1/*-
2 * Copyright (c) 1982, 1986, 1993, 1994, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
30 * $FreeBSD: head/sys/netinet/tcp_var.h 155767 2006-02-16 19:38:07Z andre $
31 */
32
33#ifndef _NETINET_TCP_VAR_H_
34#define _NETINET_TCP_VAR_H_
35
36#include <netinet/tcp.h>
37
38/*
39 * Kernel variables for tcp.
40 */
41extern int tcp_do_rfc1323;
42
43/* TCP segment queue entry */
44struct tseg_qent {
45 LIST_ENTRY(tseg_qent) tqe_q;
46 int tqe_len; /* TCP segment data length */
47 struct tcphdr *tqe_th; /* a pointer to tcp header */
48 struct mbuf *tqe_m; /* mbuf contains packet */
49};
50LIST_HEAD(tsegqe_head, tseg_qent);
51extern int tcp_reass_qsize;
52extern struct uma_zone *tcp_reass_zone;
53
54struct sackblk {
55 tcp_seq start; /* start seq no. of sack block */
56 tcp_seq end; /* end seq no. */
57};
58
59struct sackhole {
60 tcp_seq start; /* start seq no. of hole */
61 tcp_seq end; /* end seq no. */
62 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
63 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
64};
65
66struct sackhint {
67 struct sackhole *nexthole;
68 int sack_bytes_rexmit;
69};
70
71struct tcptemp {
72 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
73 struct tcphdr tt_t;
74};
75
76#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
77
78/*
79 * Tcp control block, one per tcp; fields:
80 * Organized for 16 byte cacheline efficiency.
81 */
82struct tcpcb {
83 struct tsegqe_head t_segq; /* segment reassembly queue */
84 int t_segqlen; /* segment reassembly queue length */
85 int t_dupacks; /* consecutive dup acks recd */
86
87 struct callout *tt_rexmt; /* retransmit timer */
88 struct callout *tt_persist; /* retransmit persistence */
89 struct callout *tt_keep; /* keepalive */
90 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */
91 struct callout *tt_delack; /* delayed ACK timer */
92
93 struct inpcb *t_inpcb; /* back pointer to internet pcb */
94 int t_state; /* state of this connection */
95 u_int t_flags;
96#define TF_ACKNOW 0x000001 /* ack peer immediately */
97#define TF_DELACK 0x000002 /* ack, but try to delay it */
98#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */
99#define TF_NOOPT 0x000008 /* don't use tcp options */
100#define TF_SENTFIN 0x000010 /* have sent FIN */
101#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */
102#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */
103#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */
104#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */
105#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */
106#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
107#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
108#define TF_NOPUSH 0x001000 /* don't push */
109#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */
110#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
111#define TF_LASTIDLE 0x040000 /* connection was previously idle */
112#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */
113#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */
114#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */
115#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
116#define TF_FORCEDATA 0x800000 /* force out a byte */
117
118 tcp_seq snd_una; /* send unacknowledged */
119 tcp_seq snd_max; /* highest sequence number sent;
120 * used to recognize retransmits
121 */
122 tcp_seq snd_nxt; /* send next */
123 tcp_seq snd_up; /* send urgent pointer */
124
125 tcp_seq snd_wl1; /* window update seg seq number */
126 tcp_seq snd_wl2; /* window update seg ack number */
127 tcp_seq iss; /* initial send sequence number */
128 tcp_seq irs; /* initial receive sequence number */
129
130 tcp_seq rcv_nxt; /* receive next */
131 tcp_seq rcv_adv; /* advertised window */
132 u_long rcv_wnd; /* receive window */
133 tcp_seq rcv_up; /* receive urgent pointer */
134
135 u_long snd_wnd; /* send window */
136 u_long snd_cwnd; /* congestion-controlled window */
137 u_long snd_bwnd; /* bandwidth-controlled window */
138 u_long snd_ssthresh; /* snd_cwnd size threshold for
139 * for slow start exponential to
140 * linear switch
141 */
142 u_long snd_bandwidth; /* calculated bandwidth or 0 */
143 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
144
145 u_int t_maxopd; /* mss plus options */
146
147 u_long t_rcvtime; /* inactivity time */
148 u_long t_starttime; /* time connection was established */
149 int t_rtttime; /* round trip time */
150 tcp_seq t_rtseq; /* sequence number being timed */
151
152 int t_bw_rtttime; /* used for bandwidth calculation */
153 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */
154
155 int t_rxtcur; /* current retransmit value (ticks) */
156 u_int t_maxseg; /* maximum segment size */
157 int t_srtt; /* smoothed round-trip time */
158 int t_rttvar; /* variance in round-trip time */
159
160 int t_rxtshift; /* log(2) of rexmt exp. backoff */
161 u_int t_rttmin; /* minimum rtt allowed */
162 u_int t_rttbest; /* best rtt we've seen */
163 u_long t_rttupdated; /* number of times rtt sampled */
164 u_long max_sndwnd; /* largest window peer has offered */
165
166 int t_softerror; /* possible error not yet reported */
167/* out-of-band data */
168 char t_oobflags; /* have some */
169 char t_iobc; /* input character */
170#define TCPOOB_HAVEDATA 0x01
171#define TCPOOB_HADDATA 0x02
172/* RFC 1323 variables */
173 u_char snd_scale; /* window scaling for send window */
174 u_char rcv_scale; /* window scaling for recv window */
175 u_char request_r_scale; /* pending window scaling */
176 u_char requested_s_scale;
177 u_long ts_recent; /* timestamp echo data */
178
179 u_long ts_recent_age; /* when last updated */
180 tcp_seq last_ack_sent;
181/* experimental */
182 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
183 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
184 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
185 u_long t_badrxtwin; /* window for retransmit recovery */
186 u_char snd_limited; /* segments limited transmitted */
187/* anti DoS counters */
188 u_long rcv_second; /* start of interval second */
189 u_long rcv_pps; /* received packets per second */
190 u_long rcv_byps; /* received bytes per second */
191/* SACK related state */
192 int sack_enable; /* enable SACK for this connection */
193 int snd_numholes; /* number of holes seen by sender */
194 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
195 /* SACK scoreboard (sorted) */
196 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
197 int rcv_numsacks; /* # distinct sack blks present */
198 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
199 tcp_seq sack_newdata; /* New data xmitted in this recovery
200 episode starts at this seq number */
201 struct sackhint sackhint; /* SACK scoreboard hint */
202 int t_rttlow; /* smallest observerved RTT */
203};
204
205#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY)
206#define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY
207#define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY
208
209#ifdef TCP_SIGNATURE
210/*
211 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
212 * for SADB verification and lookup.
