1 /* $NetBSD: tcp_var.h,v 1.138 2006/10/09 16:27:07 rpaulo 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_ECN_PERMIT 0x10000 /* other side said is ECN-ready */ 221 #define TF_ECN_SND_CWR 0x20000 /* ECN CWR in queue */ 222 #define TF_ECN_SND_ECE 0x40000 /* ECN ECE in queue */ 223 #define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */ 224 225 226 struct mbuf *t_template; /* skeletal packet for transmit */ 227 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 228 struct in6pcb *t_in6pcb; /* back pointer to internet pcb */ 229 struct callout t_delack_ch; /* delayed ACK callout */ 230 /* 231 * The following fields are used as in the protocol specification. 232 * See RFC793, Dec. 1981, page 21. 233 */ 234 /* send sequence variables */ 235 tcp_seq snd_una; /* send unacknowledged */ 236 tcp_seq snd_nxt; /* send next */ 237 tcp_seq snd_up; /* send urgent pointer */ 238 tcp_seq snd_wl1; /* window update seg seq number */ 239 tcp_seq snd_wl2; /* window update seg ack number */ 240 tcp_seq iss; /* initial send sequence number */ 241 u_long snd_wnd; /* send window */ 242 tcp_seq snd_recover; /* for use in fast recovery */ 243 tcp_seq snd_high; /* NewReno false fast rexmit seq */ 244 /* receive sequence variables */ 245 u_long rcv_wnd; /* receive window */ 246 tcp_seq rcv_nxt; /* receive next */ 247 tcp_seq rcv_up; /* receive urgent pointer */ 248 tcp_seq irs; /* initial receive sequence number */ 249 /* 250 * Additional variables for this implementation. 251 */ 252 /* receive variables */ 253 tcp_seq rcv_adv; /* advertised window */ 254 /* retransmit variables */ 255 tcp_seq snd_max; /* highest sequence number sent; 256 * used to recognize retransmits 257 */ 258 /* congestion control (for slow start, source quench, retransmit after loss) */ 259 u_long snd_cwnd; /* congestion-controlled window */ 260 u_long snd_ssthresh; /* snd_cwnd size threshhold for 261 * for slow start exponential to 262 * linear switch 263 */ 264 /* 265 * transmit timing stuff. See below for scale of srtt and rttvar. 266 * "Variance" is actually smoothed difference. 267 */ 268 uint32_t t_rcvtime; /* time last segment received */ 269 uint32_t t_rtttime; /* time we started measuring rtt */ 270 tcp_seq t_rtseq; /* sequence number being timed */ 271 int32_t t_srtt; /* smoothed round-trip time */ 272 int32_t t_rttvar; /* variance in round-trip time */ 273 uint32_t t_rttmin; /* minimum rtt allowed */ 274 u_long max_sndwnd; /* largest window peer has offered */ 275 276 /* out-of-band data */ 277 char t_oobflags; /* have some */ 278 char t_iobc; /* input character */ 279 #define TCPOOB_HAVEDATA 0x01 280 #define TCPOOB_HADDATA 0x02 281 short t_softerror; /* possible error not yet reported */ 282 283 /* RFC 1323 variables */ 284 u_char snd_scale; /* window scaling for send window */ 285 u_char rcv_scale; /* window scaling for recv window */ 286 u_char request_r_scale; /* pending window scaling */ 287 u_char requested_s_scale; 288 u_int32_t ts_recent; /* timestamp echo data */ 289 u_int32_t ts_recent_age; /* when last updated */ 290 u_int32_t ts_timebase; /* our timebase */ 291 tcp_seq last_ack_sent; 292 293 /* SACK stuff */ 294 #define TCP_SACK_MAX 3 295 #define TCPSACK_NONE 0 296 #define TCPSACK_HAVED 1 297 u_char rcv_sack_flags; /* SACK flags. */ 298 struct sackblk rcv_dsack_block; /* RX D-SACK block. */ 299 struct ipqehead timeq; /* time sequenced queue. */ 