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 */ 69 70 #ifndef _NETINET_TCP_VAR_H_ 71 #define _NETINET_TCP_VAR_H_ 72 73 #ifndef _NETINET_IN_PCB_H_ 74 #include <netinet/in_pcb.h> /* needed for in_conninfo, inp_gen_t */ 75 #endif 76 #ifndef _NETINET_TCP_H_ 77 #include <netinet/tcp.h> 78 #endif 79 80 /* 81 * Kernel variables for tcp. 82 */ 83 extern int tcp_do_rfc1323; 84 extern int tcp_low_rtobase; 85 extern int tcp_ncr_rxtthresh_max; 86 extern int tcp_do_sack; 87 extern int tcp_do_smartsack; 88 extern int tcp_do_rescuesack; 89 extern int tcp_aggressive_rescuesack; 90 extern int tcp_do_rfc6675; 91 extern int tcp_rfc6675_rxt; 92 extern int tcp_aggregate_acks; 93 extern int tcp_eifel_rtoinc; 94 extern int tcp_prio_synack; 95 96 /* TCP segment queue entry */ 97 struct tseg_qent { 98 TAILQ_ENTRY(tseg_qent) tqe_q; 99 int tqe_len; /* TCP segment data length */ 100 struct tcphdr *tqe_th; /* a pointer to tcp header */ 101 struct mbuf *tqe_m; /* mbuf contains packet */ 102 }; 103 TAILQ_HEAD(tsegqe_head, tseg_qent); 104 extern int tcp_reass_maxseg; 105 extern int tcp_reass_qsize; 106 #ifdef MALLOC_DECLARE 107 MALLOC_DECLARE(M_TSEGQ); 108 #endif 109 110 struct tcptemp { 111 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 112 struct tcphdr tt_t; 113 }; 114 115 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 116 117 struct raw_sackblock { /* covers [start, end) */ 118 tcp_seq rblk_start; 119 tcp_seq rblk_end; 120 }; 121 122 /* maximum number of SACK blocks that will fit in the TCP option space */ 123 #define MAX_SACK_REPORT_BLOCKS 4 124 125 TAILQ_HEAD(sackblock_list, sackblock); 126 127 struct scoreboard { 128 int nblocks; 129 struct sackblock_list sackblocks; 130 tcp_seq lostseq; /* passed SACK lost test */ 131 struct sackblock *lastfound; /* search hint */ 132 struct sackblock *freecache; /* one slot free block cache */ 133 }; 134 135 struct netmsg_tcp_timer; 136 struct netmsg_base; 137 138 /* 139 * Tcp control block, one per tcp; fields: 140 * Organized for 16 byte cacheline efficiency. 141 */ 142 struct tcpcb { 143 struct tsegqe_head t_segq; 144 int t_dupacks; /* consecutive dup acks recd */ 145 int t_rxtthresh; /* # dup acks to start fast rxt */ 146 int tt_cpu; /* sanity check the cpu */ 147 148 struct tcp_callout *tt_rexmt; /* retransmit timer */ 149 struct tcp_callout *tt_persist;/* retransmit persistence */ 150 struct tcp_callout *tt_keep; /* keepalive */ 151 struct tcp_callout *tt_2msl; /* 2*msl TIME_WAIT timer */ 152 struct tcp_callout *tt_delack; /* delayed ACK timer */ 153 struct netmsg_tcp_timer *tt_msg; /* timer message */ 154 155 struct netmsg_base *tt_sndmore;/* send more segments (fairsend) */ 156 157 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 158 int t_state; /* state of this connection */ 159 u_int t_flags; 160 #define TF_ACKNOW 0x00000001 /* ack peer immediately */ 161 #define TF_DELACK 0x00000002 /* ack, but try to delay it */ 162 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */ 163 #define TF_NOOPT 0x00000008 /* don't use tcp options */ 164 #define TF_SENTFIN 0x00000010 /* have sent FIN */ 165 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */ 166 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */ 167 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */ 168 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */ 169 #define TF_SACK_PERMITTED 0x00000200 /* other side said I could SACK */ 170 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */ 171 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */ 172 #define TF_NOPUSH 0x00001000 /* don't push */ 173 #define TF_LISTEN 0x00002000 /* listen(2) has been called */ 174 #define TF_SIGNATURE 0x00004000 /* require MD5 digests (RFC2385) */ 175 #define TF_NCR 0x00008000 /* Non-Congestion Robustness RFC4653 */ 176 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */ 177 #define TF_SAWFIN 0x00020000 /* FIN has been seen */ 178 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */ 179 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */ 180 #define TF_FASTRECOVERY 0x00100000 /* in Fast Recovery */ 181 #define TF_QUEDFIN 0x00200000 /* FIN has been received */ 182 #define TF_XMITNOW 0x00400000 /* Temporarily override Nagle */ 183 #define TF_UNUSED008 0x00800000 184 #define TF_UNUSED009 0x01000000 185 #define TF_FORCE 0x02000000 /* Set if forcing out a byte */ 186 #define TF_ONOUTPUTQ 0x04000000 /* on t_outputq list */ 187 #define TF_FAIRSEND 0x08000000 /* fairsend is requested */ 188 #define TF_UNUSED003 0x10000000 189 #define TF_UNUSED004 0x20000000 190 #define TF_KEEPALIVE 0x40000000 /* temporary keepalive */ 191 #define TF_RXRESIZED 0x80000000 /* rcvbuf was resized */ 192 tcp_seq snd_up; /* send urgent pointer */ 193 u_long snd_last; /* time last data were sent */ 194 195 tcp_seq snd_una; /* send unacknowledged */ 196 tcp_seq snd_recover; /* for use with Fast Recovery */ 197 tcp_seq snd_max; /* highest sequence number sent; 198 * used to recognize retransmits */ 199 tcp_seq snd_nxt; /* send next */ 200 201 tcp_seq snd_wl1; /* window update seg seq number */ 202 tcp_seq snd_wl2; /* window update seg ack number */ 203 tcp_seq iss; /* initial send sequence number */ 204 tcp_seq irs; /* initial receive sequence number */ 205 206 tcp_seq rcv_nxt; /* receive next */ 207 tcp_seq rcv_adv; /* advertised window */ 208 u_long rcv_wnd; /* receive window */ 209 tcp_seq rcv_up; /* receive urgent pointer */ 210 211 u_long snd_wnd; /* send window */ 212 u_long snd_cwnd; /* congestion-controlled window */ 213 u_long snd_wacked; /* bytes acked in one send window */ 214 u_long snd_ssthresh; /* snd_cwnd size threshold for 215 * for slow start exponential to 216 * linear switch */ 217 218 int t_rxtcur; /* current retransmit value (ticks) */ 219 u_int t_maxseg; /* maximum segment size */ 220 int t_srtt; /* smoothed round-trip time */ 221 int t_rttvar; /* variance in round-trip time */ 222 223 u_int t_maxopd; /* mss plus options */ 224 225 u_long t_rcvtime; /* reception inactivity time */ 226 u_long t_starttime; /* time connection was established */ 227 int t_rtttime; /* round trip time */ 228 tcp_seq t_rtseq; /* sequence number being timed */ 229 230 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 231 u_int t_rttmin; /* minimum rtt allowed */ 232 u_int t_rttbest; /* best rtt we've seen */ 233 u_long t_rttupdated; /* number of times rtt sampled */ 234 u_long max_sndwnd; /* largest window peer has offered */ 235 236 int t_softerror; /* possible error not yet reported */ 237 /* out-of-band data */ 238 char t_oobflags; /* have some */ 239 char t_iobc; /* input character */ 240 #define TCPOOB_HAVEDATA 0x01 241 #define TCPOOB_HADDATA 0x02 242 243 /* RFC 1323 variables */ 244 u_char snd_scale; /* window scaling for send window */ 245 u_char rcv_scale; /* window scaling for recv window */ 246 u_char request_r_scale; /* pending window scaling */ 247 u_long ts_recent; /* timestamp echo data */ 248 249 u_long ts_recent_age; /* when last updated */ 250 tcp_seq last_ack_sent; 251 252 /* experimental */ 253 u_int rxt_flags; 254 #define TRXT_F_REBASERTO 