213 */
214#define TCP_SIGLEN 16 /* length of computed digest in bytes */
215#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
216#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
217/*
218 * Only a single SA per host may be specified at this time. An SPI is
219 * needed in order for the KEY_ALLOCSA() lookup to work.
220 */
221#define TCP_SIG_SPI 0x1000
222#endif /* TCP_SIGNATURE */
223
224/*
225 * Structure to hold TCP options that are only used during segment
226 * processing (in tcp_input), but not held in the tcpcb.
227 * It's basically used to reduce the number of parameters
228 * to tcp_dooptions.
229 */
230struct tcpopt {
231 u_long to_flags; /* which options are present */
232#define TOF_TS 0x0001 /* timestamp */
233#define TOF_MSS 0x0010
234#define TOF_SCALE 0x0020
235#define TOF_SIGNATURE 0x0040 /* signature option present */
236#define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */
237#define TOF_SACK 0x0100 /* Peer sent SACK option */
238 u_int32_t to_tsval;
239 u_int32_t to_tsecr;
240 u_int16_t to_mss;
241 u_int8_t to_requested_s_scale;
242 u_int8_t to_nsacks; /* number of SACK blocks */
243 u_char *to_sacks; /* pointer to the first SACK blocks */
244};
245
246#ifdef _NETINET_IN_PCB_H_
247struct syncache {
248 inp_gen_t sc_inp_gencnt; /* pointer check */
249 struct tcpcb *sc_tp; /* tcb for listening socket */
250 struct mbuf *sc_ipopts; /* source route */
251 struct in_conninfo sc_inc; /* addresses */
252 u_int32_t sc_tsrecent;
253 u_int32_t sc_flowlabel; /* IPv6 flowlabel */
254 tcp_seq sc_irs; /* seq from peer */
255 tcp_seq sc_iss; /* our ISS */
256 u_long sc_rxttime; /* retransmit time */
257 u_int16_t sc_rxtslot; /* retransmit counter */
258 u_int16_t sc_peer_mss; /* peer's MSS */
259 u_int16_t sc_wnd; /* advertised window */
260 u_int8_t sc_requested_s_scale:4,
261 sc_request_r_scale:4;
262 u_int8_t sc_flags;
263#define SCF_NOOPT 0x01 /* no TCP options */
264#define SCF_WINSCALE 0x02 /* negotiated window scaling */
265#define SCF_TIMESTAMP 0x04 /* negotiated timestamps */
266#define SCF_UNREACH 0x10 /* icmp unreachable received */
267#define SCF_SIGNATURE 0x20 /* send MD5 digests */
268#define SCF_SACK 0x80 /* send SACK option */
269 TAILQ_ENTRY(syncache) sc_hash;
270 TAILQ_ENTRY(syncache) sc_timerq;
271};
272
273struct syncache_head {
274 TAILQ_HEAD(, syncache) sch_bucket;
275 u_int sch_length;
276};
277#else
278struct in_conninfo;
279#endif /* _NETINET_IN_PCB_H_ */
280
281struct hc_metrics_lite { /* must stay in sync with hc_metrics */
282 u_long rmx_mtu; /* MTU for this path */
283 u_long rmx_ssthresh; /* outbound gateway buffer limit */
284 u_long rmx_rtt; /* estimated round trip time */
285 u_long rmx_rttvar; /* estimated rtt variance */
286 u_long rmx_bandwidth; /* estimated bandwidth */
287 u_long rmx_cwnd; /* congestion window */
288 u_long rmx_sendpipe; /* outbound delay-bandwidth product */
289 u_long rmx_recvpipe; /* inbound delay-bandwidth product */
290};
291
292struct tcptw {
293 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
294 tcp_seq snd_nxt;
295 tcp_seq rcv_nxt;
296 tcp_seq iss;
297 tcp_seq irs;
298 u_short last_win; /* cached window value */
299 u_short tw_so_options; /* copy of so_options */
300 struct ucred *tw_cred; /* user credentials */
301 u_long t_recent;
302 u_long t_starttime;
303 int tw_time;
304 LIST_ENTRY(tcptw) tw_2msl;
305};
306
307#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
308#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
309#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
310
311/*
312 * The smoothed round-trip time and estimated variance
313 * are stored as fixed point numbers scaled by the values below.
314 * For convenience, these scales are also used in smoothing the average
315 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
316 * With these scales, srtt has 3 bits to the right of the binary point,
317 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
318 * binary point, and is smoothed with an ALPHA of 0.75.
319 */
320#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
321#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
322#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
323#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
324#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
325
326/*
327 * The initial retransmission should happen at rtt + 4 * rttvar.
