1 /* $NetBSD: tcp_var.h,v 1.132 2005/12/11 12:24:58 christos 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, 2005 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 * This code is derived from software contributed to The NetBSD Foundation 80 * by Charles M. Hannum. 81 * 82 * Redistribution and use in source and binary forms, with or without 83 * modification, are permitted provided that the following conditions 84 * are met: 85 * 1. Redistributions of source code must retain the above copyright 86 * notice, this list of conditions and the following disclaimer. 87 * 2. Redistributions in binary form must reproduce the above copyright 88 * notice, this list of conditions and the following disclaimer in the 89 * documentation and/or other materials provided with the distribution. 90 * 3. All advertising materials mentioning features or use of this software 91 * must display the following acknowledgement: 92 * This product includes software developed by the NetBSD 93 * Foundation, Inc. and its contributors. 94 * 4. Neither the name of The NetBSD Foundation nor the names of its 95 * contributors may be used to endorse or promote products derived 96 * from this software without specific prior written permission. 97 * 98 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS 99 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED 100 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR 101 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS 102 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 103 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 104 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS 105 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN 106 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) 107 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE 108 * POSSIBILITY OF SUCH DAMAGE. 109 */ 110 111 /* 112 * Copyright (c) 1982, 1986, 1993, 1994, 1995 113 * The Regents of the University of California. All rights reserved. 114 * 115 * Redistribution and use in source and binary forms, with or without 116 * modification, are permitted provided that the following conditions 117 * are met: 118 * 1. Redistributions of source code must retain the above copyright 119 * notice, this list of conditions and the following disclaimer. 120 * 2. Redistributions in binary form must reproduce the above copyright 121 * notice, this list of conditions and the following disclaimer in the 122 * documentation and/or other materials provided with the distribution. 123 * 3. Neither the name of the University nor the names of its contributors 124 * may be used to endorse or promote products derived from this software 125 * without specific prior written permission. 126 * 127 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 128 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 129 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 130 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 131 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 132 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 133 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 134 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 135 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 136 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 137 * SUCH DAMAGE. 138 * 139 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 140 */ 141 142 #ifndef _NETINET_TCP_VAR_H_ 143 #define _NETINET_TCP_VAR_H_ 144 145 #if defined(_KERNEL_OPT) 146 #include "opt_inet.h" 147 #include "opt_mbuftrace.h" 148 #endif 149 150 /* 151 * Kernel variables for tcp. 152 */ 153 154 #include <sys/callout.h> 155 156 #ifdef TCP_SIGNATURE 157 /* 158 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 159 * for SADB verification and lookup. 