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 #include <sys/mbuf.h> 41 42 /* 43 * Kernel variables for tcp. 44 */ 45 VNET_DECLARE(int, tcp_do_rfc1323); 46 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 47 48 #endif /* _KERNEL */ 49 50 /* TCP segment queue entry */ 51 struct tseg_qent { 52 LIST_ENTRY(tseg_qent) tqe_q; 53 int tqe_len; /* TCP segment data length */ 54 struct tcphdr *tqe_th; /* a pointer to tcp header */ 55 struct mbuf *tqe_m; /* mbuf contains packet */ 56 }; 57 LIST_HEAD(tsegqe_head, tseg_qent); 58 59 struct sackblk { 60 tcp_seq start; /* start seq no. of sack block */ 61 tcp_seq end; /* end seq no. */ 62 }; 63 64 struct sackhole { 65 tcp_seq start; /* start seq no. of hole */ 66 tcp_seq end; /* end seq no. */ 67 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 68 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 69 }; 70 71 struct sackhint { 72 struct sackhole *nexthole; 73 int sack_bytes_rexmit; 74 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 75 76 int ispare; /* explicit pad for 64bit alignment */ 77 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */ 78 }; 79 80 struct tcptemp { 81 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 82 struct tcphdr tt_t; 83 }; 84 85 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 86 87 /* 88 * Tcp control block, one per tcp; fields: 89 * Organized for 16 byte cacheline efficiency. 90 */ 91 struct tcpcb { 92 struct tsegqe_head t_segq; /* segment reassembly queue */ 93 void *t_pspare[2]; /* new reassembly queue */ 94 int t_segqlen; /* segment reassembly queue length */ 95 int t_dupacks; /* consecutive dup acks recd */ 96 97 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 98 99 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 100 int t_state; /* state of this connection */ 101 u_int t_flags; 102 103 struct vnet *t_vnet; /* back pointer to parent vnet */ 104 105 tcp_seq snd_una; /* sent but unacknowledged */ 106 tcp_seq snd_max; /* highest sequence number sent; 107 * used to recognize retransmits 108 */ 109 tcp_seq snd_nxt; /* send next */ 110 tcp_seq snd_up; /* send urgent pointer */ 111 112 tcp_seq snd_wl1; /* window update seg seq number */ 113 tcp_seq snd_wl2; /* window update seg ack number */ 114 tcp_seq iss; /* initial send sequence number */ 115 tcp_seq irs; /* initial receive sequence number */ 116 117 tcp_seq rcv_nxt; /* receive next */ 118 tcp_seq rcv_adv; /* advertised window */ 119 u_long rcv_wnd; /* receive window */ 120 tcp_seq rcv_up; /* receive urgent pointer */ 121 122 u_long snd_wnd; /* send window */ 123 u_long snd_cwnd; /* congestion-controlled window */ 124 u_long snd_spare1; /* unused */ 125 u_long snd_ssthresh; /* snd_cwnd size threshold for 126 * for slow start exponential to 127 * linear switch 128 */ 129 u_long snd_spare2; /* unused */ 130 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 131 132 u_int t_maxopd; /* mss plus options */ 133 134 u_int t_rcvtime; /* inactivity time */ 135 u_int t_starttime; /* time connection was established */ 136 u_int t_rtttime; /* RTT measurement start time */ 137 tcp_seq t_rtseq; /* sequence number being timed */ 138 139 u_int t_bw_spare1; /* unused */ 140 tcp_seq t_bw_spare2; /* unused */ 141 142 int t_rxtcur; /* current retransmit value (ticks) */ 143 u_int t_maxseg; /* maximum segment size */ 144 int t_srtt; /* smoothed round-trip time */ 145 int t_rttvar; /* variance in round-trip time */ 146 147 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 148 u_int t_rttmin; /* minimum rtt allowed */ 149 u_int t_rttbest; /* best rtt we've seen */ 150 u_long t_rttupdated; /* number of times rtt sampled */ 151 u_long max_sndwnd; /* largest window peer has offered */ 152 153 int t_softerror; /* possible error not yet reported */ 154 /* out-of-band data */ 155 char t_oobflags; /* have some */ 156 char t_iobc; /* input character */ 157 /* RFC 1323 variables */ 158 u_char snd_scale; /* window scaling for send window */ 159 u_char rcv_scale; /* window scaling for recv window */ 160 u_char request_r_scale; /* pending window scaling */ 161 u_int32_t ts_recent; /* timestamp echo data */ 162 u_int ts_recent_age; /* when last updated */ 163 u_int32_t ts_offset; /* our timestamp offset */ 164 165 tcp_seq last_ack_sent; 166 /* experimental */ 167 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 168 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 169 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 170 int t_sndzerowin; /* zero-window updates sent */ 171 u_int t_badrxtwin; /* window for retransmit recovery */ 172 u_char snd_limited; /* segments limited transmitted */ 173 /* SACK related state */ 174 int snd_numholes; /* number of holes seen by sender */ 175 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 176 /* SACK scoreboard (sorted) */ 177 