tcp_var.h revision 285976
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 285976 2015-07-28 19:58:44Z delphij $ 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 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ 77}; 78 79struct tcptemp { 80 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 81 struct tcphdr tt_t; 82}; 83 84#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 85 86/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ 87#ifdef INET6 88#define ND6_HINT(tp) \ 89do { \ 90 if ((tp) && (tp)->t_inpcb && \ 91 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ 92 nd6_nud_hint(NULL, NULL, 0); \ 93} while (0) 94#else 95#define ND6_HINT(tp) 96#endif 97 98/* 99 * Tcp control block, one per tcp; fields: 100 * Organized for 16 byte cacheline efficiency. 101 */ 102struct tcpcb { 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 */ 107 108 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 109 110 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 111 int t_state; /* state of this connection */ 112 u_int t_flags; 113 114 struct vnet *t_vnet; /* back pointer to parent vnet */ 115 116 tcp_seq snd_una; /* send unacknowledged */ 117 tcp_seq snd_max; /* highest sequence number sent; 118 * used to recognize retransmits 119 */ 120 tcp_seq snd_nxt; /* send next */ 121 tcp_seq snd_up; /* send urgent pointer */ 122 123 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 */ 127 128 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 */ 132 133 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 138 * linear switch 139 */ 140 u_long snd_spare2; /* unused */ 141 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 142 143 u_int t_maxopd; /* mss plus options */ 144 145 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 */ 149 150 u_int t_bw_spare1; /* unused */ 151 tcp_seq t_bw_spare2; /* unused */ 152 153 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 */ 157 158 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 */ 163 164 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 */ 175 176 tcp_seq last_ack_sent; 177/* experimental */ 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 */ 205 206 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 */ 210 211 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 */ 214 215 uint32_t t_ispare[6]; /* 5 UTO, 1 TBD */ 216 void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ 217 uint64_t _pad[5]; /* 5 TBD (1-2 CC/RTT?) */ 218 uint32_t t_tsomaxsegcount; /* TSO maximum segment count */ 219 uint32_t t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 220}; 221 222/* 223 * Flags and utility macros for the t_flags field. 224 */ 225#define TF_ACKNOW 0x000001 /* ack peer immediately */ 226#define TF_DELACK 0x000002 /* ack, but try to delay it */ 227#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ 228#define TF_NOOPT 0x000008 /* don't use tcp options */ 229#define TF_SENTFIN 0x000010 /* have sent FIN */ 230#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ 231#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ 232#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ 233#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ 234#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ 235#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ 236#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ 237#define TF_NOPUSH 0x001000 /* don't push */ 238#define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ 239#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ 240#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ 241#define TF_LASTIDLE 0x040000 /* connection was previously idle */ 242#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ 243#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ 244#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ 245#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 246#define TF_FORCEDATA 0x800000 /* force out a byte */ 247#define TF_TSO 0x1000000 /* TSO enabled on this connection */ 248#define TF_TOE 0x2000000 /* this connection is offloaded */ 249#define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ 250#define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ 251#define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ 252#define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 253#define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 254 255#define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 256#define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 257#define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 258 259#define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 260#define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 261#define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 262 263#define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 264#define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 265#define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 266 267#define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 268 269/* 270 * Flags for the t_oobflags field. 