160 */ 161 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 162 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 163 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 164 /* 165 * Only a single SA per host may be specified at this time. An SPI is 166 * needed in order for the KEY_ALLOCSA() lookup to work. 167 */ 168 #define TCP_SIG_SPI 0x1000 169 #endif /* TCP_SIGNATURE */ 170 171 /* 172 * SACK option block. 173 */ 174 struct sackblk { 175 tcp_seq left; /* Left edge of sack block. */ 176 tcp_seq right; /* Right edge of sack block. */ 177 }; 178 179 TAILQ_HEAD(sackhead, sackhole); 180 struct sackhole { 181 tcp_seq start; 182 tcp_seq end; 183 tcp_seq rxmit; 184 185 TAILQ_ENTRY(sackhole) sackhole_q; 186 }; 187 188 /* 189 * Tcp control block, one per tcp; fields: 190 */ 191 struct tcpcb { 192 int t_family; /* address family on the wire */ 193 struct ipqehead segq; /* sequencing queue */ 194 int t_segqlen; /* length of the above */ 195 struct callout t_timer[TCPT_NTIMERS];/* tcp timers */ 196 short t_state; /* state of this connection */ 197 short t_rxtshift; /* log(2) of rexmt exp. backoff */ 198 uint32_t t_rxtcur; /* current retransmit value */ 199 short t_dupacks; /* consecutive dup acks recd */ 200 short t_partialacks; /* partials acks during fast rexmit */ 201 u_short t_peermss; /* peer's maximum segment size */ 202 u_short t_ourmss; /* our's maximum segment size */ 203 u_short t_segsz; /* current segment size in use */ 204 char t_force; /* 1 if forcing out a byte */ 205 u_int t_flags; 206 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 207 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 208 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 209 #define TF_NOOPT 0x0008 /* don't use tcp options */ 210 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 211 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 212 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 213 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 214 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 215 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */ 216 #define TF_WILL_SACK 0x0800 /* try to use SACK */ 217 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */ 218 #define TF_DEAD 0x2000 /* dead and to-be-released */ 219 #define TF_PMTUD_PEND 0x4000 /* Path MTU Discovery pending */ 220 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 221 222 223 struct mbuf *t_template; /* skeletal packet for transmit */ 224 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 225 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */ 226 struct callout t_delack_ch; /* delayed ACK callout */ 227 /* 228 * The following fields are used as in the protocol specification. 229 * See RFC783, Dec. 1981, page 21. 230 */ 231 /* send sequence variables */ 232 tcp_seq snd_una; /* send unacknowledged */ 233 tcp_seq snd_nxt; /* send next */ 234 tcp_seq snd_up; /* send urgent pointer */ 235 tcp_seq snd_wl1; /* window update seg seq number */ 236 tcp_seq snd_wl2; /* window update seg ack number */ 237 tcp_seq iss; /* initial send sequence number */ 238 u_long snd_wnd; /* send window */ 239 tcp_seq snd_recover; /* for use in fast recovery */ 240 tcp_seq snd_high; /* NewReno false fast rexmit seq */ 241 /* receive sequence variables */ 242 u_long rcv_wnd; /* receive window */ 243 tcp_seq rcv_nxt; /* receive next */ 244 tcp_seq rcv_up; /* receive urgent pointer */ 245 tcp_seq irs; /* initial receive sequence number */ 246 /* 247 * Additional variables for this implementation. 248 */ 249 /* receive variables */ 250 tcp_seq rcv_adv; /* advertised window */ 251 /* retransmit variables */ 252 tcp_seq snd_max; /* highest sequence number sent; 253 * used to recognize retransmits 254 */ 255 /* congestion control (for slow start, source quench, retransmit after loss) */ 256 u_long snd_cwnd; /* congestion-controlled window */ 257 u_long snd_ssthresh; /* snd_cwnd size threshhold for 258 * for slow start exponential to 259 * linear switch 260 */ 261 /* 262 * transmit timing stuff. See below for scale of srtt and rttvar. 263 * "Variance" is actually smoothed difference. 