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 178 int rcv_numsacks; /* # distinct sack blks present */ 179 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 180 tcp_seq sack_newdata; /* New data xmitted in this recovery 181 episode starts at this seq number */ 182 struct sackhint sackhint; /* SACK scoreboard hint */ 183 int t_rttlow; /* smallest observerved RTT */ 184 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 185 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 186 struct toedev *tod; /* toedev handling this connection */ 187 int t_sndrexmitpack; /* retransmit packets sent */ 188 int t_rcvoopack; /* out-of-order packets received */ 189 void *t_toe; /* TOE pcb pointer */ 190 int t_bytes_acked; /* # bytes acked during current RTT */ 191 struct cc_algo *cc_algo; /* congestion control algorithm */ 192 struct cc_var *ccv; /* congestion control specific vars */ 193 struct osd *osd; /* storage for Khelp module data */ 194 195 u_int t_keepinit; /* time to establish connection */ 196 u_int t_keepidle; /* time before keepalive probes begin */ 197 u_int t_keepintvl; /* interval between keepalives */ 198 u_int t_keepcnt; /* number of keepalives before close */ 199 200 u_int t_tsomax; /* TSO total burst length limit in bytes */ 201 u_int t_tsomaxsegcount; /* TSO maximum segment count */ 202 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 203 u_int t_pmtud_saved_maxopd; /* pre-blackhole MSS */ 204 u_int t_flags2; /* More tcpcb flags storage */ 205 206 uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ 207 void *t_pspare2[4]; /* 1 TCP_SIGNATURE, 3 TBD */ 208 #if defined(_KERNEL) && defined(TCPPCAP) 209 struct mbufq t_inpkts; /* List of saved input packets. */ 210 struct mbufq t_outpkts; /* List of saved output packets. */ 211 #ifdef _LP64 212 uint64_t _pad[0]; /* all used! */ 213 #else 214 uint64_t _pad[2]; /* 2 are available */ 215 #endif /* _LP64 */ 216 #else 217 uint64_t _pad[6]; 218 #endif /* defined(_KERNEL) && defined(TCPPCAP) */ 219 }; 220 221 /* 222 * Flags and utility macros for the t_flags field. 223 */ 224 #define TF_ACKNOW 0x000001 /* ack peer immediately */ 225 #define TF_DELACK 0x000002 /* ack, but try to delay it */ 226 #define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ 227 #define TF_NOOPT 0x000008 /* don't use tcp options */ 228 #define TF_SENTFIN 0x000010 /* have sent FIN */ 229 #define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ 230 #define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ 231 #define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ 232 #define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ 233 #define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ 234 #define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ 235 #define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ 236 #define TF_NOPUSH 0x001000 /* don't push */ 237 #define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ 238 #define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ 239 #define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ 240 #define TF_LASTIDLE 0x040000 /* connection was previously idle */ 241 #define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ 242 #define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ 243 #define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ 244 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 245 #define TF_FORCEDATA 0x800000 /* force out a byte */ 246 #define TF_TSO 0x1000000 /* TSO enabled on this connection */ 247 #define TF_TOE 0x2000000 /* this connection is offloaded */ 248 #define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ 249 #define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ 250 #define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ 251 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 252 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 253 254 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 255 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 256 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 257 258 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 259 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 260 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 261 262 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 263 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 264 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 265 266 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 267 268 /* 269 * Flags for the t_oobflags field. 270 */ 271 #define TCPOOB_HAVEDATA 0x01 272 #define TCPOOB_HADDATA 0x02 273 274 #ifdef TCP_SIGNATURE 275 /* 276 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 277 * for SADB verification and lookup. 