1 /* $NetBSD: tcp_var.h,v 1.141 2006/10/21 10:08:54 yamt 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 #include "rnd.h" 149 #endif 150 151 /* 152 * Kernel variables for tcp. 153 */ 154 155 #include <sys/callout.h> 156 157 #ifdef TCP_SIGNATURE 158 /* 159 * Defines which are needed by the xform_tcp module and tcp_[in|out]put 160 * for SADB verification and lookup. 161 */ 162 #define TCP_SIGLEN 16 /* length of computed digest in bytes */ 163 #define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */ 164 #define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */ 165 /* 166 * Only a single SA per host may be specified at this time. An SPI is 167 * needed in order for the KEY_ALLOCSA() lookup to work. 168 */ 169 #define TCP_SIG_SPI 0x1000 170 #endif /* TCP_SIGNATURE */ 171 172 /* 173 * SACK option block. 174 */ 175 struct sackblk { 176 tcp_seq left; /* Left edge of sack block. */ 177 tcp_seq right; /* Right edge of sack block. */ 178 }; 179 180 TAILQ_HEAD(sackhead, sackhole); 181 struct sackhole { 182 tcp_seq start; 183 tcp_seq end; 184 tcp_seq rxmit; 185 186 TAILQ_ENTRY(sackhole) sackhole_q; 187 }; 188 189 /* 190 * Tcp control block, one per tcp; fields: 191 */ 192 struct tcpcb { 193 int t_family; /* address family on the wire */ 194 struct ipqehead segq; /* sequencing queue */ 195 int t_segqlen; /* length of the above */ 196 struct callout t_timer[TCPT_NTIMERS];/* tcp timers */ 197 short t_state; /* state of this connection */ 198 short t_rxtshift; /* log(2) of rexmt exp. backoff */ 199 uint32_t t_rxtcur; /* current retransmit value */ 200 short t_dupacks; /* consecutive dup acks recd */ 201 short t_partialacks; /* partials acks during fast rexmit */ 202 u_short t_peermss; /* peer's maximum segment size */ 203 u_short t_ourmss; /* our's maximum segment size */ 204 u_short t_segsz; /* current segment size in use */ 205 char t_force; /* 1 if forcing out a byte */ 206 u_int t_flags; 207 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 208 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 209 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 210 #define TF_NOOPT 0x0008 /* don't use tcp options */ 211 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 212 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 213 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 214 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 215 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 216 #define TF_SYN_REXMT 0x0400 /* rexmit timer fired on SYN */ 217 #define TF_WILL_SACK 0x0800 /* try to use SACK */ 218 #define TF_REASSEMBLING 0x1000 /* we're busy reassembling */ 219 #define TF_DEAD 0x2000 /* dead and to-be-released */ 220 #define TF_PMTUD_PEND 0x4000 /* Path MTU Discovery pending */ 221 #define TF_ECN_PERMIT 0x10000 /* other side said is ECN-ready */ 222 #define TF_ECN_SND_CWR 0x20000 /* ECN CWR in queue */ 223 #define TF_ECN_SND_ECE 0x40000 /* ECN ECE in queue */ 224 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 225 226 227 struct mbuf *t_template; /* skeletal packet for transmit */ 228 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 229 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */ 230 struct callout t_delack_ch; /* delayed ACK callout */ 231 /* 232 * The following fields are used as in the protocol specification. 233 * See RFC793, Dec. 1981, page 21. 