271 */ 272#define TCPOOB_HAVEDATA 0x01 273#define TCPOOB_HADDATA 0x02 274 275#ifdef TCP_SIGNATURE 276/* 277 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 278 * for SADB verification and lookup. 279 */ 280#define TCP_SIGLEN 16 /* length of computed digest in bytes */ 281#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 282#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 283/* 284 * Only a single SA per host may be specified at this time. An SPI is 285 * needed in order for the KEY_ALLOCSA() lookup to work. 286 */ 287#define TCP_SIG_SPI 0x1000 288#endif /* TCP_SIGNATURE */ 289 290/* 291 * Flags for PLPMTU handling, t_flags2 292 */ 293#define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 294#define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 295#define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 296 297/* 298 * Structure to hold TCP options that are only used during segment 299 * processing (in tcp_input), but not held in the tcpcb. 300 * It's basically used to reduce the number of parameters 301 * to tcp_dooptions and tcp_addoptions. 302 * The binary order of the to_flags is relevant for packing of the 303 * options in tcp_addoptions. 304 */ 305struct tcpopt { 306 u_int64_t to_flags; /* which options are present */ 307#define TOF_MSS 0x0001 /* maximum segment size */ 308#define TOF_SCALE 0x0002 /* window scaling */ 309#define TOF_SACKPERM 0x0004 /* SACK permitted */ 310#define TOF_TS 0x0010 /* timestamp */ 311#define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 312#define TOF_SACK 0x0080 /* Peer sent SACK option */ 313#define TOF_MAXOPT 0x0100 314 u_int32_t to_tsval; /* new timestamp */ 315 u_int32_t to_tsecr; /* reflected timestamp */ 316 u_char *to_sacks; /* pointer to the first SACK blocks */ 317 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 318 u_int16_t to_mss; /* maximum segment size */ 319 u_int8_t to_wscale; /* window scaling */ 320 u_int8_t to_nsacks; /* number of SACK blocks */ 321 u_int32_t to_spare; /* UTO */ 322}; 323 324/* 325 * Flags for tcp_dooptions. 326 */ 327#define TO_SYN 0x01 /* parse SYN-only options */ 328 329struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 330 u_long rmx_mtu; /* MTU for this path */ 331 u_long rmx_ssthresh; /* outbound gateway buffer limit */ 332 u_long rmx_rtt; /* estimated round trip time */ 333 u_long rmx_rttvar; /* estimated rtt variance */ 334 u_long rmx_bandwidth; /* estimated bandwidth */ 335 u_long rmx_cwnd; /* congestion window */ 336 u_long rmx_sendpipe; /* outbound delay-bandwidth product */ 337 u_long rmx_recvpipe; /* inbound delay-bandwidth product */ 338}; 339 340/* 341 * Used by tcp_maxmtu() to communicate interface specific features 342 * and limits at the time of connection setup. 343 */ 344struct tcp_ifcap { 345 int ifcap; 346 u_int tsomax; 347 u_int tsomaxsegcount; 348 u_int tsomaxsegsize; 349}; 350 351#ifndef _NETINET_IN_PCB_H_ 352struct in_conninfo; 353#endif /* _NETINET_IN_PCB_H_ */ 354 355struct tcptw { 356 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 357 tcp_seq snd_nxt; 358 tcp_seq rcv_nxt; 359 tcp_seq iss; 360 tcp_seq irs; 361 u_short last_win; /* cached window value */ 362 u_short tw_so_options; /* copy of so_options */ 363 struct ucred *tw_cred; /* user credentials */ 364 u_int32_t t_recent; 365 u_int32_t ts_offset; /* our timestamp offset */ 366 u_int t_starttime; 367 int tw_time; 368 TAILQ_ENTRY(tcptw) tw_2msl; 369 void *tw_pspare; /* TCP_SIGNATURE */ 370 u_int *tw_spare; /* TCP_SIGNATURE */ 371}; 372 373#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 374#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 375#define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 376 377/* 378 * The smoothed round-trip time and estimated variance 379 * are stored as fixed point numbers scaled by the values below. 380 * For convenience, these scales are also used in smoothing the average 381 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 382 * With these scales, srtt has 3 bits to the right of the binary point, 383 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 384 * binary point, and is smoothed with an ALPHA of 0.75. 385 */ 386#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 387#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 388#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 389#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 390#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 391 392/* 393 * The initial retransmission should happen at rtt + 4 * rttvar. 394 * Because of the way we do the smoothing, srtt and rttvar 395 * will each average +1/2 tick of bias. When we compute 396 * the retransmit timer, we want 1/2 tick of rounding and 397 * 1 extra tick because of +-1/2 tick uncertainty in the 398 * firing of the timer. The bias will give us exactly the 399 * 1.5 tick we need. But, because the bias is 400 * statistical, we have to test that we don't drop below 401 * the minimum feasible timer (which is 2 ticks). 