tcp_var.h revision 315514
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/11/sys/netinet/tcp_var.h 315514 2017-03-18 22:04:20Z ae $ 31 */ 32 33#ifndef _NETINET_TCP_VAR_H_ 34#define _NETINET_TCP_VAR_H_ 35 36#include <netinet/tcp.h> 37#include <netinet/tcp_fsm.h> 38 39#ifdef _KERNEL 40#include <net/vnet.h> 41#include <sys/mbuf.h> 42 43/* 44 * Kernel variables for tcp. 45 */ 46VNET_DECLARE(int, tcp_do_rfc1323); 47#define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 48 49#endif /* _KERNEL */ 50 51/* TCP segment queue entry */ 52struct tseg_qent { 53 LIST_ENTRY(tseg_qent) tqe_q; 54 int tqe_len; /* TCP segment data length */ 55 struct tcphdr *tqe_th; /* a pointer to tcp header */ 56 struct mbuf *tqe_m; /* mbuf contains packet */ 57}; 58LIST_HEAD(tsegqe_head, tseg_qent); 59 60struct sackblk { 61 tcp_seq start; /* start seq no. of sack block */ 62 tcp_seq end; /* end seq no. */ 63}; 64 65struct sackhole { 66 tcp_seq start; /* start seq no. of hole */ 67 tcp_seq end; /* end seq no. */ 68 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 69 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 70}; 71 72struct sackhint { 73 struct sackhole *nexthole; 74 int sack_bytes_rexmit; 75 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 76 77 int ispare; /* explicit pad for 64bit alignment */ 78 int sacked_bytes; /* 79 * Total sacked bytes reported by the 80 * receiver via sack option 81 */ 82 uint32_t _pad1[1]; /* TBD */ 83 uint64_t _pad[1]; /* TBD */ 84}; 85 86struct tcptemp { 87 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 88 struct tcphdr tt_t; 89}; 90 91#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 92 93/* 94 * TODO: We yet need to brave plowing in 95 * to tcp_input() and the pru_usrreq() block. 96 * Right now these go to the old standards which 97 * are somewhat ok, but in the long term may 98 * need to be changed. If we do tackle tcp_input() 99 * then we need to get rid of the tcp_do_segment() 100 * function below. 101 */ 102/* Flags for tcp functions */ 103#define TCP_FUNC_BEING_REMOVED 0x01 /* Can no longer be referenced */ 104struct tcpcb; 105struct inpcb; 106struct sockopt; 107struct socket; 108 109/* 110 * If defining the optional tcp_timers, in the 111 * tfb_tcp_timer_stop call you must use the 112 * callout_async_drain() function with the 113 * tcp_timer_discard callback. You should check 114 * the return of callout_async_drain() and if 0 115 * increment tt_draincnt. Since the timer sub-system 116 * does not know your callbacks you must provide a 117 * stop_all function that loops through and calls 118 * tcp_timer_stop() with each of your defined timers. 119 */ 120struct tcp_function_block { 121 char tfb_tcp_block_name[TCP_FUNCTION_NAME_LEN_MAX]; 122 int (*tfb_tcp_output)(struct tcpcb *); 123 void (*tfb_tcp_do_segment)(struct mbuf *, struct tcphdr *, 124 struct socket *, struct tcpcb *, 125 int, int, uint8_t, 126 int); 127 int (*tfb_tcp_ctloutput)(struct socket *so, struct sockopt *sopt, 128 struct inpcb *inp, struct tcpcb *tp); 129 /* Optional memory allocation/free routine */ 130 void (*tfb_tcp_fb_init)(struct tcpcb *); 131 void (*tfb_tcp_fb_fini)(struct tcpcb *); 132 /* Optional timers, must define all if you define one */ 133 int (*tfb_tcp_timer_stop_all)(struct tcpcb *); 134 void (*tfb_tcp_timer_activate)(struct tcpcb *, 135 uint32_t, u_int); 136 int (*tfb_tcp_timer_active)(struct tcpcb *, uint32_t); 137 void (*tfb_tcp_timer_stop)(struct tcpcb *, uint32_t); 138 void (*tfb_tcp_rexmit_tmr)(struct tcpcb *); 139 volatile uint32_t tfb_refcnt; 140 uint32_t tfb_flags; 141}; 142 143struct tcp_function { 144 TAILQ_ENTRY(tcp_function) tf_next; 145 struct tcp_function_block *tf_fb; 146}; 147 148TAILQ_HEAD(tcp_funchead, tcp_function); 149 150/* 151 * Tcp control block, one per tcp; fields: 152 * Organized for 16 byte cacheline efficiency. 153 */ 154struct tcpcb { 155 struct tsegqe_head t_segq; /* segment reassembly queue */ 156 void *t_pspare[2]; /* new reassembly queue */ 157 int t_segqlen; /* segment reassembly queue length */ 158 int t_dupacks; /* consecutive dup acks recd */ 159 160 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 161 162 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 163 int t_state; /* state of this connection */ 164 u_int t_flags; 165 166 struct vnet *t_vnet; /* back pointer to parent vnet */ 167 168 tcp_seq snd_una; /* sent but unacknowledged */ 169 tcp_seq snd_max; /* highest sequence number sent; 170 * used to recognize retransmits 171 */ 172 tcp_seq snd_nxt; /* send next */ 173 tcp_seq snd_up; /* send urgent pointer */ 174 175 tcp_seq snd_wl1; /* window update seg seq number */ 176 tcp_seq snd_wl2; /* window update seg ack number */ 177 tcp_seq iss; /* initial