1/* 2 * Copyright (c) 2000-2012 Apple Inc. All rights reserved. 3 * 4 * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ 5 * 6 * This file contains Original Code and/or Modifications of Original Code 7 * as defined in and that are subject to the Apple Public Source License 8 * Version 2.0 (the 'License'). You may not use this file except in 9 * compliance with the License. The rights granted to you under the License 10 * may not be used to create, or enable the creation or redistribution of, 11 * unlawful or unlicensed copies of an Apple operating system, or to 12 * circumvent, violate, or enable the circumvention or violation of, any 13 * terms of an Apple operating system software license agreement. 14 * 15 * Please obtain a copy of the License at 16 * http://www.opensource.apple.com/apsl/ and read it before using this file. 17 * 18 * The Original Code and all software distributed under the License are 19 * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER 20 * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, 21 * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, 22 * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. 23 * Please see the License for the specific language governing rights and 24 * limitations under the License. 25 * 26 * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ 27 */ 28/* 29 * Copyright (c) 1982, 1986, 1993, 1994, 1995 30 * The Regents of the University of California. All rights reserved. 31 * 32 * Redistribution and use in source and binary forms, with or without 33 * modification, are permitted provided that the following conditions 34 * are met: 35 * 1. Redistributions of source code must retain the above copyright 36 * notice, this list of conditions and the following disclaimer. 37 * 2. Redistributions in binary form must reproduce the above copyright 38 * notice, this list of conditions and the following disclaimer in the 39 * documentation and/or other materials provided with the distribution. 40 * 3. All advertising materials mentioning features or use of this software 41 * must display the following acknowledgement: 42 * This product includes software developed by the University of 43 * California, Berkeley and its contributors. 44 * 4. Neither the name of the University nor the names of its contributors 45 * may be used to endorse or promote products derived from this software 46 * without specific prior written permission. 47 * 48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 58 * SUCH DAMAGE. 59 * 60 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 61 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.8 2001/08/22 00:59:13 silby Exp $ 62 */ 63 64#ifndef _NETINET_TCP_VAR_H_ 65#define _NETINET_TCP_VAR_H_ 66#include <sys/appleapiopts.h> 67#include <sys/queue.h> 68#include <netinet/in_pcb.h> 69#include <netinet/tcp_timer.h> 70 71#if defined(__LP64__) 72#define _TCPCB_PTR(x) u_int32_t 73#define _TCPCB_LIST_HEAD(name, type) \ 74struct name { \ 75 u_int32_t lh_first; \ 76}; 77#else 78#define _TCPCB_PTR(x) x 79#define _TCPCB_LIST_HEAD(name, type) LIST_HEAD(name, type) 80#endif 81 82#define TCP_RETRANSHZ 1000 /* granularity of TCP timestamps, 1ms */ 83#define TCP_TIMERHZ 100 /* frequency of TCP fast timer, 100 ms */ 84 85/* Minimum time quantum within which the timers are coalesced */ 86#define TCP_FASTTIMER_QUANTUM TCP_TIMERHZ /* fast mode, once every 100ms */ 87#define TCP_SLOWTIMER_QUANTUM TCP_RETRANSHZ / PR_SLOWHZ /* slow mode, once every 500ms */ 88 89#define TCP_RETRANSHZ_TO_USEC 1000 90 91#ifdef KERNEL_PRIVATE 92#define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */ 93 94/* Base RTT is stored for N_MIN_RTT_HISTORY slots. This is used to 95 * estimate expected minimum RTT for delay based congestion control 96 * algorithms. 97 */ 98#define N_RTT_BASE 5 99 100/* Always allow at least 4 packets worth of recv window when adjusting 101 * recv window using inter-packet arrival jitter. 102 */ 103#define MIN_IAJ_WIN 4 104 105/* A variation in delay of this many milliseconds is tolerable. This limit has to 106 * be low but greater than zero. We also use standard deviation on jitter to adjust 107 * this limit for different link and connection types. 108 */ 109#define ALLOWED_IAJ 5 110 111/* Ignore the first few packets on a connection until the ACK clock gets going 112 */ 113#define IAJ_IGNORE_PKTCNT 40 114 115/* Let the accumulated IAJ value increase by this threshold at most. This limit 116 * will control how many ALLOWED_IAJ measurements a receiver will have to see 117 * before opening the receive window 118 */ 119#define ACC_IAJ_HIGH_THRESH 100 120 121/* When accumulated IAJ reaches this value, the receiver starts to react by 122 * closing the window 123 */ 124#define ACC_IAJ_REACT_LIMIT 200 125 126/* If the number of small packets (smaller than IAJ packet size) seen on a 127 * connection is more than this threshold, reset the size and learn it again. 