tcp_var.h revision 1.153 1 /* $NetBSD: tcp_var.h,v 1.153 2008/02/29 07:39:17 matt Exp $ */
2
3 /*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
6 *
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.
18 *
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.
30 */
31
32 /*
33 * @(#)COPYRIGHT 1.1 (NRL) 17 January 1995
34 *
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:
38 *
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.
53 *
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.
65 *
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).
70 */
71
72 /*-
73 * Copyright (c) 1997, 1998, 1999, 2001, 2005 The NetBSD Foundation, Inc.
74 * All rights reserved.
75 *
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.
81 *
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.
90 * 3. All advertising materials mentioning features or use of this software
91 * must display the following acknowledgement:
92 * This product includes software developed by the NetBSD
93 * Foundation, Inc. and its contributors.
94 * 4. Neither the name of The NetBSD Foundation nor the names of its
95 * contributors may be used to endorse or promote products derived
96 * from this software without specific prior written permission.
97 *
98 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
99 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
100 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
101 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
102 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
103 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
104 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
105 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
106 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
107 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
108 * POSSIBILITY OF SUCH DAMAGE.
109 */
110
111 /*
112 * Copyright (c) 1982, 1986, 1993, 1994, 1995
113 * The Regents of the University of California. All rights reserved.
114 *
115 * Redistribution and use in source and binary forms, with or without
116 * modification, are permitted provided that the following conditions
117 * are met:
118 * 1. Redistributions of source code must retain the above copyright
119 * notice, this list of conditions and the following disclaimer.
120 * 2. Redistributions in binary form must reproduce the above copyright
121 * notice, this list of conditions and the following disclaimer in the
122 * documentation and/or other materials provided with the distribution.
123 * 3. Neither the name of the University nor the names of its contributors
124 * may be used to endorse or promote products derived from this software
125 * without specific prior written permission.
126 *
127 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
128 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
129 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
130 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
131 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
132 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
133 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
134 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
135 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
136 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
137 * SUCH DAMAGE.
138 *
139 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
140 */
141
142 #ifndef _NETINET_TCP_VAR_H_
143 #define _NETINET_TCP_VAR_H_
144
145 #if defined(_KERNEL_OPT)
146 #include "opt_inet.h"
147 #include "opt_mbuftrace.h"
148 #include "rnd.h"
149 #endif
150
151 /*
152 * Kernel variables for tcp.
153 */
154
155 #include <sys/callout.h>
156
157 #ifdef TCP_SIGNATURE
158 /*
159 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
160 * for SADB verification and lookup.
161 */
162 #define TCP_SIGLEN 16 /* length of computed digest in bytes */
163 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
164 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
165 /*
166 * Only a single SA per host may be specified at this time. An SPI is
167 * needed in order for the KEY_ALLOCSA() lookup to work.
168 */
169 #define TCP_SIG_SPI 0x1000
170 #endif /* TCP_SIGNATURE */
171
172 /*
173 * SACK option block.
174 */
175 struct sackblk {
176 tcp_seq left; /* Left edge of sack block. */
177 tcp_seq right; /* Right edge of sack block. */
178 };
179
180 TAILQ_HEAD(sackhead, sackhole);
181 struct sackhole {
182 tcp_seq start;
183 tcp_seq end;
184 tcp_seq rxmit;
185
186 TAILQ_ENTRY(sackhole) sackhole_q;
187 };
188
189 /*
190 * Tcp control block, one per tcp; fields:
191 */
192 struct tcpcb {
193 int t_family; /* address family on the wire */
194 struct ipqehead segq; /* sequencing queue */
195 int t_segqlen; /* length of the above */
196 callout_t t_timer[TCPT_NTIMERS];/* tcp timers */
197 short t_state; /* state of this connection */
198 short t_rxtshift; /* log(2) of rexmt exp. backoff */
199 uint32_t t_rxtcur; /* current retransmit value */
200 short t_dupacks; /* consecutive dup acks recd */
201 short t_partialacks; /* partials acks during fast rexmit */
202 u_short t_peermss; /* peer's maximum segment size */
203 u_short t_ourmss; /* our's maximum segment size */
204 u_short t_segsz; /* current segment size in use */
205 char t_force; /* 1 if forcing out a byte */
206 u_int t_flags;
207 #define TF_ACKNOW 0x0001 /* ack peer immediately */
208 #define TF_DELACK 0x0002 /* ack, but try to delay it */
209 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */
210 #define TF_NOOPT 0x0008 /* don't use tcp options */
211 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */
212 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */
213 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */
214 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */
215 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */
216 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */
217 #define TF_WILL_SACK 0x0800 /* try to use SACK */
218 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */
219 #define TF_DEAD 0x2000 /* dead and to-be-released */
220 #define TF_PMTUD_PEND 0x4000 /* Path MTU Discovery pending */
221 #define TF_ECN_PERMIT 0x10000 /* other side said is ECN-ready */
222 #define TF_ECN_SND_CWR 0x20000 /* ECN CWR in queue */
223 #define TF_ECN_SND_ECE 0x40000 /* ECN ECE in queue */
224 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
225
226
227 struct mbuf *t_template; /* skeletal packet for transmit */
228 struct inpcb *t_inpcb; /* back pointer to internet pcb */
229 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */
230 callout_t t_delack_ch; /* delayed ACK callout */
231 /*
232 * The following fields are used as in the protocol specification.