278 */ 279 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 280 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 281 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 282 /* 283 * Only a single SA per host may be specified at this time. An SPI is 284 * needed in order for the KEY_ALLOCSA() lookup to work. 285 */ 286 #define TCP_SIG_SPI 0x1000 287 #endif /* TCP_SIGNATURE */ 288 289 /* 290 * Flags for PLPMTU handling, t_flags2 291 */ 292 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 293 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 294 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 295 296 /* 297 * Structure to hold TCP options that are only used during segment 298 * processing (in tcp_input), but not held in the tcpcb. 299 * It's basically used to reduce the number of parameters 300 * to tcp_dooptions and tcp_addoptions. 301 * The binary order of the to_flags is relevant for packing of the 302 * options in tcp_addoptions. 303 */ 304 struct tcpopt { 305 u_int64_t to_flags; /* which options are present */ 306 #define TOF_MSS 0x0001 /* maximum segment size */ 307 #define TOF_SCALE 0x0002 /* window scaling */ 308 #define TOF_SACKPERM 0x0004 /* SACK permitted */ 309 #define TOF_TS 0x0010 /* timestamp */ 310 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 311 #define TOF_SACK 0x0080 /* Peer sent SACK option */ 312 #define TOF_MAXOPT 0x0100 313 u_int32_t to_tsval; /* new timestamp */ 314 u_int32_t to_tsecr; /* reflected timestamp */ 315 u_char *to_sacks; /* pointer to the first SACK blocks */ 316 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 317 u_int16_t to_mss; /* maximum segment size */ 318 u_int8_t to_wscale; /* window scaling */ 319 u_int8_t to_nsacks; /* number of SACK blocks */ 320 u_int32_t to_spare; /* UTO */ 321 }; 322 323 /* 324 * Flags for tcp_dooptions. 325 */ 326 #define TO_SYN 0x01 /* parse SYN-only options */ 327 328 struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 329 u_long rmx_mtu; /* MTU for this path */ 330 u_long rmx_ssthresh; /* outbound gateway buffer limit */ 331 u_long rmx_rtt; /* estimated round trip time */ 332 u_long rmx_rttvar; /* estimated rtt variance */ 333 u_long rmx_bandwidth; /* estimated bandwidth */ 334 u_long rmx_cwnd; /* congestion window */ 335 u_long rmx_sendpipe; /* outbound delay-bandwidth product */ 336 u_long rmx_recvpipe; /* inbound delay-bandwidth product */ 337 }; 338 339 /* 340 * Used by tcp_maxmtu() to communicate interface specific features 341 * and limits at the time of connection setup. 342 */ 343 struct tcp_ifcap { 344 int ifcap; 345 u_int tsomax; 346 u_int tsomaxsegcount; 347 u_int tsomaxsegsize; 348 }; 349 350 #ifndef _NETINET_IN_PCB_H_ 351 struct in_conninfo; 352 #endif /* _NETINET_IN_PCB_H_ */ 353 354 struct tcptw { 355 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 356 tcp_seq snd_nxt; 357 tcp_seq rcv_nxt; 358 tcp_seq iss; 359 tcp_seq irs; 360 u_short last_win; /* cached window value */ 361 u_short tw_so_options; /* copy of so_options */ 362 struct ucred *tw_cred; /* user credentials */ 363 u_int32_t t_recent; 364 u_int32_t ts_offset; /* our timestamp offset */ 365 u_int t_starttime; 366 int tw_time; 367 TAILQ_ENTRY(tcptw) tw_2msl; 368 void *tw_pspare; /* TCP_SIGNATURE */ 369 u_int *tw_spare; /* TCP_SIGNATURE */ 370 }; 371 372 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 373 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 374 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 375 376 /* 377 * The smoothed round-trip time and estimated variance 378 * are stored as fixed point numbers scaled by the values below. 379 * For convenience, these scales are also used in smoothing the average 380 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 381 * With these scales, srtt has 3 bits to the right of the binary point, 382 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 383 * binary point, and is smoothed with an ALPHA of 0.75. 384 */ 385 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 386 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 387 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 388 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 389 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 390 391 /* 392 * The initial retransmission should happen at rtt + 4 * rttvar. 393 * Because of the way we do the smoothing, srtt and rttvar 394 * will each average +1/2 tick of bias. When we compute 395 * the retransmit timer, we want 1/2 tick of rounding and 396 * 1 extra tick because of +-1/2 tick uncertainty in the 397 * firing of the timer. The bias will give us exactly the 398 * 1.5 tick we need. But, because the bias is 399 * statistical, we have to test that we don't drop below 400 * the minimum feasible timer (which is 2 ticks). 401 * This version of the macro adapted from a paper by Lawrence 402 * Brakmo and Larry Peterson which outlines a problem caused 403 * by insufficient precision in the original implementation, 404 * which results in inappropriately large RTO values for very 405 * fast networks. 