1 /* 2 * Copyright (c) 2003, 2004 Jeffrey M. Hsu. All rights reserved. 3 * Copyright (c) 2003, 2004 The DragonFly Project. All rights reserved. 4 * 5 * This code is derived from software contributed to The DragonFly Project 6 * by Jeffrey M. Hsu. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. Neither the name of The DragonFly Project nor the names of its 17 * contributors may be used to endorse or promote products derived 18 * from this software without specific, prior written permission. 19 * 20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS 21 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT 22 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS 23 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE 24 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, 25 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING, 26 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; 27 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED 28 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, 29 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT 30 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 31 * SUCH DAMAGE. 32 */ 33 34 /* 35 * Copyright (c) 1982, 1986, 1993, 1994, 1995 36 * The Regents of the University of California. All rights reserved. 37 * 38 * Redistribution and use in source and binary forms, with or without 39 * modification, are permitted provided that the following conditions 40 * are met: 41 * 1. Redistributions of source code must retain the above copyright 42 * notice, this list of conditions and the following disclaimer. 43 * 2. Redistributions in binary form must reproduce the above copyright 44 * notice, this list of conditions and the following disclaimer in the 45 * documentation and/or other materials provided with the distribution. 46 * 3. All advertising materials mentioning features or use of this software 47 * must display the following acknowledgement: 48 * This product includes software developed by the University of 49 * California, Berkeley and its contributors. 50 * 4. Neither the name of the University nor the names of its contributors 51 * may be used to endorse or promote products derived from this software 52 * without specific prior written permission. 53 * 54 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 55 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 56 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 57 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 58 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 59 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 60 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 61 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 62 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 63 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 64 * SUCH DAMAGE. 65 * 66 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 67 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.13 2003/02/03 02:34:07 hsu Exp $ 68 * $DragonFly: src/sys/netinet/tcp_var.h,v 1.39 2007/03/04 18:51:59 swildner Exp $ 69 */ 70 71 #ifndef _NETINET_TCP_VAR_H_ 72 #define _NETINET_TCP_VAR_H_ 73 74 #ifndef _NETINET_IN_PCB_H_ 75 #include <netinet/in_pcb.h> /* needed for in_conninfo, inp_gen_t */ 76 #endif 77 #ifndef _NETINET_TCP_H_ 78 #include <netinet/tcp.h> 79 #endif 80 81 /* 82 * Kernel variables for tcp. 83 */ 84 extern int tcp_do_rfc1323; 85 extern int tcp_do_rfc1644; 86 extern int tcp_do_rfc3390; 87 extern int tcp_do_sack; 88 extern int tcp_do_smartsack; 89 extern int tcp_aggregate_acks; 90 91 /* TCP segment queue entry */ 92 struct tseg_qent { 93 LIST_ENTRY(tseg_qent) tqe_q; 94 int tqe_len; /* TCP segment data length */ 95 struct tcphdr *tqe_th; /* a pointer to tcp header */ 96 struct mbuf *tqe_m; /* mbuf contains packet */ 97 }; 98 