264 */ 265 uint32_t t_rcvtime; /* time last segment received */ 266 uint32_t t_rtttime; /* time we started measuring rtt */ 267 tcp_seq t_rtseq; /* sequence number being timed */ 268 int32_t t_srtt; /* smoothed round-trip time */ 269 int32_t t_rttvar; /* variance in round-trip time */ 270 uint32_t t_rttmin; /* minimum rtt allowed */ 271 u_long max_sndwnd; /* largest window peer has offered */ 272 273 /* out-of-band data */ 274 char t_oobflags; /* have some */ 275 char t_iobc; /* input character */ 276 #define TCPOOB_HAVEDATA 0x01 277 #define TCPOOB_HADDATA 0x02 278 short t_softerror; /* possible error not yet reported */ 279 280 /* RFC 1323 variables */ 281 u_char snd_scale; /* window scaling for send window */ 282 u_char rcv_scale; /* window scaling for recv window */ 283 u_char request_r_scale; /* pending window scaling */ 284 u_char requested_s_scale; 285 u_int32_t ts_recent; /* timestamp echo data */ 286 u_int32_t ts_recent_age; /* when last updated */ 287 u_int32_t ts_timebase; /* our timebase */ 288 tcp_seq last_ack_sent; 289 290 /* SACK stuff */ 291 #define TCP_SACK_MAX 3 292 #define TCPSACK_NONE 0 293 #define TCPSACK_HAVED 1 294 u_char rcv_sack_flags; /* SACK flags. */ 295 struct sackblk rcv_dsack_block; /* RX D-SACK block. */ 296 struct ipqehead timeq; /* time sequenced queue. */ 297 struct sackhead snd_holes; /* TX SACK holes. */ 298 int snd_numholes; /* Number of TX SACK holes. */ 299 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/ 300 tcp_seq sack_newdata; /* New data xmitted in this recovery 301 episode starts at this seq number*/ 302 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by 303 peer. */ 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 /* Path-MTU discovery blackhole detection */ 315 int t_mtudisc; /* perform mtudisc for this tcb */ 316 /* Path-MTU Discovery Information */ 317 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */ 318 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */ 319 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */ 320 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */ 321 u_short t_pmtud_ip_len; /* IP length from ICMP payload */ 322 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */ 323 }; 324 325 /* 326 * Macros to aid SACK/FACK TCP. 327 */ 328 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK) 329 #define TCP_FACK_FASTRECOV(tp) \ 330 (TCP_SACK_ENABLED(tp) && \ 331 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz))) 332 333 #ifdef _KERNEL 334 /* 335 * TCP reassembly queue locks. 336 */ 337 static __inline int tcp_reass_lock_try (struct tcpcb *) 338 __attribute__((__unused__)); 339 static __inline void tcp_reass_unlock (struct tcpcb *) 340 __attribute__((__unused__)); 341 342 static __inline int 343 tcp_reass_lock_try(tp) 344 struct tcpcb *tp; 345 { 346 int s; 347 348 /* 349 * Use splvm() -- we're blocking things that would cause 350 * mbuf allocation. 351 */ 352 s = splvm(); 353 if (tp->t_flags & TF_REASSEMBLING) { 354 splx(s); 355 return (0); 356 } 357 tp->t_flags |= TF_REASSEMBLING; 358 splx(s); 359 return (1); 360 } 361 362 static __inline void 363 tcp_reass_unlock(tp) 364 struct tcpcb *tp; 365 { 366 int s; 367 368 s = splvm(); 369 tp->t_flags &= ~TF_REASSEMBLING; 370 splx(s); 371 } 372 373 #ifdef DIAGNOSTIC 374 #define TCP_REASS_LOCK(tp) \ 375 do { \ 376 if (tcp_reass_lock_try(tp) == 0) { \ 377 printf("%s:%d: tcpcb %p reass already locked\n", \ 378 __FILE__, __LINE__, tp); \ 379 panic("tcp_reass_lock"); \ 380 } \ 381 } while (/*CONSTCOND*/ 0) 382 #define TCP_REASS_LOCK_CHECK(tp) \ 383 do { \ 384 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \ 385 printf("%s:%d: tcpcb %p reass lock not held\n", \ 386 __FILE__, __LINE__, tp); \ 387 panic("tcp reass lock check"); \ 388 } \ 389 } while (/*CONSTCOND*/ 0) 390 #else 391 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp)) 392 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */ 393 #endif 394 395 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp)) 396 #endif /* _KERNEL */ 397 398 /* 399 * Queue for delayed ACK processing. 