tcp_var.h revision 1.119
1 /* $NetBSD: tcp_var.h,v 1.119 2005/02/28 16:20:59 jonathan Exp $ */ 2 3 /* 4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project. 5 * All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the project nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 */ 31 32 /* 33 * @(#)COPYRIGHT 1.1 (NRL) 17 January 1995 34 * 35 * NRL grants permission for redistribution and use in source and binary 36 * forms, with or without modification, of the software and documentation 37 * created at NRL provided that the following conditions are met: 38 * 39 * 1. Redistributions of source code must retain the above copyright 40 * notice, this list of conditions and the following disclaimer. 41 * 2. Redistributions in binary form must reproduce the above copyright 42 * notice, this list of conditions and the following disclaimer in the 43 * documentation and/or other materials provided with the distribution. 44 * 3. All advertising materials mentioning features or use of this software 45 * must display the following acknowledgements: 46 * This product includes software developed by the University of 47 * California, Berkeley and its contributors. 48 * This product includes software developed at the Information 49 * Technology Division, US Naval Research Laboratory. 50 * 4. Neither the name of the NRL 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 * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS 55 * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 56 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A 57 * PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL NRL OR 58 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, 59 * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, 60 * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR 61 * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF 62 * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING 63 * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS 64 * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 65 * 66 * The views and conclusions contained in the software and documentation 67 * are those of the authors and should not be interpreted as representing 68 * official policies, either expressed or implied, of the US Naval 69 * Research Laboratory (NRL). 70 */ 71 72 /*- 73 * Copyright (c) 1997, 1998, 1999, 2001 The NetBSD Foundation, Inc. 74 * All rights reserved. 75 * 76 * This code is derived from software contributed to The NetBSD Foundation 77 * by Jason R. Thorpe of the Numerical Aerospace Simulation Facility, 78 * NASA Ames Research Center. 79 * 80 * Redistribution and use in source and binary forms, with or without 81 * modification, are permitted provided that the following conditions 82 * are met: 83 * 1. Redistributions of source code must retain the above copyright 84 * notice, this list of conditions and the following disclaimer. 85 * 2. Redistributions in binary form must reproduce the above copyright 86 * notice, this list of conditions and the following disclaimer in the 87 * documentation and/or other materials provided with the distribution. 88 * 3. All advertising materials mentioning features or use of this software 89 * must display the following acknowledgement: 90 * This product includes software developed by the NetBSD 91 * Foundation, Inc. and its contributors. 92 * 4. Neither the name of The NetBSD Foundation nor the names of its 93 * contributors may be used to endorse or promote products derived 94 * from this software without specific prior written permission. 95 * 96 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 97 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 98 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 99 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 100 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 101 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 102 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 103 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 104 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 105 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 106 * POSSIBILITY OF SUCH DAMAGE. 