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 * $FreeBSD$ 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 */ 44 VNET_DECLARE(int, tcp_do_rfc1323); 45 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 46 47 #endif /* _KERNEL */ 48 49 struct sackblk { 50 tcp_seq start; /* start seq no. of sack block */ 51 tcp_seq end; /* end seq no. */ 52 }; 53 54 struct sackhole { 55 tcp_seq start; /* start seq no. of hole */ 56 tcp_seq end; /* end seq no. */ 57 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 58 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 59 }; 60 61 struct sackhint { 62 struct sackhole *nexthole; 63 int sack_bytes_rexmit; 64 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 65 66 int ispare; /* explicit pad for 64bit alignment */ 67 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ 68 }; 69 70 struct tcptemp { 71 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 72 struct tcphdr tt_t; 73 }; 74 75 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 76 77 /* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */ 78 #ifdef INET6 79 #define ND6_HINT(tp) \ 80 do { \ 81 if ((tp) && (tp)->t_inpcb && \ 82 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \ 83 nd6_nud_hint(NULL, NULL, 0); \ 84 } while (0) 85 #else 86 #define ND6_HINT(tp) 87 #endif 88 89 /* 90 * Tcp control block, one per tcp; fields: 91 * Organized for 16 byte cacheline efficiency. 92 */ 93 struct tcpcb { 94 struct mbuf *t_segq; /* segment reassembly queue */ 95 void *t_pspare[2]; /* new reassembly queue */ 96 int t_segqlen; /* segment reassembly queue length */ 97 int t_dupacks; /* consecutive dup acks recd */ 98 99 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 100 101 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 102 int t_state; /* state of this connection */ 103 u_int t_flags; 104 105 struct vnet *t_vnet; /* back pointer to parent vnet */ 106 107 tcp_seq snd_una; /* send unacknowledged */ 108 tcp_seq snd_max; /* highest sequence number sent; 109 * used to recognize retransmits 110 */ 111 tcp_seq snd_nxt; /* send next */ 112 tcp_seq snd_up; /* send urgent pointer */ 113 114 tcp_seq snd_wl1; /* window update seg seq number */ 115 tcp_seq snd_wl2; /* window update seg ack number */ 116 tcp_seq iss; /* initial send sequence number */ 117 tcp_seq irs; /* initial receive sequence number */ 118 119 tcp_seq rcv_nxt; /* receive next */ 120 tcp_seq rcv_adv; /* advertised window */ 121 u_long rcv_wnd; /* receive window */ 122 tcp_seq rcv_up; /* receive urgent pointer */ 123 124 u_long snd_wnd; /* send window */ 125 u_long snd_cwnd; /* congestion-controlled window */ 126 u_long snd_spare1; /* unused */ 127 u_long snd_ssthresh; /* snd_cwnd size threshold for 128 * for slow start exponential to 129 * linear switch 130 */ 131 u_long snd_spare2; /* unused */ 132 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 133 134 u_int t_maxopd; /* mss plus options */ 135 136 u_int t_rcvtime; /* inactivity time */ 137 u_int t_starttime; /* time connection was established */ 138 u_int t_rtttime; /* RTT measurement start time */ 139 tcp_seq t_rtseq; /* sequence number being timed */ 140 141 u_int t_bw_spare1; /* unused */ 142 tcp_seq t_bw_spare2; /* unused */ 143 144 int t_rxtcur; /* current retransmit value (ticks) */ 145 u_int t_maxseg; /* maximum segment size */ 146 int t_srtt; /* smoothed round-trip time */ 147 int t_rttvar; /* variance in round-trip time */ 148 149 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 150 u_int t_rttmin; /* minimum rtt allowed */ 151 u_int t_rttbest; /* best rtt we've seen */ 152 u_long t_rttupdated; /* number of times rtt sampled */ 153 u_long max_sndwnd; /* largest window peer has offered */ 154 155 int t_softerror; /* possible error not yet reported */ 156 /* out-of-band data */ 157 char t_oobflags; /* have some */ 158 char t_iobc; /* input character */ 159 /* RFC 1323 variables */ 160 u_char snd_scale; /* window scaling for send window */ 161 u_char rcv_scale; /* window scaling for recv window */ 162 u_char request_r_scale; /* pending window scaling */ 163 u_int32_t ts_recent; /* timestamp echo data */ 164 u_int ts_recent_age; /* when last updated */ 165 u_int32_t ts_offset; /* our timestamp offset */ 166 167 tcp_seq last_ack_sent; 168 /* experimental */ 169 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 170 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 171 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 172 int t_sndzerowin; /* zero-window updates sent */ 173 u_int t_badrxtwin; /* window for retransmit recovery */ 174 u_char snd_limited; /* segments limited transmitted */ 175 /* SACK related state */ 176 int snd_numholes; /* number of holes seen by sender */ 177 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 178 /* SACK scoreboard (sorted) */ 