2 * Copyright (c) 1982, 1986, 1993, 1994, 1995
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
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.
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
29 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
33 #ifndef _NETINET_TCP_VAR_H_
34 #define _NETINET_TCP_VAR_H_
36 #include <netinet/tcp.h>
42 * Kernel variables for tcp.
44 VNET_DECLARE(int, tcp_do_rfc1323);
45 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323)
49 /* TCP segment queue entry */
51 LIST_ENTRY(tseg_qent) tqe_q;
52 int tqe_len; /* TCP segment data length */
53 struct tcphdr *tqe_th; /* a pointer to tcp header */
54 struct mbuf *tqe_m; /* mbuf contains packet */
56 LIST_HEAD(tsegqe_head, tseg_qent);
59 tcp_seq start; /* start seq no. of sack block */
60 tcp_seq end; /* end seq no. */
64 tcp_seq start; /* start seq no. of hole */
65 tcp_seq end; /* end seq no. */
66 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
67 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
71 struct sackhole *nexthole;
72 int sack_bytes_rexmit;
73 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */
75 int ispare; /* explicit pad for 64bit alignment */
76 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */
80 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
84 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
86 /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
88 #define ND6_HINT(tp) \
90 if ((tp) && (tp)->t_inpcb && \
91 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \
92 nd6_nud_hint(NULL, NULL, 0); \
99 * Tcp control block, one per tcp; fields:
100 * Organized for 16 byte cacheline efficiency.
103 struct tsegqe_head t_segq; /* segment reassembly queue */
104 void *t_pspare[2]; /* new reassembly queue */
105 int t_segqlen; /* segment reassembly queue length */
106 int t_dupacks; /* consecutive dup acks recd */
108 struct tcp_timer *t_timers; /* All the TCP timers in one struct */
110 struct inpcb *t_inpcb; /* back pointer to internet pcb */
111 int t_state; /* state of this connection */
114 struct vnet *t_vnet; /* back pointer to parent vnet */
116 tcp_seq snd_una; /* send unacknowledged */
117 tcp_seq snd_max; /* highest sequence number sent;
118 * used to recognize retransmits
120 tcp_seq snd_nxt; /* send next */
121 tcp_seq snd_up; /* send urgent pointer */
123 tcp_seq snd_wl1; /* window update seg seq number */
124 tcp_seq snd_wl2; /* window update seg ack number */
125 tcp_seq iss; /* initial send sequence number */
126 tcp_seq irs; /* initial receive sequence number */
128 tcp_seq rcv_nxt; /* receive next */
129 tcp_seq rcv_adv; /* advertised window */
130 u_long rcv_wnd; /* receive window */
131 tcp_seq rcv_up; /* receive urgent pointer */
133 u_long snd_wnd; /* send window */
134 u_long snd_cwnd; /* congestion-controlled window */
135 u_long snd_spare1; /* unused */
136 u_long snd_ssthresh; /* snd_cwnd size threshold for
137 * for slow start exponential to
140 u_long snd_spare2; /* unused */
141 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
143 u_int t_maxopd; /* mss plus options */
145 u_int t_rcvtime; /* inactivity time */
146 u_int t_starttime; /* time connection was established */
147 u_int t_rtttime; /* RTT measurement start time */
148 tcp_seq t_rtseq; /* sequence number being timed */
150 u_int t_bw_spare1; /* unused */
151 tcp_seq t_bw_spare2; /* unused */
153 int t_rxtcur; /* current retransmit value (ticks) */
154 u_int t_maxseg; /* maximum segment size */
155 int t_srtt; /* smoothed round-trip time */
156 int t_rttvar; /* variance in round-trip time */
158 int t_rxtshift; /* log(2) of rexmt exp. backoff */
159 u_int t_rttmin; /* minimum rtt allowed */
160 u_int t_rttbest; /* best rtt we've seen */
161 u_long t_rttupdated; /* number of times rtt sampled */
162 u_long max_sndwnd; /* largest window peer has offered */
164 int t_softerror; /* possible error not yet reported */
165 /* out-of-band data */
166 char t_oobflags; /* have some */
167 char t_iobc; /* input character */
168 /* RFC 1323 variables */
169 u_char snd_scale; /* window scaling for send window */
170 u_char rcv_scale; /* window scaling for recv window */
171 u_char request_r_scale; /* pending window scaling */
172 u_int32_t ts_recent; /* timestamp echo data */
173 u_int ts_recent_age; /* when last updated */
174 u_int32_t ts_offset; /* our timestamp offset */
176 tcp_seq last_ack_sent;
178 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
179 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
180 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
181 int t_sndzerowin; /* zero-window updates sent */
182 u_int t_badrxtwin; /* window for retransmit recovery */