328 * Because of the way we do the smoothing, srtt and rttvar
329 * will each average +1/2 tick of bias. When we compute
330 * the retransmit timer, we want 1/2 tick of rounding and
331 * 1 extra tick because of +-1/2 tick uncertainty in the
332 * firing of the timer. The bias will give us exactly the
333 * 1.5 tick we need. But, because the bias is
334 * statistical, we have to test that we don't drop below
335 * the minimum feasible timer (which is 2 ticks).
336 * This version of the macro adapted from a paper by Lawrence
337 * Brakmo and Larry Peterson which outlines a problem caused
338 * by insufficient precision in the original implementation,
339 * which results in inappropriately large RTO values for very
340 * fast networks.
341 */
342#define TCP_REXMTVAL(tp) \
343 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
344 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
345
346/*
347 * TCP statistics.
348 * Many of these should be kept per connection,
349 * but that's inconvenient at the moment.
350 */
351struct tcpstat {
352 u_long tcps_connattempt; /* connections initiated */
353 u_long tcps_accepts; /* connections accepted */
354 u_long tcps_connects; /* connections established */
355 u_long tcps_drops; /* connections dropped */
356 u_long tcps_conndrops; /* embryonic connections dropped */
357 u_long tcps_minmssdrops; /* average minmss too low drops */
358 u_long tcps_closed; /* conn. closed (includes drops) */
359 u_long tcps_segstimed; /* segs where we tried to get rtt */
360 u_long tcps_rttupdated; /* times we succeeded */
361 u_long tcps_delack; /* delayed acks sent */
362 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
363 u_long tcps_rexmttimeo; /* retransmit timeouts */
364 u_long tcps_persisttimeo; /* persist timeouts */
365 u_long tcps_keeptimeo; /* keepalive timeouts */
366 u_long tcps_keepprobe; /* keepalive probes sent */
367 u_long tcps_keepdrops; /* connections dropped in keepalive */
368
369 u_long tcps_sndtotal; /* total packets sent */
370 u_long tcps_sndpack; /* data packets sent */
371 u_long tcps_sndbyte; /* data bytes sent */
372 u_long tcps_sndrexmitpack; /* data packets retransmitted */
373 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
374 u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */
375 u_long tcps_sndacks; /* ack-only packets sent */
376 u_long tcps_sndprobe; /* window probes sent */
377 u_long tcps_sndurg; /* packets sent with URG only */
378 u_long tcps_sndwinup; /* window update-only packets sent */
379 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
380
381 u_long tcps_rcvtotal; /* total packets received */
382 u_long tcps_rcvpack; /* packets received in sequence */
383 u_long tcps_rcvbyte; /* bytes received in sequence */
384 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
385 u_long tcps_rcvbadoff; /* packets received with bad offset */
386 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
387 u_long tcps_rcvshort; /* packets received too short */
388 u_long tcps_rcvduppack; /* duplicate-only packets received */
389 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
390 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
391 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
392 u_long tcps_rcvoopack; /* out-of-order packets received */
393 u_long tcps_rcvoobyte; /* out-of-order bytes received */
394 u_long tcps_rcvpackafterwin; /* packets with data after window */
395 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
396 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
397 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
398 u_long tcps_rcvdupack; /* rcvd duplicate acks */
399 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
400 u_long tcps_rcvackpack; /* rcvd ack packets */
401 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
402 u_long tcps_rcvwinupd; /* rcvd window update packets */
403 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
404 u_long tcps_predack; /* times hdr predict ok for acks */
405 u_long tcps_preddat; /* times hdr predict ok for data pkts */
406 u_long tcps_pcbcachemiss;
407 u_long tcps_cachedrtt; /* times cached RTT in route updated */
408 u_long tcps_cachedrttvar; /* times cached rttvar updated */
409 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
410 u_long tcps_usedrtt; /* times RTT initialized from route */
411 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
412 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
413 u_long tcps_persistdrop; /* timeout in persist state */
414 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
415 u_long tcps_mturesent; /* resends due to MTU discovery */
416 u_long tcps_listendrop; /* listen queue overflows */
417 u_long tcps_badrst; /* ignored RSTs in the window */
418
419 u_long tcps_sc_added; /* entry added to syncache */
420 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */
421 u_long tcps_sc_dupsyn; /* duplicate SYN packet */
422 u_long tcps_sc_dropped; /* could not reply to packet */
423 u_long tcps_sc_completed; /* successful extraction of entry */
424 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
425 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */
426 u_long tcps_sc_reset; /* RST removed entry from syncache */
427 u_long tcps_sc_stale; /* timed out or listen socket gone */
428 u_long tcps_sc_aborted; /* syncache entry aborted */
429 u_long tcps_sc_badack; /* removed due to bad ACK */
430 u_long tcps_sc_unreach; /* ICMP unreachable received */
431 u_long tcps_sc_zonefail; /* zalloc() failed */
432 u_long tcps_sc_sendcookie; /* SYN cookie sent */
433 u_long tcps_sc_recvcookie; /* SYN cookie received */
434
435 u_long tcps_hc_added; /* entry added to hostcache */
436 u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
437
438 /* SACK related stats */
439 u_long tcps_sack_recovery_episode; /* SACK recovery episodes */
440 u_long tcps_sack_rexmits; /* SACK rexmit segments */
441 u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
442 u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */
443 u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */
444 u_long tcps_sack_sboverflow; /* times scoreboard overflowed */
445};
446
447/*
448 * TCB structure exported to user-land via sysctl(3).
449 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
450 * included. Not all of our clients do.