233 * See RFC793, Dec. 1981, page 21.
234 */
235 /* send sequence variables */
236 tcp_seq snd_una; /* send unacknowledged */
237 tcp_seq snd_nxt; /* send next */
238 tcp_seq snd_up; /* send urgent pointer */
239 tcp_seq snd_wl1; /* window update seg seq number */
240 tcp_seq snd_wl2; /* window update seg ack number */
241 tcp_seq iss; /* initial send sequence number */
242 u_long snd_wnd; /* send window */
243 tcp_seq snd_recover; /* for use in fast recovery */
244 tcp_seq snd_high; /* NewReno false fast rexmit seq */
245 /* receive sequence variables */
246 u_long rcv_wnd; /* receive window */
247 tcp_seq rcv_nxt; /* receive next */
248 tcp_seq rcv_up; /* receive urgent pointer */
249 tcp_seq irs; /* initial receive sequence number */
250 /*
251 * Additional variables for this implementation.
252 */
253 /* receive variables */
254 tcp_seq rcv_adv; /* advertised window */
255 /* retransmit variables */
256 tcp_seq snd_max; /* highest sequence number sent;
257 * used to recognize retransmits
258 */
259 /* congestion control (for slow start, source quench, retransmit after loss) */
260 u_long snd_cwnd; /* congestion-controlled window */
261 u_long snd_ssthresh; /* snd_cwnd size threshhold for
262 * for slow start exponential to
263 * linear switch
264 */
265 /* auto-sizing variables */
266 u_int rfbuf_cnt; /* recv buffer autoscaling byte count */
267 uint32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
268
269 /*
270 * transmit timing stuff. See below for scale of srtt and rttvar.
271 * "Variance" is actually smoothed difference.
272 */
273 uint32_t t_rcvtime; /* time last segment received */
274 uint32_t t_rtttime; /* time we started measuring rtt */
275 tcp_seq t_rtseq; /* sequence number being timed */
276 int32_t t_srtt; /* smoothed round-trip time */
277 int32_t t_rttvar; /* variance in round-trip time */
278 uint32_t t_rttmin; /* minimum rtt allowed */
279 u_long max_sndwnd; /* largest window peer has offered */
280
281 /* out-of-band data */
282 char t_oobflags; /* have some */
283 char t_iobc; /* input character */
284 #define TCPOOB_HAVEDATA 0x01
285 #define TCPOOB_HADDATA 0x02
286 short t_softerror; /* possible error not yet reported */
287
288 /* RFC 1323 variables */
289 u_char snd_scale; /* window scaling for send window */
290 u_char rcv_scale; /* window scaling for recv window */
291 u_char request_r_scale; /* pending window scaling */
292 u_char requested_s_scale;
293 u_int32_t ts_recent; /* timestamp echo data */
294 u_int32_t ts_recent_age; /* when last updated */
295 u_int32_t ts_timebase; /* our timebase */
296 tcp_seq last_ack_sent;
297
298 /* RFC 3465 variables */
299 u_long t_bytes_acked; /* ABC "bytes_acked" parameter */
300
301 /* SACK stuff */
302 #define TCP_SACK_MAX 3
303 #define TCPSACK_NONE 0
304 #define TCPSACK_HAVED 1
305 u_char rcv_sack_flags; /* SACK flags. */
306 struct sackblk rcv_dsack_block; /* RX D-SACK block. */
307 struct ipqehead timeq; /* time sequenced queue. */
308 struct sackhead snd_holes; /* TX SACK holes. */
309 int snd_numholes; /* Number of TX SACK holes. */
310 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/
311 tcp_seq sack_newdata; /* New data xmitted in this recovery
312 episode starts at this seq number*/
313 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by
314 peer. */
315
316 /* pointer for syn cache entries*/
317 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */
318
319 /* prediction of next mbuf when using large window sizes */
320 struct mbuf *t_lastm; /* last mbuf that data was sent from */