128 * This is needed because the sender might send smaller segments after PMTU 129 * discovery and the receiver has to learn the new size. 130 */ 131#define RESET_IAJ_SIZE_THRESH 20 132 133/* 134 * Kernel variables for tcp. 135 */ 136 137/* TCP segment queue entry */ 138struct tseg_qent { 139 LIST_ENTRY(tseg_qent) tqe_q; 140 int tqe_len; /* TCP segment data length */ 141 struct tcphdr *tqe_th; /* a pointer to tcp header */ 142 struct mbuf *tqe_m; /* mbuf contains packet */ 143}; 144LIST_HEAD(tsegqe_head, tseg_qent); 145extern int tcp_reass_maxseg; 146extern int tcp_reass_qsize; 147#ifdef MALLOC_DECLARE 148MALLOC_DECLARE(M_TSEGQ); 149#endif 150 151struct sackblk { 152 tcp_seq start; /* start seq no. of sack block */ 153 tcp_seq end; /* end seq no. */ 154}; 155 156struct sackhole { 157 tcp_seq start; /* start seq no. of hole */ 158 tcp_seq end; /* end seq no. */ 159 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 160 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 161}; 162 163struct sackhint { 164 struct sackhole *nexthole; 165 int sack_bytes_rexmit; 166}; 167 168struct tcptemp { 169 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 170 struct tcphdr tt_t; 171}; 172 173struct bwmeas { 174 tcp_seq bw_start; /* start of bw measurement */ 175 uint32_t bw_ts; /* timestamp when bw measurement started */ 176 uint32_t bw_size; /* burst size in bytes for this bw measurement */ 177 uint32_t bw_minsizepkts; /* Min burst size as segments */ 178 uint32_t bw_maxsizepkts; /* Max burst size as segments */ 179 uint32_t bw_minsize; /* Min size in bytes */ 180 uint32_t bw_maxsize; /* Max size in bytes */ 181 uint32_t bw_sndbw; /* Measured send bw */ 182}; 183 184#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 185 186/* 187 * Tcp control block, one per tcp; fields: 188 * Organized for 16 byte cacheline efficiency. 189 */ 190struct tcpcb { 191 struct tsegqe_head t_segq; 192 int t_dupacks; /* consecutive dup acks recd */ 193 uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */ 194 struct tcptimerentry tentry; /* entry in timer list */ 195 196 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 197 int t_state; /* state of this connection */ 198 u_int t_flags; 199#define TF_ACKNOW 0x00001 /* ack peer immediately */ 200#define TF_DELACK 0x00002 /* ack, but try to delay it */ 201#define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ 202#define TF_NOOPT 0x00008 /* don't use tcp options */ 203#define TF_SENTFIN 0x00010 /* have sent FIN */ 204#define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ 205#define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ 206#define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ 207#define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ 208#define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ 209#define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */ 210#define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ 211#define TF_NOPUSH 0x01000 /* don't push */ 212#define TF_REQ_CC 0x02000 /* have/will request CC */ 213#define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ 214#define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */ 215#define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ 216#define TF_LOCAL 0x20000 /* connection to a host on local link */ 217#define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ 218#define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ 219#define TF_LASTIDLE 0x100000 /* connection was previously idle */ 220#define TF_FASTRECOVERY 0x200000 /* in NewReno Fast Recovery */ 221#define TF_WASFRECOVERY 0x400000 /* was in NewReno Fast Recovery */ 222#define TF_SIGNATURE 0x800000 /* require MD5 digests (RFC2385) */ 223#define TF_MAXSEGSNT 0x1000000 /* last segment sent was a full segment */ 224#define TF_PMTUD 0x4000000 /* Perform Path MTU Discovery for this connection */ 225#define TF_CLOSING 0x8000000 /* pending tcp close */ 226#define TF_TSO 0x10000000 /* TCP Segment Offloading is enable on this connection */ 227#define TF_BLACKHOLE 0x20000000 /* Path MTU Discovery Black Hole detection */ 228#define TF_TIMER_ONLIST 0x40000000 /* pcb is on tcp_timer_list */ 229#define TF_STRETCHACK 0x80000000 /* receiver is going to delay acks */ 230 231 int t_force; /* 1 if forcing out a byte */ 232 233 tcp_seq snd_una; /* send unacknowledged */ 234 tcp_seq snd_max; /* highest sequence number sent; 235 * used to recognize retransmits 236 */ 237 tcp_seq snd_nxt; /* send next */ 238 tcp_seq snd_up; /* send urgent pointer */ 239 240 tcp_seq snd_wl1; /* window update seg seq number */ 241 tcp_seq snd_wl2; /* window update seg ack number */ 242 tcp_seq iss; /* initial send sequence number */ 243 tcp_seq irs; /* initial receive sequence number */ 244 245 tcp_seq rcv_nxt; /* receive next */ 246 tcp_seq rcv_adv; /* advertised window */ 247 u_int32_t rcv_wnd; /* receive window */ 248 tcp_seq rcv_up; /* receive urgent pointer */ 249 250 u_int32_t snd_wnd; /* send window */ 251 u_int32_t snd_cwnd; /* congestion-controlled window */ 252 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 253 * for slow start exponential to 254 * linear switch 255 */ 256 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 257 258 u_int t_maxopd; /* mss plus options */ 259 260 u_int32_t t_rcvtime; /* time at which a packet was received */ 261 u_int32_t t_starttime; /* time connection was established */ 262 int t_rtttime; /* tcp clock when rtt calculation was started */ 263 tcp_seq t_rtseq; /* sequence number being timed */ 264 265 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 266 u_int32_t rfbuf_cnt; /* recv buffer autoscaling byte count */ 267 268 int t_rxtcur; /* current retransmit value (ticks) */ 269 u_int t_maxseg; /* maximum segment size */ 270 int t_srtt; /* smoothed round-trip time */ 271 int t_rttvar; /* variance in round-trip time */ 272 273 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 274 u_int t_rttmin; /* minimum rtt allowed */ 275 u_int t_rttbest; /* best rtt we've seen */ 276 u_int t_rttcur; /* most recent value of rtt */ 277 u_int32_t t_rttupdated; /* number of times rtt sampled */ 278 u_int32_t rxt_conndroptime; /* retxmt conn gets dropped after this time, when set */ 