400 */ 401 #ifdef _KERNEL 402 extern int tcp_delack_ticks; 403 void tcp_delack(void *); 404 405 #define TCP_RESTART_DELACK(tp) \ 406 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \ 407 tcp_delack, tp) 408 409 #define TCP_SET_DELACK(tp) \ 410 do { \ 411 if (((tp)->t_flags & TF_DELACK) == 0) { \ 412 (tp)->t_flags |= TF_DELACK; \ 413 TCP_RESTART_DELACK(tp); \ 414 } \ 415 } while (/*CONSTCOND*/0) 416 417 #define TCP_CLEAR_DELACK(tp) \ 418 do { \ 419 if ((tp)->t_flags & TF_DELACK) { \ 420 (tp)->t_flags &= ~TF_DELACK; \ 421 callout_stop(&(tp)->t_delack_ch); \ 422 } \ 423 } while (/*CONSTCOND*/0) 424 #endif /* _KERNEL */ 425 426 /* 427 * Compute the current timestamp for a connection. 428 */ 429 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase) 430 431 /* 432 * Handy way of passing around TCP option info. 433 */ 434 struct tcp_opt_info { 435 int ts_present; 436 u_int32_t ts_val; 437 u_int32_t ts_ecr; 438 u_int16_t maxseg; 439 }; 440 441 #define TOF_SIGNATURE 0x0040 /* signature option present */ 442 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */ 443 444 /* 445 * Data for the TCP compressed state engine. 446 */ 447 union syn_cache_sa { 448 struct sockaddr sa; 449 struct sockaddr_in sin; 450 #if 1 /*def INET6*/ 451 struct sockaddr_in6 sin6; 452 #endif 453 }; 454 455 struct syn_cache { 456 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */ 457 struct callout sc_timer; /* rexmt timer */ 458 union { /* cached route */ 459 struct route route4; 460 #ifdef INET6 461 struct route_in6 route6; 462 #endif 463 } sc_route_u; 464 #define sc_route4 sc_route_u.route4 465 #ifdef INET6 466 #define sc_route6 sc_route_u.route6 467 #endif 468 long sc_win; /* advertised window */ 469 int sc_bucketidx; /* our bucket index */ 470 u_int32_t sc_hash; 471 u_int32_t sc_timestamp; /* timestamp from SYN */ 472 u_int32_t sc_timebase; /* our local timebase */ 473 union syn_cache_sa sc_src; 474 union syn_cache_sa sc_dst; 475 tcp_seq sc_irs; 476 tcp_seq sc_iss; 477 u_int sc_rxtcur; /* current rxt timeout */ 478 u_int sc_rxttot; /* total time spend on queues */ 479 u_short sc_rxtshift; /* for computing backoff */ 480 u_short sc_flags; 481 482 #define SCF_UNREACH 0x0001 /* we've had an unreach error */ 483 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */ 484 #define SCF_DEAD 0x0004 /* this entry to be released */ 485 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */ 486 #define SCF_SIGNATURE 0x40 /* send MD5 digests */ 487 488 struct mbuf *sc_ipopts; /* IP options */ 489 u_int16_t sc_peermaxseg; 490 u_int16_t sc_ourmaxseg; 491 u_int8_t sc_request_r_scale : 4, 492 sc_requested_s_scale : 4; 493 494 struct tcpcb *sc_tp; /* tcb for listening socket */ 495 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */ 496 }; 497 498 struct syn_cache_head { 499 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */ 500 u_short sch_length; /* # entries in bucket */ 501 }; 502 503 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 504 #ifdef INET6 505 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb) 506 #endif 507 #ifndef INET6 508 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 509 #else 510 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \ 511 ? intotcpcb(sotoinpcb(so)) \ 512 : in6totcpcb(sotoin6pcb(so))) 513 #endif 514 515 /* 516 * The smoothed round-trip time and estimated variance 517 * are stored as fixed point numbers scaled by the values below. 518 * For convenience, these scales are also used in smoothing the average 519 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 520 * With these scales, srtt has 3 bits to the right of the binary point, 521 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 522 * binary point, and is smoothed with an ALPHA of 0.75. 523 */ 524 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 525 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 526 527 /* 528 * The initial retransmission should happen at rtt + 4 * rttvar. 