tcp_var.h revision 317375
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: stable/10/sys/netinet/tcp_var.h 317375 2017-04-24 16:31:28Z smh $ 31 */ 32 33#ifndef _NETINET_TCP_VAR_H_ 34#define _NETINET_TCP_VAR_H_ 35 36#include <netinet/tcp.h> 37 38#ifdef _KERNEL 39#include <net/vnet.h> 40 41/* 42 * Kernel variables for tcp. 43 */ 44VNET_DECLARE(int, tcp_do_rfc1323); 45#define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 46 47#endif /* _KERNEL */ 48 49/* TCP segment queue entry */ 50struct tseg_qent { 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 */ 55}; 56LIST_HEAD(tsegqe_head, tseg_qent); 57 58struct sackblk { 59 tcp_seq start; /* start seq no. of sack block */ 60 tcp_seq end; /* end seq no. */ 61}; 62 63struct sackhole { 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 */ 68}; 69 70struct sackhint { 71 struct sackhole *nexthole; 72 int sack_bytes_rexmit; 73 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 74 75 int ispare; /* explicit pad for 64bit alignment */ 76 int sacked_bytes; /* 77 * Total sacked bytes reported by the 78 * receiver via sack option 79 */ 80 uint32_t _pad1[1]; /* TBD */ 81 uint64_t _pad[1]; /* TBD */ 82}; 83 84struct tcptemp { 85 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 86 struct tcphdr tt_t; 87}; 88 89#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 90 91/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ 92#ifdef INET6 93#define ND6_HINT(tp) \ 94do { \ 95 if ((tp) && (tp)->t_inpcb && \ 96 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ 97 nd6_nud_hint(NULL, NULL, 0); \ 98} while (0) 99#else 100#define ND6_HINT(tp) 101#endif 102 103/* 104 * Tcp control block, one per tcp; fields: 105 * Organized for 16 byte cacheline efficiency. 106 */ 107struct tcpcb { 108 struct tsegqe_head t_segq; /* segment reassembly queue */ 109 void *t_pspare[2]; /* new reassembly queue */ 110 int t_segqlen; /* segment reassembly queue length */ 111 int t_dupacks; /* consecutive dup acks recd */ 112 113 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 114 115 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 116 int t_state; /* state of this connection */ 117 u_int t_flags; 118 119 struct vnet *t_vnet; /* back pointer to parent vnet */ 120 121 tcp_seq snd_una; /* send unacknowledged */ 122 tcp_seq snd_max; /* highest sequence number sent; 123 * used to recognize retransmits 124 */ 125 tcp_seq snd_nxt; /* send next */ 126 tcp_seq snd_up; /* send urgent pointer */ 127 128 tcp_seq snd_wl1; /* window update seg seq number */ 129 tcp_seq snd_wl2; /* window update seg ack number */ 130 tcp_seq iss; /* initial send sequence number */ 131 tcp_seq irs; /* initial receive sequence number */ 132 133 tcp_seq rcv_nxt; /* receive next */ 134 tcp_seq rcv_adv; /* advertised window */ 135 u_long rcv_wnd; /* receive window */ 136 tcp_seq rcv_up; /* receive urgent pointer */ 137 138 u_long snd_wnd; /* send window */ 139 u_long snd_cwnd; /* congestion-controlled window */ 140 u_long snd_spare1; /* unused */ 141 u_long snd_ssthresh; /* snd_cwnd size threshold for 142 * for slow start exponential to 143 * linear switch 144 */ 145 u_long snd_spare2; /* unused */ 146 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 147 148 u_int t_maxopd; /* mss plus options */ 149 150 u_int t_rcvtime; /* inactivity time */ 151 u_int t_starttime; /* time connection was established */ 152 u_int t_rtttime; /* RTT measurement start time */ 153 tcp_seq t_rtseq; /* sequence number being timed */ 154 155 u_int t_bw_spare1; /* unused */ 156 tcp_seq t_bw_spare2; /* unused */ 157 158 int t_rxtcur; /* current retransmit value (ticks) */ 159 u_int t_maxseg; /* maximum segment size */ 160 int t_srtt; /* smoothed round-trip time */ 161 int t_rttvar; /* variance in round-trip time */ 162 163 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 164 u_int t_rttmin; /* minimum rtt allowed */ 165 u_int t_rttbest; /* best rtt we've seen */ 166 u_long t_rttupdated; /* number of times rtt sampled */ 167 u_long max_sndwnd; /* largest window peer has offered */ 168 169 int t_softerror; /* possible error not yet reported */ 170/* out-of-band data */ 171 char t_oobflags; /* have some */ 172 char t_iobc; /* input character */ 173/* RFC 1323 variables */ 174 u_char snd_scale; /* window scaling for send window */ 175 u_char rcv_scale; /* window scaling for recv window */ 176 u_char request_r_scale; /* pending window scaling */ 177 u_int32_t ts_recent; /* timestamp echo data */ 178 u_int ts_recent_age; /* when last updated */ 179 u_int32_t ts_offset; /* our timestamp offset */ 180 181 tcp_seq last_ack_sent; 182/* experimental */ 183 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 184 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 185 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 186 int t_sndzerowin; /* zero-window updates sent */ 187 u_int t_badrxtwin; /* window for retransmit recovery */ 188 u_char snd_limited; /* segments limited transmitted */ 189/* SACK related state */ 190 int snd_numholes; /* number of holes seen by sender */ 191 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 192 /* SACK scoreboard (sorted) */ 193 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 194 int rcv_numsacks; /* # distinct sack blks present */ 195 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 196 tcp_seq sack_newdata; /* New data xmitted in this recovery 197 episode starts at this seq number */ 198 struct sackhint sackhint; /* SACK scoreboard hint */ 199 int t_rttlow; /* smallest observerved RTT */ 200 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 201 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 202 struct toedev *tod; /* toedev handling this connection */ 203 int t_sndrexmitpack; /* retransmit packets sent */ 204 int t_rcvoopack; /* out-of-order packets received */ 205 void *t_toe; /* TOE pcb pointer */ 206 int t_bytes_acked; /* # bytes acked during current RTT */ 207 struct cc_algo *cc_algo; /* congestion control algorithm */ 208 struct cc_var *ccv; /* congestion control specific vars */ 209 struct osd *osd; /* storage for Khelp module data */ 210 211 u_int t_keepinit; /* time to establish connection */ 212 u_int t_keepidle; /* time before keepalive probes begin */ 213 u_int t_keepintvl; /* interval between keepalives */ 214 u_int t_keepcnt; /* number of keepalives before close */ 215 216 u_int t_tsomax; /* TSO total burst length limit in bytes */ 217 u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */ 218 u_int t_flags2; /* More tcpcb flags storage */ 219 220 uint32_t t_ispare[6]; /* 5 UTO, 1 TBD */ 221#if defined(_KERNEL) && defined(TCP_RFC7413) 222 void *t_pspare2[3]; /* 1 TCP_SIGNATURE, 2 TBD */ 223 unsigned int *t_tfo_pending; /* TCP Fast Open pending counter */ 224#else 225 void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ 226#endif 227#if defined(_KERNEL) && defined(TCP_RFC7413) 228 uint64_t _pad[4]; /* 4 TBD (1-2 CC/RTT?) */ 229 uint64_t t_tfo_cookie; /* TCP Fast Open cookie */ 230#else 231 uint64_t _pad[5]; /* 5 TBD (1-2 CC/RTT?) */ 232#endif 233 uint32_t t_tsomaxsegcount; /* TSO maximum segment count */ 234 uint32_t t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 235}; 236 237/* 238 * Flags and utility macros for the t_flags field. 239 */ 240#define TF_ACKNOW 0x000001 /* ack peer immediately */ 241#define TF_DELACK 0x000002 /* ack, but try to delay it */ 242#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ 243#define TF_NOOPT 0x000008 /* don't use tcp options */ 244#define TF_SENTFIN 0x000010 /* have sent FIN */ 245#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ 246#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ 247#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ 248#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ 249#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ 250#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ 251#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ 252#define TF_NOPUSH 0x001000 /* don't push */ 253#define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ 254#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ 255#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ 256#define TF_LASTIDLE 0x040000 /* connection was previously idle */ 257#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ 258#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ 259#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ 260#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 261#define TF_FORCEDATA 0x800000 /* force out a byte */ 262#define TF_TSO 0x1000000 /* TSO enabled on this connection */ 263#define TF_TOE 0x2000000 /* this connection is offloaded */ 264#define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ 265#define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ 266#define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ 267#define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 268#define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 269#define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */ 270 271#define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 272#define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 273#define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 274 275#define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 276#define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 277#define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 278 279#define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 280#define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 281#define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 282 283#define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 284 285/* 286 * Flags for the t_oobflags field. 287 */ 288#define TCPOOB_HAVEDATA 0x01 289#define TCPOOB_HADDATA 0x02 290 291#ifdef TCP_SIGNATURE 292/* 293 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 294 * for SADB verification and lookup. 