402 * This version of the macro adapted from a paper by Lawrence 403 * Brakmo and Larry Peterson which outlines a problem caused 404 * by insufficient precision in the original implementation, 405 * which results in inappropriately large RTO values for very 406 * fast networks. 407 */ 408#define TCP_REXMTVAL(tp) \ 409 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 410 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 411 412/* 413 * TCP statistics. 414 * Many of these should be kept per connection, 415 * but that's inconvenient at the moment. 416 */ 417struct tcpstat { 418 uint64_t tcps_connattempt; /* connections initiated */ 419 uint64_t tcps_accepts; /* connections accepted */ 420 uint64_t tcps_connects; /* connections established */ 421 uint64_t tcps_drops; /* connections dropped */ 422 uint64_t tcps_conndrops; /* embryonic connections dropped */ 423 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 424 uint64_t tcps_closed; /* conn. closed (includes drops) */ 425 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 426 uint64_t tcps_rttupdated; /* times we succeeded */ 427 uint64_t tcps_delack; /* delayed acks sent */ 428 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 429 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 430 uint64_t tcps_persisttimeo; /* persist timeouts */ 431 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 432 uint64_t tcps_keepprobe; /* keepalive probes sent */ 433 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 434 435 uint64_t tcps_sndtotal; /* total packets sent */ 436 uint64_t tcps_sndpack; /* data packets sent */ 437 uint64_t tcps_sndbyte; /* data bytes sent */ 438 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 439 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 440 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 441 uint64_t tcps_sndacks; /* ack-only packets sent */ 442 uint64_t tcps_sndprobe; /* window probes sent */ 443 uint64_t tcps_sndurg; /* packets sent with URG only */ 444 uint64_t tcps_sndwinup; /* window update-only packets sent */ 445 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 446 447 uint64_t tcps_rcvtotal; /* total packets received */ 448 uint64_t tcps_rcvpack; /* packets received in sequence */ 449 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 450 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 451 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 452 uint64_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 453 uint64_t tcps_rcvshort; /* packets received too short */ 454 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 455 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 456 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 457 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 458 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 459 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 460 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 461 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 462 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 463 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 464 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 465 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 466 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 467 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 468 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 469 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 470 uint64_t tcps_predack; /* times hdr predict ok for acks */ 471 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 472 uint64_t tcps_pcbcachemiss; 473 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 474 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 475 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 476 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 477 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 478 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 479 uint64_t tcps_persistdrop; /* timeout in persist state */ 480 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 481 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 482 uint64_t tcps_listendrop; /* listen queue overflows */ 483 uint64_t tcps_badrst; /* ignored RSTs in the window */ 484 485 uint64_t tcps_sc_added; /* entry added to syncache */ 486 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 487 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 488 uint64_t tcps_sc_dropped; /* could not reply to packet */ 489 uint64_t tcps_sc_completed; /* successful extraction of entry */ 490 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 491 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 492 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 493 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 494 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 