183 u_char snd_limited; /* segments limited transmitted */
184 /* SACK related state */
185 int snd_numholes; /* number of holes seen by sender */
186 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
187 /* SACK scoreboard (sorted) */
188 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
189 int rcv_numsacks; /* # distinct sack blks present */
190 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
191 tcp_seq sack_newdata; /* New data xmitted in this recovery
192 episode starts at this seq number */
193 struct sackhint sackhint; /* SACK scoreboard hint */
194 int t_rttlow; /* smallest observerved RTT */
195 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
196 int rfbuf_cnt; /* recv buffer autoscaling byte count */
197 struct toedev *tod; /* toedev handling this connection */
198 int t_sndrexmitpack; /* retransmit packets sent */
199 int t_rcvoopack; /* out-of-order packets received */
200 void *t_toe; /* TOE pcb pointer */
201 int t_bytes_acked; /* # bytes acked during current RTT */
202 struct cc_algo *cc_algo; /* congestion control algorithm */
203 struct cc_var *ccv; /* congestion control specific vars */
204 struct osd *osd; /* storage for Khelp module data */
206 u_int t_keepinit; /* time to establish connection */
207 u_int t_keepidle; /* time before keepalive probes begin */
208 u_int t_keepintvl; /* interval between keepalives */
209 u_int t_keepcnt; /* number of keepalives before close */
211 u_int t_tsomax; /* TSO total burst length limit in bytes */
212 u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */
213 u_int t_flags2; /* More tcpcb flags storage */
215 uint32_t t_ispare[6]; /* 5 UTO, 1 TBD */
216 #if defined(_KERNEL) && defined(TCP_RFC7413)
217 void *t_pspare2[3]; /* 1 TCP_SIGNATURE, 2 TBD */
218 unsigned int *t_tfo_pending; /* TCP Fast Open pending counter */
220 void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */
222 #if defined(_KERNEL) && defined(TCP_RFC7413)
223 uint64_t _pad[4]; /* 4 TBD (1-2 CC/RTT?) */
224 uint64_t t_tfo_cookie; /* TCP Fast Open cookie */
226 uint64_t _pad[5]; /* 5 TBD (1-2 CC/RTT?) */
228 uint32_t t_tsomaxsegcount; /* TSO maximum segment count */
229 uint32_t t_tsomaxsegsize; /* TSO maximum segment size in bytes */
233 * Flags and utility macros for the t_flags field.
235 #define TF_ACKNOW 0x000001 /* ack peer immediately */
236 #define TF_DELACK 0x000002 /* ack, but try to delay it */
237 #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */
238 #define TF_NOOPT 0x000008 /* don't use tcp options */
239 #define TF_SENTFIN 0x000010 /* have sent FIN */
240 #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */
241 #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */
242 #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */
243 #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */
244 #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */
245 #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
246 #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
247 #define TF_NOPUSH 0x001000 /* don't push */
248 #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */
249 #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */
250 #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
251 #define TF_LASTIDLE 0x040000 /* connection was previously idle */
252 #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */
253 #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */
254 #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */
255 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
256 #define TF_FORCEDATA 0x800000 /* force out a byte */
257 #define TF_TSO 0x1000000 /* TSO enabled on this connection */
258 #define TF_TOE 0x2000000 /* this connection is offloaded */
259 #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */
260 #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */
261 #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */
262 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */
263 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */
264 #define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */
266 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY)
267 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY
268 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY
270 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY)
271 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY
272 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY
274 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY))
275 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY)
276 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY)
278 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una)
281 * Flags for the t_oobflags field.