300 struct sackhead snd_holes; /* TX SACK holes. */ 301 int snd_numholes; /* Number of TX SACK holes. */ 302 tcp_seq rcv_lastsack; /* last seq number(+1) sack'd by rcv'r*/ 303 tcp_seq sack_newdata; /* New data xmitted in this recovery 304 episode starts at this seq number*/ 305 tcp_seq snd_fack; /* FACK TCP. Forward-most data held by 306 peer. */ 307 308 /* pointer for syn cache entries*/ 309 LIST_HEAD(, syn_cache) t_sc; /* list of entries by this tcb */ 310 311 /* prediction of next mbuf when using large window sizes */ 312 struct mbuf *t_lastm; /* last mbuf that data was sent from */ 313 int t_inoff; /* data offset in previous mbuf */ 314 int t_lastoff; /* last data address in mbuf chain */ 315 int t_lastlen; /* last length read from mbuf chain */ 316 317 /* Path-MTU discovery blackhole detection */ 318 int t_mtudisc; /* perform mtudisc for this tcb */ 319 /* Path-MTU Discovery Information */ 320 u_int t_pmtud_mss_acked; /* MSS acked, lower bound for MTU */ 321 u_int t_pmtud_mtu_sent; /* MTU used, upper bound for MTU */ 322 tcp_seq t_pmtud_th_seq; /* TCP SEQ from ICMP payload */ 323 u_int t_pmtud_nextmtu; /* Advertised Next-Hop MTU from ICMP */ 324 u_short t_pmtud_ip_len; /* IP length from ICMP payload */ 325 u_short t_pmtud_ip_hl; /* IP header length from ICMP payload */ 326 327 uint8_t t_ecn_retries; /* # of ECN setup retries */ 328 329 struct tcp_congctl *t_congctl; /* per TCB congctl algorithm */ 330 }; 331 332 /* 333 * Macros to aid ECN TCP. 334 */ 335 #define TCP_ECN_ALLOWED(tp) (tp->t_flags & TF_ECN_PERMIT) 336 337 /* 338 * Macros to aid SACK/FACK TCP. 339 */ 340 #define TCP_SACK_ENABLED(tp) (tp->t_flags & TF_WILL_SACK) 341 #define TCP_FACK_FASTRECOV(tp) \ 342 (TCP_SACK_ENABLED(tp) && \ 343 (SEQ_GT(tp->snd_fack, tp->snd_una + tcprexmtthresh * tp->t_segsz))) 344 345 #ifdef _KERNEL 346 /* 347 * TCP reassembly queue locks. 348 */ 349 static __inline int tcp_reass_lock_try (struct tcpcb *) 350 __attribute__((__unused__)); 351 static __inline void tcp_reass_unlock (struct tcpcb *) 352 __attribute__((__unused__)); 353 354 static __inline int 355 tcp_reass_lock_try(tp) 356 struct tcpcb *tp; 357 { 358 int s; 359 360 /* 361 * Use splvm() -- we're blocking things that would cause 362 * mbuf allocation. 363 */ 364 s = splvm(); 365 if (tp->t_flags & TF_REASSEMBLING) { 366 splx(s); 367 return (0); 368 } 369 tp->t_flags |= TF_REASSEMBLING; 370 splx(s); 371 return (1); 372 } 373 374 static __inline void 375 tcp_reass_unlock(tp) 376 struct tcpcb *tp; 377 { 378 int s; 379 380 s = splvm(); 381 tp->t_flags &= ~TF_REASSEMBLING; 382 splx(s); 383 } 384 385 #ifdef DIAGNOSTIC 386 #define TCP_REASS_LOCK(tp) \ 387 do { \ 388 if (tcp_reass_lock_try(tp) == 0) { \ 389 printf("%s:%d: tcpcb %p reass already locked\n", \ 390 __FILE__, __LINE__, tp); \ 391 panic("tcp_reass_lock"); \ 392 } \ 393 } while (/*CONSTCOND*/ 0) 394 #define TCP_REASS_LOCK_CHECK(tp) \ 395 do { \ 396 if (((tp)->t_flags & TF_REASSEMBLING) == 0) { \ 397 printf("%s:%d: tcpcb %p reass lock not held\n", \ 398 __FILE__, __LINE__, tp); \ 399 panic("tcp reass lock check"); \ 400 } \ 401 } while (/*CONSTCOND*/ 0) 402 #else 403 #define TCP_REASS_LOCK(tp) (void) tcp_reass_lock_try((tp)) 404 #define TCP_REASS_LOCK_CHECK(tp) /* nothing */ 405 #endif 406 407 #define TCP_REASS_UNLOCK(tp) tcp_reass_unlock((tp)) 408 #endif /* _KERNEL */ 409 410 /* 411 * Queue for delayed ACK processing. 