279 u_int32_t rxt_start; /* time at a connection starts retransmitting */ 280 u_int32_t max_sndwnd; /* largest window peer has offered */ 281 282 int t_softerror; /* possible error not yet reported */ 283/* out-of-band data */ 284 char t_oobflags; /* have some */ 285 char t_iobc; /* input character */ 286#define TCPOOB_HAVEDATA 0x01 287#define TCPOOB_HADDATA 0x02 288/* RFC 1323 variables */ 289 u_char snd_scale; /* window scaling for send window */ 290 u_char rcv_scale; /* window scaling for recv window */ 291 u_char request_r_scale; /* pending window scaling */ 292 u_char requested_s_scale; 293 u_int16_t tcp_cc_index; /* index of congestion control algorithm */ 294 u_int32_t ts_recent; /* timestamp echo data */ 295 296 u_int32_t ts_recent_age; /* when last updated */ 297 tcp_seq last_ack_sent; 298/* RFC 1644 variables */ 299 tcp_cc cc_send; /* send connection count */ 300 tcp_cc cc_recv; /* receive connection count */ 301/* RFC 3465 variables */ 302 u_int32_t t_bytes_acked; /* ABC "bytes_acked" parameter */ 303/* experimental */ 304 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 305 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 306 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 307 308 int t_keepidle; /* keepalive idle timer (override global if > 0) */ 309 int t_lastchain; /* amount of packets chained last time around */ 310 int t_unacksegs; /* received but unacked segments: used for delaying acks */ 311 u_int32_t t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */ 312 u_int32_t t_persist_stop; /* persistence limit deadline if triggered by ZWP */ 313 u_int32_t t_notsent_lowat; /* Low water for not sent data */ 314 315/* 3529618 MSS overload prevention */ 316 u_int32_t rcv_reset; 317 u_int32_t rcv_pps; 318 u_int32_t rcv_byps; 319 u_int32_t rcv_maxbyps; 320 321/* Receiver state for stretch-ack algorithm */ 322 u_int32_t rcv_unackwin; /* to measure win for stretching acks */ 323 u_int32_t rcv_by_unackwin; /* bytes seen during the last ack-stretching win */ 324 u_int16_t rcv_waitforss; /* wait for packets during slow-start */ 325 u_int16_t ecn_flags; 326#define TE_SETUPSENT 0x01 /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */ 327#define TE_SETUPRECEIVED 0x02 /* Indicate we have received ECN-SETUP SYN or SYN-ACK */ 328#define TE_SENDIPECT 0x04 /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */ 329#define TE_SENDCWR 0x08 /* Indicate that the next non-retransmit should have the TCP CWR flag set */ 330#define TE_SENDECE 0x10 /* Indicate that the next packet should have the TCP ECE flag set */ 331#define TE_ECN_ON (TE_SETUPSENT | TE_SETUPRECEIVED) /* Indicate ECN was successfully negotiated on a connection) */ 332 333 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 334/* anti DoS counters */ 335 u_int32_t rcv_second; /* start of interval second */ 336 337/* SACK related state */ 338 int sack_enable; /* enable SACK for this connection */ 339 int snd_numholes; /* number of holes seen by sender */ 340 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 341 /* SACK scoreboard (sorted) */ 342 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 343 int rcv_numsacks; /* # distinct sack blks present */ 344 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 345 tcp_seq sack_newdata; /* New data xmitted in this recovery 346 episode starts at this seq number */ 347 struct sackhint sackhint; /* SACK scoreboard hint */ 348 349 u_int32_t t_pktlist_sentlen; /* total bytes in transmit chain */ 350 struct mbuf *t_pktlist_head; /* First packet in transmit chain */ 351 struct mbuf *t_pktlist_tail; /* Last packet in transmit chain */ 352 353 int t_keepinit; /* connection timeout, i.e. idle time in SYN_SENT or SYN_RECV state */ 354 u_int32_t tso_max_segment_size; /* TCP Segment Offloading maximum segment unit for NIC */ 355 u_int t_pmtud_saved_maxopd; /* MSS saved before performing PMTU-D BlackHole detection */ 356 357 struct 358 { 359 u_int32_t rxduplicatebytes; 360 u_int32_t rxoutoforderbytes; 361 u_int32_t txretransmitbytes; 362 u_int32_t unused_pad_to_8; 363 } t_stat; 364 365 /* Background congestion related state */ 366 uint32_t rtt_hist[N_RTT_BASE]; /* history of minimum RTT */ 367 uint32_t rtt_count; /* Number of RTT samples in recent base history */ 368 uint32_t bg_ssthresh; /* Slow start threshold until delay increases */ 369 uint32_t t_flagsext; /* Another field to accommodate more flags */ 370#define TF_RXTFINDROP 0x1 /* Drop conn after retransmitting FIN 3 times */ 371#define TF_RCVUNACK_WAITSS 0x2 /* set when the receiver should not stretch acks */ 372#define TF_BWMEAS_INPROGRESS 0x4 /* Indicate BW meas is happening */ 373#define TF_MEASURESNDBW 0x8 /* Measure send bw on this connection */ 374#define TF_LRO_OFFLOADED 0x10 /* Connection LRO offloaded */ 375#if TRAFFIC_MGT 376 /* Inter-arrival jitter related state */ 377 uint32_t iaj_rcv_ts; /* tcp clock when the first packet was received */ 378 uint16_t iaj_size; /* Size of packet for iaj measurement */ 379 uint16_t iaj_small_pkt; /* Count of packets smaller than iaj_size */ 380 uint16_t iaj_pktcnt; /* packet count, to avoid throttling initially */ 381 uint16_t acc_iaj; /* Accumulated iaj */ 382 tcp_seq iaj_rwintop; /* recent max advertised window */ 383 uint32_t avg_iaj; /* Mean */ 384 uint32_t std_dev_iaj; /* Standard deviation */ 385#endif /* TRAFFIC_MGT */ 386 struct bwmeas *t_bwmeas; /* State for bandwidth measurement */ 387 uint32_t t_lropktlen; /* Bytes in a LRO frame */ 388 tcp_seq t_idleat; /* rcv_nxt at idle time */ 389}; 390 391#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) 392 393/* 394 * If the connection is in a throttled state due to advisory feedback from 395 * the interface output queue, reset that state. We do this in favor 396 * of entering recovery because the data transfer during recovery 397 * should be just a trickle and it will help to improve performance. 