295 */ 296#define TCP_SIGLEN 16 /* length of computed digest in bytes */ 297#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 298#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 299/* 300 * Only a single SA per host may be specified at this time. An SPI is 301 * needed in order for the KEY_ALLOCSA() lookup to work. 302 */ 303#define TCP_SIG_SPI 0x1000 304#endif /* TCP_SIGNATURE */ 305 306/* 307 * Flags for PLPMTU handling, t_flags2 308 */ 309#define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 310#define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 311#define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 312 313/* 314 * Structure to hold TCP options that are only used during segment 315 * processing (in tcp_input), but not held in the tcpcb. 316 * It's basically used to reduce the number of parameters 317 * to tcp_dooptions and tcp_addoptions. 318 * The binary order of the to_flags is relevant for packing of the 319 * options in tcp_addoptions. 320 */ 321struct tcpopt { 322 u_int64_t to_flags; /* which options are present */ 323#define TOF_MSS 0x0001 /* maximum segment size */ 324#define TOF_SCALE 0x0002 /* window scaling */ 325#define TOF_SACKPERM 0x0004 /* SACK permitted */ 326#define TOF_TS 0x0010 /* timestamp */ 327#define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 328#define TOF_SACK 0x0080 /* Peer sent SACK option */ 329#define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */ 330#define TOF_MAXOPT 0x0200 331 u_int32_t to_tsval; /* new timestamp */ 332 u_int32_t to_tsecr; /* reflected timestamp */ 333 u_char *to_sacks; /* pointer to the first SACK blocks */ 334 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 335 u_char *to_tfo_cookie; /* pointer to the TFO cookie */ 336 u_int16_t to_mss; /* maximum segment size */ 337 u_int8_t to_wscale; /* window scaling */ 338 u_int8_t to_nsacks; /* number of SACK blocks */ 339 u_int8_t to_tfo_len; /* TFO cookie length */ 340 u_int32_t to_spare; /* UTO */ 341}; 342 343/* 344 * Flags for tcp_dooptions. 345 */ 346#define TO_SYN 0x01 /* parse SYN-only options */ 347 348struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 349 u_long rmx_mtu; /* MTU for this path */ 350 u_long rmx_ssthresh; /* outbound gateway buffer limit */ 351 u_long rmx_rtt; /* estimated round trip time */ 352 u_long rmx_rttvar; /* estimated rtt variance */ 353 u_long rmx_bandwidth; /* estimated bandwidth */ 354 u_long rmx_cwnd; /* congestion window */ 355 u_long rmx_sendpipe; /* outbound delay-bandwidth product */ 356 u_long rmx_recvpipe; /* inbound delay-bandwidth product */ 357}; 358 359/* 360 * Used by tcp_maxmtu() to communicate interface specific features 361 * and limits at the time of connection setup. 362 */ 363struct tcp_ifcap { 364 int ifcap; 365 u_int tsomax; 366 u_int tsomaxsegcount; 367 u_int tsomaxsegsize; 368}; 369 370#ifndef _NETINET_IN_PCB_H_ 371struct in_conninfo; 372#endif /* _NETINET_IN_PCB_H_ */ 373 374struct tcptw { 375 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 376 tcp_seq snd_nxt; 377 tcp_seq rcv_nxt; 378 tcp_seq iss; 379 tcp_seq irs; 380 u_short last_win; /* cached window value */ 381 u_short tw_so_options; /* copy of so_options */ 382 struct ucred *tw_cred; /* user credentials */ 383 u_int32_t t_recent; 384 u_int32_t ts_offset; /* our timestamp offset */ 385 u_int t_starttime; 386 int tw_time; 387 TAILQ_ENTRY(tcptw) tw_2msl; 388 void *tw_pspare; /* TCP_SIGNATURE */ 389 u_int *tw_spare; /* TCP_SIGNATURE */ 390}; 391 392#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 393#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 394#define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 395 396/* 397 * The smoothed round-trip time and estimated variance 398 * are stored as fixed point numbers scaled by the values below. 399 * For convenience, these scales are also used in smoothing the average 400 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 401 * With these scales, srtt has 3 bits to the right of the binary point, 402 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 403 * binary point, and is smoothed with an ALPHA of 0.75. 404 */ 405#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 406#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 407#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 408#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 409#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 410 411/* 412 * The initial retransmission should happen at rtt + 4 * rttvar. 413 * Because of the way we do the smoothing, srtt and rttvar 414 * will each average +1/2 tick of bias. When we compute 415 * the retransmit timer, we want 1/2 tick of rounding and 416 * 1 extra tick because of +-1/2 tick uncertainty in the 417 * firing of the timer. The bias will give us exactly the 418 * 1.5 tick we need. But, because the bias is 419 * statistical, we have to test that we don't drop below 420 * the minimum feasible timer (which is 2 ticks). 