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) 2003, 2004 Jeffrey M. Hsu. All rights reserved. 36 * 37 * License terms: all terms for the DragonFly license above plus the following: 38 * 39 * 4. All advertising materials mentioning features or use of this software 40 * must display the following acknowledgement: 41 * 42 * This product includes software developed by Jeffrey M. Hsu 43 * for the DragonFly Project. 44 * 45 * This requirement may be waived with permission from Jeffrey Hsu. 46 * This requirement will sunset and may be removed on July 8 2005, 47 * after which the standard DragonFly license (as shown above) will 48 * apply. 49 */ 50 51 /* 52 * Copyright (c) 1982, 1986, 1993, 1994, 1995 53 * The Regents of the University of California. All rights reserved. 54 * 55 * Redistribution and use in source and binary forms, with or without 56 * modification, are permitted provided that the following conditions 57 * are met: 58 * 1. Redistributions of source code must retain the above copyright 59 * notice, this list of conditions and the following disclaimer. 60 * 2. Redistributions in binary form must reproduce the above copyright 61 * notice, this list of conditions and the following disclaimer in the 62 * documentation and/or other materials provided with the distribution. 63 * 3. All advertising materials mentioning features or use of this software 64 * must display the following acknowledgement: 65 * This product includes software developed by the University of 66 * California, Berkeley and its contributors. 67 * 4. Neither the name of the University nor the names of its contributors 68 * may be used to endorse or promote products derived from this software 69 * without specific prior written permission. 70 * 71 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 72 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 73 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 74 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 75 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 76 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 77 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 78 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 79 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 80 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 81 * SUCH DAMAGE. 82 * 83 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 84 * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.13 2003/02/03 02:34:07 hsu Exp $ 85 * $DragonFly: src/sys/netinet/tcp_var.h,v 1.37 2006/05/20 02:42:12 dillon Exp $ 86 */ 87 88 #ifndef _NETINET_TCP_VAR_H_ 89 #define _NETINET_TCP_VAR_H_ 90 91 #ifndef _NETINET_IN_PCB_H_ 92 #include <netinet/in_pcb.h> /* needed for in_conninfo, inp_gen_t */ 93 #endif 94 #ifndef _NETINET_TCP_H_ 95 #include <netinet/tcp.h> 96 #endif 97 98 /* 99 * Kernel variables for tcp. 100 */ 101 extern int tcp_do_rfc1323; 102 extern int tcp_do_rfc1644; 103 extern int tcp_do_rfc3390; 104 extern int tcp_do_sack; 105 extern int tcp_do_smartsack; 106 extern int tcp_aggregate_acks; 107 108 /* TCP segment queue entry */ 109 struct tseg_qent { 110 LIST_ENTRY(tseg_qent) tqe_q; 111 int tqe_len; /* TCP segment data length */ 112 struct tcphdr *tqe_th; /* a pointer to tcp header */ 113 struct mbuf *tqe_m; /* mbuf contains packet */ 114 }; 115 LIST_HEAD(tsegqe_head, tseg_qent); 116 extern int tcp_reass_maxseg; 117 extern int tcp_reass_qsize; 118 #ifdef MALLOC_DECLARE 119 MALLOC_DECLARE(M_TSEGQ); 120 #endif 121 122 struct tcptemp { 123 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 124 struct tcphdr tt_t; 125 }; 126 127 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 128 129 struct raw_sackblock { /* covers [start, end) */ 130 tcp_seq rblk_start; 131 tcp_seq rblk_end; 132 }; 133 134 /* maximum number of SACK blocks that will fit in the TCP option space */ 135 #define MAX_SACK_REPORT_BLOCKS 4 136 137 TAILQ_HEAD(sackblock_list, sackblock); 138 139 struct scoreboard { 140 int nblocks; 141 struct sackblock_list sackblocks; 142 tcp_seq lostseq; /* passed SACK lost test */ 143 struct sackblock *lastfound; /* search hint */ 144 }; 145 146 /* 147 * Tcp control block, one per tcp; fields: 148 * Organized for 16 byte cacheline efficiency. 