179 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 180 int rcv_numsacks; /* # distinct sack blks present */ 181 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 182 tcp_seq sack_newdata; /* New data xmitted in this recovery 183 episode starts at this seq number */ 184 struct sackhint sackhint; /* SACK scoreboard hint */ 185 int t_rttlow; /* smallest observerved RTT */ 186 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 187 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 188 struct toedev *tod; /* toedev handling this connection */ 189 int t_sndrexmitpack; /* retransmit packets sent */ 190 int t_rcvoopack; /* out-of-order packets received */ 191 void *t_toe; /* TOE pcb pointer */ 192 int t_bytes_acked; /* # bytes acked during current RTT */ 193 struct cc_algo *cc_algo; /* congestion control algorithm */ 194 struct cc_var *ccv; /* congestion control specific vars */ 195 struct osd *osd; /* storage for Khelp module data */ 196 197 u_int t_keepinit; /* time to establish connection */ 198 u_int t_keepidle; /* time before keepalive probes begin */ 199 u_int t_keepintvl; /* interval between keepalives */ 200 u_int t_keepcnt; /* number of keepalives before close */ 201 202 u_int t_tsomax; /* TSO total burst length limit in bytes */ 203 u_int t_tsomaxsegcount; /* TSO maximum segment count */ 204 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 205 u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */ 206 u_int t_flags2; /* More tcpcb flags storage */ 207 208 uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ 209 void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ 210 uint64_t _pad[6]; /* 6 TBD (1-2 CC/RTT?) */ 211 }; 212 213 /* 214 * Flags and utility macros for the t_flags field. 215 */ 216 #define TF_ACKNOW 0x000001 /* ack peer immediately */ 217 #define TF_DELACK 0x000002 /* ack, but try to delay it */ 218 #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ 219 #define TF_NOOPT 0x000008 /* don't use tcp options */ 220 #define TF_SENTFIN 0x000010 /* have sent FIN */ 221 #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ 222 #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ 223 #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ 224 #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ 225 #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ 226 #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ 227 #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ 228 #define TF_NOPUSH 0x001000 /* don't push */ 229 #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ 230 #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ 231 #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ 232 #define TF_LASTIDLE 0x040000 /* connection was previously idle */ 233 #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ 234 #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ 235 #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ 236 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 237 #define TF_FORCEDATA 0x800000 /* force out a byte */ 238 #define TF_TSO 0x1000000 /* TSO enabled on this connection */ 239 #define TF_TOE 0x2000000 /* this connection is offloaded */ 240 #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ 241 #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ 242 #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ 243 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 244 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 245 246 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 247 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 248 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 249 250 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 251 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 252 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 253 254 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 255 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 256 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 257 258 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 259 260 /* 261 * Flags for the t_oobflags field. 262 */ 263 #define TCPOOB_HAVEDATA 0x01 264 #define TCPOOB_HADDATA 0x02 265 266 #ifdef TCP_SIGNATURE 267 /* 268 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 269 * for SADB verification and lookup. 270 */ 271 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 272 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 273 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 274 /* 275 * Only a single SA per host may be specified at this time. An SPI is 276 * needed in order for the KEY_ALLOCSA() lookup to work. 277 */ 278 #define TCP_SIG_SPI 0x1000 279 #endif /* TCP_SIGNATURE */ 280 281 /* 282 * Flags for PLPMTU handling, t_flags2 283 */ 284 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 285 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 286 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 287 288 /* 289 * Structure to hold TCP options that are only used during segment 290 * processing (in tcp_input), but not held in the tcpcb. 