451 */
452#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
453struct xtcpcb {
454 size_t xt_len;
455 struct inpcb xt_inp;
456 struct tcpcb xt_tp;
457 struct xsocket xt_socket;
458 u_quad_t xt_alignment_hack;
459};
460#endif
461
462/*
463 * Names for TCP sysctl objects
464 */
465#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
466#define TCPCTL_MSSDFLT 3 /* MSS default */
467#define TCPCTL_STATS 4 /* statistics (read-only) */
468#define TCPCTL_RTTDFLT 5 /* default RTT estimate */
469#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
470#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
471#define TCPCTL_SENDSPACE 8 /* send buffer space */
472#define TCPCTL_RECVSPACE 9 /* receive buffer space */
473#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
474#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
475#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
476#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
477#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
478#define TCPCTL_DROP 15 /* drop tcp connection */
479#define TCPCTL_MAXID 16
480
481#define TCPCTL_NAMES { \
482 { 0, 0 }, \
483 { "rfc1323", CTLTYPE_INT }, \
484 { "mssdflt", CTLTYPE_INT }, \
485 { "stats", CTLTYPE_STRUCT }, \
486 { "rttdflt", CTLTYPE_INT }, \
487 { "keepidle", CTLTYPE_INT }, \
488 { "keepintvl", CTLTYPE_INT }, \
489 { "sendspace", CTLTYPE_INT }, \
490 { "recvspace", CTLTYPE_INT }, \
491 { "keepinit", CTLTYPE_INT }, \
492 { "pcblist", CTLTYPE_STRUCT }, \
493 { "delacktime", CTLTYPE_INT }, \
494 { "v6mssdflt", CTLTYPE_INT }, \
495 { "maxid", CTLTYPE_INT }, \
496}
497
498
499#ifdef _KERNEL
500#ifdef SYSCTL_DECL
501SYSCTL_DECL(_net_inet_tcp);
502SYSCTL_DECL(_net_inet_tcp_sack);
503#endif
504
505extern struct inpcbhead tcb; /* head of queue of active tcpcb's */
506extern struct inpcbinfo tcbinfo;
507extern struct tcpstat tcpstat; /* tcp statistics */
508extern int tcp_mssdflt; /* XXX */
509extern int tcp_minmss;
510extern int tcp_minmssoverload;
511extern int tcp_delack_enabled;
512extern int tcp_do_newreno;
513extern int path_mtu_discovery;
514extern int ss_fltsz;
515extern int ss_fltsz_local;
516
517extern int tcp_do_sack; /* SACK enabled/disabled */
518
519struct tcpcb *
520 tcp_close(struct tcpcb *);
521void tcp_twstart(struct tcpcb *);
522int tcp_twrecycleable(struct tcptw *tw);
523struct tcptw *
524 tcp_twclose(struct tcptw *_tw, int _reuse);
525void tcp_ctlinput(int, struct sockaddr *, void *);
526int tcp_ctloutput(struct socket *, struct sockopt *);
527struct tcpcb *
528 tcp_drop(struct tcpcb *, int);
529void tcp_drain(void);
530void tcp_fasttimo(void);
531void tcp_init(void);
532void tcp_fini(void *);
533void tcp_reass_init(void);
534void tcp_input(struct mbuf *, int);
535u_long tcp_maxmtu(struct in_conninfo *);
536u_long tcp_maxmtu6(struct in_conninfo *);
537void tcp_mss(struct tcpcb *, int);
538int tcp_mssopt(struct in_conninfo *);
539struct inpcb *
540 tcp_drop_syn_sent(struct inpcb *, int);
541struct inpcb *
542 tcp_mtudisc(struct inpcb *, int);
543struct tcpcb *
544 tcp_newtcpcb(struct inpcb *);
545int tcp_output(struct tcpcb *);
546void tcp_respond(struct tcpcb *, void *,
547 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
548int tcp_twrespond(struct tcptw *, int);
549void tcp_setpersist(struct tcpcb *);
550#ifdef TCP_SIGNATURE
551int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
552#endif
553void tcp_slowtimo(void);
554struct tcptemp *
555 tcpip_maketemplate(struct inpcb *);
556void tcpip_fillheaders(struct inpcb *, void *, void *);
557struct tcpcb *
558 tcp_timers(struct tcpcb *, int);
559void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
560void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
561void syncache_init(void);
562void syncache_unreach(struct in_conninfo *, struct tcphdr *);
563int syncache_expand(struct in_conninfo *, struct tcphdr *,
564 struct socket **, struct mbuf *);
565int syncache_add(struct in_conninfo *, struct tcpopt *,
566 struct tcphdr *, struct socket **, struct mbuf *);
567void syncache_chkrst(struct in_conninfo *, struct tcphdr *);
568void syncache_badack(struct in_conninfo *);
569/*
570 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
571 */
572void tcp_hc_init(void);
573void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
574u_long tcp_hc_getmtu(struct in_conninfo *);
575void tcp_hc_updatemtu(struct in_conninfo *, u_long);
576void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
577
578extern struct pr_usrreqs tcp_usrreqs;
579extern u_long tcp_sendspace;
580extern u_long tcp_recvspace;
581tcp_seq tcp_new_isn(struct tcpcb *);
582
583void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
584void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
585void tcp_clean_sackreport(struct tcpcb *tp);
586void tcp_sack_adjust(struct tcpcb *tp);
587struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
588void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
589void tcp_free_sackholes(struct tcpcb *tp);
590int tcp_newreno(struct tcpcb *, struct tcphdr *);
591u_long tcp_seq_subtract(u_long, u_long );
592
593#endif /* _KERNEL */
594
595#endif /* _NETINET_TCP_VAR_H_ */
2 * Copyright (c) 1982, 1986, 1993, 1994, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
30 * $FreeBSD: head/sys/netinet/tcp_var.h 155767 2006-02-16 19:38:07Z andre $
31 */
32
33#ifndef _NETINET_TCP_VAR_H_
34#define _NETINET_TCP_VAR_H_
35
36#include <netinet/tcp.h>
37
38/*
39 * Kernel variables for tcp.