149 */ 150 struct tcpcb { 151 struct tsegqe_head t_segq; 152 int t_dupacks; /* consecutive dup acks recd */ 153 int tt_cpu; /* sanity check the cpu */ 154 struct callout *tt_rexmt; /* retransmit timer */ 155 156 struct callout *tt_persist; /* retransmit persistence */ 157 struct callout *tt_keep; /* keepalive */ 158 struct callout *tt_2msl; /* 2*msl TIME_WAIT timer */ 159 struct callout *tt_delack; /* delayed ACK timer */ 160 161 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 162 int t_state; /* state of this connection */ 163 u_int t_flags; 164 #define TF_ACKNOW 0x00000001 /* ack peer immediately */ 165 #define TF_DELACK 0x00000002 /* ack, but try to delay it */ 166 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */ 167 #define TF_NOOPT 0x00000008 /* don't use tcp options */ 168 #define TF_SENTFIN 0x00000010 /* have sent FIN */ 169 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */ 170 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */ 171 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */ 172 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */ 173 #define TF_SACK_PERMITTED 0x00000200 /* other side said I could SACK */ 174 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */ 175 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */ 176 #define TF_NOPUSH 0x00001000 /* don't push */ 177 #define TF_REQ_CC 0x00002000 /* have/will request CC */ 178 #define TF_RCVD_CC 0x00004000 /* a CC was received in SYN */ 179 #define TF_SENDCCNEW 0x00008000 /* send CCnew instead of CC in SYN */ 180 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */ 181 #define TF_LQ_OVERFLOW 0x00020000 /* listen queue overflow */ 182 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */ 183 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */ 184 #define TF_FASTRECOVERY 0x00100000 /* in NewReno Fast Recovery */ 185 #define TF_WASFRECOVERY 0x00200000 /* was in NewReno Fast Recovery */ 186 #define TF_FIRSTACCACK 0x00400000 /* Look for 1st acceptable ACK. */ 187 #define TF_FASTREXMT 0x00800000 /* Did Fast Retransmit. */ 188 #define TF_EARLYREXMT 0x01000000 /* Did Early (Fast) Retransmit. */ 189 #define TF_FORCE 0x02000000 /* Set if forcing out a byte */ 190 #define TF_ONOUTPUTQ 0x04000000 /* on t_outputq list */ 191 #define TF_DUPSEG 0x08000000 /* last seg a duplicate */ 192 #define TF_ENCLOSESEG 0x10000000 /* enclosing SACK block */ 193 #define TF_SACKLEFT 0x20000000 /* send SACK blocks from left side */ 194 tcp_seq snd_up; /* send urgent pointer */ 195 196 tcp_seq snd_una; /* send unacknowledged */ 197 tcp_seq snd_recover; /* for use with NewReno Fast Recovery */ 198 tcp_seq snd_max; /* highest sequence number sent; 199 * used to recognize retransmits */ 200 tcp_seq snd_nxt; /* send next */ 201 202 tcp_seq snd_wl1; /* window update seg seq number */ 203 tcp_seq snd_wl2; /* window update seg ack number */ 204 tcp_seq iss; /* initial send sequence number */ 205 tcp_seq irs; /* initial receive sequence number */ 206 207 tcp_seq rcv_nxt; /* receive next */ 208 tcp_seq rcv_adv; /* advertised window */ 209 u_long rcv_wnd; /* receive window */ 210 tcp_seq rcv_up; /* receive urgent pointer */ 211 212 u_long snd_wnd; /* send window */ 213 u_long snd_cwnd; /* congestion-controlled window */ 214 u_long snd_wacked; /* bytes acked in one send