234 */ 235 /* send sequence variables */ 236 tcp_seq snd_una; /* send unacknowledged */ 237 tcp_seq snd_nxt; /* send next */ 238 tcp_seq snd_up; /* send urgent pointer */ 239 tcp_seq snd_wl1; /* window update seg seq number */ 240 tcp_seq snd_wl2; /* window update seg ack number */ 241 tcp_seq iss; /* initial send sequence number */ 242 u_long snd_wnd; /* send window */ 243 tcp_seq snd_recover; /* for use in fast recovery */ 244 tcp_seq snd_high; /* NewReno false fast rexmit seq */ 245 /* receive sequence variables */ 246 u_long rcv_wnd; /* receive window */ 247 tcp_seq rcv_nxt; /* receive next */ 248 tcp_seq rcv_up; /* receive urgent pointer */ 249 tcp_seq irs; /* initial receive sequence number */ 250 /* 251 * Additional variables for this implementation. 252 */ 253 /* receive variables */ 254 tcp_seq rcv_adv; /* advertised window */ 255 /* retransmit variables */ 256 tcp_seq snd_max; /* highest sequence number sent; 257 * used to recognize retransmits 258 */ 259 /* congestion control (for slow start, source quench, retransmit after loss) */ 260 u_long snd_cwnd; /* congestion-controlled window */ 261 u_long snd_ssthresh; /* snd_cwnd size threshhold for 262 * for slow start exponential to 263 * linear switch 264 */ 265 /* 266 * transmit timing stuff. See below for scale of srtt and rttvar. 267 * "Variance" is actually smoothed difference. 268 */ 269 uint32_t t_rcvtime; /* time last segment received */ 270 uint32_t t_rtttime; /* time we started measuring rtt */ 271 tcp_seq t_rtseq; /* sequence number being timed */ 272 int32_t t_srtt; /* smoothed round-trip time */ 273 int32_t t_rttvar; /* variance in round-trip time */ 274 uint32_t t_rttmin; /* minimum rtt allowed */ 275 u_long max_sndwnd; /* largest window peer has offered */ 276 277 /* out-of-band data */ 278 char t_oobflags; /* have some */ 279 char t_iobc; /* input character */ 280 #define TCPOOB_HAVEDATA 0x01 281 #define TCPOOB_HADDATA 0x02 282 short t_softerror; /* possible error not yet reported */ 283 284 /* RFC 1323 variables */ 285 u_char snd_scale; /* window scaling for send window */ 286 u_char rcv_scale; /* window scaling for recv window */ 287 u_char request_r_scale; /* pending window scaling */ 288 u_char requested_s_scale; 289 u_int32_t ts_recent; /* timestamp echo data */ 290 u_int32_t ts_recent_age; /* when last updated */ 291 u_int32_t ts_timebase; /* our timebase */ 292 tcp_seq last_ack_sent; 293 294 /* RFC 3465 variables */ 295 u_long t_bytes_acked; /* ABC "bytes_acked" parameter */ 296 297 /* SACK stuff */ 298 #define TCP_SACK_MAX 3 299 #define TCPSACK_NONE 0 300 #define TCPSACK_HAVED 1 301 u_char rcv_sack_flags; /* SACK flags. */ 302 struct sackblk rcv_dsack_block; /* RX D-SACK block. */ 303 struct ipqehead timeq; /* time sequenced queue. */ 304 struct sackhead snd_holes; /* TX SACK holes. */ 305 int snd_numholes; /* Number of TX SACK holes. */ 306 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/ 307 tcp_seq sack_newdata; /* New data xmitted in this recovery 308 episode starts at this seq number*/ 309 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by 310 peer. */ 311 312 /* pointer for syn cache entries*/ 313 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */ 314 315 /* prediction of next mbuf when using large window sizes */ 316 struct mbuf *t_lastm; /* last mbuf that data was sent from */ 317 int t_inoff; /* data offset in previous mbuf */ 318 int t_lastoff; /* last data address in mbuf chain */ 319 int t_lastlen; /* last length read from mbuf chain */ 320 321 /* Path-MTU discovery blackhole detection */ 322 int t_mtudisc; /* perform mtudisc for this tcb */ 323 /* Path-MTU Discovery Information */ 324 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */ 325 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */ 326 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */ 327 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */ 328 u_short t_pmtud_ip_len; /* IP length from ICMP payload */ 329 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */ 330 331 uint8_t t_ecn_retries; /* # of ECN setup retries */ 332 333 struct tcp_congctl *t_congctl; /* per TCB congctl algorithm */ 334 }; 335 336 /* 337 * Macros to aid ECN TCP. 338 */ 339 #define TCP_ECN_ALLOWED(tp) (tp->t_flags & TF_ECN_PERMIT) 340 341 /* 342 * Macros to aid SACK/FACK TCP. 343 */ 344 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK) 345 #define TCP_FACK_FASTRECOV(tp) \ 346 (TCP_SACK_ENABLED(tp) && \ 347 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz))) 348 349 #ifdef _KERNEL 350 /* 351 * TCP reassembly queue locks. 