321 int t_inoff; /* data offset in previous mbuf */
322 int t_lastoff; /* last data address in mbuf chain */
323 int t_lastlen; /* last length read from mbuf chain */
324
325 /* Path-MTU discovery blackhole detection */
326 int t_mtudisc; /* perform mtudisc for this tcb */
327 /* Path-MTU Discovery Information */
328 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */
329 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */
330 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */
331 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */
332 u_short t_pmtud_ip_len; /* IP length from ICMP payload */
333 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */
334
335 uint8_t t_ecn_retries; /* # of ECN setup retries */
336
337 const struct tcp_congctl *t_congctl; /* per TCB congctl algorithm */
338
339 /* Keepalive per socket */
340 u_int t_keepinit;
341 u_int t_keepidle;
342 u_int t_keepintvl;
343 u_int t_keepcnt;
344 u_int t_maxidle; /* t_keepcnt * t_keepintvl */
345
346 };
347
348 /*
349 * Macros to aid ECN TCP.
350 */
351 #define TCP_ECN_ALLOWED(tp) (tp->t_flags & TF_ECN_PERMIT)
352
353 /*
354 * Macros to aid SACK/FACK TCP.
355 */
356 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK)
357 #define TCP_FACK_FASTRECOV(tp) \
358 (TCP_SACK_ENABLED(tp) && \
359 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz)))
360
361 #ifdef _KERNEL
362 /*
363 * TCP reassembly queue locks.
364 */
365 static __inline int tcp_reass_lock_try (struct tcpcb *)
366 __unused;
367 static __inline void tcp_reass_unlock (struct tcpcb *)
368 __unused;
369
370 static __inline int
371 tcp_reass_lock_try(struct tcpcb *tp)
372 {
373 int s;
374
375 /*
376 * Use splvm() -- we're blocking things that would cause
377 * mbuf allocation.
378 */
379 s = splvm();
380 if (tp->t_flags & TF_REASSEMBLING) {
381 splx(s);
382 return (0);
383 }
384 tp->t_flags |= TF_REASSEMBLING;
385 splx(s);
386 return (1);
387 }
388
389 static __inline void
390 tcp_reass_unlock(struct tcpcb *tp)
391 {
392 int s;
393
394 s = splvm();
395 tp->t_flags &= ~TF_REASSEMBLING;
396 splx(s);
397 }
398
399 #ifdef DIAGNOSTIC
400 #define TCP_REASS_LOCK(tp) \
401 do { \
402 if (tcp_reass_lock_try(tp) == 0) { \
403 printf("%s:%d: tcpcb %p reass already locked\n", \
404 __FILE__, __LINE__, tp); \
405 panic("tcp_reass_lock"); \
406 } \
407 } while (/*CONSTCOND*/ 0)
408 #define TCP_REASS_LOCK_CHECK(tp) \
409 do { \
410 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \
411 printf("%s:%d: tcpcb %p reass lock not held\n", \
412 __FILE__, __LINE__, tp); \
413 panic("tcp reass lock check"); \
414 } \
415 } while (/*CONSTCOND*/ 0)
416 #else
417 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp))
418 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */
419 #endif
420
421 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp))
422 #endif /* _KERNEL */
423
424 /*
425 * Queue for delayed ACK processing.
426 */
427 #ifdef _KERNEL
428 extern int tcp_delack_ticks;
429 void tcp_delack(void *);
430
431 #define TCP_RESTART_DELACK(tp) \
432 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \
433 tcp_delack, tp)
434
435 #define TCP_SET_DELACK(tp) \
436 do { \
437 if (((tp)->t_flags & TF_DELACK) == 0) { \
438 (tp)->t_flags |= TF_DELACK; \
439 TCP_RESTART_DELACK(tp); \
440 } \
441 } while (/*CONSTCOND*/0)
442
443 #define TCP_CLEAR_DELACK(tp) \
444 do { \
445 if ((tp)->t_flags & TF_DELACK) { \
446 (tp)->t_flags &= ~TF_DELACK; \
447 callout_stop(&(tp)->t_delack_ch); \
448 } \
449 } while (/*CONSTCOND*/0)
450 #endif /* _KERNEL */
451
452 /*
453 * Compute the current timestamp for a connection.