window */ 215 u_long snd_ssthresh; /* snd_cwnd size threshold for 216 * for slow start exponential to 217 * linear switch */ 218 219 int t_rxtcur; /* current retransmit value (ticks) */ 220 u_int t_maxseg; /* maximum segment size */ 221 int t_srtt; /* smoothed round-trip time */ 222 int t_rttvar; /* variance in round-trip time */ 223 224 u_int t_maxopd; /* mss plus options */ 225 226 u_long t_rcvtime; /* inactivity time */ 227 u_long t_starttime; /* time connection was established */ 228 int t_rtttime; /* round trip time */ 229 tcp_seq t_rtseq; /* sequence number being timed */ 230 231 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 232 u_int t_rttmin; /* minimum rtt allowed */ 233 u_int t_rttbest; /* best rtt we've seen */ 234 u_long t_rttupdated; /* number of times rtt sampled */ 235 u_long max_sndwnd; /* largest window peer has offered */ 236 237 int t_softerror; /* possible error not yet reported */ 238 /* out-of-band data */ 239 char t_oobflags; /* have some */ 240 char t_iobc; /* input character */ 241 #define TCPOOB_HAVEDATA 0x01 242 #define TCPOOB_HADDATA 0x02 243 244 /* RFC 1323 variables */ 245 u_char snd_scale; /* window scaling for send window */ 246 u_char rcv_scale; /* window scaling for recv window */ 247 u_char request_r_scale; /* pending window scaling */ 248 u_char requested_s_scale; 249 u_long ts_recent; /* timestamp echo data */ 250 251 u_long ts_recent_age; /* when last updated */ 252 tcp_seq last_ack_sent; 253 254 /* RFC 1644 variables */ 255 tcp_cc cc_send; /* send connection count */ 256 tcp_cc cc_recv; /* receive connection count */ 257 258 /* experimental */ 259 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 260 u_long snd_wacked_prev; /* prior bytes acked in send window */ 261 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 262 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 263 u_long t_badrxtwin; /* window for retransmit recovery */ 264 u_long t_rexmtTS; /* timestamp of last retransmit */ 265 u_char snd_limited; /* segments limited transmitted */ 266 267 tcp_seq rexmt_high; /* highest seq # retransmitted + 1 */ 268 tcp_seq snd_max_rexmt; /* snd_max when rexmting snd_una */ 269 struct scoreboard scb; /* sack scoreboard */ 270 struct raw_sackblock reportblk; /* incoming segment or D-SACK block */ 271 struct raw_sackblock encloseblk; 272 int nsackhistory; 273 struct raw_sackblock sackhistory[MAX_SACK_REPORT_BLOCKS]; /* reported */ 274 TAILQ_ENTRY(tcpcb) t_outputq; /* tcp_output needed list */ 275 276 /* bandwith limitation */ 277 u_long snd_bandwidth; /* calculated bandwidth or 0 */ 278 u_long snd_bwnd; /* bandwidth-controlled window */ 279 int t_bw_rtttime; /* used for bandwidth calculation */ 280 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */ 281 }; 282 283 #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) 284 #define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY 285 #define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY 286 287 #ifdef _KERNEL 288 289 #if defined(SMP) 290 #define tcpstat tcpstats_percpu[mycpuid] 291 #else 292 #define tcpstat tcpstats_percpu[0] 293 #endif 294 295 struct tcp_stats; 296 struct sockopt; 297 298 extern struct tcp_stats tcpstats_percpu[MAXCPU]; 299 300 static const int tcprexmtthresh = 3; 301 #endif 302 303 /* 304 * TCP statistics. 