */
702 #define TCP_STAT_SC_DUPESYN 65 /* # of duplicate SYNs received */
703 #define TCP_STAT_SC_DROPPED 66 /* # of SYNs dropped (no route/mem) */
704 #define TCP_STAT_SC_COLLISIONS 67 /* # of sc hash collisions */
705 #define TCP_STAT_SC_RETRANSMITTED 68 /* # of sc retransmissions */
706 #define TCP_STAT_SC_DELAYED_FREE 69 /* # of delayed pool_put()s */
707 #define TCP_STAT_SELFQUENCH 70 /* # of ENOBUFS we get on output */
708 #define TCP_STAT_BADSIG 71 /* # of drops due to bad signature */
709 #define TCP_STAT_GOODSIG 72 /* # of packets with good signature */
710 #define TCP_STAT_ECN_SHS 73 /* # of successful ECN handshakes */
711 #define TCP_STAT_ECN_CE 74 /* # of packets with CE bit */
712 #define TCP_STAT_ECN_ECT 75 /* # of packets with ECT(0) bit */
714 #define TCP_NSTATS 76
717 * Names for TCP sysctl objects.
719 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */
720 #define TCPCTL_SENDSPACE 2 /* default send buffer */
721 #define TCPCTL_RECVSPACE 3 /* default recv buffer */
722 #define TCPCTL_MSSDFLT 4 /* default seg size */
723 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */
724 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */
726 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */
728 #define TCPCTL_INIT_WIN 8 /* initial window */
729 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */
730 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */
731 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */
732 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */
733 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */
734 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */
735 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */
736 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */
737 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */
738 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */
739 #define TCPCTL_KEEPCNT 19 /* keepalive count */
740 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */
741 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */
742 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */
744 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */
746 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */
747 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */
748 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */
749 #define TCPCTL_IDENT 27 /* rfc 931 identd */
750 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */
751 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */
752 #define TCPCTL_STATS 30 /* TCP statistics */
753 #define TCPCTL_DEBUG 31 /* TCP debug sockets */
754 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */
755 #define TCPCTL_DROP 33 /* drop tcp connection */
756 #define TCPCTL_MSL 34 /* Max Segment Life */
757 #define TCPCTL_MAXID 35
759 #define TCPCTL_NAMES { \
761 { "rfc1323", CTLTYPE_INT }, \
762 { "sendspace", CTLTYPE_INT }, \
763 { "recvspace", CTLTYPE_INT }, \
764 { "mssdflt", CTLTYPE_INT }, \
765 { "syn_cache_limit", CTLTYPE_INT }, \
766 { "syn_bucket_limit", CTLTYPE_INT }, \
768 { "init_win", CTLTYPE_INT }, \
769 { "mss_ifmtu", CTLTYPE_INT }, \
770 { "sack", CTLTYPE_INT }, \
771 { "win_scale", CTLTYPE_INT }, \
772 { "timestamps", CTLTYPE_INT }, \
773 { "compat_42", CTLTYPE_INT }, \
774 { "cwm", CTLTYPE_INT }, \
775 { "cwm_burstsize", CTLTYPE_INT }, \
776 { "ack_on_push", CTLTYPE_INT }, \
777 { "keepidle", CTLTYPE_INT }, \
778 { "keepintvl", CTLTYPE_INT }, \
779 { "keepcnt", CTLTYPE_INT }, \
780 { "slowhz", CTLTYPE_INT }, \
782 { "log_refused",CTLTYPE_INT }, \
784 { "rstppslimit", CTLTYPE_INT }, \
785 { "delack_ticks", CTLTYPE_INT }, \
786 { "init_win_local", CTLTYPE_INT }, \
787 { "ident", CTLTYPE_STRUCT }, \
788 { "ackdropppslimit", CTLTYPE_INT }, \
789 { "do_loopback_cksum", CTLTYPE_INT }, \
790 { "stats", CTLTYPE_STRUCT }, \
791 { "debug", CTLTYPE_STRUCT }, \
792 { "debx", CTLTYPE_INT }, \