454 */
455 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase)
456
457 /*
458 * Handy way of passing around TCP option info.
459 */
460 struct tcp_opt_info {
461 int ts_present;
462 u_int32_t ts_val;
463 u_int32_t ts_ecr;
464 u_int16_t maxseg;
465 };
466
467 #define TOF_SIGNATURE 0x0040 /* signature option present */
468 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */
469
470 /*
471 * Data for the TCP compressed state engine.
472 */
473 union syn_cache_sa {
474 struct sockaddr sa;
475 struct sockaddr_in sin;
476 #if 1 /*def INET6*/
477 struct sockaddr_in6 sin6;
478 #endif
479 };
480
481 struct syn_cache {
482 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */
483 callout_t sc_timer; /* rexmt timer */
484 struct route sc_route;
485 long sc_win; /* advertised window */
486 int sc_bucketidx; /* our bucket index */
487 u_int32_t sc_hash;
488 u_int32_t sc_timestamp; /* timestamp from SYN */
489 u_int32_t sc_timebase; /* our local timebase */
490 union syn_cache_sa sc_src;
491 union syn_cache_sa sc_dst;
492 tcp_seq sc_irs;
493 tcp_seq sc_iss;
494 u_int sc_rxtcur; /* current rxt timeout */
495 u_int sc_rxttot; /* total time spend on queues */
496 u_short sc_rxtshift; /* for computing backoff */
497 u_short sc_flags;
498
499 #define SCF_UNREACH 0x0001 /* we've had an unreach error */
500 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */
501 #define SCF_DEAD 0x0004 /* this entry to be released */
502 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */
503 #define SCF_ECN_PERMIT 0x0010 /* peer will do ECN */
504 #define SCF_SIGNATURE 0x40 /* send MD5 digests */
505
506 struct mbuf *sc_ipopts; /* IP options */
507 u_int16_t sc_peermaxseg;
508 u_int16_t sc_ourmaxseg;
509 u_int8_t sc_request_r_scale : 4,
510 sc_requested_s_scale : 4;
511
512 struct tcpcb *sc_tp; /* tcb for listening socket */
513 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */
514 };
515
516 struct syn_cache_head {
517 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */
518 u_short sch_length; /* # entries in bucket */
519 };
520
521 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
522 #ifdef INET6
523 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb)
524 #endif
525 #ifndef INET6
526 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
527 #else
528 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \
529 ? intotcpcb(sotoinpcb(so)) \
530 : in6totcpcb(sotoin6pcb(so)))
531 #endif
532
533 /*
534 * The smoothed round-trip time and estimated variance
535 * are stored as fixed point numbers scaled by the values below.
536 * For convenience, these scales are also used in smoothing the average
537 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
538 * With these scales, srtt has 3 bits to the right of the binary point,
539 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
540 * binary point, and is smoothed with an ALPHA of 0.75.
541 */
542 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */
543 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */
544
545 /*
546 * The initial retransmission should happen at rtt + 4 * rttvar.
547 * Because of the way we do the smoothing, srtt and rttvar
548 * will each average +1/2 tick of bias. When we compute
549 * the retransmit timer, we want 1/2 tick of rounding and
550 * 1 extra tick because of +-1/2 tick uncertainty in the
551 * firing of the timer. The bias will give us exactly the
552 * 1.5 tick we need. But, because the bias is
553 * statistical, we have to test that we don't drop below
554 * the minimum feasible timer (which is 2 ticks).
555 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT
556 * is the same as the multiplier for rttvar.
557 */
558 #define TCP_REXMTVAL(tp) \
559 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2)
560
561 /*
562 * Compute the initial window for slow start.
563 */
564 #define TCP_INITIAL_WINDOW(iw, segsz) \
565 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \
566 ((segsz) * (iw)))
567
568 /*
569 * TCP statistics.
570 * Many of these should be kept per connection,
571 * but that's inconvenient at the moment.