412 */ 413 #ifdef _KERNEL 414 extern int tcp_delack_ticks; 415 void tcp_delack(void *); 416 417 #define TCP_RESTART_DELACK(tp) \ 418 callout_reset(&(tp)->t_delack_ch, tcp_delack_ticks, \ 419 tcp_delack, tp) 420 421 #define TCP_SET_DELACK(tp) \ 422 do { \ 423 if (((tp)->t_flags & TF_DELACK) == 0) { \ 424 (tp)->t_flags |= TF_DELACK; \ 425 TCP_RESTART_DELACK(tp); \ 426 } \ 427 } while (/*CONSTCOND*/0) 428 429 #define TCP_CLEAR_DELACK(tp) \ 430 do { \ 431 if ((tp)->t_flags & TF_DELACK) { \ 432 (tp)->t_flags &= ~TF_DELACK; \ 433 callout_stop(&(tp)->t_delack_ch); \ 434 } \ 435 } while (/*CONSTCOND*/0) 436 #endif /* _KERNEL */ 437 438 /* 439 * Compute the current timestamp for a connection. 440 */ 441 #define TCP_TIMESTAMP(tp) (tcp_now - (tp)->ts_timebase) 442 443 /* 444 * Handy way of passing around TCP option info. 445 */ 446 struct tcp_opt_info { 447 int ts_present; 448 u_int32_t ts_val; 449 u_int32_t ts_ecr; 450 u_int16_t maxseg; 451 }; 452 453 #define TOF_SIGNATURE 0x0040 /* signature option present */ 454 #define TOF_SIGLEN 0x0080 /* sigature length valid (RFC2385) */ 455 456 /* 457 * Data for the TCP compressed state engine. 458 */ 459 union syn_cache_sa { 460 struct sockaddr sa; 461 struct sockaddr_in sin; 462 #if 1 /*def INET6*/ 463 struct sockaddr_in6 sin6; 464 #endif 465 }; 466 467 struct syn_cache { 468 TAILQ_ENTRY(syn_cache) sc_bucketq; /* link on bucket list */ 469 struct callout sc_timer; /* rexmt timer */ 470 union { /* cached route */ 471 struct route route4; 472 #ifdef INET6 473 struct route_in6 route6; 474 #endif 475 } sc_route_u; 476 #define sc_route4 sc_route_u.route4 477 #ifdef INET6 478 #define sc_route6 sc_route_u.route6 479 #endif 480 long sc_win; /* advertised window */ 481 int sc_bucketidx; /* our bucket index */ 482 u_int32_t sc_hash; 483 u_int32_t sc_timestamp; /* timestamp from SYN */ 484 u_int32_t sc_timebase; /* our local timebase */ 485 union syn_cache_sa sc_src; 486 union syn_cache_sa sc_dst; 487 tcp_seq sc_irs; 488 tcp_seq sc_iss; 489 u_int sc_rxtcur; /* current rxt timeout */ 490 u_int sc_rxttot; /* total time spend on queues */ 491 u_short sc_rxtshift; /* for computing backoff */ 492 u_short sc_flags; 493 494 #define SCF_UNREACH 0x0001 /* we've had an unreach error */ 495 #define SCF_TIMESTAMP 0x0002 /* peer will do timestamps */ 496 #define SCF_DEAD 0x0004 /* this entry to be released */ 497 #define SCF_SACK_PERMIT 0x0008 /* peer will do SACK */ 498 #define SCF_ECN_PERMIT 0x0010 /* peer will do ECN */ 499 #define SCF_SIGNATURE 0x40 /* send MD5 digests */ 500 501 struct mbuf *sc_ipopts; /* IP options */ 502 u_int16_t sc_peermaxseg; 503 u_int16_t sc_ourmaxseg; 504 u_int8_t sc_request_r_scale : 4, 505 sc_requested_s_scale : 4; 506 507 struct tcpcb *sc_tp; /* tcb for listening socket */ 508 LIST_ENTRY(syn_cache) sc_tpq; /* list of entries by same tp */ 509 }; 510 511 struct syn_cache_head { 512 TAILQ_HEAD(, syn_cache) sch_bucket; /* bucket entries */ 513 u_short sch_length; /* # entries in bucket */ 514 }; 515 516 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 517 #ifdef INET6 518 #define in6totcpcb(ip) ((struct tcpcb *)(ip)->in6p_ppcb) 519 #endif 520 #ifndef INET6 521 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 522 #else 523 #define sototcpcb(so) (((so)->so_proto->pr_domain->dom_family == AF_INET) \ 524 ? intotcpcb(sotoinpcb(so)) \ 525 : in6totcpcb(sotoin6pcb(so))) 526 #endif 527 528 /* 529 * The smoothed round-trip time and estimated variance 530 * are stored as fixed point numbers scaled by the values below. 