LIST_HEAD(tsegqe_head, tseg_qent); 99 extern int tcp_reass_maxseg; 100 extern int tcp_reass_qsize; 101 #ifdef MALLOC_DECLARE 102 MALLOC_DECLARE(M_TSEGQ); 103 #endif 104 105 struct tcptemp { 106 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 107 struct tcphdr tt_t; 108 }; 109 110 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 111 112 struct raw_sackblock { /* covers [start, end) */ 113 tcp_seq rblk_start; 114 tcp_seq rblk_end; 115 }; 116 117 /* maximum number of SACK blocks that will fit in the TCP option space */ 118 #define MAX_SACK_REPORT_BLOCKS 4 119 120 TAILQ_HEAD(sackblock_list, sackblock); 121 122 struct scoreboard { 123 int nblocks; 124 struct sackblock_list sackblocks; 125 tcp_seq lostseq; /* passed SACK lost test */ 126 struct sackblock *lastfound; /* search hint */ 127 }; 128 129 /* 130 * Tcp control block, one per tcp; fields: 131 * Organized for 16 byte cacheline efficiency. 132 */ 133 struct tcpcb { 134 struct tsegqe_head t_segq; 135 int t_dupacks; /* consecutive dup acks recd */ 136 int tt_cpu; /* sanity check the cpu */ 137 struct callout *tt_rexmt; /* retransmit timer */ 138 139 struct callout *tt_persist; /* retransmit persistence */ 140 struct callout *tt_keep; /* keepalive */ 141 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */ 142 struct callout *tt_delack; /* delayed ACK timer */ 143 144 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 145 int t_state; /* state of this connection */ 146 u_int t_flags; 147 #define TF_ACKNOW 0x00000001 /* ack peer immediately */ 148 #define TF_DELACK 0x00000002 /* ack, but try to delay it */ 149 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */ 150 #define TF_NOOPT 0x00000008 /* don't use tcp options */ 151 #define TF_SENTFIN 0x00000010 /* have sent FIN */ 152 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */ 153 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */ 154 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */ 155 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */ 156 #define TF_SACK_PERMITTED 0x00000200 /* other side said I could SACK */ 157 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */ 158 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */ 159 #define TF_NOPUSH 0x00001000 /* don't push */ 160 #define TF_REQ_CC 0x00002000 /* have/will request CC */ 161 #define TF_RCVD_CC 0x00004000 /* a CC was received in SYN */ 162 #define TF_SENDCCNEW 0x00008000 /* send CCnew instead of CC in SYN */ 163 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */ 164 #define TF_LQ_OVERFLOW 0x00020000 /* listen queue overflow */ 165 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */ 166 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */ 167 #define TF_FASTRECOVERY 0x00100000 /* in NewReno Fast Recovery */ 168 #define TF_WASFRECOVERY 0x00200000 /* was in NewReno Fast Recovery */ 169 #define TF_FIRSTACCACK 0x00400000 /* Look for 1st acceptable ACK. */ 170 #define TF_FASTREXMT 0x00800000 /* Did Fast Retransmit. */ 171 #define TF_EARLYREXMT 0x01000000 /* Did Early (Fast) Retransmit. */ 172 #define TF_FORCE 0x02000000 /* Set if forcing out a byte */ 173 #define TF_ONOUTPUTQ 0x04000000 /* on t_outputq list */ 174 #define TF_DUPSEG 0x08000000 /* last seg a duplicate */ 175 #define TF_ENCLOSESEG 0x10000000 /* enclosing SACK block */ 176 #define TF_SACKLEFT 0x20000000 /* send SACK blocks from left side */ 177 tcp_seq snd_up; /* send urgent pointer */ 178 179 tcp_seq snd_una; /* send unacknowledged */ 180 tcp_seq snd_recover; /* for use with NewReno Fast Recovery */ 181 tcp_seq snd_max; /* highest sequence number sent; 182 * used to recognize retransmits */ 183 tcp_seq snd_nxt; /* send next */ 184 185 tcp_seq snd_wl1; /* window update seg seq number */ 186 tcp_seq snd_wl2; /* window update seg ack number */ 187 tcp_seq iss; /* initial send sequence number */ 188 tcp_seq