398 * We also do not want to back off twice in the same RTT. 399 */ 400#define ENTER_FASTRECOVERY(_tp_) do { \ 401 (_tp_)->t_flags |= TF_FASTRECOVERY; \ 402 if (INP_IS_FLOW_CONTROLLED((_tp_)->t_inpcb)) \ 403 inp_reset_fc_state((_tp_)->t_inpcb); \ 404} while(0) 405 406#define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY 407 408#if CONFIG_DTRACE 409enum tcp_cc_event { 410 TCP_CC_CWND_INIT, 411 TCP_CC_INSEQ_ACK_RCVD, 412 TCP_CC_ACK_RCVD, 413 TCP_CC_ENTER_FASTRECOVERY, 414 TCP_CC_IN_FASTRECOVERY, 415 TCP_CC_EXIT_FASTRECOVERY, 416 TCP_CC_PARTIAL_ACK, 417 TCP_CC_IDLE_TIMEOUT, 418 TCP_CC_REXMT_TIMEOUT, 419 TCP_CC_ECN_RCVD, 420 TCP_CC_BAD_REXMT_RECOVERY, 421 TCP_CC_OUTPUT_ERROR, 422 TCP_CC_CHANGE_ALGO, 423 TCP_CC_FLOW_CONTROL, 424 TCP_CC_SUSPEND 425}; 426#endif /* CONFIG_DTRACE */ 427 428/* 429 * Structure to hold TCP options that are only used during segment 430 * processing (in tcp_input), but not held in the tcpcb. 431 * It's basically used to reduce the number of parameters 432 * to tcp_dooptions. 433 */ 434struct tcpopt { 435 u_int32_t to_flags; /* which options are present */ 436#define TOF_TS 0x0001 /* timestamp */ 437#define TOF_MSS 0x0010 438#define TOF_SCALE 0x0020 439#define TOF_SIGNATURE 0x0040 /* signature option present */ 440#define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */ 441#define TOF_SACK 0x0100 /* Peer sent SACK option */ 442 u_int32_t to_tsval; 443 u_int32_t to_tsecr; 444 u_int16_t to_mss; 445 u_int8_t to_requested_s_scale; 446 u_int8_t to_nsacks; /* number of SACK blocks */ 447 u_char *to_sacks; /* pointer to the first SACK blocks */ 448}; 449 450/* 451 * The TAO cache entry which is stored in the protocol family specific 452 * portion of the route metrics. 453 */ 454struct rmxp_tao { 455 tcp_cc tao_cc; /* latest CC in valid SYN */ 456 tcp_cc tao_ccsent; /* latest CC sent to peer */ 457 u_short tao_mssopt; /* peer's cached MSS */ 458#ifdef notyet 459 u_short tao_flags; /* cache status flags */ 460#define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */ 461#define TAOF_OK 0x0002 /* peer does understand rfc1644 */ 462#define TAOF_UNDEF 0 /* we don't know yet */ 463#endif /* notyet */ 464}; 465#define rmx_taop(r) ((struct rmxp_tao *)(r).rmx_filler) 466 467#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 468#define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 469 470/* 471 * The rtt measured is in milliseconds as the timestamp granularity is 472 * a millisecond. The smoothed round-trip time and estimated variance 473 * are stored as fixed point numbers scaled by the values below. 474 * For convenience, these scales are also used in smoothing the average 475 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 476 * With these scales, srtt has 5 bits to the right of the binary point, 477 * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the 478 * binary point, and is smoothed with an ALPHA of 0.75. 479 */ 480#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 481#define TCP_RTT_SHIFT 5 /* shift for srtt; 5 bits frac. */ 482#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 4 bits */ 483#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 4 bits */ 484#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 485 486/* 487 * The initial retransmission should happen at rtt + 4 * rttvar. 488 * Because of the way we do the smoothing, srtt and rttvar 489 * will each average +1/2 tick of bias. When we compute 490 * the retransmit timer, we want 1/2 tick of rounding and 491 * 1 extra tick because of +-1/2 tick uncertainty in the 492 * firing of the timer. The bias will give us exactly the 493 * 1.5 tick we need. But, because the bias is 494 * statistical, we have to test that we don't drop below 495 * the minimum feasible timer (which is 2 ticks). 496 * This version of the macro adapted from a paper by Lawrence 497 * Brakmo and Larry Peterson which outlines a problem caused 498 * by insufficient precision in the original implementation, 499 * which results in inappropriately large RTO values for very 500 * fast networks. 501 */ 502#define TCP_REXMTVAL(tp) \ 503 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 504 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 505 506/* 507 * Jaguar compatible TCP control block, for xtcpcb 508 * Does not have the old fields 509 */ 510struct otcpcb { 511#else 512struct tseg_qent; 513_TCPCB_LIST_HEAD(tsegqe_head, tseg_qent); 514 515struct tcpcb { 516#endif /* KERNEL_PRIVATE */ 517#if defined(KERNEL_PRIVATE) 518 u_int32_t t_segq; 519#else 520 struct tsegqe_head t_segq; 521#endif /* KERNEL_PRIVATE */ 522 int t_dupacks; /* consecutive dup acks recd */ 523 u_int32_t unused; /* unused now: was t_template */ 524 525 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 526 527 _TCPCB_PTR(struct inpcb *) t_inpcb; /* back pointer to internet pcb */ 528 int t_state; /* state of this connection */ 529 u_int t_flags; 530#define TF_ACKNOW 0x00001 /* ack peer immediately */ 531#define TF_DELACK 0x00002 /* ack, but try to delay it */ 532#define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ 533#define TF_NOOPT 0x00008 /* don't use tcp options */ 534#define TF_SENTFIN 0x00010 /* have sent FIN */ 535#define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ 536#define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ 537#define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ 538#define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ 539#define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ 540#define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */ 541#define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ 542#define TF_NOPUSH 0x01000 /* don't push */ 543#define TF_REQ_CC 0x02000 /* have/will request CC */ 544#define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ 545#define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */ 546#define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ 547#define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */ 548#define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ 549#define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ 550 551 int t_force; /* 1 if forcing out a byte */ 552 553 tcp_seq snd_una; /* send unacknowledged */ 554 tcp_seq snd_max; /* highest sequence number sent; 555 * used to recognize retransmits 556 */ 557 tcp_seq snd_nxt; /* send next */ 558 tcp_seq snd_up; /* send urgent pointer */ 559 560 tcp_seq snd_wl1; /* window update seg seq number */ 561 tcp_seq snd_wl2; /* window update seg ack number */ 562 tcp_seq iss; /* initial send sequence number */ 563 tcp_seq irs; /* initial receive sequence number */ 564 565 tcp_seq rcv_nxt; /* receive next */ 566 tcp_seq rcv_adv; /* advertised window */ 567 u_int32_t rcv_wnd; /* receive window */ 568 tcp_seq rcv_up; /* receive urgent pointer */ 569 570 u_int32_t snd_wnd; /* send window */ 571 u_int32_t snd_cwnd; /* congestion-controlled window */ 572 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 573 * for slow start exponential to 574 * linear switch 575 */ 576 u_int t_maxopd; /* mss plus options */ 577 578 u_int32_t t_rcvtime; /* time at which a packet was received */ 579 u_int32_t t_starttime; /* time connection was established */ 580 int t_rtttime; /* round trip time */ 581 tcp_seq t_rtseq; /* sequence number being timed */ 582 583 int t_rxtcur; /* current retransmit value (ticks) */ 584 u_int t_maxseg; /* maximum segment size */ 585 int t_srtt; /* smoothed round-trip time */ 586 int t_rttvar; /* variance in round-trip time */ 587 588 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 589 u_int t_rttmin; /* minimum rtt allowed */ 590 u_int32_t t_rttupdated; /* number of times rtt sampled */ 591 u_int32_t max_sndwnd; /* largest window peer has offered */ 592 593 int t_softerror; /* possible error not yet reported */ 594/* out-of-band data */ 595 char t_oobflags; /* have some */ 596 char t_iobc; /* input character */ 597#define TCPOOB_HAVEDATA 0x01 598#define TCPOOB_HADDATA 0x02 599/* RFC 1323 variables */ 600 u_char snd_scale; /* window scaling for send window */ 601 u_char rcv_scale; /* window scaling for recv window */ 602 u_char request_r_scale; /* pending window scaling */ 603 u_char requested_s_scale; 604 u_int32_t ts_recent; /* timestamp echo data */ 605 606 u_int32_t ts_recent_age; /* when last updated */ 607 tcp_seq last_ack_sent; 608/* RFC 1644 variables */ 609 tcp_cc cc_send; /* send connection count */ 610 tcp_cc cc_recv; /* receive connection count */ 611 tcp_seq snd_recover; /* for use in fast recovery */ 612/* experimental */ 613 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 614 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 615 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 616}; 617 618 619/* 620 * TCP statistics. 621 * Many of these should be kept per connection, 622 * but that's inconvenient at the moment. 623 */ 624struct tcpstat { 625 u_int32_t tcps_connattempt; /* connections initiated */ 626 u_int32_t tcps_accepts; /* connections accepted */ 627 u_int32_t tcps_connects; /* connections established */ 628 u_int32_t tcps_drops; /* connections dropped */ 629 u_int32_t tcps_conndrops; /* embryonic connections dropped */ 630 u_int32_t tcps_closed; /* conn. closed (includes drops) */ 631 u_int32_t tcps_segstimed; /* segs where we tried to get rtt */ 632 u_int32_t tcps_rttupdated; /* times we succeeded */ 633 u_int32_t tcps_delack; /* delayed acks sent */ 634 u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 635 u_int32_t tcps_rexmttimeo; /* retransmit timeouts */ 636 u_int32_t tcps_persisttimeo; /* persist timeouts */ 637 u_int32_t tcps_keeptimeo; /* keepalive timeouts */ 638 u_int32_t tcps_keepprobe; /* keepalive probes sent */ 639 u_int32_t tcps_keepdrops; /* connections dropped in keepalive */ 640 641 u_int32_t tcps_sndtotal; /* total packets sent */ 642 u_int32_t tcps_sndpack; /* data packets sent */ 643 u_int32_t tcps_sndbyte; /* data bytes sent */ 644 u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */ 645 u_int32_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 646 u_int32_t tcps_sndacks; /* ack-only packets sent */ 647 u_int32_t tcps_sndprobe; /* window probes sent */ 648 u_int32_t tcps_sndurg; /* packets sent with URG only */ 649 u_int32_t tcps_sndwinup; /* window update-only packets sent */ 650 u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 651 652 u_int32_t tcps_rcvtotal; /* total packets received */ 653 u_int32_t tcps_rcvpack; /* packets received in sequence */ 654 u_int32_t tcps_rcvbyte; /* bytes received in sequence */ 655 u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */ 656 u_int32_t tcps_rcvbadoff; /* packets received with bad offset */ 657 u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 658 u_int32_t tcps_rcvshort; /* packets received too short */ 659 