107 */ 108 109 /* 110 * Copyright (c) 1982, 1986, 1993, 1994, 1995 111 * The Regents of the University of California. All rights reserved. 112 * 113 * Redistribution and use in source and binary forms, with or without 114 * modification, are permitted provided that the following conditions 115 * are met: 116 * 1. Redistributions of source code must retain the above copyright 117 * notice, this list of conditions and the following disclaimer. 118 * 2. Redistributions in binary form must reproduce the above copyright 119 * notice, this list of conditions and the following disclaimer in the 120 * documentation and/or other materials provided with the distribution. 121 * 3. Neither the name of the University nor the names of its contributors 122 * may be used to endorse or promote products derived from this software 123 * without specific prior written permission. 124 * 125 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 126 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 127 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 128 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 129 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 130 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 131 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 132 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 133 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 134 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 135 * SUCH DAMAGE. 136 * 137 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 138 */ 139 140 #ifndef _NETINET_TCP_VAR_H_ 141 #define _NETINET_TCP_VAR_H_ 142 143 #if defined(_KERNEL_OPT) 144 #include "opt_inet.h" 145 #include "opt_mbuftrace.h" 146 #endif 147 148 /* 149 * Kernel variables for tcp. 150 */ 151 152 #include <sys/callout.h> 153 154 #ifdef TCP_SIGNATURE 155 /* 156 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 157 * for SADB verification and lookup. 158 */ 159 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 160 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 161 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 162 /* 163 * Only a single SA per host may be specified at this time. An SPI is 164 * needed in order for the KEY_ALLOCSA() lookup to work. 165 */ 166 #define TCP_SIG_SPI 0x1000 167 #endif /* TCP_SIGNATURE */ 168 169 /* 170 * SACK option block. 171 */ 172 struct sackblk { 173 tcp_seq left; /* Left edge of sack block. */ 174 tcp_seq right; /* Right edge of sack block. */ 175 }; 176 177 TAILQ_HEAD(sackhead, sackhole); 178 struct sackhole { 179 tcp_seq start; 180 tcp_seq end; 181 tcp_seq rxmit; 182 183 TAILQ_ENTRY(sackhole) sackhole_q; 184 }; 185 186 /* 187 * Tcp control block, one per tcp; fields: 188 */ 189 struct tcpcb { 190 int t_family; /* address family on the wire */ 191 struct ipqehead segq; /* sequencing queue */ 192 struct callout t_timer[TCPT_NTIMERS];/* tcp timers */ 193 short t_state; /* state of this connection */ 194 short t_rxtshift; /* log(2) of rexmt exp. backoff */ 195 uint32_t t_rxtcur; /* current retransmit value */ 196 short t_dupacks; /* consecutive dup acks recd */ 197 short t_partialacks; /* partials acks during fast rexmit */ 198 u_short t_peermss; /* peer's maximum segment size */ 199 u_short t_ourmss; /* our's maximum segment size */ 200 u_short t_segsz; /* current segment size in use */ 201 char t_force; /* 1 if forcing out a byte */ 202 u_int t_flags; 203 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 204 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 205 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 206 #define TF_NOOPT 0x0008 /* don't use tcp options */ 207 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 208 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 209 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 210 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 211 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 212 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */ 213 #define TF_WILL_SACK 0x0800 /* try to use SACK */ 214 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */ 215 #define TF_DEAD 0x2000 /* dead and to-be-released */ 216 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 217 218 219 struct mbuf *t_template; /* skeletal packet for transmit */ 220 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 221 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */ 222 struct callout t_delack_ch; /* delayed ACK callout */ 223 /* 224 * The following fields are used as in the protocol specification. 