send sequence number */ 178 tcp_seq irs; /* initial receive sequence number */ 179 180 tcp_seq rcv_nxt; /* receive next */ 181 tcp_seq rcv_adv; /* advertised window */ 182 u_long rcv_wnd; /* receive window */ 183 tcp_seq rcv_up; /* receive urgent pointer */ 184 185 u_long snd_wnd; /* send window */ 186 u_long snd_cwnd; /* congestion-controlled window */ 187 u_long snd_spare1; /* unused */ 188 u_long snd_ssthresh; /* snd_cwnd size threshold for 189 * for slow start exponential to 190 * linear switch 191 */ 192 u_long snd_spare2; /* unused */ 193 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 194 195 u_int t_rcvtime; /* inactivity time */ 196 u_int t_starttime; /* time connection was established */ 197 u_int t_rtttime; /* RTT measurement start time */ 198 tcp_seq t_rtseq; /* sequence number being timed */ 199 200 u_int t_bw_spare1; /* unused */ 201 tcp_seq t_bw_spare2; /* unused */ 202 203 int t_rxtcur; /* current retransmit value (ticks) */ 204 u_int t_maxseg; /* maximum segment size */ 205 u_int t_pmtud_saved_maxseg; /* pre-blackhole MSS */ 206 int t_srtt; /* smoothed round-trip time */ 207 int t_rttvar; /* variance in round-trip time */ 208 209 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 210 u_int t_rttmin; /* minimum rtt allowed */ 211 u_int t_rttbest; /* best rtt we've seen */ 212 u_long t_rttupdated; /* number of times rtt sampled */ 213 u_long max_sndwnd; /* largest window peer has offered */ 214 215 int t_softerror; /* possible error not yet reported */ 216/* out-of-band data */ 217 char t_oobflags; /* have some */ 218 char t_iobc; /* input character */ 219/* RFC 1323 variables */ 220 u_char snd_scale; /* window scaling for send window */ 221 u_char rcv_scale; /* window scaling for recv window */ 222 u_char request_r_scale; /* pending window scaling */ 223 u_int32_t ts_recent; /* timestamp echo data */ 224 u_int ts_recent_age; /* when last updated */ 225 u_int32_t ts_offset; /* our timestamp offset */ 226 227 tcp_seq last_ack_sent; 228/* experimental */ 229 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 230 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 231 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 232 int t_sndzerowin; /* zero-window updates sent */ 233 u_int t_badrxtwin; /* window for retransmit recovery */ 234 u_char snd_limited; /* segments limited transmitted */ 235/* SACK related state */ 236 int snd_numholes; /* number of holes seen by sender */ 237 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 238 /* SACK scoreboard (sorted) */ 239 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 240 int rcv_numsacks; /* # distinct sack blks present */ 241 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 242 tcp_seq sack_newdata; /* New data xmitted in this recovery 243 episode starts at this seq number */ 244 struct sackhint sackhint; /* SACK scoreboard hint */ 245 int t_rttlow; /* smallest observerved RTT */ 246 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 247 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 248 struct toedev *tod; /* toedev handling this connection */ 249 int t_sndrexmitpack; /* retransmit packets sent */ 250 int t_rcvoopack; /* out-of-order packets received */ 251 void *t_toe; /* TOE pcb pointer */ 252 int t_bytes_acked; /* # bytes acked during current RTT */ 253 struct cc_algo *cc_algo; /* congestion control algorithm */ 254 struct cc_var *ccv; /* congestion control specific vars */ 255 struct osd *osd; /* storage for Khelp module data */ 256 257 u_int t_keepinit; /* time to establish connection */ 258 u_int t_keepidle; /* time before keepalive probes begin */ 259 u_int t_keepintvl; /* interval between keepalives */ 260 u_int t_keepcnt; /* number of keepalives before close */ 261 262 u_int t_tsomax; /* TSO total burst length limit in bytes */ 263 u_int t_tsomaxsegcount; /* TSO maximum segment count */ 264 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 265 u_int t_flags2; /* More tcpcb flags storage */ 266#if defined(_KERNEL) && defined(TCP_RFC7413) 267 uint32_t t_ispare[6]; /* 5 UTO, 1 TBD */ 268 uint64_t t_tfo_cookie; /* TCP Fast Open cookie */ 269#else 270 uint32_t t_ispare[8]; /* 5 UTO, 3 TBD */ 271#endif 272 struct tcp_function_block *t_fb;/* TCP function call block */ 273 void *t_fb_ptr; /* Pointer to t_fb specific data */ 274#if defined(_KERNEL) && defined(TCP_RFC7413) 275 unsigned int *t_tfo_pending; /* TCP Fast Open pending counter */ 276 void *t_pspare2[1]; /* 1 TCP_SIGNATURE */ 277#else 278 void *t_pspare2[2]; /* 1 TCP_SIGNATURE, 1 TBD */ 279#endif 280#if defined(_KERNEL) && defined(TCPPCAP) 281 struct mbufq t_inpkts; /* List of saved input packets. */ 282 struct mbufq t_outpkts; /* List of saved output packets. */ 283#ifdef _LP64 284 uint64_t _pad[0]; /* all used! */ 285#else 286 uint64_t _pad[2]; /* 2 are available */ 287#endif /* _LP64 */ 288#else 289 uint64_t _pad[6]; 290#endif /* defined(_KERNEL) && defined(TCPPCAP) */ 291}; 292 293/* 294 * Flags and utility macros for the t_flags field. 