0x0001 /* Recalculate RTO based on new RTT */ 255 #define TRXT_F_WASFRECOVERY 0x0002 /* was in Fast Recovery */ 256 #define TRXT_F_FIRSTACCACK 0x0004 /* Look for 1st acceptable ACK. */ 257 #define TRXT_F_FASTREXMT 0x0008 /* Did Fast Retransmit. */ 258 #define TRXT_F_EARLYREXMT 0x0010 /* Did Early (Fast) Retransmit. */ 259 int t_srtt_prev; /* adjusted SRTT prior to retransmit */ 260 int t_rttvar_prev; /* RTTVAR prior to retransmit */ 261 int t_rxtcur_prev; /* rexmt timeout prior to retransmit */ 262 tcp_seq snd_max_prev; /* SND_MAX prior to retransmit */ 263 u_long snd_cwnd_prev; /* cwnd prior to retransmit */ 264 u_long snd_wacked_prev; /* prior bytes acked in send window */ 265 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ 266 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 267 u_long t_badrxtwin; /* window for retransmit recovery */ 268 u_long t_rexmtTS; /* timestamp of last retransmit */ 269 u_char snd_limited; /* segments limited transmitted */ 270 271 u_int sack_flags; 272 #define TSACK_F_SACKRESCUED 0x0001 /* sent rescue SACK recovery data */ 273 #define TSACK_F_DUPSEG 0x0002 /* last seg a duplicate */ 274 #define TSACK_F_ENCLOSESEG 0x0004 /* enclosing SACK block */ 275 #define TSACK_F_SACKLEFT 0x0008 /* send SACK blocks from left side */ 276 tcp_seq rexmt_high; /* highest seq # retransmitted + 1 */ 277 tcp_seq rexmt_rescue; /* rescue SACKED sequence number */ 278 tcp_seq snd_max_rexmt; /* snd_max when rexmting snd_una */ 279 struct scoreboard scb; /* sack scoreboard */ 280 struct raw_sackblock reportblk; /* incoming segment or D-SACK block */ 281 struct raw_sackblock encloseblk; 282 int nsackhistory; 283 struct raw_sackblock sackhistory[MAX_SACK_REPORT_BLOCKS]; /* reported */ 284 285 TAILQ_ENTRY(tcpcb) t_outputq; /* tcp_output needed list */ 286 287 /* bandwith limitation */ 288 u_long snd_bandwidth; /* calculated bandwidth or 0 */ 289 u_long snd_bwnd; /* bandwidth-controlled window */ 290 int t_bw_rtttime; /* used for bandwidth calculation */ 291 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */ 292 293 /* anti DoS counters */ 294 u_long rcv_second; /* start of interval second */ 295 u_long rcv_pps; /* received packets per second */ 296 u_long rcv_byps; /* received bytes per second */ 297 298 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 299 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 300 301 int t_keepinit; /* time to establish connection */ 302 303 int t_keepidle; /* time before keepalive probes begin */ 304 int t_keepintvl; /* time between keepalive probes */ 305 int t_keepcnt; /* maximum number of keepalive probes */ 306 int t_maxidle; /* time to drop after starting probes */ 307 308 int t_rxtsyn; /* time spent in SYN or SYN|ACK rexmt */ 309 }; 310 311 #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) 312 #define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY 313 #define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY 314 315 #ifdef TCP_SIGNATURE 316 /* 317 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 318 * for SADB verification and lookup. 319 */ 320 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 321 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 322 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 323 /* 324 * Only a single SA per host may be specified at this time. An SPI is 325 * needed in order for the KEY_ALLOCSA() lookup to work. 326 */ 327 #define TCP_SIG_SPI 0x1000 328 #endif /* TCP_SIGNATURE */ 329 330 /* 331 * TCP statistics. 