283 #define TCPOOB_HAVEDATA 0x01
284 #define TCPOOB_HADDATA 0x02
288 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
289 * for SADB verification and lookup.
291 #define TCP_SIGLEN 16 /* length of computed digest in bytes */
292 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
293 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
295 * Only a single SA per host may be specified at this time. An SPI is
296 * needed in order for the KEY_ALLOCSA() lookup to work.
298 #define TCP_SIG_SPI 0x1000
299 #endif /* TCP_SIGNATURE */
302 * Flags for PLPMTU handling, t_flags2
304 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */
305 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */
306 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */
309 * Structure to hold TCP options that are only used during segment
310 * processing (in tcp_input), but not held in the tcpcb.
311 * It's basically used to reduce the number of parameters
312 * to tcp_dooptions and tcp_addoptions.
313 * The binary order of the to_flags is relevant for packing of the
314 * options in tcp_addoptions.
317 u_int64_t to_flags; /* which options are present */
318 #define TOF_MSS 0x0001 /* maximum segment size */
319 #define TOF_SCALE 0x0002 /* window scaling */
320 #define TOF_SACKPERM 0x0004 /* SACK permitted */
321 #define TOF_TS 0x0010 /* timestamp */
322 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */
323 #define TOF_SACK 0x0080 /* Peer sent SACK option */
324 #define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */
325 #define TOF_MAXOPT 0x0200
326 u_int32_t to_tsval; /* new timestamp */
327 u_int32_t to_tsecr; /* reflected timestamp */
328 u_char *to_sacks; /* pointer to the first SACK blocks */
329 u_char *to_signature; /* pointer to the TCP-MD5 signature */
330 u_char *to_tfo_cookie; /* pointer to the TFO cookie */
331 u_int16_t to_mss; /* maximum segment size */
332 u_int8_t to_wscale; /* window scaling */
333 u_int8_t to_nsacks; /* number of SACK blocks */
334 u_int8_t to_tfo_len; /* TFO cookie length */
335 u_int32_t to_spare; /* UTO */
339 * Flags for tcp_dooptions.
341 #define TO_SYN 0x01 /* parse SYN-only options */
343 struct hc_metrics_lite { /* must stay in sync with hc_metrics */
344 u_long rmx_mtu; /* MTU for this path */
345 u_long rmx_ssthresh; /* outbound gateway buffer limit */
346 u_long rmx_rtt; /* estimated round trip time */
347 u_long rmx_rttvar; /* estimated rtt variance */
348 u_long rmx_bandwidth; /* estimated bandwidth */
349 u_long rmx_cwnd; /* congestion window */
350 u_long rmx_sendpipe; /* outbound delay-bandwidth product */
351 u_long rmx_recvpipe; /* inbound delay-bandwidth product */
355 * Used by tcp_maxmtu() to communicate interface specific features
356 * and limits at the time of connection setup.
361 u_int tsomaxsegcount;
365 #ifndef _NETINET_IN_PCB_H_
367 #endif /* _NETINET_IN_PCB_H_ */
370 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
375 u_short last_win; /* cached window value */
376 u_short tw_so_options; /* copy of so_options */
377 struct ucred *tw_cred; /* user credentials */
379 u_int32_t ts_offset; /* our timestamp offset */
382 TAILQ_ENTRY(tcptw) tw_2msl;
383 void *tw_pspare; /* TCP_SIGNATURE */
384 u_int *tw_spare; /* TCP_SIGNATURE */
387 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
388 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
389 #define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
392 * The smoothed round-trip time and estimated variance
393 * are stored as fixed point numbers scaled by the values below.
394 * For convenience, these scales are also used in smoothing the average
395 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
396 * With these scales, srtt has 3 bits to the right of the binary point,
397 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
398 * binary point, and is smoothed with an ALPHA of 0.75.
400 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
401 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
402 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
403 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
404 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
407 * The initial retransmission should happen at rtt + 4 * rttvar.