406 */ 407 #define TCP_REXMTVAL(tp) \ 408 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 409 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 410 411 /* 412 * TCP statistics. 413 * Many of these should be kept per connection, 414 * but that's inconvenient at the moment. 415 */ 416 struct tcpstat { 417 uint64_t tcps_connattempt; /* connections initiated */ 418 uint64_t tcps_accepts; /* connections accepted */ 419 uint64_t tcps_connects; /* connections established */ 420 uint64_t tcps_drops; /* connections dropped */ 421 uint64_t tcps_conndrops; /* embryonic connections dropped */ 422 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 423 uint64_t tcps_closed; /* conn. closed (includes drops) */ 424 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 425 uint64_t tcps_rttupdated; /* times we succeeded */ 426 uint64_t tcps_delack; /* delayed acks sent */ 427 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 428 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 429 uint64_t tcps_persisttimeo; /* persist timeouts */ 430 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 431 uint64_t tcps_keepprobe; /* keepalive probes sent */ 432 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 433 434 uint64_t tcps_sndtotal; /* total packets sent */ 435 uint64_t tcps_sndpack; /* data packets sent */ 436 uint64_t tcps_sndbyte; /* data bytes sent */ 437 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 438 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 439 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 440 uint64_t tcps_sndacks; /* ack-only packets sent */ 441 uint64_t tcps_sndprobe; /* window probes sent */ 442 uint64_t tcps_sndurg; /* packets sent with URG only */ 443 uint64_t tcps_sndwinup; /* window update-only packets sent */ 444 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 445 446 uint64_t tcps_rcvtotal; /* total packets received */ 447 uint64_t tcps_rcvpack; /* packets received in sequence */ 448 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 449 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 450 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 451 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ 452 uint64_t tcps_rcvshort; /* packets received too short */ 453 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 454 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 455 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 456 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 457 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 458 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 459 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 460 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 461 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 462 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 463 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 464 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 465 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 466 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 467 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 468 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 469 uint64_t tcps_predack; /* times hdr predict ok for acks */ 470 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 471 uint64_t tcps_pcbcachemiss; 472 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 473 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 474 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 475 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 476 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 477 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 478 uint64_t tcps_persistdrop; /* timeout in persist state */ 479 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 480 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 481 uint64_t tcps_listendrop; /* listen queue overflows */ 482 uint64_t tcps_badrst; /* ignored RSTs in the window */ 483 484 uint64_t tcps_sc_added; /* entry added to syncache */ 485 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 486 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 487 uint64_t tcps_sc_dropped; /* could not reply to packet */ 488 uint64_t tcps_sc_completed; /* successful extraction of entry */ 489 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 490 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 491 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 492 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 493 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 494 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 495 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 496 