irs; /* initial receive sequence number */ 189 190 tcp_seq rcv_nxt; /* receive next */ 191 tcp_seq rcv_adv; /* advertised window */ 192 u_long rcv_wnd; /* receive window */ 193 tcp_seq rcv_up; /* receive urgent pointer */ 194 195 u_long snd_wnd; /* send window */ 196 u_long snd_cwnd; /* congestion-controlled window */ 197 u_long snd_wacked; /* bytes acked in one send window */ 198 u_long snd_ssthresh; /* snd_cwnd size threshold for 199 * for slow start exponential to 200 * linear switch */ 201 202 int t_rxtcur; /* current retransmit value (ticks) */ 203 u_int t_maxseg; /* maximum segment size */ 204 int t_srtt; /* smoothed round-trip time */ 205 int t_rttvar; /* variance in round-trip time */ 206 207 u_int t_maxopd; /* mss plus options */ 208 209 u_long t_rcvtime; /* inactivity time */ 210 u_long t_starttime; /* time connection was established */ 211 int t_rtttime; /* round trip time */ 212 tcp_seq t_rtseq; /* sequence number being timed */ 213 214 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 215 u_int t_rttmin; /* minimum rtt allowed */ 216 u_int t_rttbest; /* best rtt we've seen */ 217 u_long t_rttupdated; /* number of times rtt sampled */ 218 u_long max_sndwnd; /* largest window peer has offered */ 219 220 int t_softerror; /* possible error not yet reported */ 221 /* out-of-band data */ 222 char t_oobflags; /* have some */ 223 char t_iobc; /* input character */ 224 #define TCPOOB_HAVEDATA 0x01 225 #define TCPOOB_HADDATA 0x02 226 227 /* RFC 1323 variables */ 228 u_char snd_scale; /* window scaling for send window */ 229 u_char rcv_scale; /* window scaling for recv window */ 230 u_char request_r_scale; /* pending window scaling */ 231 u_char requested_s_scale; 232 u_long ts_recent; /* timestamp echo data */ 233 234 u_long ts_recent_age; /* when last updated */ 235 tcp_seq last_ack_sent; 236 237 /* RFC 1644 variables */ 238 tcp_cc cc_send; /* send connection count */ 239 tcp_cc cc_recv; /* receive connection count */ 240 241 /* experimental */ 242 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 243 u_long snd_wacked_prev; /* prior bytes acked in send window */ 244 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 245 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 246 u_long t_badrxtwin; /* window for retransmit recovery */ 247 u_long t_rexmtTS; /* timestamp of last retransmit */ 248 u_char snd_limited; /* segments limited transmitted */ 249 250 tcp_seq rexmt_high; /* highest seq # retransmitted + 1 */ 251 tcp_seq snd_max_rexmt; /* snd_max when rexmting snd_una */ 252 struct scoreboard scb; /* sack scoreboard */ 253 struct raw_sackblock reportblk; /* incoming segment or D-SACK block */ 254 struct raw_sackblock encloseblk; 255 int nsackhistory; 256 struct raw_sackblock sackhistory[MAX_SACK_REPORT_BLOCKS]; /* reported */ 257 TAILQ_ENTRY(tcpcb) t_outputq; /* tcp_output needed list */ 258 259 /* bandwith limitation */ 260 u_long snd_bandwidth; /* calculated bandwidth or 0 */ 261 u_long snd_bwnd; /* bandwidth-controlled window */ 262 int t_bw_rtttime; /* used for bandwidth calculation */ 263 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */ 264 }; 265 266 #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) 267 #define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY 268 #define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY 269 270 /* 271 * TCP statistics. 