295 */ 296#define TF_ACKNOW 0x000001 /* ack peer immediately */ 297#define TF_DELACK 0x000002 /* ack, but try to delay it */ 298#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */ 299#define TF_NOOPT 0x000008 /* don't use tcp options */ 300#define TF_SENTFIN 0x000010 /* have sent FIN */ 301#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */ 302#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */ 303#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */ 304#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */ 305#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */ 306#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */ 307#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */ 308#define TF_NOPUSH 0x001000 /* don't push */ 309#define TF_PREVVALID 0x002000 /* saved values for bad rxmit valid */ 310#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */ 311#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */ 312#define TF_LASTIDLE 0x040000 /* connection was previously idle */ 313#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */ 314#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */ 315#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */ 316#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 317#define TF_FORCEDATA 0x800000 /* force out a byte */ 318#define TF_TSO 0x1000000 /* TSO enabled on this connection */ 319#define TF_TOE 0x2000000 /* this connection is offloaded */ 320#define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */ 321#define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */ 322#define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */ 323#define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 324#define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 325#define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */ 326 327#define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 328#define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 329#define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 330 331#define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 332#define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 333#define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 334 335#define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 336#define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 337#define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 338 339#define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 340 341/* 342 * Flags for the t_oobflags field. 343 */ 344#define TCPOOB_HAVEDATA 0x01 345#define TCPOOB_HADDATA 0x02 346 347/* 348 * Flags for PLPMTU handling, t_flags2 349 */ 350#define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 351#define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 352#define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 353 354/* 355 * Structure to hold TCP options that are only used during segment 356 * processing (in tcp_input), but not held in the tcpcb. 357 * It's basically used to reduce the number of parameters 358 * to tcp_dooptions and tcp_addoptions. 359 * The binary order of the to_flags is relevant for packing of the 360 * options in tcp_addoptions. 361 */ 362struct tcpopt { 363 u_int32_t to_flags; /* which options are present */ 364#define TOF_MSS 0x0001 /* maximum segment size */ 365#define TOF_SCALE 0x0002 /* window scaling */ 366#define TOF_SACKPERM 0x0004 /* SACK permitted */ 367#define TOF_TS 0x0010 /* timestamp */ 368#define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 369#define TOF_SACK 0x0080 /* Peer sent SACK option */ 370#define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */ 371#define TOF_MAXOPT 0x0200 372 u_int32_t to_tsval; /* new timestamp */ 373 u_int32_t to_tsecr; /* reflected timestamp */ 374 u_char *to_sacks; /* pointer to the first SACK blocks */ 375 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 376 u_char *to_tfo_cookie; /* pointer to the TFO cookie */ 377 u_int16_t to_mss; /* maximum segment size */ 378 u_int8_t to_wscale; /* window scaling */ 379 u_int8_t to_nsacks; /* number of SACK blocks */ 380 u_int8_t to_tfo_len; /* TFO cookie length */ 381 u_int32_t to_spare; /* UTO */ 382}; 383 384/* 385 * Flags for tcp_dooptions. 