332 * 333 * NOTE: Make sure this struct's size is multiple cache line size. 334 */ 335 struct tcp_stats { 336 u_long tcps_connattempt; /* connections initiated */ 337 u_long tcps_accepts; /* connections accepted */ 338 u_long tcps_connects; /* connections established */ 339 u_long tcps_drops; /* connections dropped */ 340 u_long tcps_conndrops; /* embryonic connections dropped */ 341 u_long tcps_minmssdrops; /* average minmss too low drops */ 342 u_long tcps_closed; /* conn. closed (includes drops) */ 343 u_long tcps_segstimed; /* segs where we tried to get rtt */ 344 u_long tcps_rttupdated; /* times we succeeded */ 345 u_long tcps_delack; /* delayed acks sent */ 346 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 347 u_long tcps_rexmttimeo; /* retransmit timeouts */ 348 u_long tcps_persisttimeo; /* persist timeouts */ 349 u_long tcps_keeptimeo; /* keepalive timeouts */ 350 u_long tcps_keepprobe; /* keepalive probes sent */ 351 u_long tcps_keepdrops; /* connections dropped in keepalive */ 352 353 u_long tcps_sndtotal; /* total packets sent */ 354 u_long tcps_sndpack; /* data packets sent */ 355 u_long tcps_sndbyte; /* data bytes sent */ 356 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 357 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 358 u_long tcps_sndsackrtopack; /* packets sent by SACK after RTO */ 359 u_long tcps_sndsackrtobyte; /* bytes sent by SACK after RTO */ 360 u_long tcps_sndfastrexmit; /* Fast Retransmissions */ 361 u_long tcps_sndearlyrexmit; /* early Fast Retransmissions */ 362 u_long tcps_sndlimited; /* Limited Transmit packets */ 363 u_long tcps_sndrtobad; /* spurious RTO retransmissions */ 364 u_long tcps_sndfastrexmitbad; /* spurious Fast Retransmissions */ 365 u_long tcps_sndearlyrexmitbad; /* spurious early Fast Retransmissions, 366 a subset of tcps_sndfastrexmitbad */ 367 u_long tcps_eifeldetected; /* Eifel-detected spurious rexmits */ 368 u_long tcps_rttcantdetect; /* Eifel but not 1/2 RTT-detectable */ 369 u_long tcps_rttdetected; /* RTT-detected spurious RTO rexmits */ 370 u_long tcps_sndacks; /* ack-only packets sent */ 371 u_long tcps_sndprobe; /* window probes sent */ 372 u_long tcps_sndurg; /* packets sent with URG only */ 373 u_long tcps_sndwinup; /* window update-only packets sent */ 374 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 375 u_long tcps_sndsackpack; /* packets sent by SACK recovery alg */ 376 u_long tcps_sndsackbyte; /* bytes sent by SACK recovery */ 377 u_long tcps_snduna3; /* re-retransmit snd_una on 3 new seg */ 378 u_long tcps_snduna1; /* re-retransmit snd_una on 1 new seg */ 379 u_long tcps_sndsackopt; /* SACK options sent */ 380 u_long tcps_snddsackopt; /* D-SACK options sent */ 381 u_long tcps_sndidle; /* sending idle detected */ 382 u_long tcps_sackrescue; /* SACK rescue data packets sent */ 383 u_long tcps_sackrescue_try; /* SACK rescues attempted */ 384 u_long tcps_eifelresponse; /* Eifel responses */ 385 386 u_long tcps_rcvtotal; /* total packets received */ 387 u_long tcps_rcvpack; /* packets received in sequence */ 388 u_long tcps_rcvbyte; /* bytes received in sequence */ 389 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 390 u_long tcps_rcvbadoff; /* packets received with bad offset */ 391 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 392 u_long tcps_rcvshort; /* packets received too short */ 393 u_long tcps_rcvduppack; /* duplicate-only packets received */ 394 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 395 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 396 