272 */ 273 struct tcp_stats { 274 u_long tcps_connattempt; /* connections initiated */ 275 u_long tcps_accepts; /* connections accepted */ 276 u_long tcps_connects; /* connections established */ 277 u_long tcps_drops; /* connections dropped */ 278 u_long tcps_conndrops; /* embryonic connections dropped */ 279 u_long tcps_closed; /* conn. closed (includes drops) */ 280 u_long tcps_segstimed; /* segs where we tried to get rtt */ 281 u_long tcps_rttupdated; /* times we succeeded */ 282 u_long tcps_delack; /* delayed acks sent */ 283 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 284 u_long tcps_rexmttimeo; /* retransmit timeouts */ 285 u_long tcps_persisttimeo; /* persist timeouts */ 286 u_long tcps_keeptimeo; /* keepalive timeouts */ 287 u_long tcps_keepprobe; /* keepalive probes sent */ 288 u_long tcps_keepdrops; /* connections dropped in keepalive */ 289 290 u_long tcps_sndtotal; /* total packets sent */ 291 u_long tcps_sndpack; /* data packets sent */ 292 u_long tcps_sndbyte; /* data bytes sent */ 293 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 294 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 295 u_long tcps_sndfastrexmit; /* Fast Retransmissions */ 296 u_long tcps_sndearlyrexmit; /* early Fast Retransmissions */ 297 u_long tcps_sndlimited; /* Limited Transmit packets */ 298 u_long tcps_sndrtobad; /* spurious RTO retransmissions */ 299 u_long tcps_sndfastrexmitbad; /* spurious Fast Retransmissions */ 300 u_long tcps_sndearlyrexmitbad; /* spurious early Fast Retransmissions, 301 a subset of tcps_sndfastrexmitbad */ 302 u_long tcps_eifeldetected; /* Eifel-detected spurious rexmits */ 303 u_long tcps_rttcantdetect; /* Eifel but not 1/2 RTT-detectable */ 304 u_long tcps_rttdetected; /* RTT-detected spurious RTO rexmits */ 305 u_long tcps_sndacks; /* ack-only packets sent */ 306 u_long tcps_sndprobe; /* window probes sent */ 307 u_long tcps_sndurg; /* packets sent with URG only */ 308 u_long tcps_sndwinup; /* window update-only packets sent */ 309 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 310 u_long tcps_sndsackpack; /* packets sent by SACK recovery alg */ 311 u_long tcps_sndsackbyte; /* bytes sent by SACK recovery */ 312 u_long tcps_snduna3; /* re-retransmit snd_una on 3 new seg */ 313 u_long tcps_snduna1; /* re-retransmit snd_una on 1 new seg */ 314 315 u_long tcps_rcvtotal; /* total packets received */ 316 u_long tcps_rcvpack; /* packets received in sequence */ 317 u_long tcps_rcvbyte; /* bytes received in sequence */ 318 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 319 u_long tcps_rcvbadoff; /* packets received with bad offset */ 320 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 321 u_long tcps_rcvshort; /* packets received too short */ 322 u_long tcps_rcvduppack; /* duplicate-only packets received */ 323 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 324 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 325 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 326 u_long tcps_rcvoopack; /* out-of-order packets received */ 327 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 328 u_long tcps_rcvpackafterwin; /* packets with data after window */ 329 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 330 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 331 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 332 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 333 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 334 u_long tcps_rcvackpack; /* rcvd ack packets */ 335 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 336 u_long tcps_rcvwinupd; /* rcvd window update packets */ 337 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 338 u_long tcps_predack; /* times hdr predict ok for acks */ 339 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 340 u_long tcps_pcbcachemiss; 341 u_long tcps_cachedrtt; /* times cached RTT in route updated */ 342 u_long tcps_cachedrttvar; /* times cached rttvar updated */ 343 u_long tcps_cachedssthresh; /* times cached ssthresh updated */ 344 u_long tcps_usedrtt; /* times RTT initialized from route */ 345 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ 346 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ 347 u_long tcps_persistdrop; /* timeout in persist state */ 348 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 