531 * For convenience, these scales are also used in smoothing the average 532 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 533 * With these scales, srtt has 3 bits to the right of the binary point, 534 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 535 * binary point, and is smoothed with an ALPHA of 0.75. 536 */ 537 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 538 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 539 540 /* 541 * The initial retransmission should happen at rtt + 4 * rttvar. 542 * Because of the way we do the smoothing, srtt and rttvar 543 * will each average +1/2 tick of bias. When we compute 544 * the retransmit timer, we want 1/2 tick of rounding and 545 * 1 extra tick because of +-1/2 tick uncertainty in the 546 * firing of the timer. The bias will give us exactly the 547 * 1.5 tick we need. But, because the bias is 548 * statistical, we have to test that we don't drop below 549 * the minimum feasible timer (which is 2 ticks). 550 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 551 * is the same as the multiplier for rttvar. 552 */ 553 #define TCP_REXMTVAL(tp) \ 554 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 555 556 /* 557 * Compute the initial window for slow start. 558 */ 559 #define TCP_INITIAL_WINDOW(iw, segsz) \ 560 (((iw) == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 561 ((segsz) * (iw))) 562 563 /* 564 * TCP statistics. 565 * Many of these should be kept per connection, 566 * but that's inconvenient at the moment. 567 */ 568 struct tcpstat { 569 u_quad_t tcps_connattempt; /* connections initiated */ 570 u_quad_t tcps_accepts; /* connections accepted */ 571 u_quad_t tcps_connects; /* connections established */ 572 u_quad_t tcps_drops; /* connections dropped */ 573 u_quad_t tcps_conndrops; /* embryonic connections dropped */ 574 u_quad_t tcps_closed; /* conn. closed (includes drops) */ 575 u_quad_t tcps_segstimed; /* segs where we tried to get rtt */ 576 u_quad_t tcps_rttupdated; /* times we succeeded */ 577 u_quad_t tcps_delack; /* delayed acks sent */ 578 u_quad_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 579 u_quad_t tcps_rexmttimeo; /* retransmit timeouts */ 580 u_quad_t tcps_persisttimeo; /* persist timeouts */ 581 u_quad_t tcps_keeptimeo; /* keepalive timeouts */ 582 u_quad_t tcps_keepprobe; /* keepalive probes sent */ 583 u_quad_t tcps_keepdrops; /* connections dropped in keepalive */ 584 u_quad_t tcps_persistdrops; /* connections dropped in persist */ 585 u_quad_t tcps_connsdrained; /* connections drained due to memory 586 shortage */ 587 u_quad_t tcps_pmtublackhole; /* PMTUD blackhole detected */ 588 589 u_quad_t tcps_sndtotal; /* total packets sent */ 590 u_quad_t tcps_sndpack; /* data packets sent */ 591 u_quad_t tcps_sndbyte; /* data bytes sent */ 592 u_quad_t tcps_sndrexmitpack; /* data packets retransmitted */ 593 u_quad_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 594 u_quad_t tcps_sndacks; /* ack-only packets sent */ 595 u_quad_t tcps_sndprobe; /* window probes sent */ 596 u_quad_t tcps_sndurg; /* packets sent with URG only */ 597 u_quad_t tcps_sndwinup; /* window update-only packets sent */ 598 u_quad_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 599 600 u_quad_t tcps_rcvtotal; /* total