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 397 u_long tcps_rcvoopack; /* out-of-order packets received */ 398 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 399 u_long tcps_rcvpackafterwin; /* packets with data after window */ 400 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 401 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 402 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 403 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 404 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 405 u_long tcps_rcvackpack; /* rcvd ack packets */ 406 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 407 u_long tcps_rcvwinupd; /* rcvd window update packets */ 408 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 409 u_long tcps_pawsaccept; /* segments accepted, PAWS tolerance */ 410 u_long tcps_predack; /* times hdr predict ok for acks */ 411 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 412 u_long tcps_pcbcachemiss; 413 u_long tcps_cachedrtt; /* times cached RTT in route updated */ 414 u_long tcps_cachedrttvar; /* times cached rttvar updated */ 415 u_long tcps_cachedssthresh; /* times cached ssthresh updated */ 416 u_long tcps_usedrtt; /* times RTT initialized from route */ 417 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ 418 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ 419 u_long tcps_persistdrop; /* timeout in persist state */ 420 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 421 u_long tcps_mturesent; /* resends due to MTU discovery */ 422 u_long tcps_listendrop; /* listen queue overflows */ 423 u_long tcps_rcvbadsackopt; /* rcvd illegal SACK options */ 424 u_long tcps_sackrenege; /* times other side reneged */ 425 426 u_long tcps_sacksbupdate; /* times SACK scoreboard updated */ 427 u_long tcps_sacksboverflow; /* times SACK scoreboard overflowed */ 428 u_long tcps_sacksbreused; /* times SACK sb-block reused */ 429 u_long tcps_sacksbfailed; /* times SACK sb update failed */ 430 u_long tcps_sacksbfast; /* times SACK sb-block uses cache */ 431 432 u_long tcps_sc_added; /* entry added to syncache */ 433 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */ 434 u_long tcps_sc_dupsyn; /* duplicate SYN packet */ 435 u_long tcps_sc_dropped; /* could not reply to packet */ 436 u_long tcps_sc_completed; /* successful extraction of entry */ 437 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ 438 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */ 439 u_long tcps_sc_reset; /* RST removed entry from syncache */ 440 u_long tcps_sc_stale; /* timed out or listen socket gone */ 441 u_long tcps_sc_aborted; /* syncache entry aborted */ 442 u_long tcps_sc_badack; /* removed due to bad ACK */ 443 u_long tcps_sc_unreach; /* ICMP unreachable received */ 444 u_long tcps_sc_zonefail; /* zalloc() failed */ 445 u_long tcps_sc_sendcookie; /* SYN cookie sent */ 446 u_long tcps_sc_recvcookie; /* SYN cookie received */ 447 448 u_long tcps_pad[6]; /* pad to cache line size (64B) */ 449 }; 450 451 #ifdef _KERNEL 452 453 #define tcpstat tcpstats_percpu[mycpuid] 454 455 struct sockopt; 456 457 extern struct tcp_stats tcpstats_percpu[MAXCPU]; 458 459 static const int tcprexmtthresh = 3; 460 #endif 461 462 /* 463 * Structure to hold TCP options that are only used during segment 464 * processing (in tcp_input), but not held in the tcpcb. 465 * It's basically used to reduce the number of parameters 466 * to tcp_dooptions. 