349 u_long tcps_mturesent; /* resends due to MTU discovery */ 350 u_long tcps_listendrop; /* listen queue overflows */ 351 352 u_long tcps_sc_added; /* entry added to syncache */ 353 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */ 354 u_long tcps_sc_dupsyn; /* duplicate SYN packet */ 355 u_long tcps_sc_dropped; /* could not reply to packet */ 356 u_long tcps_sc_completed; /* successful extraction of entry */ 357 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ 358 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */ 359 u_long tcps_sc_reset; /* RST removed entry from syncache */ 360 u_long tcps_sc_stale; /* timed out or listen socket gone */ 361 u_long tcps_sc_aborted; /* syncache entry aborted */ 362 u_long tcps_sc_badack; /* removed due to bad ACK */ 363 u_long tcps_sc_unreach; /* ICMP unreachable received */ 364 u_long tcps_sc_zonefail; /* zalloc() failed */ 365 u_long tcps_sc_sendcookie; /* SYN cookie sent */ 366 u_long tcps_sc_recvcookie; /* SYN cookie received */ 367 }; 368 369 #ifdef _KERNEL 370 371 #if defined(SMP) 372 #define tcpstat tcpstats_percpu[mycpuid] 373 #else 374 #define tcpstat tcpstats_percpu[0] 375 #endif 376 377 struct sockopt; 378 379 extern struct tcp_stats tcpstats_percpu[MAXCPU]; 380 381 static const int tcprexmtthresh = 3; 382 #endif 383 384 /* 385 * Structure to hold TCP options that are only used during segment 386 * processing (in tcp_input), but not held in the tcpcb. 387 * It's basically used to reduce the number of parameters 388 * to tcp_dooptions. 389 */ 390 struct tcpopt { 391 u_long to_flags; /* which options are present */ 392 #define TOF_TS 0x0001 /* timestamp */ 393 #define TOF_CC 0x0002 /* CC and CCnew are exclusive */ 394 #define TOF_CCNEW 0x0004 395 #define TOF_CCECHO 0x0008 396 #define TOF_MSS 0x0010 397 #define TOF_SCALE 0x0020 398 #define TOF_SACK_PERMITTED 0x0040 399 #define TOF_SACK 0x0080 400 u_int32_t to_tsval; 401 u_int32_t to_tsecr; 402 tcp_cc to_cc; /* holds CC or CCnew */ 403 tcp_cc to_ccecho; 404 u_int16_t to_mss; 405 u_int8_t to_requested_s_scale; 406 u_int8_t to_nsackblocks; 407 struct raw_sackblock *to_sackblocks; 408 }; 409 410 struct syncache { 411 inp_gen_t sc_inp_gencnt; /* pointer check */ 412 struct tcpcb *sc_tp; /* tcb for listening socket */ 413 struct mbuf *sc_ipopts; /* source route */ 414 struct in_conninfo sc_inc; /* addresses */ 415 #define sc_route sc_inc.inc_route 416 #define sc_route6 sc_inc.inc6_route 417 u_int32_t sc_tsrecent; 418 tcp_cc sc_cc_send; /* holds CC or CCnew */ 419 tcp_cc sc_cc_recv; 420 tcp_seq sc_irs; /* seq from peer */ 421 tcp_seq sc_iss; /* our ISS */ 422 u_long sc_rxttime; /* retransmit time */ 423 u_int16_t sc_rxtslot; /* retransmit counter */ 424 u_int16_t sc_peer_mss; /* peer's MSS */ 425 u_int16_t sc_wnd; /* advertised window */ 426 u_int8_t sc_requested_s_scale:4, 427 sc_request_r_scale:4; 428 u_int8_t sc_flags; 429 #define SCF_NOOPT 0x01 /* no TCP options */ 430 #define SCF_WINSCALE 0x02 /* negotiated window scaling */ 431 #define SCF_TIMESTAMP 0x04 /* negotiated timestamps */ 432 #define SCF_CC 0x08 /* negotiated CC */ 433 #define SCF_UNREACH 0x10 /* icmp unreachable received */ 434 #define SCF_SACK_PERMITTED 0x20 /* saw SACK permitted option */ 435 TAILQ_ENTRY(syncache) sc_hash; 436 TAILQ_ENTRY(syncache) sc_timerq; 437 }; 438 439 struct syncache_head { 440 TAILQ_HEAD(, syncache) sch_bucket; 441 u_int sch_length; 442 }; 443 444 /* 445 * The TAO cache entry which is stored in the protocol family specific 446 * portion of the route metrics. 