u_int32_t tcps_rcvduppack; /* duplicate-only packets received */ 660 u_int32_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 661 u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */ 662 u_int32_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 663 u_int32_t tcps_rcvoopack; /* out-of-order packets received */ 664 u_int32_t tcps_rcvoobyte; /* out-of-order bytes received */ 665 u_int32_t tcps_rcvpackafterwin; /* packets with data after window */ 666 u_int32_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 667 u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */ 668 u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */ 669 u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */ 670 u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 671 u_int32_t tcps_rcvackpack; /* rcvd ack packets */ 672 u_int32_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 673 u_int32_t tcps_rcvwinupd; /* rcvd window update packets */ 674 u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */ 675 u_int32_t tcps_predack; /* times hdr predict ok for acks */ 676 u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */ 677 u_int32_t tcps_pcbcachemiss; 678 u_int32_t tcps_cachedrtt; /* times cached RTT in route updated */ 679 u_int32_t tcps_cachedrttvar; /* times cached rttvar updated */ 680 u_int32_t tcps_cachedssthresh; /* times cached ssthresh updated */ 681 u_int32_t tcps_usedrtt; /* times RTT initialized from route */ 682 u_int32_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 683 u_int32_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 684 u_int32_t tcps_persistdrop; /* timeout in persist state */ 685 u_int32_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 686 u_int32_t tcps_mturesent; /* resends due to MTU discovery */ 687 u_int32_t tcps_listendrop; /* listen queue overflows */ 688 689 /* new stats from FreeBSD 5.4 sync up */ 690 u_int32_t tcps_minmssdrops; /* average minmss too low drops */ 691 u_int32_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 692 u_int32_t tcps_badrst; /* ignored RSTs in the window */ 693 694 u_int32_t tcps_sc_added; /* entry added to syncache */ 695 u_int32_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 696 u_int32_t tcps_sc_dupsyn; /* duplicate SYN packet */ 697 u_int32_t tcps_sc_dropped; /* could not reply to packet */ 698 u_int32_t tcps_sc_completed; /* successful extraction of entry */ 699 u_int32_t tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ 700 u_int32_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 701 u_int32_t tcps_sc_reset; /* RST removed entry from syncache */ 702 u_int32_t tcps_sc_stale; /* timed out or listen socket gone */ 703 u_int32_t tcps_sc_aborted; /* syncache entry aborted */ 704 u_int32_t tcps_sc_badack; /* removed due to bad ACK */ 705 u_int32_t tcps_sc_unreach; /* ICMP unreachable received */ 706 u_int32_t tcps_sc_zonefail; /* zalloc() failed */ 707 u_int32_t tcps_sc_sendcookie; /* SYN cookie sent */ 708 u_int32_t tcps_sc_recvcookie; /* SYN cookie received */ 709 710 u_int32_t tcps_hc_added; /* entry added to hostcache */ 711 u_int32_t tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ 712 713 /* SACK related stats */ 714 u_int32_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 715 u_int32_t tcps_sack_rexmits; /* SACK rexmit segments */ 716 u_int32_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 717 u_int32_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 718 u_int32_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 719 u_int32_t tcps_sack_sboverflow; /* SACK sendblock overflow */ 720 721 u_int32_t tcps_bg_rcvtotal; /* total background packets received */ 722 u_int32_t tcps_rxtfindrop; /* drop conn after retransmitting FIN */ 723 u_int32_t tcps_fcholdpacket; /* packets withheld because of flow control */ 724 725 /* LRO related stats */ 726 u_int32_t tcps_coalesced_pack; /* number of coalesced packets */ 727 u_int32_t tcps_flowtbl_full; /* times flow table was full */ 728 u_int32_t tcps_flowtbl_collision; /* collisions in flow tbl */ 729 u_int32_t tcps_lro_twopack; /* 2 packets coalesced */ 730 u_int32_t tcps_lro_multpack; /* 3 or 4 pkts coalesced */ 731 u_int32_t tcps_lro_largepack; /* 5 or more pkts coalesced */ 732}; 733 734struct tcpstat_local { 735 u_int64_t badformat; 736 u_int64_t unspecv6; 737 u_int64_t synfin; 738 u_int64_t badformatipsec; 739 u_int64_t noconnnolist; 740 u_int64_t noconnlist; 741 u_int64_t listbadsyn; 742 u_int64_t icmp6unreach; 743 u_int64_t deprecate6; 744 u_int64_t ooopacket; 745 u_int64_t rstinsynrcv; 746 u_int64_t dospacket; 747 u_int64_t cleanup; 748 u_int64_t synwindow; 749}; 750 751#pragma pack(4) 752 753/* 754 * TCB structure exported to user-land via sysctl(3). 755 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 756 * included. Not all of our clients do. 757 */ 758 759struct xtcpcb { 760 u_int32_t xt_len; 761#ifdef KERNEL_PRIVATE 762 struct inpcb_compat xt_inp; 763#else 764 struct inpcb xt_inp; 765#endif 766#ifdef KERNEL_PRIVATE 767 struct otcpcb xt_tp; 768#else 769 struct tcpcb xt_tp; 770#endif 771 struct xsocket xt_socket; 772 u_quad_t xt_alignment_hack; 773}; 774 775#if !CONFIG_EMBEDDED 776 777struct xtcpcb64 { 778 u_int32_t xt_len; 779 struct xinpcb64 xt_inpcb; 780 781 u_int64_t t_segq; 782 int t_dupacks; /* consecutive dup acks recd */ 783 784 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 785 786 int t_state; /* state of this connection */ 787 u_int t_flags; 788 789 int t_force; /* 1 if forcing out a byte */ 790 791 tcp_seq snd_una; /* send unacknowledged */ 792 tcp_seq snd_max; /* highest