467 */ 468 struct tcpopt { 469 u_int to_flags; /* which options are present */ 470 #define TOF_TS 0x0001 /* timestamp */ 471 #define TOF_MSS 0x0010 472 #define TOF_SCALE 0x0020 473 #define TOF_SACK_PERMITTED 0x0040 474 #define TOF_SACK 0x0080 475 #define TOF_SIGNATURE 0x0100 /* signature option present */ 476 #define TOF_SIGLEN 0x0200 /* signature length valid (RFC2385) */ 477 #define TOF_DSACK 0x0400 /* D-SACK */ 478 #define TOF_SACK_REDUNDANT 0x0800 /* all SACK blocks are known */ 479 u_int32_t to_tsval; 480 u_int32_t to_tsecr; 481 u_int16_t to_mss; 482 u_int8_t to_requested_s_scale; 483 u_int8_t to_nsackblocks; 484 struct raw_sackblock *to_sackblocks; 485 }; 486 487 struct syncache { 488 inp_gen_t sc_inp_gencnt; /* pointer check */ 489 struct tcpcb *sc_tp; /* tcb for listening socket */ 490 struct mbuf *sc_ipopts; /* source route */ 491 struct in_conninfo sc_inc; /* addresses */ 492 #define sc_route sc_inc.inc_route 493 #define sc_route6 sc_inc.inc6_route 494 u_int32_t sc_tsrecent; 495 tcp_seq sc_irs; /* seq from peer */ 496 tcp_seq sc_iss; /* our ISS */ 497 u_long sc_rxttime; /* retransmit time */ 498 u_int16_t sc_rxtslot; /* retransmit counter */ 499 u_int16_t sc_peer_mss; /* peer's MSS */ 500 u_int16_t sc_wnd; /* advertised window */ 501 u_int8_t sc_requested_s_scale:4, 502 sc_request_r_scale:4; 503 u_int8_t sc_flags; 504 #define SCF_NOOPT 0x01 /* no TCP options */ 505 #define SCF_WINSCALE 0x02 /* negotiated window scaling */ 506 #define SCF_TIMESTAMP 0x04 /* negotiated timestamps */ 507 #define SCF_UNUSED 0x08 /* unused */ 508 #define SCF_UNREACH 0x10 /* icmp unreachable received */ 509 #define SCF_SACK_PERMITTED 0x20 /* saw SACK permitted option */ 510 #define SCF_SIGNATURE 0x40 /* send MD5 digests */ 511 #define SCF_MARKER 0x80 /* not a real entry */ 512 int sc_rxtused; /* time spent in SYN|ACK rxt */ 513 u_long sc_sndwnd; /* send window */ 514 TAILQ_ENTRY(syncache) sc_hash; 515 TAILQ_ENTRY(syncache) sc_timerq; 516 }; 517 518 struct syncache_head { 519 TAILQ_HEAD(, syncache) sch_bucket; 520 u_int sch_length; 521 }; 522 523 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 524 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 525 526 /* 527 * The smoothed round-trip time and estimated variance 528 * are stored as fixed point numbers scaled by the values below. 529 * For convenience, these scales are also used in smoothing the average 530 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 531 * With these scales, srtt has 3 bits to the right of the binary point, 532 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 533 * binary point, and is smoothed with an ALPHA of 0.75. 534 */ 535 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 536 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 537 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 538 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 539 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 540 541 /* 542 * The initial retransmission should happen at rtt + 4 * rttvar. 543 * Because of the way we do the smoothing, srtt and rttvar 544 * will each average +1/2 tick of bias. When we compute 545 * the retransmit timer, we want 1/2 tick of rounding and 546 * 1 extra tick because of +-1/2 tick uncertainty in the 547 * firing of the timer. The bias will give us exactly the 548 * 1.5 tick we need. But, because the bias is 549 * statistical, we have to test that we don't drop below 550 * the minimum feasible timer (which is 2 ticks). 551 * This version of the macro adapted from a paper by Lawrence 552 * Brakmo and Larry Peterson which outlines a problem caused 553 * by insufficient precision in the original implementation, 554 * which results in inappropriately large RTO values for very 555 * fast networks. 