bsd/netinet/tcp_var.h

   1 /*
   2  * Copyright (c) 2000-2007 Apple Inc. All rights reserved.
   3  *
   4  * @APPLE_OSREFERENCE_LICENSE_HEADER_START@
   5  *
   6  * This file contains Original Code and/or Modifications of Original Code
   7  * as defined in and that are subject to the Apple Public Source License
   8  * Version 2.0 (the 'License'). You may not use this file except in
   9  * compliance with the License. The rights granted to you under the License
  10  * may not be used to create, or enable the creation or redistribution of,
  11  * unlawful or unlicensed copies of an Apple operating system, or to
  12  * circumvent, violate, or enable the circumvention or violation of, any
  13  * terms of an Apple operating system software license agreement.
  14  *
  15  * Please obtain a copy of the License at
  16  * http://www.opensource.apple.com/apsl/ and read it before using this file.
  17  *
  18  * The Original Code and all software distributed under the License are
  19  * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER
  20  * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES,
  21  * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY,
  22  * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT.
  23  * Please see the License for the specific language governing rights and
  24  * limitations under the License.
  25  *
  26  * @APPLE_OSREFERENCE_LICENSE_HEADER_END@
  27  */
  28 /*
  29  * Copyright (c) 1982, 1986, 1993, 1994, 1995
  30  *      The Regents of the University of California.  All rights reserved.
  31  *
  32  * Redistribution and use in source and binary forms, with or without
  33  * modification, are permitted provided that the following conditions
  34  * are met:
  35  * 1. Redistributions of source code must retain the above copyright
  36  *    notice, this list of conditions and the following disclaimer.
  37  * 2. Redistributions in binary form must reproduce the above copyright
  38  *    notice, this list of conditions and the following disclaimer in the
  39  *    documentation and/or other materials provided with the distribution.
  40  * 3. All advertising materials mentioning features or use of this software
  41  *    must display the following acknowledgement:
  42  *      This product includes software developed by the University of
  43  *      California, Berkeley and its contributors.
  44  * 4. Neither the name of the University nor the names of its contributors
  45  *    may be used to endorse or promote products derived from this software
  46  *    without specific prior written permission.
  47  *
  48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  58  * SUCH DAMAGE.
  59  *
  60  *      @(#)tcp_var.h   8.4 (Berkeley) 5/24/95
  61  * $FreeBSD: src/sys/netinet/tcp_var.h,v 1.56.2.8 2001/08/22 00:59:13 silby Exp $
  62  */
  63
  64 #ifndef _NETINET_TCP_VAR_H_
  65 #define _NETINET_TCP_VAR_H_
  66 #include <sys/appleapiopts.h>
  67 #include <sys/queue.h>
  68 #include <netinet/in_pcb.h>
  69 #include <netinet/tcp_timer.h>
  70
  71 #if defined(__LP64__)
  72 #define _TCPCB_PTR(x)                   u_int32_t
  73 #define _TCPCB_LIST_HEAD(name, type)    \
  74 struct name {                           \
  75         u_int32_t       lh_first;       \
  76 };
  77 #else
  78 #define _TCPCB_PTR(x)                   x
  79 #define _TCPCB_LIST_HEAD(name, type)    LIST_HEAD(name, type)
  80 #endif
  81
  82 #define TCP_RETRANSHZ       10          /* tcp retrans timer (100ms) per hz */
  83
  84 #ifdef KERNEL_PRIVATE
  85 #define N_TIME_WAIT_SLOTS   128         /* must be power of 2 */
  86
  87 /*
  88  * Kernel variables for tcp.