793 { "drop", CTLTYPE_STRUCT }, \
794 { "msl", CTLTYPE_INT }, \
799 extern struct inpcbtable tcbtable
; /* head of queue of active tcpcb's */
800 extern const struct pr_usrreqs tcp_usrreqs
;
802 extern u_int32_t tcp_now
; /* for RFC 1323 timestamps */
803 extern int tcp_do_rfc1323
; /* enabled/disabled? */
804 extern int tcp_do_sack
; /* SACK enabled/disabled? */
805 extern int tcp_do_win_scale
; /* RFC1323 window scaling enabled/disabled? */
806 extern int tcp_do_timestamps
; /* RFC1323 timestamps enabled/disabled? */
807 extern int tcp_mssdflt
; /* default seg size */
808 extern int tcp_minmss
; /* minimal seg size */
809 extern int tcp_msl
; /* max segment life */
810 extern int tcp_init_win
; /* initial window */
811 extern int tcp_init_win_local
; /* initial window for local nets */
812 extern int tcp_init_win_max
[11];/* max sizes for values of tcp_init_win_* */
813 extern int tcp_mss_ifmtu
; /* take MSS from interface, not in_maxmtu */
814 extern int tcp_compat_42
; /* work around ancient broken TCP peers */
815 extern int tcp_cwm
; /* enable Congestion Window Monitoring */
816 extern int tcp_cwm_burstsize
; /* burst size allowed by CWM */
817 extern int tcp_ack_on_push
; /* ACK immediately on PUSH */
818 extern int tcp_syn_cache_limit
; /* max entries for compressed state engine */
819 extern int tcp_syn_bucket_limit
;/* max entries per hash bucket */
820 extern int tcp_log_refused
; /* log refused connections */
821 extern int tcp_do_ecn
; /* TCP ECN enabled/disabled? */
822 extern int tcp_ecn_maxretries
; /* Max ECN setup retries */
823 extern int tcp_do_rfc1948
; /* ISS by cryptographic hash */
824 extern int tcp_sack_tp_maxholes
; /* Max holes per connection. */
825 extern int tcp_sack_globalmaxholes
; /* Max holes per system. */
826 extern int tcp_sack_globalholes
; /* Number of holes present. */
827 extern int tcp_do_abc
; /* RFC3465 ABC enabled/disabled? */
828 extern int tcp_abc_aggressive
; /* 1: L=2*SMSS 0: L=1*SMSS */
830 extern int tcp_msl_enable
; /* enable TIME_WAIT truncation */
831 extern int tcp_msl_loop
; /* MSL for loopback */
832 extern int tcp_msl_local
; /* MSL for 'local' */
833 extern int tcp_msl_remote
; /* MSL otherwise */
834 extern int tcp_msl_remote_threshold
; /* RTT threshold */
835 extern int tcp_rttlocal
; /* Use RTT to decide who's 'local' */
836 extern int tcp4_vtw_enable
;
837 extern int tcp6_vtw_enable
;
838 extern int tcp_vtw_was_enabled
;
839 extern int tcp_vtw_entries
;
841 extern int tcp_rst_ppslim
;
842 extern int tcp_ackdrop_ppslim
;
844 extern int tcp_syn_cache_size
;
845 extern struct syn_cache_head tcp_syn_cache
[];
846 extern u_long syn_cache_count
;
849 extern struct mowner tcp_rx_mowner
;
850 extern struct mowner tcp_tx_mowner
;
851 extern struct mowner tcp_reass_mowner
;
852 extern struct mowner tcp_sock_mowner
;
853 extern struct mowner tcp_sock_rx_mowner
;
854 extern struct mowner tcp_sock_tx_mowner
;
855 extern struct mowner tcp_mowner
;
858 extern int tcp_do_autorcvbuf
;
859 extern int tcp_autorcvbuf_inc
;
860 extern int tcp_autorcvbuf_max
;
861 extern int tcp_do_autosndbuf
;
862 extern int tcp_autosndbuf_inc
;
863 extern int tcp_autosndbuf_max
;
866 #define TCPCTL_VARIABLES { \
868 { 1, 0, &tcp_do_rfc1323 }, \
869 { 1, 0, &tcp_sendspace }, \
870 { 1, 0, &tcp_recvspace }, \
871 { 1, 0, &tcp_mssdflt }, \
872 { 1, 0, &tcp_syn_cache_limit }, \
873 { 1, 0, &tcp_syn_bucket_limit }, \
875 { 1, 0, &tcp_init_win }, \
876 { 1, 0, &tcp_mss_ifmtu }, \
877 { 1, 0, &tcp_do_sack }, \
878 { 1, 0, &tcp_do_win_scale }, \
879 { 1, 0, &tcp_do_timestamps }, \
880 { 1, 0, &tcp_compat_42 }, \
881 { 1, 0, &tcp_cwm }, \
882 { 1, 0, &tcp_cwm_burstsize }, \
883 { 1, 0, &tcp_ack_on_push }, \
884 { 1, 0, &tcp_keepidle }, \
885 { 1, 0, &tcp_keepintvl }, \
886 { 1, 0, &tcp_keepcnt }, \
887 { 1, 1, 0, PR_SLOWHZ }, \
889 { 1, 0, &tcp_log_refused }, \
891 { 1, 0, &tcp_rst_ppslim }, \
892 { 1, 0, &tcp_delack_ticks }, \