386 */ 387#define TO_SYN 0x01 /* parse SYN-only options */ 388 389struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 390 u_long rmx_mtu; /* MTU for this path */ 391 u_long rmx_ssthresh; /* outbound gateway buffer limit */ 392 u_long rmx_rtt; /* estimated round trip time */ 393 u_long rmx_rttvar; /* estimated rtt variance */ 394 u_long rmx_cwnd; /* congestion window */ 395 u_long rmx_sendpipe; /* outbound delay-bandwidth product */ 396 u_long rmx_recvpipe; /* inbound delay-bandwidth product */ 397}; 398 399/* 400 * Used by tcp_maxmtu() to communicate interface specific features 401 * and limits at the time of connection setup. 402 */ 403struct tcp_ifcap { 404 int ifcap; 405 u_int tsomax; 406 u_int tsomaxsegcount; 407 u_int tsomaxsegsize; 408}; 409 410#ifndef _NETINET_IN_PCB_H_ 411struct in_conninfo; 412#endif /* _NETINET_IN_PCB_H_ */ 413 414struct tcptw { 415 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 416 tcp_seq snd_nxt; 417 tcp_seq rcv_nxt; 418 tcp_seq iss; 419 tcp_seq irs; 420 u_short last_win; /* cached window value */ 421 u_short tw_so_options; /* copy of so_options */ 422 struct ucred *tw_cred; /* user credentials */ 423 u_int32_t t_recent; 424 u_int32_t ts_offset; /* our timestamp offset */ 425 u_int t_starttime; 426 int tw_time; 427 TAILQ_ENTRY(tcptw) tw_2msl; 428 void *tw_pspare; /* TCP_SIGNATURE */ 429 u_int *tw_spare; /* TCP_SIGNATURE */ 430}; 431 432#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 433#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 434#define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 435 436/* 437 * The smoothed round-trip time and estimated variance 438 * are stored as fixed point numbers scaled by the values below. 439 * For convenience, these scales are also used in smoothing the average 440 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 441 * With these scales, srtt has 3 bits to the right of the binary point, 442 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 443 * binary point, and is smoothed with an ALPHA of 0.75. 444 */ 445#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 446#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 447#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 448#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 449#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 450 451/* 452 * The initial retransmission should happen at rtt + 4 * rttvar. 453 * Because of the way we do the smoothing, srtt and rttvar 454 * will each average +1/2 tick of bias. When we compute 455 * the retransmit timer, we want 1/2 tick of rounding and 456 * 1 extra tick because of +-1/2 tick uncertainty in the 457 * firing of the timer. The bias will give us exactly the 458 * 1.5 tick we need. But, because the bias is 459 * statistical, we have to test that we don't drop below 460 * the minimum feasible timer (which is 2 ticks). 461 * This version of the macro adapted from a paper by Lawrence 462 * Brakmo and Larry Peterson which outlines a problem caused 463 * by insufficient precision in the original implementation, 464 * which results in inappropriately large RTO values for very 465 * fast networks. 466 */ 467#define TCP_REXMTVAL(tp) \ 468 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 469 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 470 471/* 472 * TCP statistics. 473 * Many of these should be kept per connection, 474 * but that's inconvenient at the moment. 475 */ 476struct tcpstat { 477 uint64_t tcps_connattempt; /* connections initiated */ 478 uint64_t tcps_accepts; /* connections accepted */ 479 uint64_t tcps_connects; /* connections established */ 480 uint64_t tcps_drops; /* connections dropped */ 481 uint64_t tcps_conndrops; /* embryonic connections dropped */ 482 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 483 uint64_t tcps_closed; /* conn. closed (includes drops) */ 484 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 485 uint64_t tcps_rttupdated; /* times we succeeded */ 486 uint64_t tcps_delack; /* delayed acks sent */ 487 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 488 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 489 uint64_t tcps_persisttimeo; /* persist timeouts */ 490 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 491 uint64_t tcps_keepprobe; /* keepalive probes sent */ 492 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 493 494 uint64_t tcps_sndtotal; /* total packets sent */ 495 uint64_t tcps_sndpack; /* data packets sent */ 496 uint64_t tcps_sndbyte; /* data bytes sent */ 497 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 498 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 499 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 500 uint64_t tcps_sndacks; /* ack-only packets sent */ 501 uint64_t tcps_sndprobe; /* window probes sent */ 502 uint64_t tcps_sndurg; /* packets sent with URG only */ 503 uint64_t