305 */ 306 struct tcp_stats { 307 u_long tcps_connattempt; /* connections initiated */ 308 u_long tcps_accepts; /* connections accepted */ 309 u_long tcps_connects; /* connections established */ 310 u_long tcps_drops; /* connections dropped */ 311 u_long tcps_conndrops; /* embryonic connections dropped */ 312 u_long tcps_closed; /* conn. closed (includes drops) */ 313 u_long tcps_segstimed; /* segs where we tried to get rtt */ 314 u_long tcps_rttupdated; /* times we succeeded */ 315 u_long tcps_delack; /* delayed acks sent */ 316 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 317 u_long tcps_rexmttimeo; /* retransmit timeouts */ 318 u_long tcps_persisttimeo; /* persist timeouts */ 319 u_long tcps_keeptimeo; /* keepalive timeouts */ 320 u_long tcps_keepprobe; /* keepalive probes sent */ 321 u_long tcps_keepdrops; /* connections dropped in keepalive */ 322 323 u_long tcps_sndtotal; /* total packets sent */ 324 u_long tcps_sndpack; /* data packets sent */ 325 u_long tcps_sndbyte; /* data bytes sent */ 326 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 327 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 328 u_long tcps_sndfastrexmit; /* Fast Retransmissions */ 329 u_long tcps_sndearlyrexmit; /* early Fast Retransmissions */ 330 u_long tcps_sndlimited; /* Limited Transmit packets */ 331 u_long tcps_sndrtobad; /* spurious RTO retransmissions */ 332 u_long tcps_sndfastrexmitbad; /* spurious Fast Retransmissions */ 333 u_long tcps_sndearlyrexmitbad; /* spurious early Fast Retransmissions, 334 a subset of tcps_sndfastrexmitbad */ 335 u_long tcps_eifeldetected; /* Eifel-detected spurious rexmits */ 336 u_long tcps_rttcantdetect; /* Eifel but not 1/2 RTT-detectable */ 337 u_long tcps_rttdetected; /* RTT-detected spurious RTO rexmits */ 338 u_long tcps_sndacks; /* ack-only packets sent */ 339 u_long tcps_sndprobe; /* window probes sent */ 340 u_long tcps_sndurg; /* packets sent with URG only */ 341 u_long tcps_sndwinup; /* window update-only packets sent */ 342 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 343 u_long tcps_sndsackpack; /* packets sent by SACK recovery alg */ 344 u_long tcps_sndsackbyte; /* bytes sent by SACK recovery */ 345 u_long tcps_snduna3; /* re-retransmit snd_una on 3 new seg */ 346 u_long tcps_snduna1; /* re-retransmit snd_una on 1 new seg */ 347 348 u_long tcps_rcvtotal; /* total packets received */ 349 u_long tcps_rcvpack; /* packets received in sequence */ 350 u_long tcps_rcvbyte; /* bytes received in sequence */ 351 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 352 u_long tcps_rcvbadoff; /* packets received with bad offset */ 353 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 354 u_long tcps_rcvshort; /* packets received too short */ 355 u_long tcps_rcvduppack; /* duplicate-only packets received */ 356 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 357 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 358 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 359 u_long tcps_rcvoopack; /* out-of-order packets received */ 360 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 361 u_long tcps_rcvpackafterwin; /* packets with data after window */ 362 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 363 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 364 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 365 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 366 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 367 u_long tcps_rcvackpack; /* rcvd ack packets */ 368 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 369 u_long tcps_rcvwinupd; /* rcvd window update packets */ 370 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 371 u_long tcps_predack; /* times hdr predict ok for acks */ 372 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 373 u_long tcps_pcbcachemiss; 374 u_long tcps_cachedrtt; /* times cached RTT in route updated */ 375 u_long tcps_cachedrttvar; /* times cached rttvar updated */ 376 u_long tcps_cachedssthresh; /* times cached ssthresh updated */ 377 u_long tcps_usedrtt; /* times RTT initialized from route */ 378 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ 379 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ 380 u_long