  89  */
  90
  91 /* TCP segment queue entry */
  92 struct tseg_qent {
  93         LIST_ENTRY(tseg_qent) tqe_q;
  94         int     tqe_len;                /* TCP segment data length */
  95         struct  tcphdr *tqe_th;         /* a pointer to tcp header */
  96         struct  mbuf    *tqe_m;         /* mbuf contains packet */
  97 };
  98 LIST_HEAD(tsegqe_head, tseg_qent);
  99 extern int      tcp_reass_maxseg;
 100 extern int      tcp_reass_qsize;
 101 #ifdef MALLOC_DECLARE
 102 MALLOC_DECLARE(M_TSEGQ);
 103 #endif
 104
 105 struct sackblk {
 106         tcp_seq start;          /* start seq no. of sack block */
 107         tcp_seq end;            /* end seq no. */
 108 };
 109
 110 struct sackhole {
 111         tcp_seq start;          /* start seq no. of hole */
 112         tcp_seq end;            /* end seq no. */
 113         tcp_seq rxmit;          /* next seq. no in hole to be retransmitted */
 114         TAILQ_ENTRY(sackhole) scblink;  /* scoreboard linkage */
 115 };
 116
 117 struct sackhint {
 118         struct sackhole *nexthole;
 119         int     sack_bytes_rexmit;
 120 };
 121
 122 struct tcptemp {
 123         u_char  tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
 124         struct  tcphdr tt_t;
 125 };
 126
 127 #define tcp6cb          tcpcb  /* for KAME src sync over BSD*'s */
 128
 129 /*
 130  * Tcp control block, one per tcp; fields:
 131  * Organized for 16 byte cacheline efficiency.
 132  */
 133 struct tcpcb {
 134         struct  tsegqe_head t_segq;
 135         int     t_dupacks;              /* consecutive dup acks recd */
 136         struct  tcptemp *unused;        /* unused now: was t_template */
 137
 138         int     t_timer[TCPT_NTIMERS];  /* tcp timers */
 139
 140         struct  inpcb *t_inpcb;         /* back pointer to internet pcb */
 141         int     t_state;                /* state of this connection */
 142         u_int   t_flags;
 143 #define TF_ACKNOW       0x00001         /* ack peer immediately */
 144 #define TF_DELACK       0x00002         /* ack, but try to delay it */
 145 #define TF_NODELAY      0x00004         /* don't delay packets to coalesce */
 146 #define TF_NOOPT        0x00008         /* don't use tcp options */
 147 #define TF_SENTFIN      0x00010         /* have sent FIN */
 148 #define TF_REQ_SCALE    0x00020         /* have/will request window scaling */
 149 #define TF_RCVD_SCALE   0x00040         /* other side has requested scaling */
 150 #define TF_REQ_TSTMP    0x00080         /* have/will request timestamps */
 151 #define TF_RCVD_TSTMP   0x00100         /* a timestamp was received in SYN */
 152 #define TF_SACK_PERMIT  0x00200         /* other side said I could SACK */
 153 #define TF_NEEDSYN      0x00400         /* send SYN (implicit state) */
 154 #define TF_NEEDFIN      0x00800         /* send FIN (implicit state) */
 155 #define TF_NOPUSH       0x01000         /* don't push */
 156 #define TF_REQ_CC       0x02000         /* have/will request CC */
 157 #define TF_RCVD_CC      0x04000         /* a CC was received in SYN */
 158 #define TF_SENDCCNEW    0x08000         /* send CCnew instead of CC in SYN */
 159 #define TF_MORETOCOME   0x10000         /* More data to be appended to sock */
 160 #define TF_LQ_OVERFLOW  0x20000         /* UNUSED listen queue overflow */
 161 #define TF_RXWIN0SENT   0x40000         /* sent a receiver win 0 in response */
 162 #define TF_SLOWLINK     0x80000         /* route is a on a modem speed link */
 163
 164
 165 #define TF_LASTIDLE     0x100000        /* connection was previously idle */
 166 #define TF_FASTRECOVERY 0x200000        /* in NewReno Fast Recovery */
 167 #define TF_WASFRECOVERY 0x400000        /* was in NewReno Fast Recovery */
 168 #define TF_SIGNATURE    0x800000        /* require MD5 digests (RFC2385) */
 169 #define TF_MAXSEGSNT    0x1000000       /* last segment sent was a full segment */
 170 #define TF_SENDINPROG   0x2000000       /* send is in progress */
 171 #define TF_PMTUD        0x4000000       /* Perform Path MTU Discovery for this connection */
 172 #define TF_CLOSING      0x8000000       /* pending tcp close */
 173
 174         int     t_force;                /* 1 if forcing out a byte */
 175
 176         tcp_seq snd_una;                /* send unacknowledged */
 177         tcp_seq snd_max;                /* highest sequence number sent;
 178                                          * used to recognize retransmits
 179                                          */
 180         tcp_seq snd_nxt;                /* send next */