529 * Because of the way we do the smoothing, srtt and rttvar 530 * will each average +1/2 tick of bias. When we compute 531 * the retransmit timer, we want 1/2 tick of rounding and 532 * 1 extra tick because of +-1/2 tick uncertainty in the 533 * firing of the timer. The bias will give us exactly the 534 * 1.5 tick we need. But, because the bias is 535 * statistical, we have to test that we don't drop below 536 * the minimum feasible timer (which is 2 ticks). 537 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 538 * is the same as the multiplier for rttvar. 539 */ 540 #define TCP_REXMTVAL(tp) \ 541 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 542 543 /* 544 * Compute the initial window for slow start. 545 */ 546 #define TCP_INITIAL_WINDOW(iw, segsz) \ 547 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 548 ((segsz) * (iw))) 549 550 /* 551 * TCP statistics. 552 * Many of these should be kept per connection, 553 * but that's inconvenient at the moment. 554 */ 555 struct tcpstat { 556 u_quad_t tcps_connattempt; /* connections initiated */ 557 u_quad_t tcps_accepts; /* connections accepted */ 558 u_quad_t tcps_connects; /* connections established */ 559 u_quad_t tcps_drops; /* connections dropped */ 560 u_quad_t tcps_conndrops; /* embryonic connections dropped */ 561 u_quad_t tcps_closed; /* conn. closed (includes drops) */ 562 u_quad_t tcps_segstimed; /* segs where we tried to get rtt */ 563 u_quad_t tcps_rttupdated; /* times we succeeded */ 564 u_quad_t tcps_delack; /* delayed acks sent */ 565 u_quad_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 566 u_quad_t tcps_rexmttimeo; /* retransmit timeouts */ 567 u_quad_t tcps_persisttimeo; /* persist timeouts */ 568 u_quad_t tcps_keeptimeo; /* keepalive timeouts */ 569 u_quad_t tcps_keepprobe; /* keepalive probes sent */ 570 u_quad_t tcps_keepdrops; /* connections dropped in keepalive */ 571 u_quad_t tcps_persistdrops; /* connections dropped in persist */ 572 u_quad_t tcps_connsdrained; /* connections drained due to memory 573 shortage */ 574 u_quad_t tcps_pmtublackhole; /* PMTUD blackhole detected */ 575 576 u_quad_t tcps_sndtotal; /* total packets sent */ 577 u_quad_t tcps_sndpack; /* data packets sent */ 578 u_quad_t tcps_sndbyte; /* data bytes sent */ 579 u_quad_t tcps_sndrexmitpack; /* data packets retransmitted */ 580 u_quad_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 581 u_quad_t tcps_sndacks; /* ack-only packets sent */ 582 u_quad_t tcps_sndprobe; /* window probes sent */ 583 u_quad_t tcps_sndurg; /* packets sent with URG only */ 584 u_quad_t tcps_sndwinup; /* window update-only packets sent */ 585 u_quad_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 586 587 u_quad_t tcps_rcvtotal; /* total packets received */ 588 u_quad_t tcps_rcvpack; /* packets received in sequence */ 589 u_quad_t tcps_rcvbyte; /* bytes received in sequence */ 590 u_quad_t tcps_rcvbadsum; /* packets received with ccksum errs */ 591 u_quad_t tcps_rcvbadoff; /* packets received with bad offset */ 592 u_quad_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 593 u_quad_t tcps_rcvshort; /* packets received too short */ 594 u_quad_t tcps_rcvduppack; /* duplicate-only packets received */ 595 u_quad_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 596 u_quad_t tcps_rcvpartduppack; /* packets with some duplicate data */ 597 u_quad_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 598 u_quad_t tcps_rcvoopack; /* out-of-order packets received */ 599 u_quad_t tcps_rcvoobyte; /* out-of-order bytes received */ 600 u_quad_t tcps_rcvpackafterwin; /* packets with data after window */ 601 u_quad_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 602 u_quad_t tcps_rcvafterclose; /* packets rcvd after "close" */ 603 u_quad_t tcps_rcvwinprobe; /* rcvd window probe packets */ 604 u_quad_t tcps_rcvdupack; /* rcvd duplicate acks */ 605 u_quad_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 606 u_quad_t tcps_rcvackpack; /* rcvd ack packets */ 607 