tcps_sndwinup; /* window update-only packets sent */ 504 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 505 506 uint64_t tcps_rcvtotal; /* total packets received */ 507 uint64_t tcps_rcvpack; /* packets received in sequence */ 508 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 509 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 510 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 511 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ 512 uint64_t tcps_rcvshort; /* packets received too short */ 513 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 514 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 515 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 516 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 517 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 518 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 519 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 520 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 521 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 522 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 523 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 524 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 525 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 526 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 527 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 528 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 529 uint64_t tcps_predack; /* times hdr predict ok for acks */ 530 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 531 uint64_t tcps_pcbcachemiss; 532 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 533 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 534 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 535 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 536 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 537 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 538 uint64_t tcps_persistdrop; /* timeout in persist state */ 539 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 540 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 541 uint64_t tcps_listendrop; /* listen queue overflows */ 542 uint64_t tcps_badrst; /* ignored RSTs in the window */ 543 544 uint64_t tcps_sc_added; /* entry added to syncache */ 545 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 546 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 547 uint64_t tcps_sc_dropped; /* could not reply to packet */ 548 uint64_t tcps_sc_completed; /* successful extraction of entry */ 549 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 550 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 551 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 552 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 553 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 554 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 555 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 556 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 557 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 558 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 559 560 uint64_t tcps_hc_added; /* entry added to hostcache */ 561 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 562 563 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 564 565 /* SACK related stats */ 566 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 567 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 568 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 569 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 570 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 571 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 572 573 /* ECN related stats */ 574 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 575 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 576 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 577 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 578 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 579 580 /* TCP_SIGNATURE related stats */ 581 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 582 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 583 uint64_t tcps_sig_err_buildsig; /* Failed to make signature */ 584 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 585 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 586 587 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 588}; 589 590#define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ 591 592#ifdef _KERNEL 593#define TI_UNLOCKED 1 594#define TI_RLOCKED 2 595#include <sys/counter.h> 596 597VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 598/* 599 * In-kernel consumers can use these accessor macros directly to update 600 * stats. 