packets received */ 601 u_quad_t tcps_rcvpack; /* packets received in sequence */ 602 u_quad_t tcps_rcvbyte; /* bytes received in sequence */ 603 u_quad_t tcps_rcvbadsum; /* packets received with ccksum errs */ 604 u_quad_t tcps_rcvbadoff; /* packets received with bad offset */ 605 u_quad_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 606 u_quad_t tcps_rcvshort; /* packets received too short */ 607 u_quad_t tcps_rcvduppack; /* duplicate-only packets received */ 608 u_quad_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 609 u_quad_t tcps_rcvpartduppack; /* packets with some duplicate data */ 610 u_quad_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 611 u_quad_t tcps_rcvoopack; /* out-of-order packets received */ 612 u_quad_t tcps_rcvoobyte; /* out-of-order bytes received */ 613 u_quad_t tcps_rcvpackafterwin; /* packets with data after window */ 614 u_quad_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 615 u_quad_t tcps_rcvafterclose; /* packets rcvd after "close" */ 616 u_quad_t tcps_rcvwinprobe; /* rcvd window probe packets */ 617 u_quad_t tcps_rcvdupack; /* rcvd duplicate acks */ 618 u_quad_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 619 u_quad_t tcps_rcvackpack; /* rcvd ack packets */ 620 u_quad_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 621 u_quad_t tcps_rcvwinupd; /* rcvd window update packets */ 622 u_quad_t tcps_pawsdrop; /* segments dropped due to PAWS */ 623 u_quad_t tcps_predack; /* times hdr predict ok for acks */ 624 u_quad_t tcps_preddat; /* times hdr predict ok for data pkts */ 625 626 u_quad_t tcps_pcbhashmiss; /* input packets missing pcb hash */ 627 u_quad_t tcps_noport; /* no socket on port */ 628 u_quad_t tcps_badsyn; /* received ack for which we have 629 no SYN in compressed state */ 630 u_quad_t tcps_delayed_free; /* delayed pool_put() of tcpcb */ 631 632 /* These statistics deal with the SYN cache. */ 633 u_quad_t tcps_sc_added; /* # of entries added */ 634 u_quad_t tcps_sc_completed; /* # of connections completed */ 635 u_quad_t tcps_sc_timed_out; /* # of entries timed out */ 636 u_quad_t tcps_sc_overflowed; /* # dropped due to overflow */ 637 u_quad_t tcps_sc_reset; /* # dropped due to RST */ 638 u_quad_t tcps_sc_unreach; /* # dropped due to ICMP unreach */ 639 u_quad_t tcps_sc_bucketoverflow;/* # dropped due to bucket overflow */ 640 u_quad_t tcps_sc_aborted; /* # of entries aborted (no mem) */ 641 u_quad_t tcps_sc_dupesyn; /* # of duplicate SYNs received */ 642 u_quad_t tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 643 u_quad_t tcps_sc_collisions; /* # of hash collisions */ 644 u_quad_t tcps_sc_retransmitted; /* # of retransmissions */ 645 u_quad_t tcps_sc_delayed_free; /* # of delayed pool_put()s */ 646 647 u_quad_t tcps_selfquench; /* # of ENOBUFS we get on output */ 648 u_quad_t tcps_badsig; /* # of drops due to bad signature */ 649 u_quad_t tcps_goodsig; /* # of packets with good signature */ 650 651 u_quad_t tcps_ecn_shs; /* # of sucessful ECN handshakes */ 652 u_quad_t tcps_ecn_ce; /* # of packets with CE bit */ 653 u_quad_t tcps_ecn_ect; /* # of packets with ECT(0) bit */ 654 }; 655 656 /* 657 * Names for TCP sysctl objects. 