556 */ 557 #define TCP_REXMTVAL(tp) \ 558 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 559 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 560 561 /* 562 * TCB structure exported to user-land via sysctl(3). 563 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 564 * included. Not all of our clients do. 565 */ 566 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 567 struct xtcpcb { 568 size_t xt_len; 569 struct inpcb xt_inp; 570 struct tcpcb xt_tp; 571 struct xsocket xt_socket; 572 u_quad_t xt_alignment_hack; 573 }; 574 #endif 575 576 /* 577 * Names for TCP sysctl objects 578 */ 579 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 580 /* 2 was TCPCTL_DO_RFC1644 */ 581 #define TCPCTL_MSSDFLT 3 /* MSS default */ 582 #define TCPCTL_STATS 4 /* statistics (read-only) */ 583 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 584 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 585 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 586 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 587 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 588 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 589 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 590 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 591 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 592 #define TCPCTL_MAXID 14 593 594 #define TCPCTL_NAMES { \ 595 { 0, 0 }, \ 596 { "rfc1323", CTLTYPE_INT }, \ 597 { "reserved", CTLTYPE_INT}, /* was rfc1644 */ \ 598 { "mssdflt", CTLTYPE_INT }, \ 599 { "stats", CTLTYPE_STRUCT }, \ 600 { "rttdflt", CTLTYPE_INT }, \ 601 { "keepidle", CTLTYPE_INT }, \ 602 { "keepintvl", CTLTYPE_INT }, \ 603 { "sendspace", CTLTYPE_INT }, \ 604 { "recvspace", CTLTYPE_INT }, \ 605 { "keepinit", CTLTYPE_INT }, \ 606 { "pcblist", CTLTYPE_STRUCT }, \ 607 { "delacktime", CTLTYPE_INT }, \ 608 { "v6mssdflt", CTLTYPE_INT }, \ 609 } 610 611 #ifdef _KERNEL 612 #ifdef SYSCTL_DECL 613 SYSCTL_DECL(_net_inet_tcp); 614 #endif 615 616 #define TCP_DO_SACK(tp) ((tp)->t_flags & TF_SACK_PERMITTED) 617 #define TCP_DO_NCR(tp) (((tp)->t_flags & TF_NCR) && TCP_DO_SACK((tp))) 618 #define TCP_SACK_BLKEND(len, thflags) \ 619 ((len) + (((thflags) & TH_FIN) != 0)) 620 621 TAILQ_HEAD(tcpcbackqhead,tcpcb); 622 623 extern struct inpcbinfo tcbinfo[]; 624 extern struct tcpcbackqhead tcpcbackq[]; 625 626 extern int tcp_mssdflt; /* XXX */ 627 extern int tcp_minmss; 628 extern int tcp_delack_enabled; 629 extern int path_mtu_discovery; 630 631 struct ip; 632 union netmsg; 633 634 int tcp_addrcpu(in_addr_t faddr, in_port_t fport, 635 in_addr_t laddr, in_port_t lport); 636 struct lwkt_port * 637 tcp_addrport(in_addr_t faddr, in_port_t fport, 638 in_addr_t laddr, in_port_t lport); 639 struct lwkt_port *tcp_addrport0(void); 640 void tcp_canceltimers (struct tcpcb *); 641 struct tcpcb * 642 tcp_close (struct tcpcb *); 643 void tcp_ctlinput(union netmsg *); 644 void tcp_ctloutput(union netmsg *); 645 struct tcpcb * 646 tcp_drop (struct tcpcb *, int); 647 void tcp_drain (void); 648 void tcp_init (void); 649 void tcp_thread_init (void); 650 int tcp_input (struct mbuf **, int *, int); 651 void tcp_mss (struct tcpcb *, int); 652 int tcp_mssopt (struct tcpcb *); 653 void tcp_drop_syn_sent (struct inpcb *, int); 654 void tcp_mtudisc (struct inpcb *, int); 655 struct tcpcb * 656 tcp_newtcpcb (struct inpcb *); 657 int tcp_output(struct tcpcb *); 658 int tcp_output_fair(struct