u_quad_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 608 u_quad_t tcps_rcvwinupd; /* rcvd window update packets */ 609 u_quad_t tcps_pawsdrop; /* segments dropped due to PAWS */ 610 u_quad_t tcps_predack; /* times hdr predict ok for acks */ 611 u_quad_t tcps_preddat; /* times hdr predict ok for data pkts */ 612 613 u_quad_t tcps_pcbhashmiss; /* input packets missing pcb hash */ 614 u_quad_t tcps_noport; /* no socket on port */ 615 u_quad_t tcps_badsyn; /* received ack for which we have 616 no SYN in compressed state */ 617 u_quad_t tcps_delayed_free; /* delayed pool_put() of tcpcb */ 618 619 /* These statistics deal with the SYN cache. */ 620 u_quad_t tcps_sc_added; /* # of entries added */ 621 u_quad_t tcps_sc_completed; /* # of connections completed */ 622 u_quad_t tcps_sc_timed_out; /* # of entries timed out */ 623 u_quad_t tcps_sc_overflowed; /* # dropped due to overflow */ 624 u_quad_t tcps_sc_reset; /* # dropped due to RST */ 625 u_quad_t tcps_sc_unreach; /* # dropped due to ICMP unreach */ 626 u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */ 627 u_quad_t tcps_sc_aborted; /* # of entries aborted (no mem) */ 628 u_quad_t tcps_sc_dupesyn; /* # of duplicate SYNs received */ 629 u_quad_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 630 u_quad_t tcps_sc_collisions; /* # of hash collisions */ 631 u_quad_t tcps_sc_retransmitted; /* # of retransmissions */ 632 u_quad_t tcps_sc_delayed_free; /* # of delayed pool_put()s */ 633 634 u_quad_t tcps_selfquench; /* # of ENOBUFS we get on output */ 635 u_quad_t tcps_badsig; /* # of drops due to bad signature */ 636 u_quad_t tcps_goodsig; /* # of packets with good signature */ 637 }; 638 639 /* 640 * Names for TCP sysctl objects. 641 */ 642 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 643 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 644 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 645 #define TCPCTL_MSSDFLT 4 /* default seg size */ 646 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 647 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 648 #if 0 /*obsoleted*/ 649 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 650 #endif 651 #define TCPCTL_INIT_WIN 8 /* initial window */ 652 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */ 653 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */ 654 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */ 655 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */ 656 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */ 657 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */ 658 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */ 659 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */ 660 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */ 661 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */ 662 #define TCPCTL_KEEPCNT 19 /* keepalive count */ 663 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */ 664 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */ 665 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */ 666 #if 0 /*obsoleted*/ 667 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */ 668 #endif 669 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */ 670 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */ 671 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */ 672 #define TCPCTL_IDENT 27 /* rfc 931 identd */ 673 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */ 674 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */ 675 #define TCPCTL_STATS 30 /* TCP statistics */ 676 #define TCPCTL_DEBUG 31 /* TCP debug sockets */ 677 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */ 678 #define TCPCTL_MAXID 33 679 680 #define TCPCTL_NAMES { \ 681 { 0, 0 }, \ 682 { "rfc1323", CTLTYPE_INT }, \ 683 { "sendspace", CTLTYPE_INT }, \ 684 { "recvspace", CTLTYPE_INT }, \ 685 { "mssdflt", CTLTYPE_INT }, \ 686 { "syn_cache_limit", CTLTYPE_INT }, \ 687 { "syn_bucket_limit", CTLTYPE_INT }, \ 688 { 0, 0 },\ 689 { "init_win", CTLTYPE_INT }, \ 690 { "mss_ifmtu", CTLTYPE_INT }, \ 691 { "sack", CTLTYPE_INT }, \ 692 { "win_scale", CTLTYPE_INT }, \ 693 { "timestamps", CTLTYPE_INT }, \ 694 { "compat_42", CTLTYPE_INT }, \ 695 { "cwm", CTLTYPE_INT }, \ 696 { "cwm_burstsize", CTLTYPE_INT }, \ 697 { "ack_on_push", CTLTYPE_INT }, \ 698 { "keepidle", CTLTYPE_INT }, \ 699 { "keepintvl", CTLTYPE_INT }, \ 700 { "keepcnt", CTLTYPE_INT }, \ 701 { "slowhz", CTLTYPE_INT }, \ 702 { "newreno", CTLTYPE_INT }, \ 703 { "log_refused",CTLTYPE_INT }, \ 704 { 0, 0 }, \ 705 { "rstppslimit", CTLTYPE_INT }, \ 706 { "delack_ticks", CTLTYPE_INT }, \ 707 { "init_win_local", CTLTYPE_INT }, \ 708 { "ident", CTLTYPE_STRUCT }, \ 709 { "ackdropppslimit", CTLTYPE_INT }, \ 710 { "do_loopback_cksum", CTLTYPE_INT }, \ 711 { "stats", CTLTYPE_STRUCT }, \ 712 { "debug", CTLTYPE_STRUCT }, \ 713 { "debx", CTLTYPE_INT }, \ 714 } 715 716 #ifdef _KERNEL 717 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 718 extern struct tcpstat tcpstat; /* tcp statistics */ 719 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 720 extern int tcp_do_rfc1323; /* enabled/disabled? */ 721 extern int tcp_do_sack; /* SACK enabled/disabled? */ 722 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */ 723 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */ 724 extern int tcp_do_newreno; /* Use the New Reno algorithms */ 725 extern int tcp_mssdflt; /* default seg size */ 726 extern int tcp_init_win; /* initial window */ 727 extern int tcp_init_win_local; /* initial window for local nets */ 728 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */ 729 extern int tcp_compat_42; /* work around ancient broken TCP peers */ 730 extern int tcp_cwm; /* enable Congestion Window Monitoring */ 731 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */ 732 extern int tcp_ack_on_push; /* ACK immediately on PUSH */ 733 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 734 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 735 extern int tcp_log_refused; /* log refused connections */ 736 extern int tcp_sack_tp_maxholes; /* Max holes per connection. */ 737 extern int tcp_sack_globalmaxholes; /* Max holes per system. */ 738 extern int tcp_sack_globalholes; /* Number of holes present. */ 739 740 extern int tcp_rst_ppslim; 741 extern int tcp_ackdrop_ppslim; 742 743 extern int tcp_syn_cache_size; 744 extern struct syn_cache_head tcp_syn_cache[]; 745 extern u_long syn_cache_count; 746 747 #ifdef MBUFTRACE 748 extern struct mowner tcp_rx_mowner; 749 extern struct mowner tcp_tx_mowner; 750 extern struct mowner tcp_mowner; 751 #endif 752 753 #define TCPCTL_VARIABLES { \ 754 { 0 }, \ 755 { 1, 0, &tcp_do_rfc1323 }, \ 756 { 1, 0, &tcp_sendspace }, \ 757 { 1, 0, &tcp_recvspace }, \ 758 { 1, 0, &tcp_mssdflt }, \ 759 { 1, 0, &tcp_syn_cache_limit }, \ 760 { 1, 0, &tcp_syn_bucket_limit }, \ 761 { 0 }, \ 762 { 1, 0, &tcp_init_win }, \ 763 { 1, 0, &tcp_mss_ifmtu }, \ 764 { 1, 0, &tcp_do_sack }, \ 765 { 1, 0, &tcp_do_win_scale }, \ 766 { 1, 0, &tcp_do_timestamps }, \ 767 { 1, 0, &tcp_compat_42 }, \ 768 { 1, 0, &tcp_cwm }, \ 769 { 1, 0, &tcp_cwm_burstsize }, \ 770 { 1, 0, &tcp_ack_on_push }, \ 771 { 1, 0, &tcp_keepidle }, \ 772 { 1, 0, &tcp_keepintvl }, \ 773 { 1, 0, &tcp_keepcnt }, \ 774 { 1, 1, 0, PR_SLOWHZ }, \ 775 { 1, 0, &tcp_do_newreno }, \ 776 { 1, 0, &tcp_log_refused }, \ 777 { 0 }, \ 778 { 1, 0, &tcp_rst_ppslim }, \ 779 { 1, 0, &tcp_delack_ticks }, \ 780 { 1, 0, &tcp_init_win_local }, \ 781 { 1, 0, &tcp_ackdrop_ppslim }, \ 782 } 783 784 #ifdef __NO_STRICT_ALIGNMENT 785 #define TCP_HDR_ALIGNED_P(th) 1 786 #else 787 #define TCP_HDR_ALIGNED_P(th) ((((vaddr_t)(th)) & 3) == 0) 788 #endif 789 790 struct secasvar; 791 792 int tcp_attach(struct socket *); 793 void tcp_canceltimers(struct tcpcb *); 794 int tcp_timers_invoking(struct tcpcb*); 795 struct tcpcb * 796 tcp_close(struct tcpcb *); 797 int tcp_isdead(struct tcpcb *); 798 #ifdef INET6 799 void tcp6_ctlinput(int, struct sockaddr *, void *); 800 #endif 