572 */
573 struct tcpstat {
574 u_quad_t tcps_connattempt; /* connections initiated */
575 u_quad_t tcps_accepts; /* connections accepted */
576 u_quad_t tcps_connects; /* connections established */
577 u_quad_t tcps_drops; /* connections dropped */
578 u_quad_t tcps_conndrops; /* embryonic connections dropped */
579 u_quad_t tcps_closed; /* conn. closed (includes drops) */
580 u_quad_t tcps_segstimed; /* segs where we tried to get rtt */
581 u_quad_t tcps_rttupdated; /* times we succeeded */
582 u_quad_t tcps_delack; /* delayed acks sent */
583 u_quad_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
584 u_quad_t tcps_rexmttimeo; /* retransmit timeouts */
585 u_quad_t tcps_persisttimeo; /* persist timeouts */
586 u_quad_t tcps_keeptimeo; /* keepalive timeouts */
587 u_quad_t tcps_keepprobe; /* keepalive probes sent */
588 u_quad_t tcps_keepdrops; /* connections dropped in keepalive */
589 u_quad_t tcps_persistdrops; /* connections dropped in persist */
590 u_quad_t tcps_connsdrained; /* connections drained due to memory
591 shortage */
592 u_quad_t tcps_pmtublackhole; /* PMTUD blackhole detected */
593
594 u_quad_t tcps_sndtotal; /* total packets sent */
595 u_quad_t tcps_sndpack; /* data packets sent */
596 u_quad_t tcps_sndbyte; /* data bytes sent */
597 u_quad_t tcps_sndrexmitpack; /* data packets retransmitted */
598 u_quad_t tcps_sndrexmitbyte; /* data bytes retransmitted */
599 u_quad_t tcps_sndacks; /* ack-only packets sent */
600 u_quad_t tcps_sndprobe; /* window probes sent */
601 u_quad_t tcps_sndurg; /* packets sent with URG only */
602 u_quad_t tcps_sndwinup; /* window update-only packets sent */
603 u_quad_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
604
605 u_quad_t tcps_rcvtotal; /* total packets received */
606 u_quad_t tcps_rcvpack; /* packets received in sequence */
607 u_quad_t tcps_rcvbyte; /* bytes received in sequence */
608 u_quad_t tcps_rcvbadsum; /* packets received with ccksum errs */
609 u_quad_t tcps_rcvbadoff; /* packets received with bad offset */
610 u_quad_t tcps_rcvmemdrop; /* packets dropped for lack of memory */
611 u_quad_t tcps_rcvshort; /* packets received too short */
612 u_quad_t tcps_rcvduppack; /* duplicate-only packets received */
613 u_quad_t tcps_rcvdupbyte; /* duplicate-only bytes received */
614 u_quad_t tcps_rcvpartduppack; /* packets with some duplicate data */
615 u_quad_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
616 u_quad_t tcps_rcvoopack; /* out-of-order packets received */
617 u_quad_t tcps_rcvoobyte; /* out-of-order bytes received */
618 u_quad_t tcps_rcvpackafterwin; /* packets with data after window */
619 u_quad_t tcps_rcvbyteafterwin; /* bytes rcvd after window */
620 u_quad_t tcps_rcvafterclose; /* packets rcvd after "close" */
621 u_quad_t tcps_rcvwinprobe; /* rcvd window probe packets */
622 u_quad_t tcps_rcvdupack; /* rcvd duplicate acks */
623 u_quad_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */
624 u_quad_t tcps_rcvackpack; /* rcvd ack packets */
625 u_quad_t tcps_rcvackbyte; /* bytes acked by rcvd acks */
626 u_quad_t tcps_rcvwinupd; /* rcvd window update packets */
627 u_quad_t tcps_pawsdrop; /* segments dropped due to PAWS */
628 u_quad_t tcps_predack; /* times hdr predict ok for acks */
629 u_quad_t tcps_preddat; /* times hdr predict ok for data pkts */
630
631 u_quad_t tcps_pcbhashmiss; /* input packets missing pcb hash */
632 u_quad_t tcps_noport; /* no socket on port */
633 u_quad_t tcps_badsyn; /* received ack for which we have
634 no SYN in compressed state */
635 u_quad_t tcps_delayed_free; /* delayed pool_put() of tcpcb */
636
637 /* These statistics deal with the SYN cache. */
638 u_quad_t tcps_sc_added; /* # of entries added */
639 u_quad_t tcps_sc_completed; /* # of connections completed */
640 u_quad_t tcps_sc_timed_out; /* # of entries timed out */
641 u_quad_t tcps_sc_overflowed; /* # dropped due to overflow */
642 u_quad_t tcps_sc_reset; /* # dropped due to RST */
643 u_quad_t tcps_sc_unreach; /* # dropped due to ICMP unreach */
644 u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */
645 u_quad_t tcps_sc_aborted; /* # of entries aborted (no mem) */
646 u_quad_t tcps_sc_dupesyn; /* # of duplicate SYNs received */
647 u_quad_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */
648 u_quad_t tcps_sc_collisions; /* # of hash collisions */
649 u_quad_t tcps_sc_retransmitted; /* # of retransmissions */
650 u_quad_t tcps_sc_delayed_free; /* # of delayed pool_put()s */
651
652 u_quad_t tcps_selfquench; /* # of ENOBUFS we get on output */
653 u_quad_t tcps_badsig; /* # of drops due to bad signature */
654 u_quad_t tcps_goodsig; /* # of packets with good signature */
655
656 u_quad_t tcps_ecn_shs; /* # of sucessful ECN handshakes */
657 u_quad_t tcps_ecn_ce; /* # of packets with CE bit */
658 u_quad_t tcps_ecn_ect; /* # of packets with ECT(0) bit */
659 };
660
661 /*
662 * Names for TCP sysctl objects.