658 */ 659 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 660 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 661 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 662 #define TCPCTL_MSSDFLT 4 /* default seg size */ 663 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 664 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 665 #if 0 /*obsoleted*/ 666 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 667 #endif 668 #define TCPCTL_INIT_WIN 8 /* initial window */ 669 #define TCPCTL_MSS_IFMTU 9 /* mss from interface, not in_maxmtu */ 670 #define TCPCTL_SACK 10 /* RFC2018 selective acknowledgement */ 671 #define TCPCTL_WSCALE 11 /* RFC1323 window scaling */ 672 #define TCPCTL_TSTAMP 12 /* RFC1323 timestamps */ 673 #define TCPCTL_COMPAT_42 13 /* 4.2BSD TCP bug work-arounds */ 674 #define TCPCTL_CWM 14 /* Congestion Window Monitoring */ 675 #define TCPCTL_CWM_BURSTSIZE 15 /* burst size allowed by CWM */ 676 #define TCPCTL_ACK_ON_PUSH 16 /* ACK immediately on PUSH */ 677 #define TCPCTL_KEEPIDLE 17 /* keepalive idle time */ 678 #define TCPCTL_KEEPINTVL 18 /* keepalive probe interval */ 679 #define TCPCTL_KEEPCNT 19 /* keepalive count */ 680 #define TCPCTL_SLOWHZ 20 /* PR_SLOWHZ (read-only) */ 681 #define TCPCTL_NEWRENO 21 /* NewReno Congestion Control */ 682 #define TCPCTL_LOG_REFUSED 22 /* Log refused connections */ 683 #if 0 /*obsoleted*/ 684 #define TCPCTL_RSTRATELIMIT 23 /* RST rate limit */ 685 #endif 686 #define TCPCTL_RSTPPSLIMIT 24 /* RST pps limit */ 687 #define TCPCTL_DELACK_TICKS 25 /* # ticks to delay ACK */ 688 #define TCPCTL_INIT_WIN_LOCAL 26 /* initial window for local nets */ 689 #define TCPCTL_IDENT 27 /* rfc 931 identd */ 690 #define TCPCTL_ACKDROPRATELIMIT 28 /* SYN/RST -> ACK rate limit */ 691 #define TCPCTL_LOOPBACKCKSUM 29 /* do TCP checksum on loopback */ 692 #define TCPCTL_STATS 30 /* TCP statistics */ 693 #define TCPCTL_DEBUG 31 /* TCP debug sockets */ 694 #define TCPCTL_DEBX 32 /* # of tcp debug sockets */ 695 #define TCPCTL_MAXID 33 696 697 #define TCPCTL_NAMES { \ 698 { 0, 0 }, \ 699 { "rfc1323", CTLTYPE_INT }, \ 700 { "sendspace", CTLTYPE_INT }, \ 701 { "recvspace", CTLTYPE_INT }, \ 702 { "mssdflt", CTLTYPE_INT }, \ 703 { "syn_cache_limit", CTLTYPE_INT }, \ 704 { "syn_bucket_limit", CTLTYPE_INT }, \ 705 { 0, 0 },\ 706 { "init_win", CTLTYPE_INT }, \ 707 { "mss_ifmtu", CTLTYPE_INT }, \ 708 { "sack", CTLTYPE_INT }, \ 709 { "win_scale", CTLTYPE_INT }, \ 710 { "timestamps", CTLTYPE_INT }, \ 711 { "compat_42", CTLTYPE_INT }, \ 712 { "cwm", CTLTYPE_INT }, \ 713 { "cwm_burstsize", CTLTYPE_INT }, \ 714 { "ack_on_push", CTLTYPE_INT }, \ 715 { "keepidle", CTLTYPE_INT }, \ 716 { "keepintvl", CTLTYPE_INT }, \ 717 { "keepcnt", CTLTYPE_INT }, \ 718 { "slowhz", CTLTYPE_INT }, \ 719 { 0, 0 }, \ 720 { "log_refused",CTLTYPE_INT }, \ 721 { 0, 0 }, \ 722 { "rstppslimit", CTLTYPE_INT }, \ 723 { "delack_ticks", CTLTYPE_INT }, \ 724 { "init_win_local", CTLTYPE_INT }, \ 725 { "ident", CTLTYPE_STRUCT }, \ 726 { "ackdropppslimit", CTLTYPE_INT }, \ 727 { "do_loopback_cksum", CTLTYPE_INT }, \ 728 { "stats", CTLTYPE_STRUCT }, \ 729 { "debug", CTLTYPE_STRUCT }, \ 730 { "debx", CTLTYPE_INT }, \ 731 } 732 733 #ifdef _KERNEL 734 extern struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 735 extern struct tcpstat tcpstat; /* tcp statistics */ 736 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 737 extern int tcp_do_rfc1323; /* enabled/disabled? */ 738 extern int tcp_do_sack; /* SACK enabled/disabled? */ 739 extern int tcp_do_win_scale; /* RFC1323 window scaling enabled/disabled? */ 740 extern int tcp_do_timestamps; /* RFC1323 