225 * See RFC783, Dec. 1981, page 21. 226 */ 227 /* send sequence variables */ 228 tcp_seq snd_una; /* send unacknowledged */ 229 tcp_seq snd_nxt; /* send next */ 230 tcp_seq snd_up; /* send urgent pointer */ 231 tcp_seq snd_wl1; /* window update seg seq number */ 232 tcp_seq snd_wl2; /* window update seg ack number */ 233 tcp_seq iss; /* initial send sequence number */ 234 u_long snd_wnd; /* send window */ 235 tcp_seq snd_recover; /* for use in fast recovery */ 236 tcp_seq snd_high; /* NewReno false fast rexmit seq */ 237 /* receive sequence variables */ 238 u_long rcv_wnd; /* receive window */ 239 tcp_seq rcv_nxt; /* receive next */ 240 tcp_seq rcv_up; /* receive urgent pointer */ 241 tcp_seq irs; /* initial receive sequence number */ 242 /* 243 * Additional variables for this implementation. 244 */ 245 /* receive variables */ 246 tcp_seq rcv_adv; /* advertised window */ 247 /* retransmit variables */ 248 tcp_seq snd_max; /* highest sequence number sent; 249 * used to recognize retransmits 250 */ 251 /* congestion control (for slow start, source quench, retransmit after loss) */ 252 u_long snd_cwnd; /* congestion-controlled window */ 253 u_long snd_ssthresh; /* snd_cwnd size threshhold for 254 * for slow start exponential to 255 * linear switch 256 */ 257 /* 258 * transmit timing stuff. See below for scale of srtt and rttvar. 259 * "Variance" is actually smoothed difference. 260 */ 261 uint32_t t_rcvtime; /* time last segment received */ 262 uint32_t t_rtttime; /* time we started measuring rtt */ 263 tcp_seq t_rtseq; /* sequence number being timed */ 264 int32_t t_srtt; /* smoothed round-trip time */ 265 int32_t t_rttvar; /* variance in round-trip time */ 266 uint32_t t_rttmin; /* minimum rtt allowed */ 267 u_long max_sndwnd; /* largest window peer has offered */ 268 269 /* out-of-band data */ 270 char t_oobflags; /* have some */ 271 char t_iobc; /* input character */ 272 #define TCPOOB_HAVEDATA 0x01 273 #define TCPOOB_HADDATA 0x02 274 short t_softerror; /* possible error not yet reported */ 275 276 /* RFC 1323 variables */ 277 u_char snd_scale; /* window scaling for send window */ 278 u_char rcv_scale; /* window scaling for recv window */ 279 u_char request_r_scale; /* pending window scaling */ 280 u_char requested_s_scale; 281 u_int32_t ts_recent; /* timestamp echo data */ 282 u_int32_t ts_recent_age; /* when last updated */ 283 u_int32_t ts_timebase; /* our timebase */ 284 tcp_seq last_ack_sent; 285 286 /* SACK stuff */ 287 #define TCP_SACK_MAX 3 288 #define TCPSACK_NONE 0 289 #define TCPSACK_HAVED 1 290 u_char rcv_sack_flags; /* SACK flags. */ 291 u_int32_t rcv_sack_num; /* Num of RX SACK blocks. */ 292 struct sackblk rcv_sack_block[TCP_SACK_MAX]; /* RX SACK blocks. */ 293 struct sackblk rcv_dsack_block; /* RX D-SACK block. */ 294 struct ipqehead timeq; /* time sequenced queue. */ 295 struct sackhead snd_holes; /* TX SACK holes. */ 296 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/ 297 tcp_seq sack_newdata; /* New data xmitted in this recovery 298 episode starts at this seq number*/ 299 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by 300 peer. */ 301 302 /* path MTU discovery blackhole detection */ 303 int t_mtudisc; /* perform mtudisc for this tcb */ 304 305 /* pointer for syn cache entries*/ 306 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */ 307 308 /* prediction of next mbuf when using large window sizes */ 309 struct mbuf *t_lastm; /* last mbuf that data was sent from */ 310 int t_inoff; /* data offset in previous mbuf */ 311 int t_lastoff; /* last data address in mbuf chain */ 312 int t_lastlen; /* last length read from mbuf chain */ 313 }; 314 315 /* 316 * Macros to aid SACK/FACK TCP. 317 */ 318 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK) 319 #define TCP_FACK_FASTRECOV(tp) \ 320 (TCP_SACK_ENABLED(tp) && \ 321 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz))) 322 323 #ifdef _KERNEL 324 /* 325 * TCP reassembly queue locks. 