291 * It's basically used to reduce the number of parameters 292 * to tcp_dooptions and tcp_addoptions. 293 * The binary order of the to_flags is relevant for packing of the 294 * options in tcp_addoptions. 295 */ 296 struct tcpopt { 297 u_int64_t to_flags; /* which options are present */ 298 #define TOF_MSS 0x0001 /* maximum segment size */ 299 #define TOF_SCALE 0x0002 /* window scaling */ 300 #define TOF_SACKPERM 0x0004 /* SACK permitted */ 301 #define TOF_TS 0x0010 /* timestamp */ 302 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 303 #define TOF_SACK 0x0080 /* Peer sent SACK option */ 304 #define TOF_MAXOPT 0x0100 305 u_int32_t to_tsval; /* new timestamp */ 306 u_int32_t to_tsecr; /* reflected timestamp */ 307 u_char *to_sacks; /* pointer to the first SACK blocks */ 308 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 309 u_int16_t to_mss; /* maximum segment size */ 310 u_int8_t to_wscale; /* window scaling */ 311 u_int8_t to_nsacks; /* number of SACK blocks */ 312 u_int32_t to_spare; /* UTO */ 313 }; 314 315 /* 316 * Flags for tcp_dooptions. 317 */ 318 #define TO_SYN 0x01 /* parse SYN-only options */ 319 320 struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 321 u_long rmx_mtu; /* MTU for this path */ 322 u_long rmx_ssthresh; /* outbound gateway buffer limit */ 323 u_long rmx_rtt; /* estimated round trip time */ 324 u_long rmx_rttvar; /* estimated rtt variance */ 325 u_long rmx_bandwidth; /* estimated bandwidth */ 326 u_long rmx_cwnd; /* congestion window */ 327 u_long rmx_sendpipe; /* outbound delay-bandwidth product */ 328 u_long rmx_recvpipe; /* inbound delay-bandwidth product */ 329 }; 330 331 /* 332 * Used by tcp_maxmtu() to communicate interface specific features 333 * and limits at the time of connection setup. 334 */ 335 struct tcp_ifcap { 336 int ifcap; 337 u_int tsomax; 338 u_int tsomaxsegcount; 339 u_int tsomaxsegsize; 340 }; 341 342 #ifndef _NETINET_IN_PCB_H_ 343 struct in_conninfo; 344 #endif /* _NETINET_IN_PCB_H_ */ 345 346 struct tcptw { 347 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 348 tcp_seq snd_nxt; 349 tcp_seq rcv_nxt; 350 tcp_seq iss; 351 tcp_seq irs; 352 u_short last_win; /* cached window value */ 353 u_short tw_so_options; /* copy of so_options */ 354 struct ucred *tw_cred; /* user credentials */ 355 u_int32_t t_recent; 356 u_int32_t ts_offset; /* our timestamp offset */ 357 u_int t_starttime; 358 int tw_time; 359 TAILQ_ENTRY(tcptw) tw_2msl; 360 }; 361 362 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 363 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 364 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 365 366 /* 367 * The smoothed round-trip time and estimated variance 368 * are stored as fixed point numbers scaled by the values below. 369 * For convenience, these scales are also used in smoothing the average 370 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 371 * With these scales, srtt has 3 bits to the right of the binary point, 372 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 373 * binary point, and is smoothed with an ALPHA of 0.75. 374 */ 375 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 376 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 377 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 378 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 379 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 380 381 /* 382 * The initial retransmission should happen at rtt + 4 * rttvar. 383 * Because of the way we do the smoothing, srtt and rttvar 384 * will each average +1/2 tick of bias. When we compute 385 * the retransmit timer, we want 1/2 tick of rounding and 386 * 1 extra tick because of +-1/2 tick uncertainty in the 387 * firing of the timer. The bias will give us exactly the 388 * 1.5 tick we need. But, because the bias is 389 * statistical, we have to test that we don't drop below 390 * the minimum feasible timer (which is 2 ticks). 391 * This version of the macro adapted from a paper by Lawrence 392 * Brakmo and Larry Peterson which outlines a problem caused 393 * by insufficient precision in the original implementation, 394 * which results in inappropriately large RTO values for very 395 * fast networks. 396 */ 397 #define TCP_REXMTVAL(tp) \ 398 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 399 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 400 401 /* 402 * TCP statistics. 403 * Many of these should be kept per connection, 404 * but that's inconvenient at the moment. 