421 * This version of the macro adapted from a paper by Lawrence 422 * Brakmo and Larry Peterson which outlines a problem caused 423 * by insufficient precision in the original implementation, 424 * which results in inappropriately large RTO values for very 425 * fast networks. 426 */ 427#define TCP_REXMTVAL(tp) \ 428 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 429 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 430 431/* 432 * TCP statistics. 433 * Many of these should be kept per connection, 434 * but that's inconvenient at the moment. 435 */ 436struct tcpstat { 437 uint64_t tcps_connattempt; /* connections initiated */ 438 uint64_t tcps_accepts; /* connections accepted */ 439 uint64_t tcps_connects; /* connections established */ 440 uint64_t tcps_drops; /* connections dropped */ 441 uint64_t tcps_conndrops; /* embryonic connections dropped */ 442 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 443 uint64_t tcps_closed; /* conn. closed (includes drops) */ 444 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 445 uint64_t tcps_rttupdated; /* times we succeeded */ 446 uint64_t tcps_delack; /* delayed acks sent */ 447 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 448 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 449 uint64_t tcps_persisttimeo; /* persist timeouts */ 450 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 451 uint64_t tcps_keepprobe; /* keepalive probes sent */ 452 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 453 454 uint64_t tcps_sndtotal; /* total packets sent */ 455 uint64_t tcps_sndpack; /* data packets sent */ 456 uint64_t tcps_sndbyte; /* data bytes sent */ 457 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 458 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 459 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 460 uint64_t tcps_sndacks; /* ack-only packets sent */ 461 uint64_t tcps_sndprobe; /* window probes sent */ 462 uint64_t tcps_sndurg; /* packets sent with URG only */ 463 uint64_t tcps_sndwinup; /* window update-only packets sent */ 464 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 465 466 uint64_t tcps_rcvtotal; /* total packets received */ 467 uint64_t tcps_rcvpack; /* packets received in sequence */ 468 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 469 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 470 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 471 uint64_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 472 uint64_t tcps_rcvshort; /* packets received too short */ 473 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 474 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 475 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 476 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 477 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 478 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 479 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 480 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 481 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 482 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 483 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 484 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 485 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 486 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 487 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 488 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 489 uint64_t tcps_predack; /* times hdr predict ok for acks */ 490 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 491 uint64_t tcps_pcbcachemiss; 492 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 493 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 494 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 495 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 496 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 497 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 498 uint64_t tcps_persistdrop; /* timeout in persist state */ 499 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 500 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 501 uint64_t tcps_listendrop; /* listen queue overflows */ 502 uint64_t tcps_badrst; /* ignored RSTs in the window */ 503 504 uint64_t tcps_sc_added; /* entry added to syncache */ 505 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 506 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 507 uint64_t tcps_sc_dropped; /* could not reply to packet */ 508 uint64_t tcps_sc_completed; /* successful extraction of entry */ 509 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 510 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 511 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 512 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 513 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 514 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 515 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 516 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 