495 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 496 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 497 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 498 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 499 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 500 501 uint64_t tcps_hc_added; /* entry added to hostcache */ 502 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 503 504 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 505 506 /* SACK related stats */ 507 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 508 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 509 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 510 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 511 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 512 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 513 514 /* ECN related stats */ 515 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 516 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 517 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 518 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 519 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 520 521 /* TCP_SIGNATURE related stats */ 522 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 523 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 524 uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */ 525 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 526 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 527 528 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 529}; 530 531#ifdef _KERNEL 532#include <sys/counter.h> 533 534VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 535/* 536 * In-kernel consumers can use these accessor macros directly to update 537 * stats. 538 */ 539#define TCPSTAT_ADD(name, val) \ 540 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 541#define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 542 543/* 544 * Kernel module consumers must use this accessor macro. 545 */ 546void kmod_tcpstat_inc(int statnum); 547#define KMOD_TCPSTAT_INC(name) \ 548 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) 549 550/* 551 * TCP specific helper hook point identifiers. 552 */ 553#define HHOOK_TCP_EST_IN 0 554#define HHOOK_TCP_EST_OUT 1 555#define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 556 557struct tcp_hhook_data { 558 struct tcpcb *tp; 559 struct tcphdr *th; 560 struct tcpopt *to; 561 long len; 562 int tso; 563 tcp_seq curack; 564}; 565#endif 566 567/* 568 * TCB structure exported to user-land via sysctl(3). 569 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 570 * included. Not all of our clients do. 571 */ 572#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 573struct xtcp_timer { 574 int tt_rexmt; /* retransmit timer */ 575 int tt_persist; /* retransmit persistence */ 576 int tt_keep; /* keepalive */ 577 int tt_2msl; /* 2*msl TIME_WAIT timer */ 578 int tt_delack; /* delayed ACK timer */ 579 int t_rcvtime; /* Time since last packet received */ 580}; 581struct xtcpcb { 582 size_t xt_len; 583 struct inpcb xt_inp; 584 struct tcpcb xt_tp; 585 struct xsocket xt_socket; 586 struct xtcp_timer xt_timer; 587 u_quad_t xt_alignment_hack; 588}; 589#endif 590 591/* 592 * Names for TCP sysctl objects 593 */ 594#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 595#define TCPCTL_MSSDFLT 3 /* MSS default */ 596#define TCPCTL_STATS 4 /* statistics (read-only) */ 597#define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 598#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 599#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 600#define TCPCTL_SENDSPACE 8 /* send buffer space */ 601#define TCPCTL_RECVSPACE 9 /* receive buffer space */ 602#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 603#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 604#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 605#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 606#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 607#define TCPCTL_DROP 15 /* drop tcp connection */ 608#define TCPCTL_MAXID 16 609#define TCPCTL_FINWAIT2_TIMEOUT 17 610 611#ifdef _KERNEL 612#ifdef SYSCTL_DECL 613SYSCTL_DECL(_net_inet_tcp); 614SYSCTL_DECL(_net_inet_tcp_sack); 615MALLOC_DECLARE(M_TCPLOG); 616#endif 617 618VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ 619VNET_DECLARE(struct inpcbinfo, tcbinfo); 620extern int tcp_log_in_vain; 621VNET_DECLARE(int, tcp_mssdflt); /* XXX */ 622VNET_DECLARE(int, tcp_minmss); 623VNET_DECLARE(int, tcp_delack_enabled); 624VNET_DECLARE(int, tcp_do_rfc3390); 625VNET_DECLARE(int, tcp_do_initcwnd10); 626VNET_DECLARE(int, tcp_sendspace); 627VNET_DECLARE(int, tcp_recvspace); 628VNET_DECLARE(int, path_mtu_discovery); 629VNET_DECLARE(int, tcp_do_rfc3465); 630VNET_DECLARE(int, tcp_abc_l_var); 631#define V_tcb VNET(tcb) 632#define V_tcbinfo VNET(tcbinfo) 633#define V_tcp_mssdflt VNET(tcp_mssdflt) 634#define