tcpcb *); 659 void tcp_output_init(struct tcpcb *); 660 void tcp_output_cancel(struct tcpcb *); 661 boolean_t 662 tcp_output_pending(struct tcpcb *); 663 void tcp_quench (struct inpcb *, int); 664 void tcp_respond (struct tcpcb *, void *, 665 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 666 struct rtentry * 667 tcp_rtlookup (struct in_conninfo *); 668 int tcp_sack_bytes_below(const struct scoreboard *scb, tcp_seq seq); 669 void tcp_sack_destroy(struct scoreboard *scb); 670 void tcp_sack_discard(struct tcpcb *tp); 671 void tcp_sack_report_cleanup(struct tcpcb *tp); 672 boolean_t 673 tcp_sack_report_needed(const struct tcpcb *tp); 674 int tcp_sack_ndsack_blocks(const struct raw_sackblock *blocks, 675 const int numblocks, tcp_seq snd_una); 676 void tcp_sack_fill_report(struct tcpcb *tp, u_char *opt, u_int *plen); 677 boolean_t 678 tcp_sack_has_sacked(const struct scoreboard *scb, u_int amount); 679 void tcp_sack_tcpcb_init(struct tcpcb *tp); 680 uint32_t tcp_sack_compute_pipe(const struct tcpcb *tp); 681 boolean_t 682 tcp_sack_nextseg(struct tcpcb *tp, tcp_seq *nextrexmt, uint32_t *len, 683 boolean_t *rescue); 684 boolean_t 685 tcp_sack_islost(const struct scoreboard *scb, tcp_seq seq); 686 void tcp_sack_update_lostseq(struct scoreboard *scb, tcp_seq snd_una, 687 u_int maxseg, int rxtthresh); 688 #ifdef later 689 void tcp_sack_revert_scoreboard(struct scoreboard *scb, tcp_seq snd_una, 690 u_int maxseg); 691 void tcp_sack_save_scoreboard(struct scoreboard *scb); 692 #endif 693 void tcp_sack_skip_sacked(struct scoreboard *scb, tcp_seq *prexmt); 694 uint32_t tcp_sack_first_unsacked_len(const struct tcpcb *tp); 695 void tcp_sack_update_scoreboard(struct tcpcb *tp, struct tcpopt *to); 696 void tcp_save_congestion_state(struct tcpcb *tp); 697 void tcp_revert_congestion_state(struct tcpcb *tp); 698 void tcp_setpersist (struct tcpcb *); 699 struct tcptemp *tcp_maketemplate (struct tcpcb *); 700 void tcp_freetemplate (struct tcptemp *); 701 void tcp_fillheaders (struct tcpcb *, void *, void *, boolean_t); 702 struct lwkt_port * 703 tcp_soport(struct socket *, struct sockaddr *, struct mbuf **); 704 struct lwkt_port * 705 tcp_ctlport(int, struct sockaddr *, void *); 706 struct lwkt_port * 707 tcp_initport(void); 708 struct tcpcb * 709 tcp_timers (struct tcpcb *, int); 710 void tcp_trace (short, short, struct tcpcb *, void *, struct tcphdr *, 711 int); 712 void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq); 713 u_long tcp_initial_window(struct tcpcb *tp); 714 void tcp_timer_keep_activity(struct tcpcb *tp, int thflags); 715 void syncache_init(void); 716 void syncache_unreach(struct in_conninfo *, struct tcphdr *); 717 int syncache_expand(struct in_conninfo *, struct tcphdr *, 718 struct socket **, struct mbuf *); 719 int syncache_add(struct in_conninfo *, struct tcpopt *, 720 struct tcphdr *, struct socket *, struct mbuf *); 721 void syncache_chkrst(struct in_conninfo *, struct tcphdr *); 722 void syncache_badack(struct in_conninfo *); 723 void syncache_destroy(struct tcpcb *tp, struct tcpcb *new_tp); 724 725 #ifdef TCP_SIGNATURE 726 int tcpsignature_apply(void *fstate, void *data, unsigned int len); 727 int tcpsignature_compute(struct mbuf *m, int len, int tcpoptlen, 728 u_char *buf, u_int direction); 729 #endif /* TCP_SIGNATURE */ 730 731 extern struct pr_usrreqs tcp_usrreqs; 732 extern u_long tcp_sendspace; 733 extern u_long tcp_recvspace; 734 tcp_seq tcp_new_isn (struct tcpcb *); 735 736 #endif /* _KERNEL */ 737 738 #endif /* _NETINET_TCP_VAR_H_ */ 739