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 497 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 498 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 499 500 uint64_t tcps_hc_added; /* entry added to hostcache */ 501 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 502 503 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 504 505 /* SACK related stats */ 506 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 507 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 508 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 509 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 510 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 511 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 512 513 /* ECN related stats */ 514 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 515 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 516 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 517 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 518 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 519 520 /* TCP_SIGNATURE related stats */ 521 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 522 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 523 uint64_t tcps_sig_err_buildsig; /* Mismatching signature received */ 524 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 525 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 526 527 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 528 }; 529 530 #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ 531 532 #ifdef _KERNEL 533 #include <sys/counter.h> 534 535 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 536 /* 537 * In-kernel consumers can use these accessor macros directly to update 538 * stats. 539 */ 540 #define TCPSTAT_ADD(name, val) \ 541 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 542 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 543 544 /* 545 * Kernel module consumers must use this accessor macro. 546 */ 547 void kmod_tcpstat_inc(int statnum); 548 #define KMOD_TCPSTAT_INC(name) \ 549 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) 550 551 /* 552 * TCP specific helper hook point identifiers. 553 */ 554 #define HHOOK_TCP_EST_IN 0 555 #define HHOOK_TCP_EST_OUT 1 556 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 557 558 struct tcp_hhook_data { 559 struct tcpcb *tp; 560 struct tcphdr *th; 561 struct tcpopt *to; 562 long len; 563 int tso; 564 tcp_seq curack; 565 }; 566 #endif 567 568 /* 569 * TCB structure exported to user-land via sysctl(3). 570 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 571 * included. Not all of our clients do. 572 */ 573 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 574 struct xtcp_timer { 575 int tt_rexmt; /* retransmit timer */ 576 int tt_persist; /* retransmit persistence */ 577 int tt_keep; /* keepalive */ 578 int tt_2msl; /* 2*msl TIME_WAIT timer */ 579 int tt_delack; /* delayed ACK timer */ 580 int t_rcvtime; /* Time since last packet received */ 581 }; 582 struct xtcpcb { 583 size_t xt_len; 584 struct inpcb xt_inp; 585 struct tcpcb xt_tp; 586 struct xsocket xt_socket; 587 struct xtcp_timer xt_timer; 588 u_quad_t xt_alignment_hack; 589 }; 590 #endif 591 592 /* 593 * Identifiers for TCP sysctl nodes 594 */ 595 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 596 #define TCPCTL_MSSDFLT 3 /* MSS default */ 597 #define TCPCTL_STATS 4 /* statistics (read-only) */ 598 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 599 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 600 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 601 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 602 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 603 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 604 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 605 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 606 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 607 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 608 #define TCPCTL_DROP 15 /* drop tcp connection */ 609 610 #ifdef _KERNEL 611 #ifdef SYSCTL_DECL 612 SYSCTL_DECL(_net_inet_tcp); 613 SYSCTL_DECL(_net_inet_tcp_sack); 614 MALLOC_DECLARE(M_TCPLOG); 615 #endif 616 617 VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ 618 VNET_DECLARE(struct inpcbinfo, tcbinfo); 619 extern int tcp_log_in_vain; 620 VNET_DECLARE(int, tcp_mssdflt); /* XXX */ 621 VNET_DECLARE(int, tcp_minmss); 622 VNET_DECLARE(int, tcp_delack_enabled); 623 VNET_DECLARE(int, tcp_do_rfc3390); 624 VNET_DECLARE(int, tcp_do_initcwnd10); 625 VNET_DECLARE(int, tcp_sendspace); 626 VNET_DECLARE(int, tcp_recvspace); 627 VNET_DECLARE(int, path_mtu_discovery); 628 VNET_DECLARE(int, tcp_do_rfc3465); 629 VNET_DECLARE(int, tcp_abc_l_var); 630 #define V_tcb VNET(tcb) 631 #define V_tcbinfo VNET(tcbinfo) 632 #define V_tcp_mssdflt VNET(tcp_mssdflt) 