893 { 1, 0, &tcp_init_win_local }, \
894 { 1, 0, &tcp_ackdrop_ppslim }, \
899 void tcp_canceltimers(struct tcpcb
*);
901 tcp_close(struct tcpcb
*);
902 int tcp_isdead(struct tcpcb
*);
904 void *tcp6_ctlinput(int, const struct sockaddr
*, void *);
906 void *tcp_ctlinput(int, const struct sockaddr
*, void *);
907 int tcp_ctloutput(int, struct socket
*, struct sockopt
*);
909 tcp_disconnect1(struct tcpcb
*);
911 tcp_drop(struct tcpcb
*, int);
913 int tcp_signature_apply(void *, void *, u_int
);
914 struct secasvar
*tcp_signature_getsav(struct mbuf
*, struct tcphdr
*);
915 int tcp_signature(struct mbuf
*, struct tcphdr
*, int, struct secasvar
*,
918 void tcp_drain(void);
919 void tcp_drainstub(void);
920 void tcp_established(struct tcpcb
*);
922 void tcp_init_common(unsigned);
924 int tcp6_input(struct mbuf
**, int *, int);
926 void tcp_input(struct mbuf
*, ...);
927 u_int
tcp_hdrsz(struct tcpcb
*);
928 u_long
tcp_mss_to_advertise(const struct ifnet
*, int);
929 void tcp_mss_from_peer(struct tcpcb
*, int);
930 void tcp_tcpcb_template(void);
932 tcp_newtcpcb(int, void *);
933 void tcp_notify(struct inpcb
*, int);
935 void tcp6_notify(struct in6pcb
*, int);
937 u_int
tcp_optlen(struct tcpcb
*);
938 int tcp_output(struct tcpcb
*);
939 void tcp_pulloutofband(struct socket
*,
940 struct tcphdr
*, struct mbuf
*, int);
941 void tcp_quench(struct inpcb
*, int);
943 void tcp6_quench(struct in6pcb
*, int);
945 void tcp_mtudisc(struct inpcb
*, int);
947 void tcp6_mtudisc_callback(struct in6_addr
*);
950 void tcpipqent_init(void);
951 struct ipqent
*tcpipqent_alloc(void);
952 void tcpipqent_free(struct ipqent
*);
954 int tcp_respond(struct tcpcb
*, struct mbuf
*, struct mbuf
*,
955 struct tcphdr
*, tcp_seq
, tcp_seq
, int);
956 void tcp_rmx_rtt(struct tcpcb
*);
957 void tcp_setpersist(struct tcpcb
*);
959 int tcp_signature_compute(struct mbuf
*, struct tcphdr
*, int, int,
960 int, u_char
*, u_int
);
962 void tcp_slowtimo(void *);
963 extern callout_t tcp_slowtimo_ch
;
964 void tcp_fasttimo(void);
966 tcp_template(struct tcpcb
*);
967 void tcp_trace(short, short, struct tcpcb
*, struct mbuf
*, int);
969 tcp_usrclosed(struct tcpcb
*);
970 void tcp_usrreq_init(void);
971 void tcp_xmit_timer(struct tcpcb
*, uint32_t);
972 tcp_seq
tcp_new_iss(struct tcpcb
*, tcp_seq
);
973 tcp_seq
tcp_new_iss1(void *, void *, u_int16_t
, u_int16_t
, size_t,
976 void tcp_sack_init(void);
977 void tcp_new_dsack(struct tcpcb
*, tcp_seq
, u_int32_t
);
978 void tcp_sack_option(struct tcpcb
*, const struct tcphdr
*,
979 const u_char
*, int);
980 void tcp_del_sackholes(struct tcpcb
*, const struct tcphdr
*);
981 void tcp_free_sackholes(struct tcpcb
*);
982 void tcp_sack_adjust(struct tcpcb
*tp
);
983 struct sackhole
*tcp_sack_output(struct tcpcb
*tp
, int *sack_bytes_rexmt
);
984 int tcp_sack_numblks(const struct tcpcb
*);
985 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2)
987 void tcp_statinc(u_int
);
988 void tcp_statadd(u_int
, uint64_t);
990 int syn_cache_add(struct sockaddr
*, struct sockaddr
*,
991 struct tcphdr
*, unsigned int, struct socket
*,
992 struct mbuf
*, u_char
*, int, struct tcp_opt_info
*);
993 void syn_cache_unreach(const struct sockaddr
*, const struct sockaddr
*,
995 struct socket
*syn_cache_get(struct sockaddr
*, struct sockaddr
*,
996 struct tcphdr
*, unsigned int, unsigned int,
997 struct socket
*so
, struct mbuf
*);
998 void syn_cache_init(void);
999 void syn_cache_insert(struct syn_cache
*, struct tcpcb
*);
1000 struct syn_cache
*syn_cache_lookup(const struct sockaddr
*, const struct sockaddr
*,
1001 struct syn_cache_head
**);
1002 void syn_cache_reset(struct sockaddr
*, struct sockaddr
*,
1004 int syn_cache_respond(struct syn_cache
*, struct mbuf
*);
1005 void syn_cache_timer(void *);
1006 void syn_cache_cleanup(struct tcpcb
*);
1008 int tcp_input_checksum(int, struct mbuf
*, const struct tcphdr
*, int, int,
1012 #endif /* !_NETINET_TCP_VAR_H_ */