663 */
664 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */
665 #define TCPCTL_SENDSPACE 2 /* default send buffer */
666 #define TCPCTL_RECVSPACE 3 /* default recv buffer */
667 #define TCPCTL_MSSDFLT 4 /* default seg size */
668 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */
669 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */
670 #if 0 /*obsoleted*/
671 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */
672 #endif
673 #define TCPCTL_INIT_WIN 8 /* initial window */
674 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */
675 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */
676 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */
677 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */
678 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */
679 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */
680 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */
681 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */
682 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */
683 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */
684 #define TCPCTL_KEEPCNT 19 /* keepalive count */
685 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */
686 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */
687 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */
688 #if 0 /*obsoleted*/
689 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */
690 #endif
691 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */
692 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */
693 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */
694 #define TCPCTL_IDENT 27 /* rfc 931 identd */
695 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */
696 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */
697 #define TCPCTL_STATS 30 /* TCP statistics */
698 #define TCPCTL_DEBUG 31 /* TCP debug sockets */
699 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */
700 #define TCPCTL_DROP 33 /* drop tcp connection */
701 #define TCPCTL_MAXID 34
702
703 #define TCPCTL_NAMES { \
704 { 0, 0 }, \
705 { "rfc1323", CTLTYPE_INT }, \
706 { "sendspace", CTLTYPE_INT }, \
707 { "recvspace", CTLTYPE_INT }, \
708 { "mssdflt", CTLTYPE_INT }, \
709 { "syn_cache_limit", CTLTYPE_INT }, \
710 { "syn_bucket_limit", CTLTYPE_INT }, \
711 { 0, 0 },\
712 { "init_win", CTLTYPE_INT }, \
713 { "mss_ifmtu", CTLTYPE_INT }, \
714 { "sack", CTLTYPE_INT }, \
715 { "win_scale", CTLTYPE_INT }, \
716 { "timestamps", CTLTYPE_INT }, \
717 { "compat_42", CTLTYPE_INT }, \
718 { "cwm", CTLTYPE_INT }, \
719 { "cwm_burstsize", CTLTYPE_INT }, \
720 { "ack_on_push", CTLTYPE_INT }, \
721 { "keepidle", CTLTYPE_INT }, \
722 { "keepintvl", CTLTYPE_INT }, \
723 { "keepcnt", CTLTYPE_INT }, \
724 { "slowhz", CTLTYPE_INT }, \
725 { 0, 0 }, \
726 { "log_refused",CTLTYPE_INT }, \
727 { 0, 0 }, \
728 { "rstppslimit", CTLTYPE_INT }, \
729 { "delack_ticks", CTLTYPE_INT }, \
730 { "init_win_local", CTLTYPE_INT }, \
731 { "ident", CTLTYPE_STRUCT }, \
732 { "ackdropppslimit", CTLTYPE_INT }, \
733 { "do_loopback_cksum", CTLTYPE_INT }, \
734 { "stats", CTLTYPE_STRUCT }, \
735 { "debug", CTLTYPE_STRUCT }, \
736 { "debx", CTLTYPE_INT }, \
737 { "drop", CTLTYPE_STRUCT }, \
738 }
739
740 #ifdef _KERNEL
741 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */
742 extern struct tcpstat tcpstat; /* tcp statistics */
743 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */
744 extern int tcp_do_rfc1323; /* enabled/disabled? */
745 extern int tcp_do_sack; /* SACK enabled/disabled? */
746 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */
747 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */
748 extern int tcp_mssdflt; /* default seg size */
749 extern int tcp_minmss; /* minimal seg size */
750 extern int tcp_init_win; /* initial window */
751 extern int tcp_init_win_local; /* initial window for local nets */
752 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */
753 extern int tcp_compat_42; /* work around ancient broken TCP peers */
754 extern int tcp_cwm; /* enable Congestion Window Monitoring */
755 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */
756 extern int tcp_ack_on_push; /* ACK immediately on PUSH */
757 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */
758 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */
759 extern int tcp_log_refused; /* log refused connections */
760 extern int tcp_do_ecn; /* TCP ECN enabled/disabled? */
761 extern int tcp_ecn_maxretries; /* Max ECN setup retries */
762 #if NRND > 0
763 extern int tcp_do_rfc1948; /* ISS by cryptographic hash */
764 #endif
765 extern int tcp_sack_tp_maxholes; /* Max holes per connection. */
766 extern int tcp_sack_globalmaxholes; /* Max holes per system. */
767 extern int tcp_sack_globalholes; /* Number of holes present. */
768 extern int tcp_do_abc; /* RFC3465 ABC enabled/disabled? */
769 extern int tcp_abc_aggressive; /* 1: L=2*SMSS 0: L=1*SMSS */
770
771 extern int tcp_rst_ppslim;
772 extern int tcp_ackdrop_ppslim;
773
774 extern int tcp_syn_cache_size;
775 extern struct syn_cache_head tcp_syn_cache[];
776 extern u_long syn_cache_count;
777
778 #ifdef MBUFTRACE
779 extern struct mowner tcp_rx_mowner;
780 extern struct mowner tcp_tx_mowner;
781 extern struct mowner tcp_reass_mowner;
782 extern struct mowner tcp_sock_mowner;
783 extern struct mowner tcp_sock_rx_mowner;
784 extern struct mowner tcp_sock_tx_mowner;
785 extern struct mowner tcp_mowner;
786 #endif
787
788 extern int tcp_do_autorcvbuf;
789 extern int tcp_autorcvbuf_inc;
790 extern int tcp_autorcvbuf_max;
791 extern int tcp_do_autosndbuf;
792 extern int tcp_autosndbuf_inc;
793 extern int tcp_autosndbuf_max;
794
795
796 #define TCPCTL_VARIABLES { \
797 { 0 }, \
798 { 1, 0, &tcp_do_rfc1323 }, \
799 { 1, 0, &tcp_sendspace }, \
800 { 1, 0, &tcp_recvspace }, \
801 { 1, 0, &tcp_mssdflt }, \
802 { 1, 0, &tcp_syn_cache_limit }, \
803 { 1, 0, &tcp_syn_bucket_limit }, \
804 { 0 }, \
805 { 1, 0, &tcp_init_win }, \
806 { 1, 0, &tcp_mss_ifmtu }, \
807 { 1, 0, &tcp_do_sack }, \
808 { 1, 0, &tcp_do_win_scale }, \
809 { 1, 0, &tcp_do_timestamps }, \
810 { 1, 0, &tcp_compat_42 }, \
811 { 1, 0, &tcp_cwm }, \
812 { 1, 0, &tcp_cwm_burstsize }, \
813 { 1, 0, &tcp_ack_on_push }, \
814 { 1, 0, &tcp_keepidle }, \
815 { 1, 0, &tcp_keepintvl }, \
816 { 1, 0, &tcp_keepcnt }, \
817 { 1, 1, 0, PR_SLOWHZ }, \
818 { 0 }, \
819 { 1, 0, &tcp_log_refused }, \
820 { 0 }, \
821 { 1, 0, &tcp_rst_ppslim }, \
822 { 1, 0, &tcp_delack_ticks }, \
823 { 1, 0, &tcp_init_win_local }, \
824 { 1, 0, &tcp_ackdrop_ppslim }, \
825 }
826
827 #ifdef __NO_STRICT_ALIGNMENT
828 #define TCP_HDR_ALIGNED_P(th) 1
829 #else
830 #define TCP_HDR_ALIGNED_P(th) ((((vaddr_t)(th)) & 3) == 0)
831 #endif
832
833 struct secasvar;
834
835 int tcp_attach(struct socket *);
836 void tcp_canceltimers(struct tcpcb *);
837 int tcp_timers_invoking(struct tcpcb*);
838 struct tcpcb *
839 tcp_close(struct tcpcb *);
840 int tcp_isdead(struct tcpcb *);
841 #ifdef INET6
842 void tcp6_ctlinput(int, const struct sockaddr *, void *);
843 #endif
844 void *tcp_ctlinput(int, const struct sockaddr *, void *);
845 int tcp_ctloutput(int, struct socket *, int, int, struct mbuf **);