40 */
41extern int tcp_do_rfc1323;
42
43/* TCP segment queue entry */
44struct tseg_qent {
45 LIST_ENTRY(tseg_qent) tqe_q;
46 int tqe_len; /* TCP segment data length */
47 struct tcphdr *tqe_th; /* a pointer to tcp header */
48 struct mbuf *tqe_m; /* mbuf contains packet */
49};
50LIST_HEAD(tsegqe_head, tseg_qent);
51extern int tcp_reass_qsize;
52extern struct uma_zone *tcp_reass_zone;
53
54struct sackblk {
55 tcp_seq start; /* start seq no. of sack block */
56 tcp_seq end; /* end seq no. */
57};
58
59struct sackhole {
60 tcp_seq start; /* start seq no. of hole */
61 tcp_seq end; /* end seq no. */
62 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
63 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
64};
65
66struct sackhint {
67 struct sackhole *nexthole;
68 int sack_bytes_rexmit;
69};
70
71struct tcptemp {
72 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
73 struct tcphdr tt_t;
74};
75
76#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
77
78/*
79 * Tcp control block, one per tcp; fields:
80 * Organized for 16 byte cacheline efficiency.
81 */
82struct tcpcb {
83 struct tsegqe_head t_segq; /* segment reassembly queue */
84 int t_segqlen; /* segment reassembly queue length */
85 int t_dupacks; /* consecutive dup acks recd */
86
87 struct callout *tt_rexmt; /* retransmit timer */
88 struct callout *tt_persist; /* retransmit persistence */
89 struct callout *tt_keep; /* keepalive */
90 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */
91 struct callout *tt_delack; /* delayed ACK timer */
92
93 struct inpcb *t_inpcb; /* back pointer to internet pcb */
94 int t_state; /* state of this connection */
95 u_int t_flags;
96#define TF_ACKNOW 0x000001 /* ack peer immediately */
97#define TF_DELACK 0x000002 /* ack, but try to delay it */
98#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */
99#define TF_NOOPT 0x000008 /* don't use tcp options */
100#define TF_SENTFIN 0x000010 /* have sent FIN */
101#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */
102#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */
103#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */
104#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */
105#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */
106#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
107#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
108#define TF_NOPUSH 0x001000 /* don't push */
109#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */
110#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
111#define TF_LASTIDLE 0x040000 /* connection was previously idle */
112#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */
113#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */
114#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */
115#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
116#define TF_FORCEDATA 0x800000 /* force out a byte */
117
118 tcp_seq snd_una; /* send unacknowledged */
119 tcp_seq snd_max; /* highest sequence number sent;
120 * used to recognize retransmits
121 */
122 tcp_seq snd_nxt; /* send next */
123 tcp_seq snd_up; /* send urgent pointer */
124
125 tcp_seq snd_wl1; /* window update seg seq number */
126 tcp_seq snd_wl2; /* window update seg ack number */
127 tcp_seq iss; /* initial send sequence number */
128 tcp_seq irs; /* initial receive sequence number */
129
130 tcp_seq rcv_nxt; /* receive next */
131 tcp_seq rcv_adv; /* advertised window */
132 u_long rcv_wnd; /* receive window */
133 tcp_seq rcv_up; /* receive urgent pointer */
134
135 u_long snd_wnd; /* send window */
136 u_long snd_cwnd; /* congestion-controlled window */
137 u_long snd_bwnd; /* bandwidth-controlled window */
138 u_long snd_ssthresh; /* snd_cwnd size threshold for
139 * for slow start exponential to
140 * linear switch
141 */
142 u_long snd_bandwidth; /* calculated bandwidth or 0 */
143 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
144
145 u_int t_maxopd; /* mss plus options */
146
147 u_long t_rcvtime; /* inactivity time */
148 u_long t_starttime; /* time connection was established */
149 int t_rtttime; /* round trip time */
150 tcp_seq t_rtseq; /* sequence number being timed */
151
152 int t_bw_rtttime; /* used for bandwidth calculation */
153 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */
154
155 int t_rxtcur; /* current retransmit value (ticks) */
156 u_int t_maxseg; /* maximum segment size */
157 int t_srtt; /* smoothed round-trip time */
158 int t_rttvar; /* variance in round-trip time */
159
160 int t_rxtshift; /* log(2) of rexmt exp. backoff */
161 u_int t_rttmin; /* minimum rtt allowed */
162 u_int t_rttbest; /* best rtt we've seen */
163 u_long t_rttupdated; /* number of times rtt sampled */
164 u_long max_sndwnd; /* largest window peer has offered */
165
166 int t_softerror; /* possible error not yet reported */
167/* out-of-band data */
168 char t_oobflags; /* have some */
169 char t_iobc; /* input character */
170#define TCPOOB_HAVEDATA 0x01
171#define TCPOOB_HADDATA 0x02
172/* RFC 1323 variables */
173 u_char snd_scale; /* window scaling for send window */
174 u_char rcv_scale; /* window scaling for recv window */
175 u_char request_r_scale; /* pending window scaling */
176 u_char requested_s_scale;
177 u_long ts_recent; /* timestamp echo data */
178
179 u_long ts_recent_age; /* when last updated */
180 tcp_seq last_ack_sent;
181/* experimental */
182 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
183 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
184 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
185 u_long t_badrxtwin; /* window for retransmit recovery */
186 u_char snd_limited; /* segments limited transmitted */
187/* anti DoS counters */
188 u_long rcv_second; /* start of interval second */
189 u_long rcv_pps; /* received packets per second */
190 u_long rcv_byps; /* received bytes per second */
191/* SACK related state */
192 int sack_enable; /* enable SACK for this connection */
193 int snd_numholes; /* number of holes seen by sender */
194 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
195 /* SACK scoreboard (sorted) */
196 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
197 int rcv_numsacks; /* # distinct sack blks present */
198 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
199 tcp_seq sack_newdata; /* New data xmitted in this recovery
200 episode starts at this seq number */
201 struct sackhint sackhint; /* SACK scoreboard hint */
202 int t_rttlow; /* smallest observerved RTT */
203};
204
205#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY)
206#define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY
207#define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY
208
209#ifdef TCP_SIGNATURE
210/*
211 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
212 * for SADB verification and lookup.