326 */ 327 static __inline int tcp_reass_lock_try (struct tcpcb *) 328 __attribute__((__unused__)); 329 static __inline void tcp_reass_unlock (struct tcpcb *) 330 __attribute__((__unused__)); 331 332 static __inline int 333 tcp_reass_lock_try(tp) 334 struct tcpcb *tp; 335 { 336 int s; 337 338 /* 339 * Use splvm() -- we're blocking things that would cause 340 * mbuf allocation. 341 */ 342 s = splvm(); 343 if (tp->t_flags & TF_REASSEMBLING) { 344 splx(s); 345 return (0); 346 } 347 tp->t_flags |= TF_REASSEMBLING; 348 splx(s); 349 return (1); 350 } 351 352 static __inline void 353 tcp_reass_unlock(tp) 354 struct tcpcb *tp; 355 { 356 int s; 357 358 s = splvm(); 359 tp->t_flags &= ~TF_REASSEMBLING; 360 splx(s); 361 } 362 363 #ifdef DIAGNOSTIC 364 #define TCP_REASS_LOCK(tp) \ 365 do { \ 366 if (tcp_reass_lock_try(tp) == 0) { \ 367 printf("%s:%d: tcpcb %p reass already locked\n", \ 368 __FILE__, __LINE__, tp); \ 369 panic("tcp_reass_lock"); \ 370 } \ 371 } while (/*CONSTCOND*/ 0) 372 #define TCP_REASS_LOCK_CHECK(tp) \ 373 do { \ 374 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \ 375 printf("%s:%d: tcpcb %p reass lock not held\n", \ 376 __FILE__, __LINE__, tp); \ 377 panic("tcp reass lock check"); \ 378 } \ 379 } while (/*CONSTCOND*/ 0) 380 #else 381 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp)) 382 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */ 383 #endif 384 385 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp)) 386 #endif /* _KERNEL */ 387 388 /* 389 * Queue for delayed ACK processing. 390 */ 391 #ifdef _KERNEL 392 extern int tcp_delack_ticks; 393 void tcp_delack(void *); 394 395 #define TCP_RESTART_DELACK(tp) \ 396 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \ 397 tcp_delack, tp) 398 399 #define TCP_SET_DELACK(tp) \ 400 do { \ 401 if (((tp)->t_flags & TF_DELACK) == 0) { \ 402 (tp)->t_flags |= TF_DELACK; \ 403 TCP_RESTART_DELACK(tp); \ 404 } \ 405 } while (/*CONSTCOND*/0) 406 407 #define TCP_CLEAR_DELACK(tp) \ 408 do { \ 409 if ((tp)->t_flags & TF_DELACK) { \ 410 (tp)->t_flags &= ~TF_DELACK; \ 411 callout_stop(&(tp)->t_delack_ch); \ 412 } \ 413 } while (/*CONSTCOND*/0) 414 #endif /* _KERNEL */ 415 416 /* 417 * Compute the current timestamp for a connection. 418 */ 419 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase) 420 421 /* 422 * Handy way of passing around TCP option info. 423 */ 424 struct tcp_opt_info { 425 int ts_present; 426 u_int32_t ts_val; 427 u_int32_t ts_ecr; 428 u_int16_t maxseg; 429 }; 430 431 #define TOF_SIGNATURE 0x0040 /* signature option present */ 432 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */ 433 434 /* 435 * Data for the TCP compressed state engine. 436 */ 437 union syn_cache_sa { 438 struct sockaddr sa; 439 struct sockaddr_in sin; 440 #if 1 /*def INET6*/ 441 struct sockaddr_in6 sin6; 442 #endif 443 }; 444 445 struct syn_cache { 446 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */ 447 struct callout sc_timer; /* rexmt timer */ 448 union { /* cached route */ 449 struct route route4; 450 #ifdef INET6 451 struct route_in6 route6; 452 #endif 453 } sc_route_u; 454 #define sc_route4 sc_route_u.route4 455 #ifdef INET6 456 #define sc_route6 sc_route_u.route6 457 #endif 458 long sc_win; /* advertised window */ 459 int sc_bucketidx; /* our bucket index */ 460 u_int32_t sc_hash; 461 u_int32_t sc_timestamp; /* timestamp from SYN */ 462 u_int32_t sc_timebase; /* our local timebase */ 463 union syn_cache_sa sc_src; 464 union syn_cache_sa sc_dst; 465 tcp_seq sc_irs; 466 tcp_seq sc_iss; 467 u_int sc_rxtcur; /* current rxt timeout */ 468 u_int sc_rxttot; /* total time spend on queues */ 469 u_short sc_rxtshift; /* for computing backoff */ 470 u_short sc_flags; 471 472 #define SCF_UNREACH 0x0001 /* we've had an unreach error */ 473 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */ 474 #define SCF_DEAD 0x0004 /* this entry to be released */ 475 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */ 476 #define SCF_SIGNATURE 0x40 /* send MD5 digests */ 477 478 struct mbuf *sc_ipopts; /* IP options */ 479 u_int16_t sc_peermaxseg; 480 u_int16_t sc_ourmaxseg; 481 u_int8_t sc_request_r_scale : 4, 