 181         tcp_seq snd_up;                 /* send urgent pointer */
 182
 183         tcp_seq snd_wl1;                /* window update seg seq number */
 184         tcp_seq snd_wl2;                /* window update seg ack number */
 185         tcp_seq iss;                    /* initial send sequence number */
 186         tcp_seq irs;                    /* initial receive sequence number */
 187
 188         tcp_seq rcv_nxt;                /* receive next */
 189         tcp_seq rcv_adv;                /* advertised window */
 190         u_long  rcv_wnd;                /* receive window */
 191         tcp_seq rcv_up;                 /* receive urgent pointer */
 192
 193         u_long  snd_wnd;                /* send window */
 194         u_long  snd_cwnd;               /* congestion-controlled window */
 195         u_long  snd_bwnd;               /* bandwidth-controlled window */
 196         u_long  snd_ssthresh;           /* snd_cwnd size threshold for
 197                                          * for slow start exponential to
 198                                          * linear switch
 199                                          */
 200         u_long  snd_bandwidth;          /* calculated bandwidth or 0 */
 201         tcp_seq snd_recover;            /* for use in NewReno Fast Recovery */
 202
 203         u_int   t_maxopd;               /* mss plus options */
 204
 205         u_long  t_rcvtime;              /* inactivity time */
 206         u_long  t_starttime;            /* time connection was established */
 207         int     t_rtttime;              /* round trip time */
 208         tcp_seq t_rtseq;                /* sequence number being timed */
 209
 210         int     t_bw_rtttime;           /* used for bandwidth calculation */
 211         tcp_seq t_bw_rtseq;             /* used for bandwidth calculation */
 212
 213         int     t_rxtcur;               /* current retransmit value (ticks) */
 214         u_int   t_maxseg;               /* maximum segment size */
 215         int     t_srtt;                 /* smoothed round-trip time */
 216         int     t_rttvar;               /* variance in round-trip time */
 217
 218         int     t_rxtshift;             /* log(2) of rexmt exp. backoff */
 219         u_int   t_rttmin;               /* minimum rtt allowed */
 220         u_int   t_rttbest;              /* best rtt we've seen */
 221         u_long  t_rttupdated;           /* number of times rtt sampled */
 222         u_long  max_sndwnd;             /* largest window peer has offered */
 223
 224         int     t_softerror;            /* possible error not yet reported */
 225 /* out-of-band data */
 226         char    t_oobflags;             /* have some */
 227         char    t_iobc;                 /* input character */
 228 #define TCPOOB_HAVEDATA 0x01
 229 #define TCPOOB_HADDATA  0x02
 230 /* RFC 1323 variables */
 231         u_char  snd_scale;              /* window scaling for send window */
 232         u_char  rcv_scale;              /* window scaling for recv window */
 233         u_char  request_r_scale;        /* pending window scaling */
 234         u_char  requested_s_scale;
 235         u_long  ts_recent;              /* timestamp echo data */
 236
 237         u_long  ts_recent_age;          /* when last updated */
 238         tcp_seq last_ack_sent;
 239 /* RFC 1644 variables */
 240         tcp_cc  cc_send;                /* send connection count */
 241         tcp_cc  cc_recv;                /* receive connection count */
 242 /* RFC 3465 variables */
 243         u_long  t_bytes_acked;          /* ABC "bytes_acked" parameter */
 244 /* experimental */
 245         u_long  snd_cwnd_prev;          /* cwnd prior to retransmit */
 246         u_long  snd_ssthresh_prev;      /* ssthresh prior to retransmit */
 247         u_long  t_badrxtwin;            /* window for retransmit recovery */
 248
 249         int     t_keepidle;             /* keepalive idle timer (override global if > 0) */
 250         int     t_lastchain;            /* amount of packets chained last time around */
 251         int     t_unacksegs;            /* received but unacked segments: used for delaying acks */
 252
 253
 254 /* 3529618 MSS overload prevention */
 255         u_long  rcv_reset;
 256         u_long  rcv_pps;
 257         u_long  rcv_byps;
 258         u_long  rcv_maxbyps;
 259         tcp_seq snd_high;               /* for use in NewReno Fast Recovery */
 260         tcp_seq snd_high_prev;  /* snd_high prior to retransmit */
 261