timestamps enabled/disabled? */ 741 extern int tcp_mssdflt; /* default seg size */ 742 extern int tcp_init_win; /* initial window */ 743 extern int tcp_init_win_local; /* initial window for local nets */ 744 extern int tcp_mss_ifmtu; /* take MSS from interface, not in_maxmtu */ 745 extern int tcp_compat_42; /* work around ancient broken TCP peers */ 746 extern int tcp_cwm; /* enable Congestion Window Monitoring */ 747 extern int tcp_cwm_burstsize; /* burst size allowed by CWM */ 748 extern int tcp_ack_on_push; /* ACK immediately on PUSH */ 749 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 750 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 751 extern int tcp_log_refused; /* log refused connections */ 752 extern int tcp_do_ecn; /* TCP ECN enabled/disabled? */ 753 extern int tcp_ecn_maxretries; /* Max ECN setup retries */ 754 extern int tcp_sack_tp_maxholes; /* Max holes per connection. */ 755 extern int tcp_sack_globalmaxholes; /* Max holes per system. */ 756 extern int tcp_sack_globalholes; /* Number of holes present. */ 757 758 extern int tcp_rst_ppslim; 759 extern int tcp_ackdrop_ppslim; 760 761 extern int tcp_syn_cache_size; 762 extern struct syn_cache_head tcp_syn_cache[]; 763 extern u_long syn_cache_count; 764 765 #ifdef MBUFTRACE 766 extern struct mowner tcp_rx_mowner; 767 extern struct mowner tcp_tx_mowner; 768 extern struct mowner tcp_mowner; 769 #endif 770 771 #define TCPCTL_VARIABLES { \ 772 { 0 }, \ 773 { 1, 0, &tcp_do_rfc1323 }, \ 774 { 1, 0, &tcp_sendspace }, \ 775 { 1, 0, &tcp_recvspace }, \ 776 { 1, 0, &tcp_mssdflt }, \ 777 { 1, 0, &tcp_syn_cache_limit }, \ 778 { 1, 0, &tcp_syn_bucket_limit }, \ 779 { 0 }, \ 780 { 1, 0, &tcp_init_win }, \ 781 { 1, 0, &tcp_mss_ifmtu }, \ 782 { 1, 0, &tcp_do_sack }, \ 783 { 1, 0, &tcp_do_win_scale }, \ 784 { 1, 0, &tcp_do_timestamps }, \ 785 { 1, 0, &tcp_compat_42 }, \ 786 { 1, 0, &tcp_cwm }, \ 787 { 1, 0, &tcp_cwm_burstsize }, \ 788 { 1, 0, &tcp_ack_on_push }, \ 789 { 1, 0, &tcp_keepidle }, \ 790 { 1, 0, &tcp_keepintvl }, \ 791 { 1, 0, &tcp_keepcnt }, \ 792 { 1, 1, 0, PR_SLOWHZ }, \ 793 { 0 }, \ 794 { 1, 0, &tcp_log_refused }, \ 795 { 0 }, \ 796 { 1, 0, &tcp_rst_ppslim }, \ 797 { 1, 0, &tcp_delack_ticks }, \ 798 { 1, 0, &tcp_init_win_local }, \ 799 { 1, 0, &tcp_ackdrop_ppslim }, \ 800 } 801 802 #ifdef __NO_STRICT_ALIGNMENT 803 #define TCP_HDR_ALIGNED_P(th) 1 804 #else 805 #define TCP_HDR_ALIGNED_P(th) ((((vaddr_t)(th)) & 3) == 0) 806 #endif 807 808 struct secasvar; 809 810 int tcp_attach(struct socket *); 811 void tcp_canceltimers(struct tcpcb *); 812 int tcp_timers_invoking(struct tcpcb*); 813 struct tcpcb * 814 tcp_close(struct tcpcb *); 815 int tcp_isdead(struct tcpcb *); 816 #ifdef INET6 817 void tcp6_ctlinput(int, struct sockaddr *, void *); 818 #endif 819 void *tcp_ctlinput(int, struct sockaddr *, void *); 820 int tcp_ctloutput(int, struct socket *, int, int, struct mbuf **); 821 struct tcpcb * 822 tcp_disconnect(struct tcpcb *); 823 struct tcpcb * 824 tcp_drop(struct tcpcb *, int); 825 #ifdef TCP_SIGNATURE 826 int tcp_signature_apply(void *, caddr_t, u_int); 827 struct secasvar *tcp_signature_getsav(struct mbuf *, struct tcphdr *); 828 int tcp_signature(struct mbuf *, struct tcphdr *, int, struct secasvar *, 829 char *); 830 #endif 831 int tcp_dooptions(struct tcpcb *, u_char *, int, struct tcphdr *, 832 struct mbuf *, int, struct tcp_opt_info *); 