801 void *tcp_ctlinput(int, struct sockaddr *, void *); 802 int tcp_ctloutput(int, struct socket *, int, int, struct mbuf **); 803 struct tcpcb * 804 tcp_disconnect(struct tcpcb *); 805 struct tcpcb * 806 tcp_drop(struct tcpcb *, int); 807 #ifdef TCP_SIGNATURE 808 int tcp_signature_apply(void *, caddr_t, u_int); 809 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *); 810 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *, 811 char *); 812 #endif 813 int tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *, 814 struct mbuf *, int, struct tcp_opt_info *); 815 void tcp_drain(void); 816 void tcp_established(struct tcpcb *); 817 void tcp_init(void); 818 #ifdef INET6 819 int tcp6_input(struct mbuf **, int *, int); 820 #endif 821 void tcp_input(struct mbuf *, ...); 822 u_int tcp_hdrsz(struct tcpcb *); 823 u_long tcp_mss_to_advertise(const struct ifnet *, int); 824 void tcp_mss_from_peer(struct tcpcb *, int); 825 void tcp_tcpcb_template(void); 826 struct tcpcb * 827 tcp_newtcpcb(int, void *); 828 void tcp_notify(struct inpcb *, int); 829 #ifdef INET6 830 void tcp6_notify(struct in6pcb *, int); 831 #endif 832 u_int tcp_optlen(struct tcpcb *); 833 int tcp_output(struct tcpcb *); 834 void tcp_pulloutofband(struct socket *, 835 struct tcphdr *, struct mbuf *, int); 836 void tcp_quench(struct inpcb *, int); 837 #ifdef INET6 838 void tcp6_quench(struct in6pcb *, int); 839 #endif 840 void tcp_mtudisc(struct inpcb *, int); 841 842 struct ipqent *tcpipqent_alloc(void); 843 void tcpipqent_free(struct ipqent *); 844 845 int tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *); 846 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *, 847 struct tcphdr *, tcp_seq, tcp_seq, int); 848 void tcp_rmx_rtt(struct tcpcb *); 849 void tcp_setpersist(struct tcpcb *); 850 #ifdef TCP_SIGNATURE 851 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int, 852 int, u_char *, u_int); 853 #endif 854 void tcp_slowtimo(void); 855 struct mbuf * 856 tcp_template(struct tcpcb *); 857 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int); 858 struct tcpcb * 859 tcp_usrclosed(struct tcpcb *); 860 int tcp_usrreq(struct socket *, 861 int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *); 862 void tcp_xmit_timer(struct tcpcb *, uint32_t); 863 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq); 864 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t, 865 tcp_seq); 866 867 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t); 868 void tcp_sack_option(struct tcpcb *, struct tcphdr *, u_char *, int); 869 void tcp_del_sackholes(struct tcpcb *, struct tcphdr *); 870 void tcp_free_sackholes(struct tcpcb *); 871 void tcp_sack_adjust(struct tcpcb *tp); 872 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 873 void tcp_sack_newack(struct tcpcb *, struct tcphdr *); 874 int tcp_sack_numblks(const struct tcpcb *); 875 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2) 876 877 int syn_cache_add(struct sockaddr *, struct sockaddr *, 878 struct tcphdr *, unsigned int, struct socket *, 879 struct mbuf *, u_char *, int, struct tcp_opt_info *); 880 void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *, 881 struct tcphdr *); 882 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *, 883 struct tcphdr *, unsigned int, unsigned int, 884 struct socket *so, struct mbuf *); 885 void syn_cache_init(void); 886 void syn_cache_insert(struct syn_cache *, struct tcpcb *); 887 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *, 888 struct syn_cache_head **); 889 void syn_cache_reset(struct sockaddr *, struct sockaddr *, 890 struct tcphdr *); 891 int syn_cache_respond(struct syn_cache *, struct mbuf *); 892 void syn_cache_timer(void *); 893 void syn_cache_cleanup(struct tcpcb *); 894 895 void tcp_reno_newack(struct tcpcb *, struct tcphdr *); 896 void tcp_newreno_newack(struct tcpcb *, struct tcphdr *); 897 898 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int, 899 int); 900 #endif 901 902 #endif /* !_NETINET_TCP_VAR_H_ */ 903