352 */ 353 static __inline int tcp_reass_lock_try (struct tcpcb *) 354 __attribute__((__unused__)); 355 static __inline void tcp_reass_unlock (struct tcpcb *) 356 __attribute__((__unused__)); 357 358 static __inline int 359 tcp_reass_lock_try(tp) 360 struct tcpcb *tp; 361 { 362 int s; 363 364 /* 365 * Use splvm() -- we're blocking things that would cause 366 * mbuf allocation. 367 */ 368 s = splvm(); 369 if (tp->t_flags & TF_REASSEMBLING) { 370 splx(s); 371 return (0); 372 } 373 tp->t_flags |= TF_REASSEMBLING; 374 splx(s); 375 return (1); 376 } 377 378 static __inline void 379 tcp_reass_unlock(tp) 380 struct tcpcb *tp; 381 { 382 int s; 383 384 s = splvm(); 385 tp->t_flags &= ~TF_REASSEMBLING; 386 splx(s); 387 } 388 389 #ifdef DIAGNOSTIC 390 #define TCP_REASS_LOCK(tp) \ 391 do { \ 392 if (tcp_reass_lock_try(tp) == 0) { \ 393 printf("%s:%d: tcpcb %p reass already locked\n", \ 394 __FILE__, __LINE__, tp); \ 395 panic("tcp_reass_lock"); \ 396 } \ 397 } while (/*CONSTCOND*/ 0) 398 #define TCP_REASS_LOCK_CHECK(tp) \ 399 do { \ 400 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \ 401 printf("%s:%d: tcpcb %p reass lock not held\n", \ 402 __FILE__, __LINE__, tp); \ 403 panic("tcp reass lock check"); \ 404 } \ 405 } while (/*CONSTCOND*/ 0) 406 #else 407 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp)) 408 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */ 409 #endif 410 411 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp)) 412 #endif /* _KERNEL */ 413 414 /* 415 * Queue for delayed ACK processing. 416 */ 417 #ifdef _KERNEL 418 extern int tcp_delack_ticks; 419 void tcp_delack(void *); 420 421 #define TCP_RESTART_DELACK(tp) \ 422 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \ 423 tcp_delack, tp) 424 425 #define TCP_SET_DELACK(tp) \ 426 do { \ 427 if (((tp)->t_flags & TF_DELACK) == 0) { \ 428 (tp)->t_flags |= TF_DELACK; \ 429 TCP_RESTART_DELACK(tp); \ 430 } \ 431 } while (/*CONSTCOND*/0) 432 433 #define TCP_CLEAR_DELACK(tp) \ 434 do { \ 435 if ((tp)->t_flags & TF_DELACK) { \ 436 (tp)->t_flags &= ~TF_DELACK; \ 437 callout_stop(&(tp)->t_delack_ch); \ 438 } \ 439 } while (/*CONSTCOND*/0) 440 #endif /* _KERNEL */ 441 442 /* 443 * Compute the current timestamp for a connection. 444 */ 445 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase) 446 447 /* 448 * Handy way of passing around TCP option info. 449 */ 450 struct tcp_opt_info { 451 int ts_present; 452 u_int32_t ts_val; 453 u_int32_t ts_ecr; 454 u_int16_t maxseg; 455 }; 456 457 #define TOF_SIGNATURE 0x0040 /* signature option present */ 458 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */ 459 460 /* 461 * Data for the TCP compressed state engine. 