213 */
214#define TCP_SIGLEN 16 /* length of computed digest in bytes */
215#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
216#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
217/*
218 * Only a single SA per host may be specified at this time. An SPI is
219 * needed in order for the KEY_ALLOCSA() lookup to work.
220 */
221#define TCP_SIG_SPI 0x1000
222#endif /* TCP_SIGNATURE */
223
224/*
225 * Structure to hold TCP options that are only used during segment
226 * processing (in tcp_input), but not held in the tcpcb.
227 * It's basically used to reduce the number of parameters
228 * to tcp_dooptions.
229 */
230struct tcpopt {
231 u_long to_flags; /* which options are present */
232#define TOF_TS 0x0001 /* timestamp */
233#define TOF_MSS 0x0010
234#define TOF_SCALE 0x0020
235#define TOF_SIGNATURE 0x0040 /* signature option present */
236#define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */
237#define TOF_SACK 0x0100 /* Peer sent SACK option */
238 u_int32_t to_tsval;
239 u_int32_t to_tsecr;
240 u_int16_t to_mss;
241 u_int8_t to_requested_s_scale;
242 u_int8_t to_nsacks; /* number of SACK blocks */
243 u_char *to_sacks; /* pointer to the first SACK blocks */
244};
245
246#ifdef _NETINET_IN_PCB_H_
247struct syncache {
248 inp_gen_t sc_inp_gencnt; /* pointer check */
249 struct tcpcb *sc_tp; /* tcb for listening socket */
250 struct mbuf *sc_ipopts; /* source route */
251 struct in_conninfo sc_inc; /* addresses */
252 u_int32_t sc_tsrecent;
253 u_int32_t sc_flowlabel; /* IPv6 flowlabel */
254 tcp_seq sc_irs; /* seq from peer */
255 tcp_seq sc_iss; /* our ISS */
256 u_long sc_rxttime; /* retransmit time */
257 u_int16_t sc_rxtslot; /* retransmit counter */
258 u_int16_t sc_peer_mss; /* peer's MSS */
259 u_int16_t sc_wnd; /* advertised window */
260 u_int8_t sc_requested_s_scale:4,
261 sc_request_r_scale:4;
262 u_int8_t sc_flags;
263#define SCF_NOOPT 0x01 /* no TCP options */
264#define SCF_WINSCALE 0x02 /* negotiated window scaling */
265#define SCF_TIMESTAMP 0x04 /* negotiated timestamps */
266#define SCF_UNREACH 0x10 /* icmp unreachable received */
267#define SCF_SIGNATURE 0x20 /* send MD5 digests */
268#define SCF_SACK 0x80 /* send SACK option */
269 TAILQ_ENTRY(syncache) sc_hash;
270 TAILQ_ENTRY(syncache) sc_timerq;
271};
272
273struct syncache_head {
274 TAILQ_HEAD(, syncache) sch_bucket;
275 u_int sch_length;
276};
277#else
278struct in_conninfo;
279#endif /* _NETINET_IN_PCB_H_ */
280
281struct hc_metrics_lite { /* must stay in sync with hc_metrics */
282 u_long rmx_mtu; /* MTU for this path */
283 u_long rmx_ssthresh; /* outbound gateway buffer limit */
284 u_long rmx_rtt; /* estimated round trip time */
285 u_long rmx_rttvar; /* estimated rtt variance */
286 u_long rmx_bandwidth; /* estimated bandwidth */
287 u_long rmx_cwnd; /* congestion window */
288 u_long rmx_sendpipe; /* outbound delay-bandwidth product */
289 u_long rmx_recvpipe; /* inbound delay-bandwidth product */
290};
291
292struct tcptw {
293 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
294 tcp_seq snd_nxt;
295 tcp_seq rcv_nxt;
296 tcp_seq iss;
297 tcp_seq irs;
298 u_short last_win; /* cached window value */
299 u_short tw_so_options; /* copy of so_options */
300 struct ucred *tw_cred; /* user credentials */
301 u_long t_recent;
302 u_long t_starttime;
303 int tw_time;
304 LIST_ENTRY(tcptw) tw_2msl;
305};
306
307#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
308#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
309#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
310
311/*
312 * The smoothed round-trip time and estimated variance
313 * are stored as fixed point numbers scaled by the values below.
314 * For convenience, these scales are also used in smoothing the average
315 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
316 * With these scales, srtt has 3 bits to the right of the binary point,
317 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
318 * binary point, and is smoothed with an ALPHA of 0.75.
319 */
320#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
321#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
322#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
323#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
324#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
325
326/*
327 * The initial retransmission should happen at rtt + 4 * rttvar.