 262         tcp_seq snd_recover_prev;       /* snd_recover prior to retransmit */
 263         u_char  snd_limited;            /* segments limited transmitted */
 264 /* anti DoS counters */
 265         u_long  rcv_second;             /* start of interval second */
 266 /* SACK related state */
 267         int     sack_enable;            /* enable SACK for this connection */
 268         int     snd_numholes;           /* number of holes seen by sender */
 269
 270         TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
 271                                                 /* SACK scoreboard (sorted) */
 272         tcp_seq snd_fack;               /* last seq number(+1) sack'd by rcv'r*/
 273         int     rcv_numsacks;           /* # distinct sack blks present */
 274         struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
 275         tcp_seq sack_newdata;           /* New data xmitted in this recovery
 276                                            episode starts at this seq number */
 277         struct sackhint sackhint;       /* SACK scoreboard hint */
 278         int     t_rttlow;               /* smallest observerved RTT */
 279         u_long          ecn_flags;
 280 #define TE_SETUPSENT            0x01    /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */
 281 #define TE_SETUPRECEIVED        0x02    /* Indicate we have received ECN-SETUP SYN or SYN-ACK */
 282 #define TE_SENDIPECT            0x04    /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */
 283 #define TE_SENDCWR                      0x08    /* Indicate that the next non-retransmit should have the TCP CWR flag set */
 284 #define TE_SENDECE                      0x10    /* Indicate that the next packet should have the TCP ECE flag set */
 285
 286 #if TRAFFIC_MGT
 287         u_int32_t               tot_recv_snapshot;      /* snapshot of global total pkts received */
 288         u_int32_t               bg_recv_snapshot;       /* snapshot of global background pkts received */
 289 #endif /* TRAFFIC_MGT */
 290         u_int32_t       t_pktlist_sentlen; /* total bytes in transmit chain */
 291         struct mbuf     *t_pktlist_head; /* First packet in transmit chain */
 292         struct mbuf     *t_pktlist_tail; /* Last packet in transmit chain */
 293 };
 294
 295 #define IN_FASTRECOVERY(tp)     (tp->t_flags & TF_FASTRECOVERY)
 296 #define ENTER_FASTRECOVERY(tp)  tp->t_flags |= TF_FASTRECOVERY
 297 #define EXIT_FASTRECOVERY(tp)   tp->t_flags &= ~TF_FASTRECOVERY
 298
 299
 300 /*
 301  * Structure to hold TCP options that are only used during segment
 302  * processing (in tcp_input), but not held in the tcpcb.
 303  * It's basically used to reduce the number of parameters
 304  * to tcp_dooptions.
 305  */
 306 struct tcpopt {
 307         u_long  to_flags;               /* which options are present */
 308 #define TOF_TS          0x0001          /* timestamp */
 309 #define TOF_MSS         0x0010
 310 #define TOF_SCALE       0x0020
 311 #define TOF_SIGNATURE   0x0040  /* signature option present */
 312 #define TOF_SIGLEN      0x0080  /* signature length valid (RFC2385) */
 313 #define TOF_SACK        0x0100          /* Peer sent SACK option */
 314         u_long          to_tsval;
 315         u_long          to_tsecr;
 316         u_int16_t       to_mss;
 317         u_int8_t        to_requested_s_scale;
 318         u_int8_t        to_nsacks;      /* number of SACK blocks */
 319         u_char          *to_sacks;      /* pointer to the first SACK blocks */
 320 };
 321
 322 /*
 323  * The TAO cache entry which is stored in the protocol family specific
 324  * portion of the route metrics.
 325  */
 326 struct rmxp_tao {
 327         tcp_cc  tao_cc;                 /* latest CC in valid SYN */
 328         tcp_cc  tao_ccsent;             /* latest CC sent to peer */
 329         u_short tao_mssopt;             /* peer's cached MSS */
 330 #ifdef notyet
 331         u_short tao_flags;              /* cache status flags */
 332 #define TAOF_DONT       0x0001          /* peer doesn't understand rfc1644 */
 333 #define TAOF_OK         0x0002          /* peer does understand rfc1644 */
 334 #define TAOF_UNDEF      0               /* we don't know yet */
 335 #endif /* notyet */
 336 };
 337 #define rmx_taop(r)     ((struct rmxp_tao *)(r).rmx_filler)
 338
 339 #define intotcpcb(ip)   ((struct tcpcb *)(ip)->inp_ppcb)
 340 #define sototcpcb(so)   (intotcpcb(sotoinpcb(so)))
 341
 342 /*
 343  * The smoothed round-trip time and estimated variance
 344  * are stored as fixed point numbers scaled by the values below.