846 struct tcpcb *
847 tcp_disconnect(struct tcpcb *);
848 struct tcpcb *
849 tcp_drop(struct tcpcb *, int);
850 #ifdef TCP_SIGNATURE
851 int tcp_signature_apply(void *, void *, u_int);
852 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *);
853 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *,
854 char *);
855 #endif
856 void tcp_drain(void);
857 void tcp_established(struct tcpcb *);
858 void tcp_init(void);
859 #ifdef INET6
860 int tcp6_input(struct mbuf **, int *, int);
861 #endif
862 void tcp_input(struct mbuf *, ...);
863 u_int tcp_hdrsz(struct tcpcb *);
864 u_long tcp_mss_to_advertise(const struct ifnet *, int);
865 void tcp_mss_from_peer(struct tcpcb *, int);
866 void tcp_tcpcb_template(void);
867 struct tcpcb *
868 tcp_newtcpcb(int, void *);
869 void tcp_notify(struct inpcb *, int);
870 #ifdef INET6
871 void tcp6_notify(struct in6pcb *, int);
872 #endif
873 u_int tcp_optlen(struct tcpcb *);
874 int tcp_output(struct tcpcb *);
875 void tcp_pulloutofband(struct socket *,
876 struct tcphdr *, struct mbuf *, int);
877 void tcp_quench(struct inpcb *, int);
878 #ifdef INET6
879 void tcp6_quench(struct in6pcb *, int);
880 #endif
881 void tcp_mtudisc(struct inpcb *, int);
882
883 struct ipqent *tcpipqent_alloc(void);
884 void tcpipqent_free(struct ipqent *);
885
886 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *,
887 struct tcphdr *, tcp_seq, tcp_seq, int);
888 void tcp_rmx_rtt(struct tcpcb *);
889 void tcp_setpersist(struct tcpcb *);
890 #ifdef TCP_SIGNATURE
891 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int,
892 int, u_char *, u_int);
893 #endif
894 void tcp_slowtimo(void);
895 struct mbuf *
896 tcp_template(struct tcpcb *);
897 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int);
898 struct tcpcb *
899 tcp_usrclosed(struct tcpcb *);
900 int tcp_usrreq(struct socket *,
901 int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *);
902 void tcp_xmit_timer(struct tcpcb *, uint32_t);
903 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq);
904 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t,
905 tcp_seq);
906
907 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t);
908 void tcp_sack_option(struct tcpcb *, const struct tcphdr *,
909 const u_char *, int);
910 void tcp_del_sackholes(struct tcpcb *, const struct tcphdr *);
911 void tcp_free_sackholes(struct tcpcb *);
912 void tcp_sack_adjust(struct tcpcb *tp);
913 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
914 void tcp_sack_newack(struct tcpcb *, const struct tcphdr *);
915 int tcp_sack_numblks(const struct tcpcb *);
916 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2)
917
918 int syn_cache_add(struct sockaddr *, struct sockaddr *,
919 struct tcphdr *, unsigned int, struct socket *,
920 struct mbuf *, u_char *, int, struct tcp_opt_info *);
921 void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *,
922 struct tcphdr *);
923 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *,
924 struct tcphdr *, unsigned int, unsigned int,
925 struct socket *so, struct mbuf *);
926 void syn_cache_init(void);
927 void syn_cache_insert(struct syn_cache *, struct tcpcb *);
928 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *,
929 struct syn_cache_head **);
930 void syn_cache_reset(struct sockaddr *, struct sockaddr *,
931 struct tcphdr *);
932 int syn_cache_respond(struct syn_cache *, struct mbuf *);
933 void syn_cache_timer(void *);
934 void syn_cache_cleanup(struct tcpcb *);
935
936 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int,
937 int);
938 #endif
939
940 #endif /* !_NETINET_TCP_VAR_H_ */
941