405 */ 406 struct tcpstat { 407 uint64_t tcps_connattempt; /* connections initiated */ 408 uint64_t tcps_accepts; /* connections accepted */ 409 uint64_t tcps_connects; /* connections established */ 410 uint64_t tcps_drops; /* connections dropped */ 411 uint64_t tcps_conndrops; /* embryonic connections dropped */ 412 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 413 uint64_t tcps_closed; /* conn. closed (includes drops) */ 414 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 415 uint64_t tcps_rttupdated; /* times we succeeded */ 416 uint64_t tcps_delack; /* delayed acks sent */ 417 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 418 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 419 uint64_t tcps_persisttimeo; /* persist timeouts */ 420 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 421 uint64_t tcps_keepprobe; /* keepalive probes sent */ 422 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 423 424 uint64_t tcps_sndtotal; /* total packets sent */ 425 uint64_t tcps_sndpack; /* data packets sent */ 426 uint64_t tcps_sndbyte; /* data bytes sent */ 427 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 428 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 429 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 430 uint64_t tcps_sndacks; /* ack-only packets sent */ 431 uint64_t tcps_sndprobe; /* window probes sent */ 432 uint64_t tcps_sndurg; /* packets sent with URG only */ 433 uint64_t tcps_sndwinup; /* window update-only packets sent */ 434 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 435 436 uint64_t tcps_rcvtotal; /* total packets received */ 437 uint64_t tcps_rcvpack; /* packets received in sequence */ 438 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 439 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 440 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 441 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ 442 uint64_t tcps_rcvshort; /* packets received too short */ 443 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 444 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 445 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 446 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 447 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 448 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 449 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 450 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 451 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 452 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 453 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 454 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 455 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 456 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 457 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 458 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 459 uint64_t tcps_predack; /* times hdr predict ok for acks */ 460 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 461 uint64_t tcps_pcbcachemiss; 462 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 463 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 464 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 465 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 466 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 467 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 468 uint64_t tcps_persistdrop; /* timeout in persist state */ 469 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 470 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 471 uint64_t tcps_listendrop; /* listen queue overflows */ 472 uint64_t tcps_badrst; /* ignored RSTs in the window */ 473 474 uint64_t tcps_sc_added; /* entry added to syncache */ 475 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 476 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 477 uint64_t tcps_sc_dropped; /* could not reply to packet */ 478 uint64_t tcps_sc_completed; /* successful extraction of entry */ 479 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 480 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 481 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 482 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 483 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 484 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 485 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 486 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 487 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 488 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 489 490 uint64_t tcps_hc_added; /* entry added to hostcache */ 491 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 492 493 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 494 495 /* SACK related stats */ 496 