V_tcp_minmss VNET(tcp_minmss) 635#define V_tcp_delack_enabled VNET(tcp_delack_enabled) 636#define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 637#define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) 638#define V_tcp_sendspace VNET(tcp_sendspace) 639#define V_tcp_recvspace VNET(tcp_recvspace) 640#define V_path_mtu_discovery VNET(path_mtu_discovery) 641#define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 642#define V_tcp_abc_l_var VNET(tcp_abc_l_var) 643 644VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ 645VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ 646#define V_tcp_do_sack VNET(tcp_do_sack) 647#define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 648 649VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ 650VNET_DECLARE(int, tcp_ecn_maxretries); 651#define V_tcp_do_ecn VNET(tcp_do_ecn) 652#define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 653 654VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 655#define V_tcp_hhh VNET(tcp_hhh) 656 657int tcp_addoptions(struct tcpopt *, u_char *); 658int tcp_ccalgounload(struct cc_algo *unload_algo); 659struct tcpcb * 660 tcp_close(struct tcpcb *); 661void tcp_discardcb(struct tcpcb *); 662void tcp_twstart(struct tcpcb *); 663#if 0 664int tcp_twrecycleable(struct tcptw *tw); 665#endif 666void tcp_twclose(struct tcptw *, int); 667void tcp_ctlinput(int, struct sockaddr *, void *); 668int tcp_ctloutput(struct socket *, struct sockopt *); 669struct tcpcb * 670 tcp_drop(struct tcpcb *, int); 671void tcp_drain(void); 672void tcp_init(void); 673#ifdef VIMAGE 674void tcp_destroy(void); 675#endif 676void tcp_fini(void *); 677char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 678 const void *); 679char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 680 const void *); 681int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); 682void tcp_reass_global_init(void); 683void tcp_reass_flush(struct tcpcb *); 684void tcp_input(struct mbuf *, int); 685u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 686u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 687void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 688 struct tcp_ifcap *); 689void tcp_mss(struct tcpcb *, int); 690int tcp_mssopt(struct in_conninfo *); 691struct inpcb * 692 tcp_drop_syn_sent(struct inpcb *, int); 693struct inpcb * 694 tcp_mtudisc(struct inpcb *, int); 695struct tcpcb * 696 tcp_newtcpcb(struct inpcb *); 697int tcp_output(struct tcpcb *); 698void tcp_state_change(struct tcpcb *, int); 699void tcp_respond(struct tcpcb *, void *, 700 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 701void tcp_tw_init(void); 702#ifdef VIMAGE 703void tcp_tw_destroy(void); 704#endif 705void tcp_tw_zone_change(void); 706int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 707 struct mbuf *, int); 708void tcp_setpersist(struct tcpcb *); 709#ifdef TCP_SIGNATURE 710int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); 711int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, 712 struct tcphdr *, u_int); 713#endif 714void tcp_slowtimo(void); 715struct tcptemp * 716 tcpip_maketemplate(struct inpcb *); 717void tcpip_fillheaders(struct inpcb *, void *, void *); 718void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 719int tcp_timer_active(struct tcpcb *, uint32_t); 720void tcp_timer_stop(struct tcpcb *, uint32_t); 721void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 722/* 723 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 724 */ 725void tcp_hc_init(void); 726#ifdef VIMAGE 727void tcp_hc_destroy(void); 728#endif 729void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 730u_long tcp_hc_getmtu(struct in_conninfo *); 731void tcp_hc_updatemtu(struct in_conninfo *, u_long); 732void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 733 734extern struct pr_usrreqs tcp_usrreqs; 735tcp_seq tcp_new_isn(struct tcpcb *); 736 737void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 738void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 739void tcp_clean_sackreport(struct tcpcb *tp); 740void tcp_sack_adjust(struct tcpcb *tp); 741struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 742void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 743void tcp_free_sackholes(struct tcpcb *tp); 744int tcp_newreno(struct tcpcb *, struct tcphdr *); 745u_long tcp_seq_subtract(u_long, u_long ); 746 747void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 748 749static inline void 750tcp_fields_to_host(struct tcphdr *th) 751{ 752 753 th->th_seq = ntohl(th->th_seq); 754 th->th_ack = ntohl(th->th_ack); 755 th->th_win = ntohs(th->th_win); 756 th->th_urp = ntohs(th->th_urp); 757} 758 759#ifdef TCP_SIGNATURE 760static inline void 761tcp_fields_to_net(struct tcphdr *th) 762{ 763 764 th->th_seq = htonl(th->th_seq); 765 th->th_ack = htonl(th->th_ack); 766 th->th_win = htons(th->th_win); 767 th->th_urp = htons(th->th_urp); 768} 769#endif 770#endif /* _KERNEL */ 771 772#endif /* _NETINET_TCP_VAR_H_ */ 773