328 * Because of the way we do the smoothing, srtt and rttvar
329 * will each average +1/2 tick of bias. When we compute
330 * the retransmit timer, we want 1/2 tick of rounding and
331 * 1 extra tick because of +-1/2 tick uncertainty in the
332 * firing of the timer. The bias will give us exactly the
333 * 1.5 tick we need. But, because the bias is
334 * statistical, we have to test that we don't drop below
335 * the minimum feasible timer (which is 2 ticks).
336 * This version of the macro adapted from a paper by Lawrence
337 * Brakmo and Larry Peterson which outlines a problem caused
338 * by insufficient precision in the original implementation,
339 * which results in inappropriately large RTO values for very
340 * fast networks.
341 */
342#define TCP_REXMTVAL(tp) \
343 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
344 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
345
346/*
347 * TCP statistics.
348 * Many of these should be kept per connection,
349 * but that's inconvenient at the moment.
350 */
351struct tcpstat {
352 u_long tcps_connattempt; /* connections initiated */
353 u_long tcps_accepts; /* connections accepted */
354 u_long tcps_connects; /* connections established */
355 u_long tcps_drops; /* connections dropped */
356 u_long tcps_conndrops; /* embryonic connections dropped */
357 u_long tcps_minmssdrops; /* average minmss too low drops */
358 u_long tcps_closed; /* conn. closed (includes drops) */
359 u_long tcps_segstimed; /* segs where we tried to get rtt */
360 u_long tcps_rttupdated; /* times we succeeded */
361 u_long tcps_delack; /* delayed acks sent */
362 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
363 u_long tcps_rexmttimeo; /* retransmit timeouts */
364 u_long tcps_persisttimeo; /* persist timeouts */
365 u_long tcps_keeptimeo; /* keepalive timeouts */
366 u_long tcps_keepprobe; /* keepalive probes sent */
367 u_long tcps_keepdrops; /* connections dropped in keepalive */
368
369 u_long tcps_sndtotal; /* total packets sent */
370 u_long tcps_sndpack; /* data packets sent */
371 u_long tcps_sndbyte; /* data bytes sent */
372 u_long tcps_sndrexmitpack; /* data packets retransmitted */
373 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
374 u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */
375 u_long tcps_sndacks; /* ack-only packets sent */
376 u_long tcps_sndprobe; /* window probes sent */
377 u_long tcps_sndurg; /* packets sent with URG only */
378 u_long tcps_sndwinup; /* window update-only packets sent */
379 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
380
381 u_long tcps_rcvtotal; /* total packets received */
382 u_long tcps_rcvpack; /* packets received in sequence */
383 u_long tcps_rcvbyte; /* bytes received in sequence */
384 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
385 u_long tcps_rcvbadoff; /* packets received with bad offset */
386 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
387 u_long tcps_rcvshort; /* packets received too short */
388 u_long tcps_rcvduppack; /* duplicate-only packets received */
389 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
390 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
391 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
392 u_long tcps_rcvoopack; /* out-of-order packets received */
393 u_long tcps_rcvoobyte; /* out-of-order bytes received */
394 u_long tcps_rcvpackafterwin; /* packets with data after window */
395 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
396 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
397 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
398 u_long tcps_rcvdupack; /* rcvd duplicate acks */
399 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
400 u_long tcps_rcvackpack; /* rcvd ack packets */
401 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
402 u_long tcps_rcvwinupd; /* rcvd window update packets */
403 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
404 u_long tcps_predack; /* times hdr predict ok for acks */
405 u_long tcps_preddat; /* times hdr predict ok for data pkts */
406 u_long tcps_pcbcachemiss;
407 u_long tcps_cachedrtt; /* times cached RTT in route updated */
408 u_long tcps_cachedrttvar; /* times cached rttvar updated */
409 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
410 u_long tcps_usedrtt; /* times RTT initialized from route */
411 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
412 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
413 u_long tcps_persistdrop; /* timeout in persist state */
414 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
415 u_long tcps_mturesent; /* resends due to MTU discovery */
416 u_long tcps_listendrop; /* listen queue overflows */
417 u_long tcps_badrst; /* ignored RSTs in the window */
418
419 u_long tcps_sc_added; /* entry added to syncache */
420 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */
421 u_long tcps_sc_dupsyn; /* duplicate SYN packet */
422 u_long tcps_sc_dropped; /* could not reply to packet */
423 u_long tcps_sc_completed; /* successful extraction of entry */
424 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
425 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */
426 u_long tcps_sc_reset; /* RST removed entry from syncache */
427 u_long tcps_sc_stale; /* timed out or listen socket gone */
428 u_long tcps_sc_aborted; /* syncache entry aborted */
429 u_long tcps_sc_badack; /* removed due to bad ACK */
430 u_long tcps_sc_unreach; /* ICMP unreachable received */
431 u_long tcps_sc_zonefail; /* zalloc() failed */
432 u_long tcps_sc_sendcookie; /* SYN cookie sent */
433 u_long tcps_sc_recvcookie; /* SYN cookie received */
434
435 u_long tcps_hc_added; /* entry added to hostcache */
436 u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
437
438 /* SACK related stats */
439 u_long tcps_sack_recovery_episode; /* SACK recovery episodes */
440 u_long tcps_sack_rexmits; /* SACK rexmit segments */
441 u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
442 u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */
443 u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */
444 u_long tcps_sack_sboverflow; /* times scoreboard overflowed */
445};
446
447/*
448 * TCB structure exported to user-land via sysctl(3).
449 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
450 * included. Not all of our clients do.