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 497 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 498 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 499 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 500 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 501 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 502 503 /* ECN related stats */ 504 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 505 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 506 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 507 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 508 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 509 510 /* TCP_SIGNATURE related stats */ 511 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 512 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 513 uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */ 514 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 515 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 516 517 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 518 }; 519 520 #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ 521 522 #ifdef _KERNEL 523 #include <sys/counter.h> 524 525 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 526 /* 527 * In-kernel consumers can use these accessor macros directly to update 528 * stats. 529 */ 530 #define TCPSTAT_ADD(name, val) \ 531 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 532 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 533 534 /* 535 * Kernel module consumers must use this accessor macro. 536 */ 537 void kmod_tcpstat_inc(int statnum); 538 #define KMOD_TCPSTAT_INC(name) \ 539 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) 540 541 /* 542 * TCP specific helper hook point identifiers. 543 */ 544 #define HHOOK_TCP_EST_IN 0 545 #define HHOOK_TCP_EST_OUT 1 546 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 547 548 struct tcp_hhook_data { 549 struct tcpcb *tp; 550 struct tcphdr *th; 551 struct tcpopt *to; 552 long len; 553 int tso; 554 tcp_seq curack; 555 }; 556 #endif 557 558 /* 559 * TCB structure exported to user-land via sysctl(3). 560 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 561 * included. Not all of our clients do. 562 */ 563 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 564 struct xtcp_timer { 565 int tt_rexmt; /* retransmit timer */ 566 int tt_persist; /* retransmit persistence */ 567 int tt_keep; /* keepalive */ 568 int tt_2msl; /* 2*msl TIME_WAIT timer */ 569 int tt_delack; /* delayed ACK timer */ 570 int t_rcvtime; /* Time since last packet received */ 571 }; 572 struct xtcpcb { 573 size_t xt_len; 574 struct inpcb xt_inp; 575 struct tcpcb xt_tp; 576 struct xsocket xt_socket; 577 struct xtcp_timer xt_timer; 578 u_quad_t xt_alignment_hack; 579 }; 580 #endif 581 582 /* 583 * Identifiers for TCP sysctl nodes 584 */ 585 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 586 #define TCPCTL_MSSDFLT 3 /* MSS default */ 587 #define TCPCTL_STATS 4 /* statistics (read-only) */ 588 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 589 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 590 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 591 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 592 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 593 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 594 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 595 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 596 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 597 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 598 #define TCPCTL_DROP 15 /* drop tcp connection */ 599 600 #ifdef _KERNEL 601 #ifdef SYSCTL_DECL 602 SYSCTL_DECL(_net_inet_tcp); 603 SYSCTL_DECL(_net_inet_tcp_sack); 604 MALLOC_DECLARE(M_TCPLOG); 605 #endif 606 607 VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ 608 VNET_DECLARE(struct inpcbinfo, tcbinfo); 609 extern int tcp_log_in_vain; 610 VNET_DECLARE(int, tcp_mssdflt); /* XXX */ 611 VNET_DECLARE(int, tcp_minmss); 612 VNET_DECLARE(int, tcp_delack_enabled); 613 VNET_DECLARE(int, tcp_do_rfc3390); 614 VNET_DECLARE(int, tcp_do_initcwnd10); 615 VNET_DECLARE(int, tcp_sendspace); 616 VNET_DECLARE(int, tcp_recvspace); 617 VNET_DECLARE(int, path_mtu_discovery); 618 VNET_DECLARE(int, tcp_do_rfc3465); 619 VNET_DECLARE(int, tcp_abc_l_var); 620 #define V_tcb VNET(tcb) 621 #define V_tcbinfo VNET(tcbinfo) 622 #define V_tcp_mssdflt VNET(tcp_mssdflt) 623 #define V_tcp_minmss VNET(tcp_minmss) 624 #define V_tcp_delack_enabled VNET(tcp_delack_enabled) 625 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 626 #define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) 627 #define V_tcp_sendspace VNET(tcp_sendspace) 628 #define V_tcp_recvspace VNET(tcp_recvspace) 629 #define V_path_mtu_discovery