462 */ 463 union syn_cache_sa { 464 struct sockaddr sa; 465 struct sockaddr_in sin; 466 #if 1 /*def INET6*/ 467 struct sockaddr_in6 sin6; 468 #endif 469 }; 470 471 struct syn_cache { 472 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */ 473 struct callout sc_timer; /* rexmt timer */ 474 union { /* cached route */ 475 struct route route4; 476 #ifdef INET6 477 struct route_in6 route6; 478 #endif 479 } sc_route_u; 480 #define sc_route4 sc_route_u.route4 481 #ifdef INET6 482 #define sc_route6 sc_route_u.route6 483 #endif 484 long sc_win; /* advertised window */ 485 int sc_bucketidx; /* our bucket index */ 486 u_int32_t sc_hash; 487 u_int32_t sc_timestamp; /* timestamp from SYN */ 488 u_int32_t sc_timebase; /* our local timebase */ 489 union syn_cache_sa sc_src; 490 union syn_cache_sa sc_dst; 491 tcp_seq sc_irs; 492 tcp_seq sc_iss; 493 u_int sc_rxtcur; /* current rxt timeout */ 494 u_int sc_rxttot; /* total time spend on queues */ 495 u_short sc_rxtshift; /* for computing backoff */ 496 u_short sc_flags; 497 498 #define SCF_UNREACH 0x0001 /* we've had an unreach error */ 499 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */ 500 #define SCF_DEAD 0x0004 /* this entry to be released */ 501 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */ 502 #define SCF_ECN_PERMIT 0x0010 /* peer will do ECN */ 503 #define SCF_SIGNATURE 0x40 /* send MD5 digests */ 504 505 struct mbuf *sc_ipopts; /* IP options */ 506 u_int16_t sc_peermaxseg; 507 u_int16_t sc_ourmaxseg; 508 u_int8_t sc_request_r_scale : 4, 509 sc_requested_s_scale : 4; 510 511 struct tcpcb *sc_tp; /* tcb for listening socket */ 512 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */ 513 }; 514 515 struct syn_cache_head { 516 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */ 517 u_short sch_length; /* # entries in bucket */ 518 }; 519 520 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 521 #ifdef INET6 522 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb) 523 #endif 524 #ifndef INET6 525 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 526 #else 527 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \ 528 ? intotcpcb(sotoinpcb(so)) \ 529 : in6totcpcb(sotoin6pcb(so))) 530 #endif 531 532 /* 533 * The smoothed round-trip time and estimated variance 534 * are stored as fixed point numbers scaled by the values below. 535 * For convenience, these scales are also used in smoothing the average 536 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 537 * With these scales, srtt has 3 bits to the right of the binary point, 538 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 539 * binary point, and is smoothed with an ALPHA of 0.75. 540 */ 541 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 542 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 543 544 /* 545 * The initial retransmission should happen at rtt + 4 * rttvar. 546 * Because of the way we do the smoothing, srtt and rttvar 547 * will each average +1/2 tick of bias. When we compute 548 * the retransmit timer, we want 1/2 tick of rounding and 549 * 1 extra tick because of +-1/2 tick uncertainty in the 550 * firing of the timer. The bias will give us exactly the 551 * 1.5 tick we need. But, because the bias is 552 * statistical, we have to test that we don't drop below 553 * the minimum feasible timer (which is 2 ticks). 554 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 555 * is the same as the multiplier for rttvar. 556 */ 557 #define TCP_REXMTVAL(tp) \ 558 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 559 560 /* 561 * Compute the initial window for slow start. 562 */ 563 #define TCP_INITIAL_WINDOW(iw, segsz) \ 564 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 565 ((segsz) * (iw))) 566 567 /* 568 * TCP statistics. 569 * Many of these should be kept per connection, 570 * but that's inconvenient at the moment. 