sequence number sent; 793 * used to recognize retransmits 794 */ 795 tcp_seq snd_nxt; /* send next */ 796 tcp_seq snd_up; /* send urgent pointer */ 797 798 tcp_seq snd_wl1; /* window update seg seq number */ 799 tcp_seq snd_wl2; /* window update seg ack number */ 800 tcp_seq iss; /* initial send sequence number */ 801 tcp_seq irs; /* initial receive sequence number */ 802 803 tcp_seq rcv_nxt; /* receive next */ 804 tcp_seq rcv_adv; /* advertised window */ 805 u_int32_t rcv_wnd; /* receive window */ 806 tcp_seq rcv_up; /* receive urgent pointer */ 807 808 u_int32_t snd_wnd; /* send window */ 809 u_int32_t snd_cwnd; /* congestion-controlled window */ 810 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 811 * for slow start exponential to 812 * linear switch 813 */ 814 u_int t_maxopd; /* mss plus options */ 815 816 u_int32_t t_rcvtime; /* time at which a packet was received */ 817 u_int32_t t_starttime; /* time connection was established */ 818 int t_rtttime; /* round trip time */ 819 tcp_seq t_rtseq; /* sequence number being timed */ 820 821 int t_rxtcur; /* current retransmit value (ticks) */ 822 u_int t_maxseg; /* maximum segment size */ 823 int t_srtt; /* smoothed round-trip time */ 824 int t_rttvar; /* variance in round-trip time */ 825 826 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 827 u_int t_rttmin; /* minimum rtt allowed */ 828 u_int32_t t_rttupdated; /* number of times rtt sampled */ 829 u_int32_t max_sndwnd; /* largest window peer has offered */ 830 831 int t_softerror; /* possible error not yet reported */ 832/* out-of-band data */ 833 char t_oobflags; /* have some */ 834 char t_iobc; /* input character */ 835/* RFC 1323 variables */ 836 u_char snd_scale; /* window scaling for send window */ 837 u_char rcv_scale; /* window scaling for recv window */ 838 u_char request_r_scale; /* pending window scaling */ 839 u_char requested_s_scale; 840 u_int32_t ts_recent; /* timestamp echo data */ 841 842 u_int32_t ts_recent_age; /* when last updated */ 843 tcp_seq last_ack_sent; 844/* RFC 1644 variables */ 845 tcp_cc cc_send; /* send connection count */ 846 tcp_cc cc_recv; /* receive connection count */ 847 tcp_seq snd_recover; /* for use in fast recovery */ 848/* experimental */ 849 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 850 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 851 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 852 853 u_quad_t xt_alignment_hack; 854}; 855 856#endif /* !CONFIG_EMBEDDED */ 857 858#ifdef PRIVATE 859 860struct xtcpcb_n { 861 u_int32_t xt_len; 862 u_int32_t xt_kind; /* XSO_TCPCB */ 863 864 u_int64_t t_segq; 865 int t_dupacks; /* consecutive dup acks recd */ 866 867 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 868 869 int t_state; /* state of this connection */ 870 u_int t_flags; 871 872 int t_force; /* 1 if forcing out a byte */ 873 874 tcp_seq snd_una; /* send unacknowledged */ 875 tcp_seq snd_max; /* highest sequence number sent; 876 * used to recognize retransmits 877 */ 878 tcp_seq snd_nxt; /* send next */ 879 tcp_seq snd_up; /* send urgent pointer */ 880 881 tcp_seq snd_wl1; /* window update seg seq number */ 882 tcp_seq snd_wl2; /* window update seg ack number */ 883 tcp_seq iss; /* initial send sequence number */ 884 tcp_seq irs; /* initial receive sequence number */ 885 886 tcp_seq rcv_nxt; /* receive next */ 887 tcp_seq rcv_adv; /* advertised window */ 888 u_int32_t rcv_wnd; /* receive window */ 889 tcp_seq rcv_up; /* receive urgent pointer */ 890 891 u_int32_t snd_wnd; /* send window */ 892 u_int32_t snd_cwnd; /* congestion-controlled window */ 893 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 894 * for slow start exponential to 895 * linear switch 896 */ 897 u_int t_maxopd; /* mss plus options */ 898 899 u_int32_t t_rcvtime; /* time at which a packet was received */ 900 u_int32_t t_starttime; /* time connection was established */ 901 int t_rtttime; /* round trip time */ 902 tcp_seq t_rtseq; /* sequence number being timed */ 903 904 int t_rxtcur; /* current retransmit value (ticks) */ 905 u_int t_maxseg; /* maximum segment size */ 906 int t_srtt; /* smoothed round-trip time */ 907 int t_rttvar; /* variance in round-trip time */ 908 909 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 910 u_int t_rttmin; /* minimum rtt allowed */ 911 u_int32_t t_rttupdated; /* number of times rtt sampled */ 912 u_int32_t max_sndwnd; /* largest window peer has offered */ 913 914 int t_softerror; /* possible error not yet reported */ 915 /* out-of-band data */ 916 char t_oobflags; /* have some */ 917 char t_iobc; /* input character */ 918 /* RFC 1323 variables */ 919 u_char snd_scale; /* window scaling for send window */ 920 u_char rcv_scale; /* window scaling for recv window */ 921 u_char request_r_scale; /* pending window scaling */ 922 u_char requested_s_scale; 923 u_int32_t ts_recent; /* timestamp echo data */ 924 925 u_int32_t ts_recent_age; /* when last updated */ 926 tcp_seq last_ack_sent; 927 /* RFC 1644 variables */ 928 tcp_cc cc_send; /* send connection count */ 929 tcp_cc cc_recv; /* receive connection count */ 930 tcp_seq snd_recover; /* for use in fast recovery */ 931 /* experimental */ 932 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 933 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 934 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 935}; 936 937#endif /* PRIVATE */ 938 939#pragma pack() 940 941/* 942 * Names for TCP sysctl objects 943 */ 944#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 945#define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ 946#define TCPCTL_MSSDFLT 3 /* MSS default */ 947#define TCPCTL_STATS 4 /* statistics (read-only) */ 948#define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 949#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 950#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 951#define TCPCTL_SENDSPACE 8 /* send buffer space */ 952#define TCPCTL_RECVSPACE 9 /* receive buffer space */ 953#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 954#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 955#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 956#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 957#define TCPCTL_MAXID 14 958 959#ifdef KERNEL_PRIVATE 960#define TCP_PKTLIST_CLEAR(tp) { \ 961 (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL; \ 962 (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0; \ 963} 964 965#define TCPCTL_NAMES { \ 966 { 0, 0 }, \ 967 { "rfc1323", CTLTYPE_INT }, \ 968 { "rfc1644", CTLTYPE_INT }, \ 969 { "mssdflt", CTLTYPE_INT }, \ 970 { "stats", CTLTYPE_STRUCT }, \ 971 { "rttdflt", CTLTYPE_INT }, \ 972 { "keepidle", CTLTYPE_INT }, \ 973 { "keepintvl", CTLTYPE_INT }, \ 974 { "sendspace", CTLTYPE_INT }, \ 975 { "recvspace", CTLTYPE_INT }, \ 976 { "keepinit", CTLTYPE_INT }, \ 977 { "pcblist", CTLTYPE_STRUCT }, \ 978 { "delacktime", CTLTYPE_INT }, \ 979 { "v6mssdflt", CTLTYPE_INT }, \ 980} 981 982#ifdef SYSCTL_DECL 983SYSCTL_DECL(_net_inet_tcp); 984#endif /* SYSCTL_DECL */ 985 986extern struct inpcbhead tcb; /* head of queue of active tcpcb's */ 987extern struct inpcbinfo tcbinfo; 988extern struct tcpstat tcpstat; /* tcp statistics */ 989extern int tcp_mssdflt; /* XXX */ 990extern int tcp_minmss; 991extern int tcp_minmssoverload; 992extern int ss_fltsz; 993extern int ss_fltsz_local; 994extern int tcp_do_rfc3390; /* Calculate ss_fltsz according to RFC 3390 */ 995#ifdef __APPLE__ 996extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 997extern struct timeval tcp_uptime; 998extern lck_spin_t *tcp_uptime_lock; 999 1000extern int tcp_delack_enabled; 1001#endif /* __APPLE__ */ 1002 1003extern int tcp_do_sack; /* SACK enabled/disabled */ 1004 1005#if CONFIG_IFEF_NOWINDOWSCALE 1006extern int tcp_obey_ifef_nowindowscale; 1007#endif 1008 1009void tcp_canceltimers(struct tcpcb *); 1010struct tcpcb * 1011 tcp_close(struct tcpcb *); 1012void tcp_ctlinput(int, struct sockaddr *, void *); 1013int tcp_ctloutput(struct socket *, struct sockopt *); 1014struct tcpcb * 1015 tcp_drop(struct tcpcb *, int); 1016void tcp_drain(void); 1017void tcp_getrt_rtt(struct tcpcb *tp, struct rtentry *rt); 1018struct rmxp_tao * 1019 tcp_gettaocache(struct inpcb *); 1020void tcp_init(void) __attribute__((section("__TEXT, initcode"))); 1021void tcp_input(struct mbuf *, int); 1022void tcp_mss(struct tcpcb *, int, unsigned int); 1023int tcp_mssopt(struct tcpcb *); 1024void tcp_drop_syn_sent(struct inpcb *, int); 1025void tcp_mtudisc(struct inpcb *, int); 1026struct tcpcb * 1027 tcp_newtcpcb(struct inpcb *); 1028int tcp_output(struct tcpcb *); 1029void tcp_respond(struct tcpcb *, void *, 1030 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int, 1031 unsigned int, unsigned int); 1032struct rtentry *tcp_rtlookup(struct inpcb *, unsigned int); 1033void tcp_setpersist(struct tcpcb *); 1034void tcp_slowtimo(void); 1035void tcp_check_timer_state(struct tcpcb *tp); 1036void tcp_run_timerlist(void *arg1, void *arg2); 1037 1038struct tcptemp * 1039 tcp_maketemplate(struct tcpcb *); 1040void tcp_fillheaders(struct tcpcb *, void *, void *); 1041struct tcpcb * 1042 tcp_timers(struct tcpcb *, int); 1043void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int); 1044 1045void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 1046void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 1047void tcp_clean_sackreport(struct tcpcb *tp); 1048void tcp_sack_adjust(struct tcpcb *tp); 1049struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 1050void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 1051void tcp_free_sackholes(struct tcpcb *tp); 1052int32_t tcp_sbspace(struct tcpcb *tp); 1053void tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp); 1054void tcp_reset_stretch_ack(struct tcpcb *tp); 1055void tcp_get_ports_used(unsigned int , uint8_t *); 1056uint32_t tcp_count_opportunistic(unsigned int ifindex, u_int32_t flags); 1057void tcp_set_max_rwinscale(struct tcpcb *tp, struct socket *so); 1058u_int8_t tcp_cansbgrow(struct sockbuf *sb); 1059struct bwmeas* tcp_bwmeas_alloc(struct tcpcb *tp); 1060void tcp_bwmeas_free(struct tcpcb *tp); 1061 1062extern void tcp_set_background_cc(struct socket *); 1063extern void tcp_set_foreground_cc(struct socket *); 1064extern void tcp_set_recv_bg(struct socket *); 1065extern void tcp_clear_recv_bg(struct socket *); 1066#define IS_TCP_RECV_BG(_so) \ 1067 ((_so)->so_traffic_mgt_flags & TRAFFIC_MGT_TCP_RECVBG) 1068 1069#if TRAFFIC_MGT 1070#define CLEAR_IAJ_STATE(_tp_) (_tp_)->iaj_rcv_ts = 0 1071void reset_acc_iaj(struct tcpcb *tp); 1072#endif /* TRAFFIC_MGT */ 1073 1074int tcp_lock (struct socket *, int, void *); 1075int tcp_unlock (struct socket *, int, void *); 1076void calculate_tcp_clock(void); 1077 1078#ifdef _KERN_LOCKS_H_ 1079lck_mtx_t * tcp_getlock (struct socket *, int); 1080#else 1081void * tcp_getlock (struct socket *, int); 1082#endif 1083 1084 1085extern struct pr_usrreqs tcp_usrreqs; 1086extern u_int32_t tcp_sendspace; 1087extern u_int32_t tcp_recvspace; 1088tcp_seq tcp_new_isn(struct tcpcb *); 1089 1090#endif /* KERNEL_RPIVATE */ 1091 1092#endif /* _NETINET_TCP_VAR_H_ */ 1093