408 * Because of the way we do the smoothing, srtt and rttvar
409 * will each average +1/2 tick of bias. When we compute
410 * the retransmit timer, we want 1/2 tick of rounding and
411 * 1 extra tick because of +-1/2 tick uncertainty in the
412 * firing of the timer. The bias will give us exactly the
413 * 1.5 tick we need. But, because the bias is
414 * statistical, we have to test that we don't drop below
415 * the minimum feasible timer (which is 2 ticks).
416 * This version of the macro adapted from a paper by Lawrence
417 * Brakmo and Larry Peterson which outlines a problem caused
418 * by insufficient precision in the original implementation,
419 * which results in inappropriately large RTO values for very
422 #define TCP_REXMTVAL(tp) \
423 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
424 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
428 * Many of these should be kept per connection,
429 * but that's inconvenient at the moment.
432 uint64_t tcps_connattempt; /* connections initiated */
433 uint64_t tcps_accepts; /* connections accepted */
434 uint64_t tcps_connects; /* connections established */
435 uint64_t tcps_drops; /* connections dropped */
436 uint64_t tcps_conndrops; /* embryonic connections dropped */
437 uint64_t tcps_minmssdrops; /* average minmss too low drops */
438 uint64_t tcps_closed; /* conn. closed (includes drops) */
439 uint64_t tcps_segstimed; /* segs where we tried to get rtt */
440 uint64_t tcps_rttupdated; /* times we succeeded */
441 uint64_t tcps_delack; /* delayed acks sent */
442 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
443 uint64_t tcps_rexmttimeo; /* retransmit timeouts */
444 uint64_t tcps_persisttimeo; /* persist timeouts */
445 uint64_t tcps_keeptimeo; /* keepalive timeouts */
446 uint64_t tcps_keepprobe; /* keepalive probes sent */
447 uint64_t tcps_keepdrops; /* connections dropped in keepalive */
449 uint64_t tcps_sndtotal; /* total packets sent */
450 uint64_t tcps_sndpack; /* data packets sent */
451 uint64_t tcps_sndbyte; /* data bytes sent */
452 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */
453 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */
454 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */
455 uint64_t tcps_sndacks; /* ack-only packets sent */
456 uint64_t tcps_sndprobe; /* window probes sent */
457 uint64_t tcps_sndurg; /* packets sent with URG only */
458 uint64_t tcps_sndwinup; /* window update-only packets sent */
459 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
461 uint64_t tcps_rcvtotal; /* total packets received */
462 uint64_t tcps_rcvpack; /* packets received in sequence */
463 uint64_t tcps_rcvbyte; /* bytes received in sequence */
464 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */
465 uint64_t tcps_rcvbadoff; /* packets received with bad offset */
466 uint64_t tcps_rcvmemdrop; /* packets dropped for lack of memory */
467 uint64_t tcps_rcvshort; /* packets received too short */
468 uint64_t tcps_rcvduppack; /* duplicate-only packets received */
469 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */
470 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */
471 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
472 uint64_t tcps_rcvoopack; /* out-of-order packets received */
473 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */
474 uint64_t tcps_rcvpackafterwin; /* packets with data after window */
475 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */
476 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */
477 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */
478 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */
479 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */
480 uint64_t tcps_rcvackpack; /* rcvd ack packets */
481 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */
482 uint64_t tcps_rcvwinupd; /* rcvd window update packets */
483 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */
484 uint64_t tcps_predack; /* times hdr predict ok for acks */
485 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */
486 uint64_t tcps_pcbcachemiss;
487 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */
488 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */
489 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */
490 uint64_t tcps_usedrtt; /* times RTT initialized from route */
491 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */
492 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/
493 uint64_t tcps_persistdrop; /* timeout in persist state */
494 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */
495 uint64_t tcps_mturesent; /* resends due to MTU discovery */
496 uint64_t tcps_listendrop; /* listen queue overflows */
497 uint64_t tcps_badrst; /* ignored RSTs in the window */
499 uint64_t tcps_sc_added; /* entry added to syncache */