517 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 518 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 519 520 uint64_t tcps_hc_added; /* entry added to hostcache */ 521 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 522 523 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 524 525 /* SACK related stats */ 526 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 527 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 528 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 529 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 530 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 531 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 532 533 /* ECN related stats */ 534 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 535 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 536 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 537 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 538 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 539 540 /* TCP_SIGNATURE related stats */ 541 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 542 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 543 uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */ 544 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 545 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 546 547 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 548}; 549 550#ifdef _KERNEL 551#include <sys/counter.h> 552 553VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 554/* 555 * In-kernel consumers can use these accessor macros directly to update 556 * stats. 557 */ 558#define TCPSTAT_ADD(name, val) \ 559 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 560#define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 561 562/* 563 * Kernel module consumers must use this accessor macro. 564 */ 565void kmod_tcpstat_inc(int statnum); 566#define KMOD_TCPSTAT_INC(name) \ 567 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) 568 569/* 570 * TCP specific helper hook point identifiers. 571 */ 572#define HHOOK_TCP_EST_IN 0 573#define HHOOK_TCP_EST_OUT 1 574#define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 575 576struct tcp_hhook_data { 577 struct tcpcb *tp; 578 struct tcphdr *th; 579 struct tcpopt *to; 580 long len; 581 int tso; 582 tcp_seq curack; 583}; 584#endif 585 586/* 587 * TCB structure exported to user-land via sysctl(3). 588 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 589 * included. Not all of our clients do. 590 */ 591#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 592struct xtcp_timer { 593 int tt_rexmt; /* retransmit timer */ 594 int tt_persist; /* retransmit persistence */ 595 int tt_keep; /* keepalive */ 596 int tt_2msl; /* 2*msl TIME_WAIT timer */ 597 int tt_delack; /* delayed ACK timer */ 598 int t_rcvtime; /* Time since last packet received */ 599}; 600struct xtcpcb { 601 size_t xt_len; 602 struct inpcb xt_inp; 603 struct tcpcb xt_tp; 604 struct xsocket xt_socket; 605 struct xtcp_timer xt_timer; 606 u_quad_t xt_alignment_hack; 607}; 608#endif 609 610/* 611 * Names for TCP sysctl objects 612 */ 613#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 614#define TCPCTL_MSSDFLT 3 /* MSS default */ 615#define TCPCTL_STATS 4 /* statistics (read-only) */ 616#define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 617#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 618#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 619#define TCPCTL_SENDSPACE 8 /* send buffer space */ 620#define TCPCTL_RECVSPACE 9 /* receive buffer space */ 621#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 622#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 623#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 624#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 625#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 626#define TCPCTL_DROP 15 /* drop tcp connection */ 627#define TCPCTL_MAXID 16 628#define TCPCTL_FINWAIT2_TIMEOUT 17 629 630#ifdef _KERNEL 631#ifdef SYSCTL_DECL 632SYSCTL_DECL(_net_inet_tcp); 633SYSCTL_DECL(_net_inet_tcp_sack); 634MALLOC_DECLARE(M_TCPLOG); 635#endif 636 637VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ 638VNET_DECLARE(struct inpcbinfo, tcbinfo); 639extern int tcp_log_in_vain; 640VNET_DECLARE(int, tcp_mssdflt); /* XXX */ 641VNET_DECLARE(int, tcp_minmss); 642VNET_DECLARE(int, tcp_delack_enabled); 643VNET_DECLARE(int, tcp_do_rfc3390); 644VNET_DECLARE(int, tcp_do_initcwnd10); 645VNET_DECLARE(int, tcp_sendspace); 646VNET_DECLARE(int, tcp_recvspace); 647VNET_DECLARE(int, path_mtu_discovery); 648VNET_DECLARE(int, tcp_do_rfc3465); 649VNET_DECLARE(int, tcp_abc_l_var); 650#define V_tcb VNET(tcb) 651#define V_tcbinfo VNET(tcbinfo) 652#define V_tcp_mssdflt VNET(tcp_mssdflt) 653#define V_tcp_minmss VNET(tcp_minmss) 654#define V_tcp_delack_enabled VNET(tcp_delack_enabled) 655#define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 656#define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) 