633 #define V_tcp_minmss VNET(tcp_minmss) 634 #define V_tcp_delack_enabled VNET(tcp_delack_enabled) 635 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 636 #define V_tcp_do_initcwnd10 VNET(tcp_do_initcwnd10) 637 #define V_tcp_sendspace VNET(tcp_sendspace) 638 #define V_tcp_recvspace VNET(tcp_recvspace) 639 #define V_path_mtu_discovery VNET(path_mtu_discovery) 640 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 641 #define V_tcp_abc_l_var VNET(tcp_abc_l_var) 642 643 VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ 644 VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ 645 #define V_tcp_do_sack VNET(tcp_do_sack) 646 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 647 648 VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ 649 VNET_DECLARE(int, tcp_ecn_maxretries); 650 #define V_tcp_do_ecn VNET(tcp_do_ecn) 651 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 652 653 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 654 #define V_tcp_hhh VNET(tcp_hhh) 655 656 int tcp_addoptions(struct tcpopt *, u_char *); 657 int tcp_ccalgounload(struct cc_algo *unload_algo); 658 struct tcpcb * 659 tcp_close(struct tcpcb *); 660 void tcp_discardcb(struct tcpcb *); 661 void tcp_twstart(struct tcpcb *); 662 void tcp_twclose(struct tcptw *, int); 663 void tcp_ctlinput(int, struct sockaddr *, void *); 664 int tcp_ctloutput(struct socket *, struct sockopt *); 665 struct tcpcb * 666 tcp_drop(struct tcpcb *, int); 667 void tcp_drain(void); 668 void tcp_init(void); 669 #ifdef VIMAGE 670 void tcp_destroy(void); 671 #endif 672 void tcp_fini(void *); 673 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 674 const void *); 675 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 676 const void *); 677 int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); 678 void tcp_reass_global_init(void); 679 void tcp_reass_flush(struct tcpcb *); 680 int tcp_input(struct mbuf **, int *, int); 681 u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 682 u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 683 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 684 struct tcp_ifcap *); 685 void tcp_mss(struct tcpcb *, int); 686 int tcp_mssopt(struct in_conninfo *); 687 struct inpcb * 688 tcp_drop_syn_sent(struct inpcb *, int); 689 struct tcpcb * 690 tcp_newtcpcb(struct inpcb *); 691 int tcp_output(struct tcpcb *); 692 void tcp_state_change(struct tcpcb *, int); 693 void tcp_respond(struct tcpcb *, void *, 694 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 695 void tcp_tw_init(void); 696 #ifdef VIMAGE 697 void tcp_tw_destroy(void); 698 #endif 699 void tcp_tw_zone_change(void); 700 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 701 struct mbuf *, int); 702 void tcp_setpersist(struct tcpcb *); 703 #ifdef TCP_SIGNATURE 704 struct secasvar; 705 struct secasvar *tcp_get_sav(struct mbuf *, u_int); 706 int tcp_signature_do_compute(struct mbuf *, int, int, u_char *, 707 struct secasvar *); 708 int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int); 709 int tcp_signature_verify(struct mbuf *, int, int, int, struct tcpopt *, 710 struct tcphdr *, u_int); 711 int tcp_signature_check(struct mbuf *m, int off0, int tlen, int optlen, 712 struct tcpopt *to, struct tcphdr *th, u_int tcpbflag); 713 #endif 714 void tcp_slowtimo(void); 715 struct tcptemp * 716 tcpip_maketemplate(struct inpcb *); 717 void tcpip_fillheaders(struct inpcb *, void *, void *); 718 void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 719 int tcp_timer_active(struct tcpcb *, uint32_t); 720 void tcp_timer_stop(struct tcpcb *, uint32_t); 721 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 722 /* 723 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 724 */ 725 void tcp_hc_init(void); 726 #ifdef VIMAGE 727 void tcp_hc_destroy(void); 728 #endif 729 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 730 u_long tcp_hc_getmtu(struct in_conninfo *); 731 void tcp_hc_updatemtu(struct in_conninfo *, u_long); 732 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 733 734 extern struct pr_usrreqs tcp_usrreqs; 735 tcp_seq tcp_new_isn(struct tcpcb *); 736 737 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 738 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 739 void tcp_clean_sackreport(struct tcpcb *tp); 740 void tcp_sack_adjust(struct tcpcb *tp); 741 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 742 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 743 void tcp_free_sackholes(struct tcpcb *tp); 744 int tcp_newreno(struct tcpcb *, struct tcphdr *); 745 u_long tcp_seq_subtract(u_long, u_long ); 746 747 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 748 749 static inline void 750 tcp_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 760 static inline void 761 tcp_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