601 */ 602#define TCPSTAT_ADD(name, val) \ 603 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 604#define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 605 606/* 607 * Kernel module consumers must use this accessor macro. 608 */ 609void kmod_tcpstat_inc(int statnum); 610#define KMOD_TCPSTAT_INC(name) \ 611 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(uint64_t)) 612 613/* 614 * Running TCP connection count by state. 615 */ 616VNET_DECLARE(counter_u64_t, tcps_states[TCP_NSTATES]); 617#define V_tcps_states VNET(tcps_states) 618#define TCPSTATES_INC(state) counter_u64_add(V_tcps_states[state], 1) 619#define TCPSTATES_DEC(state) counter_u64_add(V_tcps_states[state], -1) 620 621/* 622 * TCP specific helper hook point identifiers. 623 */ 624#define HHOOK_TCP_EST_IN 0 625#define HHOOK_TCP_EST_OUT 1 626#define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 627 628struct tcp_hhook_data { 629 struct tcpcb *tp; 630 struct tcphdr *th; 631 struct tcpopt *to; 632 long len; 633 int tso; 634 tcp_seq curack; 635}; 636#endif 637 638/* 639 * TCB structure exported to user-land via sysctl(3). 640 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 641 * included. Not all of our clients do. 642 */ 643#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 644struct xtcp_timer { 645 int tt_rexmt; /* retransmit timer */ 646 int tt_persist; /* retransmit persistence */ 647 int tt_keep; /* keepalive */ 648 int tt_2msl; /* 2*msl TIME_WAIT timer */ 649 int tt_delack; /* delayed ACK timer */ 650 int t_rcvtime; /* Time since last packet received */ 651}; 652struct xtcpcb { 653 size_t xt_len; 654 struct inpcb xt_inp; 655 struct tcpcb xt_tp; 656 struct xsocket xt_socket; 657 struct xtcp_timer xt_timer; 658 u_quad_t xt_alignment_hack; 659}; 660#endif 661 662/* 663 * Identifiers for TCP sysctl nodes 664 */ 665#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 666#define TCPCTL_MSSDFLT 3 /* MSS default */ 667#define TCPCTL_STATS 4 /* statistics */ 668#define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 669#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 670#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 671#define TCPCTL_SENDSPACE 8 /* send buffer space */ 672#define TCPCTL_RECVSPACE 9 /* receive buffer space */ 673#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 674#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 675#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 676#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 677#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 678#define TCPCTL_DROP 15 /* drop tcp connection */ 679#define TCPCTL_STATES 16 /* connection counts by TCP state */ 680 681#ifdef _KERNEL 682#ifdef SYSCTL_DECL 683SYSCTL_DECL(_net_inet_tcp); 684SYSCTL_DECL(_net_inet_tcp_sack); 685MALLOC_DECLARE(M_TCPLOG); 686#endif 687 688VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */ 689VNET_DECLARE(struct inpcbinfo, tcbinfo); 690extern int tcp_log_in_vain; 691VNET_DECLARE(int, tcp_mssdflt); /* XXX */ 692VNET_DECLARE(int, tcp_minmss); 693VNET_DECLARE(int, tcp_delack_enabled); 694VNET_DECLARE(int, tcp_do_rfc3390); 695VNET_DECLARE(int, tcp_initcwnd_segments); 696VNET_DECLARE(int, tcp_sendspace); 697VNET_DECLARE(int, tcp_recvspace); 698VNET_DECLARE(int, path_mtu_discovery); 699VNET_DECLARE(int, tcp_do_rfc3465); 700VNET_DECLARE(int, tcp_abc_l_var); 701#define V_tcb VNET(tcb) 702#define V_tcbinfo VNET(tcbinfo) 703#define V_tcp_mssdflt VNET(tcp_mssdflt) 704#define V_tcp_minmss VNET(tcp_minmss) 705#define V_tcp_delack_enabled VNET(tcp_delack_enabled) 706#define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 707#define V_tcp_initcwnd_segments VNET(tcp_initcwnd_segments) 708#define V_tcp_sendspace VNET(tcp_sendspace) 709#define V_tcp_recvspace VNET(tcp_recvspace) 710#define V_path_mtu_discovery VNET(path_mtu_discovery) 711#define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 712#define V_tcp_abc_l_var VNET(tcp_abc_l_var) 713 714VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */ 715VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */ 716#define V_tcp_do_sack VNET(tcp_do_sack) 717#define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 718 719VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */ 720VNET_DECLARE(int, tcp_ecn_maxretries); 721#define V_tcp_do_ecn VNET(tcp_do_ecn) 722#define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 723 724VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 725#define V_tcp_hhh VNET(tcp_hhh) 726 727VNET_DECLARE(int, tcp_do_rfc6675_pipe); 728#define V_tcp_do_rfc6675_pipe VNET(tcp_do_rfc6675_pipe) 729 730int tcp_addoptions(struct tcpopt *, u_char *); 731int tcp_ccalgounload(struct