482 sc_requested_s_scale : 4; 483 484 struct tcpcb *sc_tp; /* tcb for listening socket */ 485 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */ 486 }; 487 488 struct syn_cache_head { 489 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */ 490 u_short sch_length; /* # entries in bucket */ 491 }; 492 493 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 494 #ifdef INET6 495 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb) 496 #endif 497 #ifndef INET6 498 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 499 #else 500 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \ 501 ? intotcpcb(sotoinpcb(so)) \ 502 : in6totcpcb(sotoin6pcb(so))) 503 #endif 504 505 /* 506 * The smoothed round-trip time and estimated variance 507 * are stored as fixed point numbers scaled by the values below. 508 * For convenience, these scales are also used in smoothing the average 509 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 510 * With these scales, srtt has 3 bits to the right of the binary point, 511 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 512 * binary point, and is smoothed with an ALPHA of 0.75. 513 */ 514 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 515 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 516 517 /* 518 * The initial retransmission should happen at rtt + 4 * rttvar. 519 * Because of the way we do the smoothing, srtt and rttvar 520 * will each average +1/2 tick of bias. When we compute 521 * the retransmit timer, we want 1/2 tick of rounding and 522 * 1 extra tick because of +-1/2 tick uncertainty in the 523 * firing of the timer. The bias will give us exactly the 524 * 1.5 tick we need. But, because the bias is 525 * statistical, we have to test that we don't drop below 526 * the minimum feasible timer (which is 2 ticks). 527 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 528 * is the same as the multiplier for rttvar. 529 */ 530 #define TCP_REXMTVAL(tp) \ 531 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 532 533 /* 534 * Compute the initial window for slow start. 535 */ 536 #define TCP_INITIAL_WINDOW(iw, segsz) \ 537 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 538 ((segsz) * (iw))) 539 540 /* 541 * TCP statistics. 542 * Many of these should be kept per connection, 543 * but that's inconvenient at the moment. 544 */ 545 struct tcpstat { 546 u_quad_t tcps_connattempt; /* connections initiated */ 547 u_quad_t tcps_accepts; /* connections accepted */ 548 u_quad_t tcps_connects; /* connections established */ 549 u_quad_t tcps_drops; /* connections dropped */ 550 u_quad_t tcps_conndrops; /* embryonic connections dropped */ 551 u_quad_t tcps_closed; /* conn. closed (includes drops) */ 552 u_quad_t tcps_segstimed; /* segs where we tried to get rtt */ 553 u_quad_t tcps_rttupdated; /* times we succeeded */ 554 u_quad_t tcps_delack; /* delayed acks sent */ 555 u_quad_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 556 u_quad_t tcps_rexmttimeo; /* retransmit timeouts */ 557 u_quad_t tcps_persisttimeo; /* persist timeouts */ 558 u_quad_t tcps_keeptimeo; /* keepalive timeouts */ 559 u_quad_t tcps_keepprobe; /* keepalive probes sent */ 560 u_quad_t tcps_keepdrops; /* connections dropped in keepalive */ 561 u_quad_t tcps_persistdrops; /* connections dropped in persist */ 562 u_quad_t tcps_connsdrained; /* connections drained due to memory 563 shortage */ 564 u_quad_t tcps_pmtublackhole; /* PMTUD blackhole detected */ 565 566 u_quad_t tcps_sndtotal; /* total packets sent */ 567 u_quad_t tcps_sndpack; /* data packets sent */ 568 u_quad_t tcps_sndbyte; /* data bytes sent */ 569 u_quad_t tcps_sndrexmitpack; /* data packets retransmitted */ 570 u_quad_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 571 u_quad_t tcps_sndacks; /* ack-only packets sent */ 572 u_quad_t tcps_sndprobe; /* window probes sent */ 573 u_quad_t tcps_sndurg; /* packets sent with URG only */ 574 u_quad_t tcps_sndwinup; /* window update-only packets sent */ 575 u_quad_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 576 577 u_quad_t tcps_rcvtotal; /* total packets received */ 578 u_quad_t tcps_rcvpack; /* packets received in sequence */ 579 u_quad_t tcps_rcvbyte; /* bytes received