tcps_persistdrop; /* timeout in persist state */ 381 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 382 u_long tcps_mturesent; /* resends due to MTU discovery */ 383 u_long tcps_listendrop; /* listen queue overflows */ 384 385 u_long tcps_sc_added; /* entry added to syncache */ 386 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */ 387 u_long tcps_sc_dupsyn; /* duplicate SYN packet */ 388 u_long tcps_sc_dropped; /* could not reply to packet */ 389 u_long tcps_sc_completed; /* successful extraction of entry */ 390 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ 391 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */ 392 u_long tcps_sc_reset; /* RST removed entry from syncache */ 393 u_long tcps_sc_stale; /* timed out or listen socket gone */ 394 u_long tcps_sc_aborted; /* syncache entry aborted */ 395 u_long tcps_sc_badack; /* removed due to bad ACK */ 396 u_long tcps_sc_unreach; /* ICMP unreachable received */ 397 u_long tcps_sc_zonefail; /* zalloc() failed */ 398 u_long tcps_sc_sendcookie; /* SYN cookie sent */ 399 u_long tcps_sc_recvcookie; /* SYN cookie received */ 400 }; 401 402 /* 403 * Structure to hold TCP options that are only used during segment 404 * processing (in tcp_input), but not held in the tcpcb. 405 * It's basically used to reduce the number of parameters 406 * to tcp_dooptions. 407 */ 408 struct tcpopt { 409 u_long to_flags; /* which options are present */ 410 #define TOF_TS 0x0001 /* timestamp */ 411 #define TOF_CC 0x0002 /* CC and CCnew are exclusive */ 412 #define TOF_CCNEW 0x0004 413 #define TOF_CCECHO 0x0008 414 #define TOF_MSS 0x0010 415 #define TOF_SCALE 0x0020 416 #define TOF_SACK_PERMITTED 0x0040 417 #define TOF_SACK 0x0080 418 u_int32_t to_tsval; 419 u_int32_t to_tsecr; 420 tcp_cc to_cc; /* holds CC or CCnew */ 421 tcp_cc to_ccecho; 422 u_int16_t to_mss; 423 u_int8_t to_requested_s_scale; 424 u_int8_t to_nsackblocks; 425 struct raw_sackblock *to_sackblocks; 426 }; 427 428 struct syncache { 429 inp_gen_t sc_inp_gencnt; /* pointer check */ 430 struct tcpcb *sc_tp; /* tcb for listening socket */ 431 struct mbuf *sc_ipopts; /* source route */ 432 struct in_conninfo sc_inc; /* addresses */ 433 #define sc_route sc_inc.inc_route 434 #define sc_route6 sc_inc.inc6_route 435 u_int32_t sc_tsrecent; 436 tcp_cc sc_cc_send; /* holds CC or CCnew */ 437 tcp_cc sc_cc_recv; 438 tcp_seq sc_irs; /* seq from peer */ 439 tcp_seq sc_iss; /* our ISS */ 440 u_long sc_rxttime; /* retransmit time */ 441 u_int16_t sc_rxtslot; /* retransmit counter */ 442 u_int16_t sc_peer_mss; /* peer's MSS */ 443 u_int16_t sc_wnd; /* advertised window */ 444 u_int8_t sc_requested_s_scale:4, 445 sc_request_r_scale:4; 446 u_int8_t sc_flags; 447 #define SCF_NOOPT 0x01 /* no TCP options */ 448 #define SCF_WINSCALE 0x02 /* negotiated window scaling */ 449 #define SCF_TIMESTAMP 0x04 /* negotiated timestamps */ 450 #define SCF_CC 0x08 /* negotiated CC */ 451 #define SCF_UNREACH 0x10 /* icmp unreachable received */ 452 #define SCF_SACK_PERMITTED 0x20 /* saw SACK permitted option */ 453 TAILQ_ENTRY(syncache) sc_hash; 454 TAILQ_ENTRY(syncache) sc_timerq; 455 }; 456 457 struct syncache_head { 458 TAILQ_HEAD(, syncache) sch_bucket; 459 u_int sch_length; 460 }; 461 462 /* 463 * The TAO cache entry which is stored in the protocol family specific 464 * portion of the route metrics. 465 */ 466 struct rmxp_tao { 467 tcp_cc tao_cc; /* latest CC in valid SYN */ 468 tcp_cc tao_ccsent; /* latest CC sent to peer */ 469 u_short tao_mssopt; /* peer's cached MSS */ 470 471 #ifdef notyet 472 u_short tao_flags; /* cache status flags */ 473 #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */ 474 #define TAOF_OK 0x0002 /* peer does understand rfc1644 */ 475 #define TAOF_UNDEF 0 /* we don't know yet */ 476 #endif 477 }; 478 479 #define rmx_taop(rt) ((struct rmxp_tao *)(rt).rmx_filler) 480 481 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 482 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 483 484 /* 485 * The smoothed round-trip time and estimated variance 486 * are stored as fixed point numbers scaled by the values below. 