VNET(path_mtu_discovery) 630 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 631 #define V_tcp_abc_l_var VNET(tcp_abc_l_var) 632 633 VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ 634 VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ 635 #define V_tcp_do_sack VNET(tcp_do_sack) 636 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 637 638 VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ 639 VNET_DECLARE(int, tcp_ecn_maxretries); 640 #define V_tcp_do_ecn VNET(tcp_do_ecn) 641 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 642 643 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 644 #define V_tcp_hhh VNET(tcp_hhh) 645 646 int tcp_addoptions(struct tcpopt *, u_char *); 647 int tcp_ccalgounload(struct cc_algo *unload_algo); 648 struct tcpcb * 649 tcp_close(struct tcpcb *); 650 void tcp_discardcb(struct tcpcb *); 651 void tcp_twstart(struct tcpcb *); 652 void tcp_twclose(struct tcptw *, int); 653 void tcp_ctlinput(int, struct sockaddr *, void *); 654 int tcp_ctloutput(struct socket *, struct sockopt *); 655 struct tcpcb * 656 tcp_drop(struct tcpcb *, int); 657 void tcp_drain(void); 658 void tcp_init(void); 659 #ifdef VIMAGE 660 void tcp_destroy(void); 661 #endif 662 void tcp_fini(void *); 663 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 664 const void *); 665 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 666 const void *); 667 int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); 668 void tcp_reass_flush(struct tcpcb *); 669 int tcp_input(struct mbuf **, int *, int); 670 u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 671 u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 672 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 673 struct tcp_ifcap *); 674 void tcp_mss(struct tcpcb *, int); 675 int tcp_mssopt(struct in_conninfo *); 676 struct inpcb * 677 tcp_drop_syn_sent(struct inpcb *, int); 678 struct inpcb * 679 tcp_mtudisc(struct inpcb *, int); 680 struct tcpcb * 681 tcp_newtcpcb(struct inpcb *); 682 int tcp_output(struct tcpcb *); 683 void tcp_state_change(struct tcpcb *, int); 684 void tcp_respond(struct tcpcb *, void *, 685 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 686 void tcp_tw_init(void); 687 #ifdef VIMAGE 688 void tcp_tw_destroy(void); 689 #endif 690 void tcp_tw_zone_change(void); 691 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 692 struct mbuf *, int); 693 void tcp_setpersist(struct tcpcb *); 694 #ifdef TCP_SIGNATURE 695 struct secasvar; 696 struct secasvar *tcp_get_sav(struct mbuf *, u_int); 697 int tcp_signature_do_compute(struct mbuf *, int, int, u_char *, 698 struct secasvar *); 699 int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); 700 int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, 701 struct tcphdr *, u_int); 702 int tcp_signature_check(struct mbuf *m, int off0, int tlen, int optlen, 703 struct tcpopt *to, struct tcphdr *th, u_int tcpbflag); 704 #endif 705 void tcp_slowtimo(void); 706 struct tcptemp * 707 tcpip_maketemplate(struct inpcb *); 708 void tcpip_fillheaders(struct inpcb *, void *, void *); 709 void tcp_timer_activate(struct tcpcb *, int, u_int); 710 int tcp_timer_active(struct tcpcb *, int); 711 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 712 /* 713 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 714 */ 715 void tcp_hc_init(void); 716 #ifdef VIMAGE 717 void tcp_hc_destroy(void); 718 #endif 719 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 720 u_long tcp_hc_getmtu(struct in_conninfo *); 721 void tcp_hc_updatemtu(struct in_conninfo *, u_long); 722 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 723 724 extern struct pr_usrreqs tcp_usrreqs; 725 tcp_seq tcp_new_isn(struct tcpcb *); 726 727 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 728 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 729 void tcp_clean_sackreport(struct tcpcb *tp); 730 void tcp_sack_adjust(struct tcpcb *tp); 731 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 732 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 733 void tcp_free_sackholes(struct tcpcb *tp); 734 int tcp_newreno(struct tcpcb *, struct tcphdr *); 735 u_long tcp_seq_subtract(u_long, u_long ); 736 737 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 738 739 static inline void 740 tcp_fields_to_host(struct tcphdr *th) 741 { 742 743 th->th_seq = ntohl(th->th_seq); 744 th->th_ack = ntohl(th->th_ack); 745 th->th_win = ntohs(th->th_win); 746 th->th_urp = ntohs(th->th_urp); 747 } 748 749 #ifdef TCP_SIGNATURE 750 static inline void 751 tcp_fields_to_net(struct tcphdr *th) 752 { 753 754 th->th_seq = htonl(th->th_seq); 755 th->th_ack = htonl(th->th_ack); 756 th->th_win = htons(th->th_win); 757 th->th_urp = htons(th->th_urp); 758 } 759 #endif 760 #endif /* _KERNEL */ 761 762 #endif /* _NETINET_TCP_VAR_H_ */ 763