in sequence */ 580 u_quad_t tcps_rcvbadsum; /* packets received with ccksum errs */ 581 u_quad_t tcps_rcvbadoff; /* packets received with bad offset */ 582 u_quad_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 583 u_quad_t tcps_rcvshort; /* packets received too short */ 584 u_quad_t tcps_rcvduppack; /* duplicate-only packets received */ 585 u_quad_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 586 u_quad_t tcps_rcvpartduppack; /* packets with some duplicate data */ 587 u_quad_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 588 u_quad_t tcps_rcvoopack; /* out-of-order packets received */ 589 u_quad_t tcps_rcvoobyte; /* out-of-order bytes received */ 590 u_quad_t tcps_rcvpackafterwin; /* packets with data after window */ 591 u_quad_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 592 u_quad_t tcps_rcvafterclose; /* packets rcvd after "close" */ 593 u_quad_t tcps_rcvwinprobe; /* rcvd window probe packets */ 594 u_quad_t tcps_rcvdupack; /* rcvd duplicate acks */ 595 u_quad_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 596 u_quad_t tcps_rcvackpack; /* rcvd ack packets */ 597 u_quad_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 598 u_quad_t tcps_rcvwinupd; /* rcvd window update packets */ 599 u_quad_t tcps_pawsdrop; /* segments dropped due to PAWS */ 600 u_quad_t tcps_predack; /* times hdr predict ok for acks */ 601 u_quad_t tcps_preddat; /* times hdr predict ok for data pkts */ 602 603 u_quad_t tcps_pcbhashmiss; /* input packets missing pcb hash */ 604 u_quad_t tcps_noport; /* no socket on port */ 605 u_quad_t tcps_badsyn; /* received ack for which we have 606 no SYN in compressed state */ 607 u_quad_t tcps_delayed_free; /* delayed pool_put() of tcpcb */ 608 609 /* These statistics deal with the SYN cache. */ 610 u_quad_t tcps_sc_added; /* # of entries added */ 611 u_quad_t tcps_sc_completed; /* # of connections completed */ 612 u_quad_t tcps_sc_timed_out; /* # of entries timed out */ 613 u_quad_t tcps_sc_overflowed; /* # dropped due to overflow */ 614 u_quad_t tcps_sc_reset; /* # dropped due to RST */ 615 u_quad_t tcps_sc_unreach; /* # dropped due to ICMP unreach */ 616 u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */ 617 u_quad_t tcps_sc_aborted; /* # of entries aborted (no mem) */ 618 u_quad_t tcps_sc_dupesyn; /* # of duplicate SYNs received */ 619 u_quad_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 620 u_quad_t tcps_sc_collisions; /* # of hash collisions */ 621 u_quad_t tcps_sc_retransmitted; /* # of retransmissions */ 622 u_quad_t tcps_sc_delayed_free; /* # of delayed pool_put()s */ 623 624 u_quad_t tcps_selfquench; /* # of ENOBUFS we get on output */ 625 u_quad_t tcps_badsig; /* # of drops due to bad signature */ 626 u_quad_t tcps_goodsig; /* # of packets with good signature */ 627 }; 628 629 /* 630 * Names for TCP sysctl objects. 631 */ 632 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 633 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 634 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 635 #define TCPCTL_MSSDFLT 4 /* default seg size */ 636 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 637 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 638 #if 0 /*obsoleted*/ 639 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 640 #endif 641 #define TCPCTL_INIT_WIN 8 /* initial window */ 642 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */ 643 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */ 644 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */ 645 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */ 646 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */ 647 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */ 648 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */ 649 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */ 650 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */ 651 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */ 652 #define TCPCTL_KEEPCNT 19 /* keepalive count */ 653 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */ 654 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */ 655 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */ 656 #if 0 /*obsoleted*/ 657 