447 */ 448 struct rmxp_tao { 449 tcp_cc tao_cc; /* latest CC in valid SYN */ 450 tcp_cc tao_ccsent; /* latest CC sent to peer */ 451 u_short tao_mssopt; /* peer's cached MSS */ 452 453 #ifdef notyet 454 u_short tao_flags; /* cache status flags */ 455 #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */ 456 #define TAOF_OK 0x0002 /* peer does understand rfc1644 */ 457 #define TAOF_UNDEF 0 /* we don't know yet */ 458 #endif 459 }; 460 461 #define rmx_taop(rt) ((struct rmxp_tao *)(rt).rmx_filler) 462 463 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 464 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 465 466 /* 467 * The smoothed round-trip time and estimated variance 468 * are stored as fixed point numbers scaled by the values below. 469 * For convenience, these scales are also used in smoothing the average 470 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 471 * With these scales, srtt has 3 bits to the right of the binary point, 472 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 473 * binary point, and is smoothed with an ALPHA of 0.75. 474 */ 475 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 476 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 477 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 478 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 479 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 480 481 /* 482 * The initial retransmission should happen at rtt + 4 * rttvar. 483 * Because of the way we do the smoothing, srtt and rttvar 484 * will each average +1/2 tick of bias. When we compute 485 * the retransmit timer, we want 1/2 tick of rounding and 486 * 1 extra tick because of +-1/2 tick uncertainty in the 487 * firing of the timer. The bias will give us exactly the 488 * 1.5 tick we need. But, because the bias is 489 * statistical, we have to test that we don't drop below 490 * the minimum feasible timer (which is 2 ticks). 491 * This version of the macro adapted from a paper by Lawrence 492 * Brakmo and Larry Peterson which outlines a problem caused 493 * by insufficient precision in the original implementation, 494 * which results in inappropriately large RTO values for very 495 * fast networks. 496 */ 497 #define TCP_REXMTVAL(tp) \ 498 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 499 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 500 501 /* 502 * TCB structure exported to user-land via sysctl(3). 503 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 504 * included. Not all of our clients do. 505 */ 506 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 507 struct xtcpcb { 508 size_t xt_len; 509 struct inpcb xt_inp; 510 struct tcpcb xt_tp; 511 struct xsocket xt_socket; 512 u_quad_t xt_alignment_hack; 513 }; 514 #endif 515 516 /* 517 * Names for TCP sysctl objects 518 */ 519 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 520 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ 521 #define TCPCTL_MSSDFLT 3 /* MSS default */ 522 #define TCPCTL_STATS 4 /* statistics (read-only) */ 523 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 524 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 525 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 526 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 527 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 528 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 529 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 530 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 531 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 532 #define TCPCTL_MAXID 14 533 534 #define TCPCTL_NAMES { \ 535 { 0, 0 }, \ 536 { "rfc1323", CTLTYPE_INT }, \ 537 { "rfc1644", CTLTYPE_INT }, \ 538 { "mssdflt", CTLTYPE_INT }, \ 539 { "stats", CTLTYPE_STRUCT }, \ 540 { "rttdflt", CTLTYPE_INT }, \ 541 { "keepidle", CTLTYPE_INT }, \ 542 { "keepintvl", CTLTYPE_INT }, \ 543 { "sendspace", CTLTYPE_INT }, \ 544 { "recvspace", CTLTYPE_INT }, \ 545 { "keepinit", CTLTYPE_INT }, \ 546 { "pcblist", CTLTYPE_STRUCT }, \ 547 { "delacktime", CTLTYPE_INT }, \ 548 { "v6mssdflt", CTLTYPE_INT }, \ 549 } 550 551 #ifdef _KERNEL 552 #ifdef SYSCTL_DECL 553 