571 */ 572 struct tcpstat { 573 u_quad_t tcps_connattempt; /* connections initiated */ 574 u_quad_t tcps_accepts; /* connections accepted */ 575 u_quad_t tcps_connects; /* connections established */ 576 u_quad_t tcps_drops; /* connections dropped */ 577 u_quad_t tcps_conndrops; /* embryonic connections dropped */ 578 u_quad_t tcps_closed; /* conn. closed (includes drops) */ 579 u_quad_t tcps_segstimed; /* segs where we tried to get rtt */ 580 u_quad_t tcps_rttupdated; /* times we succeeded */ 581 u_quad_t tcps_delack; /* delayed acks sent */ 582 u_quad_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 583 u_quad_t tcps_rexmttimeo; /* retransmit timeouts */ 584 u_quad_t tcps_persisttimeo; /* persist timeouts */ 585 u_quad_t tcps_keeptimeo; /* keepalive timeouts */ 586 u_quad_t tcps_keepprobe; /* keepalive probes sent */ 587 u_quad_t tcps_keepdrops; /* connections dropped in keepalive */ 588 u_quad_t tcps_persistdrops; /* connections dropped in persist */ 589 u_quad_t tcps_connsdrained; /* connections drained due to memory 590 shortage */ 591 u_quad_t tcps_pmtublackhole; /* PMTUD blackhole detected */ 592 593 u_quad_t tcps_sndtotal; /* total packets sent */ 594 u_quad_t tcps_sndpack; /* data packets sent */ 595 u_quad_t tcps_sndbyte; /* data bytes sent */ 596 u_quad_t tcps_sndrexmitpack; /* data packets retransmitted */ 597 u_quad_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 598 u_quad_t tcps_sndacks; /* ack-only packets sent */ 599 u_quad_t tcps_sndprobe; /* window probes sent */ 600 u_quad_t tcps_sndurg; /* packets sent with URG only */ 601 u_quad_t tcps_sndwinup; /* window update-only packets sent */ 602 u_quad_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 603 604 u_quad_t tcps_rcvtotal; /* total packets received */ 605 u_quad_t tcps_rcvpack; /* packets received in sequence */ 606 u_quad_t tcps_rcvbyte; /* bytes received in sequence */ 607 u_quad_t tcps_rcvbadsum; /* packets received with ccksum errs */ 608 u_quad_t tcps_rcvbadoff; /* packets received with bad offset */ 609 u_quad_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 610 u_quad_t tcps_rcvshort; /* packets received too short */ 611 u_quad_t tcps_rcvduppack; /* duplicate-only packets received */ 612 u_quad_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 613 u_quad_t tcps_rcvpartduppack; /* packets with some duplicate data */ 614 u_quad_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 615 u_quad_t tcps_rcvoopack; /* out-of-order packets received */ 616 u_quad_t tcps_rcvoobyte; /* out-of-order bytes received */ 617 u_quad_t tcps_rcvpackafterwin; /* packets with data after window */ 618 u_quad_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 619 u_quad_t tcps_rcvafterclose; /* packets rcvd after "close" */ 620 u_quad_t tcps_rcvwinprobe; /* rcvd window probe packets */ 621 u_quad_t tcps_rcvdupack; /* rcvd duplicate acks */ 622 u_quad_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 623 u_quad_t tcps_rcvackpack; /* rcvd ack packets */ 624 u_quad_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 625 u_quad_t tcps_rcvwinupd; /* rcvd window update packets */ 626 u_quad_t tcps_pawsdrop; /* segments dropped due to PAWS */ 627 u_quad_t tcps_predack; /* times hdr predict ok for acks */ 628 u_quad_t tcps_preddat; /* times hdr predict ok for data pkts */ 629 630 u_quad_t tcps_pcbhashmiss; /* input packets missing pcb hash */ 631 u_quad_t tcps_noport; /* no socket on port */ 632 u_quad_t tcps_badsyn; /* received ack for which we have 633 no SYN in compressed state */ 634 u_quad_t tcps_delayed_free; /* delayed pool_put() of tcpcb */ 635 636 /* These statistics deal with the SYN cache. */ 637 u_quad_t tcps_sc_added; /* # of entries added */ 638 u_quad_t tcps_sc_completed; /* # of connections completed */ 639 u_quad_t tcps_sc_timed_out; /* # of entries timed out */ 640 u_quad_t tcps_sc_overflowed; /* # dropped due to overflow */ 641 u_quad_t tcps_sc_reset; /* # dropped due to RST */ 642 u_quad_t tcps_sc_unreach; /* # dropped due to ICMP unreach */ 643 u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */ 644 u_quad_t tcps_sc_aborted; /* # of entries aborted (no mem) */ 645 u_quad_t tcps_sc_dupesyn; /* # of duplicate SYNs received */ 646 u_quad_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 647 u_quad_t tcps_sc_collisions; /* # of hash collisions */ 648 u_quad_t tcps_sc_retransmitted; /* # of retransmissions */ 649 u_quad_t tcps_sc_delayed_free; /* # of delayed pool_put()s */ 650 651 u_quad_t tcps_selfquench; /* # of ENOBUFS we get on output */ 652 u_quad_t tcps_badsig; /* # of drops due to bad signature */ 653 u_quad_t tcps_goodsig; /* # of packets with good signature */ 654 655 u_quad_t tcps_ecn_shs; /* # of sucessful ECN handshakes */ 656 u_quad_t tcps_ecn_ce; /* # of packets with CE bit */ 657 u_quad_t tcps_ecn_ect; /* # of packets with ECT(0) bit */ 658 }; 659 660 /* 661 * Names for TCP sysctl objects. 662 */ 663 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 664 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 665 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 666 #define TCPCTL_MSSDFLT 4 /* default seg size */ 667 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 668 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 669 #if 0 /*obsoleted*/ 670 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 671 #endif 672 #define TCPCTL_INIT_WIN 8 /* initial window */ 673 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */ 674 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */ 675 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */ 676 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */ 677 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */ 678 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */ 679 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */ 680 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */ 681 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */ 682 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */ 683 #define TCPCTL_KEEPCNT 19 /* keepalive count */ 684 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */ 685 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */ 686 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */ 687 #if 0 /*obsoleted*/ 688 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */ 689 #endif 690 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */ 691 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */ 692 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */ 693 #define TCPCTL_IDENT 27 /* rfc 931 identd */ 694 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */ 695 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */ 696 #define TCPCTL_STATS 30 /* TCP statistics */ 697 #define TCPCTL_DEBUG 31 /* TCP debug sockets */ 698 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */ 699 #define TCPCTL_MAXID 33 700 701 #define TCPCTL_NAMES { \ 702 { 0, 0 }, \ 703 { "rfc1323", CTLTYPE_INT }, \ 704 { "sendspace", CTLTYPE_INT }, \ 705 { "recvspace", CTLTYPE_INT }, \ 706 { "mssdflt", CTLTYPE_INT }, \ 707 { "syn_cache_limit", CTLTYPE_INT }, \ 708 { "syn_bucket_limit", CTLTYPE_INT }, \ 709 { 0, 0 },\ 710 { "init_win", CTLTYPE_INT }, \ 711 { "mss_ifmtu", CTLTYPE_INT }, \ 712 { "sack", CTLTYPE_INT }, \ 713 { "win_scale", CTLTYPE_INT }, \ 714 { "timestamps", CTLTYPE_INT }, \ 715 { "compat_42", CTLTYPE_INT }, \ 716 { "cwm", CTLTYPE_INT }, \ 717 { "cwm_burstsize", CTLTYPE_INT }, \ 718 { "ack_on_push", CTLTYPE_INT }, \ 719 { "keepidle", CTLTYPE_INT }, \ 720 { "keepintvl", CTLTYPE_INT }, \ 721 { "keepcnt", CTLTYPE_INT }, \ 722 { "slowhz", CTLTYPE_INT }, \ 723 { 0, 0 }, \ 724 { "log_refused",CTLTYPE_INT }, \ 725 { 0, 0 }, \ 726 { "rstppslimit", CTLTYPE_INT }, \ 727 { "delack_ticks", CTLTYPE_INT }, \ 