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */ 658 #endif 659 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */ 660 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */ 661 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */ 662 #define TCPCTL_IDENT 27 /* rfc 931 identd */ 663 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */ 664 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */ 665 #define TCPCTL_MAXID 30 666 667 #define TCPCTL_NAMES { \ 668 { 0, 0 }, \ 669 { "rfc1323", CTLTYPE_INT }, \ 670 { "sendspace", CTLTYPE_INT }, \ 671 { "recvspace", CTLTYPE_INT }, \ 672 { "mssdflt", CTLTYPE_INT }, \ 673 { "syn_cache_limit", CTLTYPE_INT }, \ 674 { "syn_bucket_limit", CTLTYPE_INT }, \ 675 { 0, 0 },\ 676 { "init_win", CTLTYPE_INT }, \ 677 { "mss_ifmtu", CTLTYPE_INT }, \ 678 { "sack", CTLTYPE_INT }, \ 679 { "win_scale", CTLTYPE_INT }, \ 680 { "timestamps", CTLTYPE_INT }, \ 681 { "compat_42", CTLTYPE_INT }, \ 682 { "cwm", CTLTYPE_INT }, \ 683 { "cwm_burstsize", CTLTYPE_INT }, \ 684 { "ack_on_push", CTLTYPE_INT }, \ 685 { "keepidle", CTLTYPE_INT }, \ 686 { "keepintvl", CTLTYPE_INT }, \ 687 { "keepcnt", CTLTYPE_INT }, \ 688 { "slowhz", CTLTYPE_INT }, \ 689 { "newreno", CTLTYPE_INT }, \ 690 { "log_refused",CTLTYPE_INT }, \ 691 { 0, 0 }, \ 692 { "rstppslimit", CTLTYPE_INT }, \ 693 { "delack_ticks", CTLTYPE_INT }, \ 694 { "init_win_local", CTLTYPE_INT }, \ 695 { "ident", CTLTYPE_STRUCT }, \ 696 { "ackdropppslimit", CTLTYPE_INT }, \ 697 { "do_loopback_cksum", CTLTYPE_INT }, \ 698 } 699 700 #ifdef _KERNEL 701 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 702 extern struct tcpstat tcpstat; /* tcp statistics */ 703 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 704 extern int tcp_do_rfc1323; /* enabled/disabled? */ 705 extern int tcp_do_sack; /* SACK enabled/disabled? */ 706 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */ 707 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */ 708 extern int tcp_do_newreno; /* Use the New Reno algorithms */ 709 extern int tcp_mssdflt; /* default seg size */ 710 extern int tcp_init_win; /* initial window */ 711 extern int tcp_init_win_local; /* initial window for local nets */ 712 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */ 713 extern int tcp_compat_42; /* work around ancient broken TCP peers */ 714 extern int tcp_cwm; /* enable Congestion Window Monitoring */ 715 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */ 716 extern int tcp_ack_on_push; /* ACK immediately on PUSH */ 717 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 718 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 719 extern int tcp_log_refused; /* log refused connections */ 720 extern int tcp_do_loopback_cksum;/* do TCP checksum on loopback? */ 721 722 extern int tcp_rst_ppslim; 723 extern int tcp_ackdrop_ppslim; 724 725 extern int tcp_syn_cache_size; 726 extern struct syn_cache_head tcp_syn_cache[]; 727 extern u_long syn_cache_count; 728 729 extern struct pool tcpipqent_pool; 730 731 #ifdef MBUFTRACE 732 extern struct mowner tcp_rx_mowner; 733 extern struct mowner tcp_tx_mowner; 734 extern struct mowner tcp_mowner; 735 #endif 736 737 #define TCPCTL_VARIABLES { \ 738 { 0 }, \ 739 { 1, 0, &tcp_do_rfc1323 }, \ 740 { 1, 0, &tcp_sendspace }, \ 741 { 1, 0, &tcp_recvspace }, \ 742 { 1, 0, &tcp_mssdflt }, \ 743 { 1, 0, &tcp_syn_cache_limit }, \ 744 { 1, 0, &tcp_syn_bucket_limit }, \ 745 { 0 }, \ 746 { 1, 0, &tcp_init_win }, \ 747 { 1, 0, &tcp_mss_ifmtu }, \ 748 { 1, 0, &tcp_do_sack }, \ 749 { 1, 0, &tcp_do_win_scale }, \ 750 { 1, 0, &tcp_do_timestamps }, \ 751 { 1, 0, &tcp_compat_42 }, \ 752 { 1, 0, &tcp_cwm }, \ 753 { 1, 0, &tcp_cwm_burstsize }, \ 754 { 1, 0, &tcp_ack_on_push }, \ 755 { 1, 0, &tcp_keepidle }, \ 756 { 1, 0, &tcp_keepintvl }, \ 757 { 1, 0, &tcp_keepcnt }, \ 758 { 1, 1, 0, PR_SLOWHZ }, \ 759 { 1, 0, &tcp_do_newreno }, \ 760 { 1, 0, &tcp_log_refused }, \ 761 { 0 }, \ 762 { 1, 0, &tcp_rst_ppslim }, \ 763 { 1, 0, &tcp_delack_ticks }, \ 764 { 1, 0, &tcp_init_win_local }, \ 765 { 1, 0, &tcp_ackdrop_ppslim }, \ 766 } 767 768 #ifdef __NO_STRICT_ALIGNMENT 769 #define TCP_HDR_ALIGNED_P(th) 1 770 #else 771 #define TCP_HDR_ALIGNED_P(th) ((((vaddr_t)(th)) & 3) == 0) 772 #endif 773 