SYSCTL_DECL(_net_inet_tcp); 554 #endif 555 556 #define TCP_DO_SACK(tp) ((tp)->t_flags & TF_SACK_PERMITTED) 557 558 TAILQ_HEAD(tcpcbackqhead,tcpcb); 559 560 extern struct inpcbinfo tcbinfo[]; 561 extern struct tcpcbackqhead tcpcbackq[]; 562 563 extern int tcp_mssdflt; /* XXX */ 564 extern int tcp_delack_enabled; 565 extern int path_mtu_discovery; 566 567 int tcp_addrcpu(in_addr_t faddr, in_port_t fport, 568 in_addr_t laddr, in_port_t lport); 569 struct lwkt_port * 570 tcp_addrport(in_addr_t faddr, in_port_t fport, 571 in_addr_t laddr, in_port_t lport); 572 void tcp_canceltimers (struct tcpcb *); 573 struct tcpcb * 574 tcp_close (struct tcpcb *); 575 void tcpmsg_service_loop (void *); 576 void tcp_ctlinput (int, struct sockaddr *, void *); 577 int tcp_ctloutput (struct socket *, struct sockopt *); 578 struct lwkt_port * 579 tcp_cport(int cpu); 580 struct tcpcb * 581 tcp_drop (struct tcpcb *, int); 582 void tcp_drain (void); 583 void tcp_fasttimo (void); 584 struct rmxp_tao * 585 tcp_gettaocache (struct in_conninfo *); 586 void tcp_init (void); 587 void tcp_thread_init (void); 588 void tcp_input (struct mbuf *, ...); 589 void tcp_mss (struct tcpcb *, int); 590 int tcp_mssopt (struct tcpcb *); 591 void tcp_drop_syn_sent (struct inpcb *, int); 592 void tcp_mtudisc (struct inpcb *, int); 593 struct tcpcb * 594 tcp_newtcpcb (struct inpcb *); 595 int tcp_output (struct tcpcb *); 596 void tcp_quench (struct inpcb *, int); 597 void tcp_respond (struct tcpcb *, void *, 598 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 599 struct rtentry * 600 tcp_rtlookup (struct in_conninfo *); 601 int tcp_sack_bytes_below(struct scoreboard *scb, tcp_seq seq); 602 void tcp_sack_cleanup(struct scoreboard *scb); 603 int tcp_sack_ndsack_blocks(struct raw_sackblock *blocks, 604 const int numblocks, tcp_seq snd_una); 605 void tcp_sack_fill_report(struct tcpcb *tp, u_char *opt, u_int *plen); 606 boolean_t 607 tcp_sack_has_sacked(struct scoreboard *scb, u_int amount); 608 void tcp_sack_init(void); 609 void tcp_sack_tcpcb_init(struct tcpcb *tp); 610 uint32_t tcp_sack_compute_pipe(struct tcpcb *tp); 611 boolean_t 612 tcp_sack_nextseg(struct tcpcb *tp, tcp_seq *nextrexmt, uint32_t *len, 613 boolean_t *losdup); 614 #ifdef later 615 void tcp_sack_revert_scoreboard(struct scoreboard *scb, tcp_seq snd_una, 616 u_int maxseg); 617 void tcp_sack_save_scoreboard(struct scoreboard *scb); 618 #endif 619 void tcp_sack_skip_sacked(struct scoreboard *scb, tcp_seq *prexmt); 620 void tcp_sack_update_scoreboard(struct tcpcb *tp, struct tcpopt *to); 621 void tcp_save_congestion_state(struct tcpcb *tp); 622 void tcp_revert_congestion_state(struct tcpcb *tp); 623 void tcp_setpersist (struct tcpcb *); 624 void tcp_slowtimo (void); 625 struct tcptemp *tcp_maketemplate (struct tcpcb *); 626 void tcp_freetemplate (struct tcptemp *); 627 void tcp_fillheaders (struct tcpcb *, void *, void *); 628 struct lwkt_port * 629 tcp_soport(struct socket *, struct sockaddr *nam, int req); 630 struct tcpcb * 631 tcp_timers (struct tcpcb *, int); 632 void tcp_trace (short, short, struct tcpcb *, void *, struct tcphdr *, 633 int); 634 void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq); 635 void syncache_init(void); 636 void syncache_unreach(struct in_conninfo *, struct tcphdr *); 637 int syncache_expand(struct in_conninfo *, struct tcphdr *, 638 struct socket **, struct mbuf *); 639 int syncache_add(struct in_conninfo *, struct tcpopt *, 640 struct tcphdr *, struct socket **, struct mbuf *); 641 void syncache_chkrst(struct in_conninfo *, struct tcphdr *); 642 void syncache_badack(struct in_conninfo *); 643 644 extern struct pr_usrreqs tcp_usrreqs; 645 extern u_long tcp_sendspace; 646 extern u_long tcp_recvspace; 647 tcp_seq tcp_new_isn (struct tcpcb *); 648 649 #endif /* _KERNEL */ 650 651 #endif /* _NETINET_TCP_VAR_H_ */ 652