728 { "init_win_local", CTLTYPE_INT }, \ 729 { "ident", CTLTYPE_STRUCT }, \ 730 { "ackdropppslimit", CTLTYPE_INT }, \ 731 { "do_loopback_cksum", CTLTYPE_INT }, \ 732 { "stats", CTLTYPE_STRUCT }, \ 733 { "debug", CTLTYPE_STRUCT }, \ 734 { "debx", CTLTYPE_INT }, \ 735 } 736 737 #ifdef _KERNEL 738 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 739 extern struct tcpstat tcpstat; /* tcp statistics */ 740 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 741 extern int tcp_do_rfc1323; /* enabled/disabled? */ 742 extern int tcp_do_sack; /* SACK enabled/disabled? */ 743 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */ 744 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */ 745 extern int tcp_mssdflt; /* default seg size */ 746 extern int tcp_init_win; /* initial window */ 747 extern int tcp_init_win_local; /* initial window for local nets */ 748 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */ 749 extern int tcp_compat_42; /* work around ancient broken TCP peers */ 750 extern int tcp_cwm; /* enable Congestion Window Monitoring */ 751 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */ 752 extern int tcp_ack_on_push; /* ACK immediately on PUSH */ 753 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 754 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 755 extern int tcp_log_refused; /* log refused connections */ 756 extern int tcp_do_ecn; /* TCP ECN enabled/disabled? */ 757 extern int tcp_ecn_maxretries; /* Max ECN setup retries */ 758 #if NRND > 0 759 extern int tcp_do_rfc1948; /* ISS by cryptographic hash */ 760 #endif 761 extern int tcp_sack_tp_maxholes; /* Max holes per connection. */ 762 extern int tcp_sack_globalmaxholes; /* Max holes per system. */ 763 extern int tcp_sack_globalholes; /* Number of holes present. */ 764 extern int tcp_do_abc; /* RFC3465 ABC enabled/disabled? */ 765 extern int tcp_abc_aggressive; /* 1: L=2*SMSS 0: L=1*SMSS */ 766 767 extern int tcp_rst_ppslim; 768 extern int tcp_ackdrop_ppslim; 769 770 extern int tcp_syn_cache_size; 771 extern struct syn_cache_head tcp_syn_cache[]; 772 extern u_long syn_cache_count; 773 774 #ifdef MBUFTRACE 775 extern struct mowner tcp_rx_mowner; 776 extern struct mowner tcp_tx_mowner; 777 extern struct mowner tcp_mowner; 778 #endif 779 780 #define TCPCTL_VARIABLES { \ 781 { 0 }, \ 782 { 1, 0, &tcp_do_rfc1323 }, \ 783 { 1, 0, &tcp_sendspace }, \ 784 { 1, 0, &tcp_recvspace }, \ 785 { 1, 0, &tcp_mssdflt }, \ 786 { 1, 0, &tcp_syn_cache_limit }, \ 787 { 1, 0, &tcp_syn_bucket_limit }, \ 788 { 0 }, \ 789 { 1, 0, &tcp_init_win }, \ 790 { 1, 0, &tcp_mss_ifmtu }, \ 791 { 1, 0, &tcp_do_sack }, \ 792 { 1, 0, &tcp_do_win_scale }, \ 793 { 1, 0, &tcp_do_timestamps }, \ 794 { 1, 0, &tcp_compat_42 }, \ 795 { 1, 0, &tcp_cwm }, \ 796 { 1, 0, &tcp_cwm_burstsize }, \ 797 { 1, 0, &tcp_ack_on_push }, \ 798 { 1, 0, &tcp_keepidle }, \ 799 { 1, 0, &tcp_keepintvl }, \ 800 { 1, 0, &tcp_keepcnt }, \ 801 { 1, 1, 0, PR_SLOWHZ }, \ 802 { 0 }, \ 803 { 1, 0, &tcp_log_refused }, \ 804 { 0 }, \ 805 { 1, 0, &tcp_rst_ppslim }, \ 806 { 1, 0, &tcp_delack_ticks }, \ 807 { 1, 0, &tcp_init_win_local }, \ 808 { 1, 0, &tcp_ackdrop_ppslim }, \ 809 } 810 811 #ifdef __NO_STRICT_ALIGNMENT 812 #define TCP_HDR_ALIGNED_P(th) 1 813 #else 814 #define TCP_HDR_ALIGNED_P(th) ((((vaddr_t)(th)) & 3) == 0) 815 #endif 816 817 struct secasvar; 818 819 int tcp_attach(struct socket *); 820 void tcp_canceltimers(struct tcpcb *); 821 int tcp_timers_invoking(struct tcpcb*); 822 struct tcpcb * 823 tcp_close(struct tcpcb *); 824 int tcp_isdead(struct tcpcb *); 825 #ifdef INET6 826 void tcp6_ctlinput(int, struct sockaddr *, void *); 827 #endif 828 void *tcp_ctlinput(int, struct sockaddr *, void *); 829 int