487 * For convenience, these scales are also used in smoothing the average 488 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 489 * With these scales, srtt has 3 bits to the right of the binary point, 490 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 491 * binary point, and is smoothed with an ALPHA of 0.75. 492 */ 493 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 494 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 495 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 496 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 497 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 498 499 /* 500 * The initial retransmission should happen at rtt + 4 * rttvar. 501 * Because of the way we do the smoothing, srtt and rttvar 502 * will each average +1/2 tick of bias. When we compute 503 * the retransmit timer, we want 1/2 tick of rounding and 504 * 1 extra tick because of +-1/2 tick uncertainty in the 505 * firing of the timer. The bias will give us exactly the 506 * 1.5 tick we need. But, because the bias is 507 * statistical, we have to test that we don't drop below 508 * the minimum feasible timer (which is 2 ticks). 509 * This version of the macro adapted from a paper by Lawrence 510 * Brakmo and Larry Peterson which outlines a problem caused 511 * by insufficient precision in the original implementation, 512 * which results in inappropriately large RTO values for very 513 * fast networks. 514 */ 515 #define TCP_REXMTVAL(tp) \ 516 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 517 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 518 519 /* 520 * TCB structure exported to user-land via sysctl(3). 521 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 522 * included. Not all of our clients do. 523 */ 524 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 525 struct xtcpcb { 526 size_t xt_len; 527 struct inpcb xt_inp; 528 struct tcpcb xt_tp; 529 struct xsocket xt_socket; 530 u_quad_t xt_alignment_hack; 531 }; 532 #endif 533 534 /* 535 * Names for TCP sysctl objects 536 */ 537 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 538 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ 539 #define TCPCTL_MSSDFLT 3 /* MSS default */ 540 #define TCPCTL_STATS 4 /* statistics (read-only) */ 541 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 542 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 543 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 544 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 545 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 546 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 547 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 548 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 549 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 550 #define TCPCTL_MAXID 14 551 552 #define TCPCTL_NAMES { \ 553 { 0, 0 }, \ 554 { "rfc1323", CTLTYPE_INT }, \ 555 { "rfc1644", CTLTYPE_INT }, \ 556 { "mssdflt", CTLTYPE_INT }, \ 557 { "stats", CTLTYPE_STRUCT }, \ 558 { "rttdflt", CTLTYPE_INT }, \ 559 { "keepidle", CTLTYPE_INT }, \ 560 { "keepintvl", CTLTYPE_INT }, \ 561 { "sendspace", CTLTYPE_INT }, \ 562 { "recvspace", CTLTYPE_INT }, \ 563 { "keepinit", CTLTYPE_INT }, \ 564 { "pcblist", CTLTYPE_STRUCT }, \ 565 { "delacktime", CTLTYPE_INT }, \ 566 { "v6mssdflt", CTLTYPE_INT }, \ 567 } 568 569 #ifdef _KERNEL 570 #ifdef SYSCTL_DECL 571 SYSCTL_DECL(_net_inet_tcp); 572 #endif 573 574 #define