cc_algo *unload_algo); 732struct tcpcb * 733 tcp_close(struct tcpcb *); 734void tcp_discardcb(struct tcpcb *); 735void tcp_twstart(struct tcpcb *); 736void tcp_twclose(struct tcptw *, int); 737void tcp_ctlinput(int, struct sockaddr *, void *); 738int tcp_ctloutput(struct socket *, struct sockopt *); 739struct tcpcb * 740 tcp_drop(struct tcpcb *, int); 741void tcp_drain(void); 742void tcp_init(void); 743void tcp_fini(void *); 744char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 745 const void *); 746char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 747 const void *); 748int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *); 749void tcp_reass_global_init(void); 750void tcp_reass_flush(struct tcpcb *); 751void tcp_dooptions(struct tcpopt *, u_char *, int, int); 752void tcp_dropwithreset(struct mbuf *, struct tcphdr *, 753 struct tcpcb *, int, int); 754void tcp_pulloutofband(struct socket *, 755 struct tcphdr *, struct mbuf *, int); 756void tcp_xmit_timer(struct tcpcb *, int); 757void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *); 758void cc_ack_received(struct tcpcb *tp, struct tcphdr *th, 759 uint16_t type); 760void cc_conn_init(struct tcpcb *tp); 761void cc_post_recovery(struct tcpcb *tp, struct tcphdr *th); 762void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 763void hhook_run_tcp_est_in(struct tcpcb *tp, 764 struct tcphdr *th, struct tcpopt *to); 765 766int tcp_input(struct mbuf **, int *, int); 767void tcp_do_segment(struct mbuf *, struct tcphdr *, 768 struct socket *, struct tcpcb *, int, int, uint8_t, 769 int); 770 771int register_tcp_functions(struct tcp_function_block *blk, int wait); 772int deregister_tcp_functions(struct tcp_function_block *blk); 773struct tcp_function_block *find_and_ref_tcp_functions(struct tcp_function_set *fs); 774struct tcp_function_block *find_and_ref_tcp_fb(struct tcp_function_block *blk); 775int tcp_default_ctloutput(struct socket *so, struct sockopt *sopt, struct inpcb *inp, struct tcpcb *tp); 776 777u_long tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 778u_long tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 779u_int tcp_maxseg(const struct tcpcb *); 780void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 781 struct tcp_ifcap *); 782void tcp_mss(struct tcpcb *, int); 783int tcp_mssopt(struct in_conninfo *); 784struct inpcb * 785 tcp_drop_syn_sent(struct inpcb *, int); 786struct tcpcb * 787 tcp_newtcpcb(struct inpcb *); 788int tcp_output(struct tcpcb *); 789void tcp_state_change(struct tcpcb *, int); 790void tcp_respond(struct tcpcb *, void *, 791 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 792void tcp_tw_init(void); 793#ifdef VIMAGE 794void tcp_tw_destroy(void); 795#endif 796void tcp_tw_zone_change(void); 797int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 798 struct mbuf *, int); 799void tcp_setpersist(struct tcpcb *); 800void tcp_slowtimo(void); 801struct tcptemp * 802 tcpip_maketemplate(struct inpcb *); 803void tcpip_fillheaders(struct inpcb *, void *, void *); 804void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 805int tcp_timer_active(struct tcpcb *, uint32_t); 806void tcp_timer_stop(struct tcpcb *, uint32_t); 807void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 808/* 809 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 810 */ 811void tcp_hc_init(void); 812#ifdef VIMAGE 813void tcp_hc_destroy(void); 814#endif 815void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 816u_long tcp_hc_getmtu(struct in_conninfo *); 817void tcp_hc_updatemtu(struct in_conninfo *, u_long); 818void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 819 820extern struct pr_usrreqs tcp_usrreqs; 821tcp_seq tcp_new_isn(struct tcpcb *); 822 823int tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 824void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 825void tcp_clean_sackreport(struct tcpcb *tp); 826void tcp_sack_adjust(struct tcpcb *tp); 827struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 828void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 829void tcp_free_sackholes(struct tcpcb *tp); 830int tcp_newreno(struct tcpcb *, struct tcphdr *); 831int tcp_compute_pipe(struct tcpcb *); 832 833static inline void 834tcp_fields_to_host(struct tcphdr *th) 835{ 836 837 th->th_seq = ntohl(th->th_seq); 838 th->th_ack = ntohl(th->th_ack); 839 th->th_win = ntohs(th->th_win); 840 th->th_urp = ntohs(th->th_urp); 841} 842 843static inline void 844tcp_fields_to_net(struct tcphdr *th) 845{ 846 847 th->th_seq = htonl(th->th_seq); 848 th->th_ack = htonl(th->th_ack); 849 th->th_win = htons(th->th_win); 850 th->th_urp = htons(th->th_urp); 851} 852#endif /* _KERNEL */ 853 854#endif /* _NETINET_TCP_VAR_H_ */ 855