451 */
452#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
453struct xtcpcb {
454 size_t xt_len;
455 struct inpcb xt_inp;
456 struct tcpcb xt_tp;
457 struct xsocket xt_socket;
458 u_quad_t xt_alignment_hack;
459};
460#endif
461
462/*
463 * Names for TCP sysctl objects
464 */
465#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
466#define TCPCTL_MSSDFLT 3 /* MSS default */
467#define TCPCTL_STATS 4 /* statistics (read-only) */
468#define TCPCTL_RTTDFLT 5 /* default RTT estimate */
469#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
470#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
471#define TCPCTL_SENDSPACE 8 /* send buffer space */
472#define TCPCTL_RECVSPACE 9 /* receive buffer space */
473#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
474#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
475#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
476#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
477#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
478#define TCPCTL_DROP 15 /* drop tcp connection */
479#define TCPCTL_MAXID 16
480
481#define TCPCTL_NAMES { \
482 { 0, 0 }, \
483 { "rfc1323", CTLTYPE_INT }, \
484 { "mssdflt", CTLTYPE_INT }, \
485 { "stats", CTLTYPE_STRUCT }, \
486 { "rttdflt", CTLTYPE_INT }, \
487 { "keepidle", CTLTYPE_INT }, \
488 { "keepintvl", CTLTYPE_INT }, \
489 { "sendspace", CTLTYPE_INT }, \
490 { "recvspace", CTLTYPE_INT }, \
491 { "keepinit", CTLTYPE_INT }, \
492 { "pcblist", CTLTYPE_STRUCT }, \
493 { "delacktime", CTLTYPE_INT }, \
494 { "v6mssdflt", CTLTYPE_INT }, \
495 { "maxid", CTLTYPE_INT }, \
496}
497
498
499#ifdef _KERNEL
500#ifdef SYSCTL_DECL
501SYSCTL_DECL(_net_inet_tcp);
502SYSCTL_DECL(_net_inet_tcp_sack);
503#endif
504
505extern struct inpcbhead tcb; /* head of queue of active tcpcb's */
506extern struct inpcbinfo tcbinfo;
507extern struct tcpstat tcpstat; /* tcp statistics */
508extern int tcp_mssdflt; /* XXX */
509extern int tcp_minmss;
510extern int tcp_minmssoverload;
511extern int tcp_delack_enabled;
512extern int tcp_do_newreno;
513extern int path_mtu_discovery;
514extern int ss_fltsz;
515extern int ss_fltsz_local;
516
517extern int tcp_do_sack; /* SACK enabled/disabled */
518
519struct tcpcb *
520 tcp_close(struct tcpcb *);
521void tcp_twstart(struct tcpcb *);
522int tcp_twrecycleable(struct tcptw *tw);
523struct tcptw *
524 tcp_twclose(struct tcptw *_tw, int _reuse);
525void tcp_ctlinput(int, struct sockaddr *, void *);
526int tcp_ctloutput(struct socket *, struct sockopt *);
527struct tcpcb *
528 tcp_drop(struct tcpcb *, int);
529void tcp_drain(void);
530void tcp_fasttimo(void);
531void tcp_init(void);
532void tcp_fini(void *);
533void tcp_reass_init(void);
534void tcp_input(struct mbuf *, int);
535u_long tcp_maxmtu(struct in_conninfo *);
536u_long tcp_maxmtu6(struct in_conninfo *);
537void tcp_mss(struct tcpcb *, int);
538int tcp_mssopt(struct in_conninfo *);
539struct inpcb *
540 tcp_drop_syn_sent(struct inpcb *, int);
541struct inpcb *
542 tcp_mtudisc(struct inpcb *, int);
543struct tcpcb *
544 tcp_newtcpcb(struct inpcb *);
545int tcp_output(struct tcpcb *);
546void tcp_respond(struct tcpcb *, void *,
547 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
548int tcp_twrespond(struct tcptw *, int);
549void tcp_setpersist(struct tcpcb *);
550#ifdef TCP_SIGNATURE
551int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
552#endif
553void tcp_slowtimo(void);
554struct tcptemp *
555 tcpip_maketemplate(struct inpcb *);
556void tcpip_fillheaders(struct inpcb *, void *, void *);
557struct tcpcb *
558 tcp_timers(struct tcpcb *, int);
559void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
560void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
561void syncache_init(void);
562void syncache_unreach(struct in_conninfo *, struct tcphdr *);
563int syncache_expand(struct in_conninfo *, struct tcphdr *,
564 struct socket **, struct mbuf *);
565int syncache_add(struct in_conninfo *, struct tcpopt *,
566 struct tcphdr *, struct socket **, struct mbuf *);
567void syncache_chkrst(struct in_conninfo *, struct tcphdr *);
568void syncache_badack(struct in_conninfo *);
569/*
570 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
571 */
572void tcp_hc_init(void);
573void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
574u_long tcp_hc_getmtu(struct in_conninfo *);
575void tcp_hc_updatemtu(struct in_conninfo *, u_long);
576void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
577
578extern struct pr_usrreqs tcp_usrreqs;
579extern u_long tcp_sendspace;
580extern u_long tcp_recvspace;
581tcp_seq tcp_new_isn(struct tcpcb *);
582
583void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
584void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
585void tcp_clean_sackreport(struct tcpcb *tp);
586void tcp_sack_adjust(struct tcpcb *tp);
587struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
588void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
589void tcp_free_sackholes(struct tcpcb *tp);
590int tcp_newreno(struct tcpcb *, struct tcphdr *);
591u_long tcp_seq_subtract(u_long, u_long );
592
593#endif /* _KERNEL */
594
595#endif /* _NETINET_TCP_VAR_H_ */