TCP_DO_SACK(tp) ((tp)->t_flags & TF_SACK_PERMITTED) 575 576 TAILQ_HEAD(tcpcbackqhead,tcpcb); 577 578 extern struct inpcbinfo tcbinfo[]; 579 extern struct tcpcbackqhead tcpcbackq[]; 580 581 extern int tcp_mssdflt; /* XXX */ 582 extern int tcp_delack_enabled; 583 extern int path_mtu_discovery; 584 585 int tcp_addrcpu(in_addr_t faddr, in_port_t fport, 586 in_addr_t laddr, in_port_t lport); 587 struct lwkt_port * 588 tcp_addrport(in_addr_t faddr, in_port_t fport, 589 in_addr_t laddr, in_port_t lport); 590 void tcp_canceltimers (struct tcpcb *); 591 struct tcpcb * 592 tcp_close (struct tcpcb *); 593 void tcpmsg_service_loop (void *); 594 void tcp_ctlinput (int, struct sockaddr *, void *); 595 int tcp_ctloutput (struct socket *, struct sockopt *); 596 struct lwkt_port * 597 tcp_cport(int cpu); 598 struct tcpcb * 599 tcp_drop (struct tcpcb *, int); 600 void tcp_drain (void); 601 void tcp_fasttimo (void); 602 struct rmxp_tao * 603 tcp_gettaocache (struct in_conninfo *); 604 void tcp_init (void); 605 void tcp_thread_init (void); 606 void tcp_input (struct mbuf *, ...); 607 void tcp_mss (struct tcpcb *, int); 608 int tcp_mssopt (struct tcpcb *); 609 void tcp_drop_syn_sent (struct inpcb *, int); 610 void tcp_mtudisc (struct inpcb *, int); 611 struct tcpcb * 612 tcp_newtcpcb (struct inpcb *); 613 int tcp_output (struct tcpcb *); 614 void tcp_quench (struct inpcb *, int); 615 void tcp_respond (struct tcpcb *, void *, 616 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 617 struct rtentry * 618 tcp_rtlookup (struct in_conninfo *); 619 int tcp_sack_bytes_below(struct scoreboard *scb, tcp_seq seq); 620 void tcp_sack_cleanup(struct scoreboard *scb); 621 int tcp_sack_ndsack_blocks(struct raw_sackblock *blocks, 622 const int numblocks, tcp_seq snd_una); 623 void tcp_sack_fill_report(struct tcpcb *tp, u_char *opt, u_int *plen); 624 boolean_t 625 tcp_sack_has_sacked(struct scoreboard *scb, u_int amount); 626 void tcp_sack_init(void); 627 void tcp_sack_tcpcb_init(struct tcpcb *tp); 628 uint32_t tcp_sack_compute_pipe(struct tcpcb *tp); 629 boolean_t 630 tcp_sack_nextseg(struct tcpcb *tp, tcp_seq *nextrexmt, uint32_t *len, 631 boolean_t *losdup); 632 #ifdef later 633 void tcp_sack_revert_scoreboard(struct scoreboard *scb, tcp_seq snd_una, 634 u_int maxseg); 635 void tcp_sack_save_scoreboard(struct scoreboard *scb); 636 #endif 637 void tcp_sack_skip_sacked(struct scoreboard *scb, tcp_seq *prexmt); 638 void tcp_sack_update_scoreboard(struct tcpcb *tp, struct tcpopt *to); 639 void tcp_save_congestion_state(struct tcpcb *tp); 640 void tcp_revert_congestion_state(struct tcpcb *tp); 641 void tcp_setpersist (struct tcpcb *); 642 void tcp_slowtimo (void); 643 struct tcptemp *tcp_maketemplate (struct tcpcb *); 644 void tcp_freetemplate (struct tcptemp *); 645 void tcp_fillheaders (struct tcpcb *, void *, void *); 646 struct lwkt_port * 647 tcp_soport(struct socket *, struct sockaddr *nam, int req); 648 struct tcpcb * 649 tcp_timers (struct tcpcb *, int); 650 void tcp_trace (short, short, struct tcpcb *, void *, struct tcphdr *, 651 int); 652 void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq); 653 void syncache_init(void); 654 void syncache_unreach(struct in_conninfo *, struct tcphdr *); 655 int syncache_expand(struct in_conninfo *, struct tcphdr *, 656 struct socket **, struct mbuf *); 657 int syncache_add(struct in_conninfo *, struct tcpopt *, 658 struct tcphdr *, struct socket **, struct mbuf *); 659 void syncache_chkrst(struct in_conninfo *, struct tcphdr *); 660 void syncache_badack(struct in_conninfo *); 661 662 extern struct pr_usrreqs tcp_usrreqs; 663 extern u_long tcp_sendspace; 664 extern u_long tcp_recvspace; 665 tcp_seq tcp_new_isn (struct tcpcb *); 666 667 #endif /* _KERNEL */ 668 669 #endif /* _NETINET_TCP_VAR_H_ */ 670