500 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */
501 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */
502 uint64_t tcps_sc_dropped; /* could not reply to packet */
503 uint64_t tcps_sc_completed; /* successful extraction of entry */
504 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */
505 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */
506 uint64_t tcps_sc_reset; /* RST removed entry from syncache */
507 uint64_t tcps_sc_stale; /* timed out or listen socket gone */
508 uint64_t tcps_sc_aborted; /* syncache entry aborted */
509 uint64_t tcps_sc_badack; /* removed due to bad ACK */
510 uint64_t tcps_sc_unreach; /* ICMP unreachable received */
511 uint64_t tcps_sc_zonefail; /* zalloc() failed */
512 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */
513 uint64_t tcps_sc_recvcookie; /* SYN cookie received */
515 uint64_t tcps_hc_added; /* entry added to hostcache */
516 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */
518 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */
520 /* SACK related stats */
521 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */
522 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */
523 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
524 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */
525 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */
526 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */
528 /* ECN related stats */
529 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */
530 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */
531 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */
532 uint64_t tcps_ecn_shs; /* ECN successful handshakes */
533 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */
535 /* TCP_SIGNATURE related stats */
536 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */
537 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */
538 uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */
539 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */
540 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */
542 uint64_t _pad[12]; /* 6 UTO, 6 TBD */
546 #include <sys/counter.h>
548 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */
550 * In-kernel consumers can use these accessor macros directly to update
553 #define TCPSTAT_ADD(name, val) \
554 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val))
555 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1)
558 * Kernel module consumers must use this accessor macro.
560 void kmod_tcpstat_inc(int statnum);
561 #define KMOD_TCPSTAT_INC(name) \
562 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t))
565 * TCP specific helper hook point identifiers.
567 #define HHOOK_TCP_EST_IN 0
568 #define HHOOK_TCP_EST_OUT 1
569 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT
571 struct tcp_hhook_data {
582 * TCB structure exported to user-land via sysctl(3).
583 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
584 * included. Not all of our clients do.
586 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
588 int tt_rexmt; /* retransmit timer */
589 int tt_persist; /* retransmit persistence */
590 int tt_keep; /* keepalive */
591 int tt_2msl; /* 2*msl TIME_WAIT timer */
592 int tt_delack; /* delayed ACK timer */
593 int t_rcvtime; /* Time since last packet received */
599 struct xsocket xt_socket;
600 struct xtcp_timer xt_timer;
601 u_quad_t xt_alignment_hack;
606 * Names for TCP sysctl objects
608 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
609 #define TCPCTL_MSSDFLT 3 /* MSS default */
610 #define TCPCTL_STATS 4 /* statistics (read-only) */
611 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */
612 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
613 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
614 #define TCPCTL_SENDSPACE 8 /* send buffer space */
615 #define TCPCTL_RECVSPACE 9 /* receive buffer space */
616 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
617 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
618 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
619 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
620 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
621 #define TCPCTL_DROP 15 /* drop tcp connection */
622 #define TCPCTL_MAXID 16
623 #define TCPCTL_FINWAIT2_TIMEOUT 17
627 SYSCTL_DECL(_net_inet_tcp);
628 SYSCTL_DECL(_net_inet_tcp_sack);
629 MALLOC_DECLARE(M_TCPLOG);
632 VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */
633 VNET_DECLARE(struct inpcbinfo, tcbinfo);
634 extern int tcp_log_in_vain;
635 VNET_DECLARE(int, tcp_mssdflt); /* XXX */
636 VNET_DECLARE(int, tcp_minmss);
637 VNET_DECLARE(int, tcp_delack_enabled);
638 VNET_DECLARE(int, tcp_do_rfc3390);
639 VNET_DECLARE(int, tcp_do_initcwnd10);