774 struct secasvar; 775 776 int tcp_attach(struct socket *); 777 void tcp_canceltimers(struct tcpcb *); 778 int tcp_timers_invoking(struct tcpcb*); 779 struct tcpcb * 780 tcp_close(struct tcpcb *); 781 int tcp_isdead(struct tcpcb *); 782 #ifdef INET6 783 void tcp6_ctlinput(int, struct sockaddr *, void *); 784 #endif 785 void *tcp_ctlinput(int, struct sockaddr *, void *); 786 int tcp_ctloutput(int, struct socket *, int, int, struct mbuf **); 787 struct tcpcb * 788 tcp_disconnect(struct tcpcb *); 789 struct tcpcb * 790 tcp_drop(struct tcpcb *, int); 791 #ifdef TCP_SIGNATURE 792 int tcp_signature_apply(void *, caddr_t, u_int); 793 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *); 794 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *, 795 char *); 796 #endif 797 int tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *, 798 struct mbuf *, int, struct tcp_opt_info *); 799 void tcp_drain(void); 800 void tcp_established(struct tcpcb *); 801 void tcp_init(void); 802 #ifdef INET6 803 int tcp6_input(struct mbuf **, int *, int); 804 #endif 805 void tcp_input(struct mbuf *, ...); 806 u_long tcp_mss_to_advertise(const struct ifnet *, int); 807 void tcp_mss_from_peer(struct tcpcb *, int); 808 void tcp_tcpcb_template(void); 809 struct tcpcb * 810 tcp_newtcpcb(int, void *); 811 void tcp_notify(struct inpcb *, int); 812 #ifdef INET6 813 void tcp6_notify(struct in6pcb *, int); 814 #endif 815 u_int tcp_optlen(struct tcpcb *); 816 int tcp_output(struct tcpcb *); 817 void tcp_pulloutofband(struct socket *, 818 struct tcphdr *, struct mbuf *, int); 819 void tcp_quench(struct inpcb *, int); 820 #ifdef INET6 821 void tcp6_quench(struct in6pcb *, int); 822 #endif 823 int tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *); 824 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *, 825 struct tcphdr *, tcp_seq, tcp_seq, int); 826 void tcp_rmx_rtt(struct tcpcb *); 827 void tcp_setpersist(struct tcpcb *); 828 #ifdef TCP_SIGNATURE 829 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int, 830 int, u_char *, u_int); 831 #endif 832 void tcp_slowtimo(void); 833 struct mbuf * 834 tcp_template(struct tcpcb *); 835 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int); 836 struct tcpcb * 837 tcp_usrclosed(struct tcpcb *); 838 int tcp_sysctl(int *, u_int, void *, size_t *, void *, size_t); 839 int tcp_usrreq(struct socket *, 840 int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *); 841 void tcp_xmit_timer(struct tcpcb *, uint32_t); 842 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq); 843 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t, 844 tcp_seq); 845 846 void tcp_update_sack_list(struct tcpcb *); 847 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t); 848 void tcp_sack_option(struct tcpcb *, struct tcphdr *, u_char *, int); 849 void tcp_del_sackholes(struct tcpcb *, struct tcphdr *); 850 void tcp_free_sackholes(struct tcpcb *); 851 void tcp_sack_adjust(struct tcpcb *tp); 852 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 853 void tcp_sack_newack(struct tcpcb *, struct tcphdr *); 854 855 856 int syn_cache_add(struct sockaddr *, struct sockaddr *, 857 struct tcphdr *, unsigned int, struct socket *, 858 struct mbuf *, u_char *, int, struct tcp_opt_info *); 859 void syn_cache_unreach(struct sockaddr *, struct sockaddr *, 860 struct tcphdr *); 861 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *, 862 struct tcphdr *, unsigned int, unsigned int, 863 struct socket *so, struct mbuf *); 864 void syn_cache_init(void); 865 void syn_cache_insert(struct syn_cache *, struct tcpcb *); 866 struct syn_cache *syn_cache_lookup(struct sockaddr *, struct sockaddr *, 867 struct syn_cache_head **); 868 void syn_cache_reset(struct sockaddr *, struct sockaddr *, 869 struct tcphdr *); 870 int syn_cache_respond(struct syn_cache *, struct mbuf *); 871 void syn_cache_timer(void *); 872 void syn_cache_cleanup(struct tcpcb *); 873 874 void tcp_reno_newack(struct tcpcb *, struct tcphdr *); 875 void tcp_newreno_newack(struct tcpcb *, struct tcphdr *); 876 877 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int, 878 int); 879 #endif 880 881 #endif /* _NETINET_TCP_VAR_H_ */ 882
Indexes created Tue Sep 23 23:09:58 GMT 2025