tcp_ctloutput(int, struct socket *, int, int, struct mbuf **); 830 struct tcpcb * 831 tcp_disconnect(struct tcpcb *); 832 struct tcpcb * 833 tcp_drop(struct tcpcb *, int); 834 #ifdef TCP_SIGNATURE 835 int tcp_signature_apply(void *, caddr_t, u_int); 836 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *); 837 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *, 838 char *); 839 #endif 840 void tcp_drain(void); 841 void tcp_established(struct tcpcb *); 842 void tcp_init(void); 843 #ifdef INET6 844 int tcp6_input(struct mbuf **, int *, int); 845 #endif 846 void tcp_input(struct mbuf *, ...); 847 u_int tcp_hdrsz(struct tcpcb *); 848 u_long tcp_mss_to_advertise(const struct ifnet *, int); 849 void tcp_mss_from_peer(struct tcpcb *, int); 850 void tcp_tcpcb_template(void); 851 struct tcpcb * 852 tcp_newtcpcb(int, void *); 853 void tcp_notify(struct inpcb *, int); 854 #ifdef INET6 855 void tcp6_notify(struct in6pcb *, int); 856 #endif 857 u_int tcp_optlen(struct tcpcb *); 858 int tcp_output(struct tcpcb *); 859 void tcp_pulloutofband(struct socket *, 860 struct tcphdr *, struct mbuf *, int); 861 void tcp_quench(struct inpcb *, int); 862 #ifdef INET6 863 void tcp6_quench(struct in6pcb *, int); 864 #endif 865 void tcp_mtudisc(struct inpcb *, int); 866 867 struct ipqent *tcpipqent_alloc(void); 868 void tcpipqent_free(struct ipqent *); 869 870 int tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *); 871 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *, 872 struct tcphdr *, tcp_seq, tcp_seq, int); 873 void tcp_rmx_rtt(struct tcpcb *); 874 void tcp_setpersist(struct tcpcb *); 875 #ifdef TCP_SIGNATURE 876 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int, 877 int, u_char *, u_int); 878 #endif 879 void tcp_slowtimo(void); 880 struct mbuf * 881 tcp_template(struct tcpcb *); 882 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int); 883 struct tcpcb * 884 tcp_usrclosed(struct tcpcb *); 885 int tcp_usrreq(struct socket *, 886 int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *); 887 void tcp_xmit_timer(struct tcpcb *, uint32_t); 888 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq); 889 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t, 890 tcp_seq); 891 892 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t); 893 void tcp_sack_option(struct tcpcb *, const struct tcphdr *, 894 const u_char *, int); 895 void tcp_del_sackholes(struct tcpcb *, const struct tcphdr *); 896 void tcp_free_sackholes(struct tcpcb *); 897 void tcp_sack_adjust(struct tcpcb *tp); 898 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 899 void tcp_sack_newack(struct tcpcb *, const struct tcphdr *); 900 int tcp_sack_numblks(const struct tcpcb *); 901 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2) 902 903 int syn_cache_add(struct sockaddr *, struct sockaddr *, 904 struct tcphdr *, unsigned int, struct socket *, 905 struct mbuf *, u_char *, int, struct tcp_opt_info *); 906 void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *, 907 struct tcphdr *); 908 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *, 909 struct tcphdr *, unsigned int, unsigned int, 910 struct socket *so, struct mbuf *); 911 void syn_cache_init(void); 912 void syn_cache_insert(struct syn_cache *, struct tcpcb *); 913 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *, 914 struct syn_cache_head **); 915 void syn_cache_reset(struct sockaddr *, struct sockaddr *, 916 struct tcphdr *); 917 int syn_cache_respond(struct syn_cache *, struct mbuf *); 918 void syn_cache_timer(void *); 919 void syn_cache_cleanup(struct tcpcb *); 920 921 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int, 922 int); 923 #endif 924 925 #endif /* !_NETINET_TCP_VAR_H_ */ 926