640 VNET_DECLARE(int, tcp_sendspace);
641 VNET_DECLARE(int, tcp_recvspace);
642 VNET_DECLARE(int, path_mtu_discovery);
643 VNET_DECLARE(int, tcp_do_rfc3465);
644 VNET_DECLARE(int, tcp_abc_l_var);
645 #define V_tcb VNET(tcb)
646 #define V_tcbinfo VNET(tcbinfo)
647 #define V_tcp_mssdflt VNET(tcp_mssdflt)
648 #define V_tcp_minmss VNET(tcp_minmss)
649 #define V_tcp_delack_enabled VNET(tcp_delack_enabled)
650 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390)
651 #define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10)
652 #define V_tcp_sendspace VNET(tcp_sendspace)
653 #define V_tcp_recvspace VNET(tcp_recvspace)
654 #define V_path_mtu_discovery VNET(path_mtu_discovery)
655 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465)
656 #define V_tcp_abc_l_var VNET(tcp_abc_l_var)
658 VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */
659 VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */
660 #define V_tcp_do_sack VNET(tcp_do_sack)
661 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail)
663 VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */
664 VNET_DECLARE(int, tcp_ecn_maxretries);
665 #define V_tcp_do_ecn VNET(tcp_do_ecn)
666 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries)
668 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]);
669 #define V_tcp_hhh VNET(tcp_hhh)
671 int tcp_addoptions(struct tcpopt *, u_char *);
672 int tcp_ccalgounload(struct cc_algo *unload_algo);
674 tcp_close(struct tcpcb *);
675 void tcp_discardcb(struct tcpcb *);
676 void tcp_twstart(struct tcpcb *);
678 int tcp_twrecycleable(struct tcptw *tw);
680 void tcp_twclose(struct tcptw *, int);
681 void tcp_ctlinput(int, struct sockaddr *, void *);
682 int tcp_ctloutput(struct socket *, struct sockopt *);
684 tcp_drop(struct tcpcb *, int);
685 void tcp_drain(void);
688 void tcp_destroy(void);
690 void tcp_fini(void *);
691 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
693 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *,
695 int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *);
696 void tcp_reass_global_init(void);
697 void tcp_reass_flush(struct tcpcb *);
698 void tcp_input(struct mbuf *, int);
699 u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *);
700 u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *);
701 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *,
703 void tcp_mss(struct tcpcb *, int);
704 int tcp_mssopt(struct in_conninfo *);
706 tcp_drop_syn_sent(struct inpcb *, int);
708 tcp_mtudisc(struct inpcb *, int);
710 tcp_newtcpcb(struct inpcb *);
711 int tcp_output(struct tcpcb *);
712 void tcp_state_change(struct tcpcb *, int);
713 void tcp_respond(struct tcpcb *, void *,
714 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
715 void tcp_tw_init(void);
717 void tcp_tw_destroy(void);
719 void tcp_tw_zone_change(void);
720 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
722 void tcp_setpersist(struct tcpcb *);
724 int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
725 int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *,
726 struct tcphdr *, u_int);
728 void tcp_slowtimo(void);
730 tcpip_maketemplate(struct inpcb *);
731 void tcpip_fillheaders(struct inpcb *, void *, void *);
732 void tcp_timer_activate(struct tcpcb *, uint32_t, u_int);
733 int tcp_timer_active(struct tcpcb *, uint32_t);
734 void tcp_timer_stop(struct tcpcb *, uint32_t);
735 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
737 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
739 void tcp_hc_init(void);
741 void tcp_hc_destroy(void);
743 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
744 u_long tcp_hc_getmtu(struct in_conninfo *);
745 void tcp_hc_updatemtu(struct in_conninfo *, u_long);
746 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
748 extern struct pr_usrreqs tcp_usrreqs;
749 tcp_seq tcp_new_isn(struct tcpcb *);
751 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
752 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
753 void tcp_clean_sackreport(struct tcpcb *tp);
754 void tcp_sack_adjust(struct tcpcb *tp);
755 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
756 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
757 void tcp_free_sackholes(struct tcpcb *tp);
758 int tcp_newreno(struct tcpcb *, struct tcphdr *);
759 u_long tcp_seq_subtract(u_long, u_long );
761 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type);
764 tcp_fields_to_host(struct tcphdr *th)
767 th->th_seq = ntohl(th->th_seq);
768 th->th_ack = ntohl(th->th_ack);
769 th->th_win = ntohs(th->th_win);
770 th->th_urp = ntohs(th->th_urp);
775 tcp_fields_to_net(struct tcphdr *th)
778 th->th_seq = htonl(th->th_seq);
779 th->th_ack = htonl(th->th_ack);
780 th->th_win = htons(th->th_win);
781 th->th_urp = htons(th->th_urp);
786 #endif /* _NETINET_TCP_VAR_H_ */