657#define V_tcp_sendspace VNET(tcp_sendspace) 658#define V_tcp_recvspace VNET(tcp_recvspace) 659#define V_path_mtu_discovery VNET(path_mtu_discovery) 660#define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 661#define V_tcp_abc_l_var VNET(tcp_abc_l_var) 662 663VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ 664VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ 665#define V_tcp_do_sack VNET(tcp_do_sack) 666#define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 667 668VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ 669VNET_DECLARE(int, tcp_ecn_maxretries); 670#define V_tcp_do_ecn VNET(tcp_do_ecn) 671#define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 672 673VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 674#define V_tcp_hhh VNET(tcp_hhh) 675 676VNET_DECLARE(int, tcp_do_rfc6675_pipe); 677#define V_tcp_do_rfc6675_pipe VNET(tcp_do_rfc6675_pipe) 678 679int tcp_addoptions(struct tcpopt *, u_char *); 680int tcp_ccalgounload(struct cc_algo *unload_algo); 681struct tcpcb * 682 tcp_close(struct tcpcb *); 683void tcp_discardcb(struct tcpcb *); 684void tcp_twstart(struct tcpcb *); 685#if 0 686int tcp_twrecycleable(struct tcptw *tw); 687#endif 688void tcp_twclose(struct tcptw *, int); 689void tcp_ctlinput(int, struct sockaddr *, void *); 690int tcp_ctloutput(struct socket *, struct sockopt *); 691struct tcpcb * 692 tcp_drop(struct tcpcb *, int); 693void tcp_drain(void); 694void tcp_init(void); 695#ifdef VIMAGE 696void tcp_destroy(void); 697#endif 698void tcp_fini(void *); 699char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 700 const void *); 701char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 702 const void *); 703int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); 704void tcp_reass_global_init(void); 705void tcp_reass_flush(struct tcpcb *); 706void tcp_input(struct mbuf *, int); 707int tcp_autorcvbuf(struct mbuf *, struct tcphdr *, struct socket *, 708 struct tcpcb *, int); 709u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 710u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 711void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 712 struct tcp_ifcap *); 713void tcp_mss(struct tcpcb *, int); 714int tcp_mssopt(struct in_conninfo *); 715struct inpcb * 716 tcp_drop_syn_sent(struct inpcb *, int); 717struct inpcb * 718 tcp_mtudisc(struct inpcb *, int); 719struct tcpcb * 720 tcp_newtcpcb(struct inpcb *); 721int tcp_output(struct tcpcb *); 722void tcp_state_change(struct tcpcb *, int); 723void tcp_respond(struct tcpcb *, void *, 724 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 725void tcp_tw_init(void); 726#ifdef VIMAGE 727void tcp_tw_destroy(void); 728#endif 729void tcp_tw_zone_change(void); 730int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 731 struct mbuf *, int); 732void tcp_setpersist(struct tcpcb *); 733#ifdef TCP_SIGNATURE 734int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); 735int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, 736 struct tcphdr *, u_int); 737#endif 738void tcp_slowtimo(void); 739struct tcptemp * 740 tcpip_maketemplate(struct inpcb *); 741void tcpip_fillheaders(struct inpcb *, void *, void *); 742void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 743int tcp_timer_active(struct tcpcb *, uint32_t); 744void tcp_timer_stop(struct tcpcb *, uint32_t); 745void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 746/* 747 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 748 */ 749void tcp_hc_init(void); 750#ifdef VIMAGE 751void tcp_hc_destroy(void); 752#endif 753void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 754u_long tcp_hc_getmtu(struct in_conninfo *); 755void tcp_hc_updatemtu(struct in_conninfo *, u_long); 756void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 757 758extern struct pr_usrreqs tcp_usrreqs; 759tcp_seq tcp_new_isn(struct tcpcb *); 760 761int tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 762void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 763void tcp_clean_sackreport(struct tcpcb *tp); 764void tcp_sack_adjust(struct tcpcb *tp); 765struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 766void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 767void tcp_free_sackholes(struct tcpcb *tp); 768int tcp_newreno(struct tcpcb *, struct tcphdr *); 769int tcp_compute_pipe(struct tcpcb *); 770 771void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 772 773static inline void 774tcp_fields_to_host(struct tcphdr *th) 775{ 776 777 th->th_seq = ntohl(th->th_seq); 778 th->th_ack = ntohl(th->th_ack); 779 th->th_win = ntohs(th->th_win); 780 th->th_urp = ntohs(th->th_urp); 781} 782 783#ifdef TCP_SIGNATURE 784static inline void 785tcp_fields_to_net(struct tcphdr *th) 786{ 787 788 th->th_seq = htonl(th->th_seq); 789 th->th_ack = htonl(th->th_ack); 790 th->th_win = htons(th->th_win); 791 th->th_urp = htons(th->th_urp); 792} 793#endif 794#endif /* _KERNEL */ 795 796#endif /* _NETINET_TCP_VAR_H_ */ 797