833 void tcp_drain(void); 834 void tcp_established(struct tcpcb *); 835 void tcp_init(void); 836 #ifdef INET6 837 int tcp6_input(struct mbuf **, int *, int); 838 #endif 839 void tcp_input(struct mbuf *, ...); 840 u_int tcp_hdrsz(struct tcpcb *); 841 u_long tcp_mss_to_advertise(const struct ifnet *, int); 842 void tcp_mss_from_peer(struct tcpcb *, int); 843 void tcp_tcpcb_template(void); 844 struct tcpcb * 845 tcp_newtcpcb(int, void *); 846 void tcp_notify(struct inpcb *, int); 847 #ifdef INET6 848 void tcp6_notify(struct in6pcb *, int); 849 #endif 850 u_int tcp_optlen(struct tcpcb *); 851 int tcp_output(struct tcpcb *); 852 void tcp_pulloutofband(struct socket *, 853 struct tcphdr *, struct mbuf *, int); 854 void tcp_quench(struct inpcb *, int); 855 #ifdef INET6 856 void tcp6_quench(struct in6pcb *, int); 857 #endif 858 void tcp_mtudisc(struct inpcb *, int); 859 860 struct ipqent *tcpipqent_alloc(void); 861 void tcpipqent_free(struct ipqent *); 862 863 int tcp_reass(struct tcpcb *, struct tcphdr *, struct mbuf *, int *); 864 int tcp_respond(struct tcpcb *, struct mbuf *, struct mbuf *, 865 struct tcphdr *, tcp_seq, tcp_seq, int); 866 void tcp_rmx_rtt(struct tcpcb *); 867 void tcp_setpersist(struct tcpcb *); 868 #ifdef TCP_SIGNATURE 869 int tcp_signature_compute(struct mbuf *, struct tcphdr *, int, int, 870 int, u_char *, u_int); 871 #endif 872 void tcp_slowtimo(void); 873 struct mbuf * 874 tcp_template(struct tcpcb *); 875 void tcp_trace(short, short, struct tcpcb *, struct mbuf *, int); 876 struct tcpcb * 877 tcp_usrclosed(struct tcpcb *); 878 int tcp_usrreq(struct socket *, 879 int, struct mbuf *, struct mbuf *, struct mbuf *, struct lwp *); 880 void tcp_xmit_timer(struct tcpcb *, uint32_t); 881 tcp_seq tcp_new_iss(struct tcpcb *, tcp_seq); 882 tcp_seq tcp_new_iss1(void *, void *, u_int16_t, u_int16_t, size_t, 883 tcp_seq); 884 885 void tcp_new_dsack(struct tcpcb *, tcp_seq, u_int32_t); 886 void tcp_sack_option(struct tcpcb *, struct tcphdr *, u_char *, int); 887 void tcp_del_sackholes(struct tcpcb *, struct tcphdr *); 888 void tcp_free_sackholes(struct tcpcb *); 889 void tcp_sack_adjust(struct tcpcb *tp); 890 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 891 void tcp_sack_newack(struct tcpcb *, struct tcphdr *); 892 int tcp_sack_numblks(const struct tcpcb *); 893 #define TCP_SACK_OPTLEN(nblks) ((nblks) * 8 + 2 + 2) 894 895 int syn_cache_add(struct sockaddr *, struct sockaddr *, 896 struct tcphdr *, unsigned int, struct socket *, 897 struct mbuf *, u_char *, int, struct tcp_opt_info *); 898 void syn_cache_unreach(const struct sockaddr *, const struct sockaddr *, 899 struct tcphdr *); 900 struct socket *syn_cache_get(struct sockaddr *, struct sockaddr *, 901 struct tcphdr *, unsigned int, unsigned int, 902 struct socket *so, struct mbuf *); 903 void syn_cache_init(void); 904 void syn_cache_insert(struct syn_cache *, struct tcpcb *); 905 struct syn_cache *syn_cache_lookup(const struct sockaddr *, const struct sockaddr *, 906 struct syn_cache_head **); 907 void syn_cache_reset(struct sockaddr *, struct sockaddr *, 908 struct tcphdr *); 909 int syn_cache_respond(struct syn_cache *, struct mbuf *); 910 void syn_cache_timer(void *); 911 void syn_cache_cleanup(struct tcpcb *); 912 913 int tcp_input_checksum(int, struct mbuf *, const struct tcphdr *, int, int, 914 int); 915 #endif 916 917 #endif /* !_NETINET_TCP_VAR_H_ */ 918