 345  * For convenience, these scales are also used in smoothing the average
 346  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
 347  * With these scales, srtt has 3 bits to the right of the binary point,
 348  * and thus an "ALPHA" of 0.875.  rttvar has 2 bits to the right of the
 349  * binary point, and is smoothed with an ALPHA of 0.75.
 350  */
 351 #define TCP_RTT_SCALE           32      /* multiplier for srtt; 3 bits frac. */
 352 #define TCP_RTT_SHIFT           5       /* shift for srtt; 3 bits frac. */
 353 #define TCP_RTTVAR_SCALE        16      /* multiplier for rttvar; 2 bits */
 354 #define TCP_RTTVAR_SHIFT        4       /* shift for rttvar; 2 bits */
 355 #define TCP_DELTA_SHIFT         2       /* see tcp_input.c */
 356
 357 /*
 358  * The initial retransmission should happen at rtt + 4 * rttvar.
 359  * Because of the way we do the smoothing, srtt and rttvar
 360  * will each average +1/2 tick of bias.  When we compute
 361  * the retransmit timer, we want 1/2 tick of rounding and
 362  * 1 extra tick because of +-1/2 tick uncertainty in the
 363  * firing of the timer.  The bias will give us exactly the
 364  * 1.5 tick we need.  But, because the bias is
 365  * statistical, we have to test that we don't drop below
 366  * the minimum feasible timer (which is 2 ticks).
 367  * This version of the macro adapted from a paper by Lawrence
 368  * Brakmo and Larry Peterson which outlines a problem caused
 369  * by insufficient precision in the original implementation,
 370  * which results in inappropriately large RTO values for very
 371  * fast networks.
 372  */
 373 #define TCP_REXMTVAL(tp) \
 374         max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
 375           + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
 376
 377 /*
 378  * Jaguar compatible TCP control block, for xtcpcb
 379  * Does not have the old fields
 380  */
 381 struct otcpcb {
 382 #else
 383 struct tseg_qent;
 384 _TCPCB_LIST_HEAD(tsegqe_head, tseg_qent);
 385
 386 struct tcpcb {
 387 #endif /* KERNEL_PRIVATE */
 388         struct  tsegqe_head t_segq;
 389         int     t_dupacks;              /* consecutive dup acks recd */
 390         u_int32_t unused;               /* unused now: was t_template */
 391
 392         int     t_timer[TCPT_NTIMERS];  /* tcp timers */
 393
 394         _TCPCB_PTR(struct inpcb *) t_inpcb;     /* back pointer to internet pcb */
 395         int     t_state;                /* state of this connection */
 396         u_int   t_flags;
 397 #define TF_ACKNOW       0x00001         /* ack peer immediately */
 398 #define TF_DELACK       0x00002         /* ack, but try to delay it */
 399 #define TF_NODELAY      0x00004         /* don't delay packets to coalesce */
 400 #define TF_NOOPT        0x00008         /* don't use tcp options */
 401 #define TF_SENTFIN      0x00010         /* have sent FIN */
 402 #define TF_REQ_SCALE    0x00020         /* have/will request window scaling */
 403 #define TF_RCVD_SCALE   0x00040         /* other side has requested scaling */
 404 #define TF_REQ_TSTMP    0x00080         /* have/will request timestamps */
 405 #define TF_RCVD_TSTMP   0x00100         /* a timestamp was received in SYN */
 406 #define TF_SACK_PERMIT  0x00200         /* other side said I could SACK */
 407 #define TF_NEEDSYN      0x00400         /* send SYN (implicit state) */
 408 #define TF_NEEDFIN      0x00800         /* send FIN (implicit state) */
 409 #define TF_NOPUSH       0x01000         /* don't push */
 410 #define TF_REQ_CC       0x02000         /* have/will request CC */
 411 #define TF_RCVD_CC      0x04000         /* a CC was received in SYN */
 412 #define TF_SENDCCNEW    0x08000         /* send CCnew instead of CC in SYN */
 413 #define TF_MORETOCOME   0x10000         /* More data to be appended to sock */
 414 #define TF_LQ_OVERFLOW  0x20000         /* listen queue overflow */
 415 #define TF_RXWIN0SENT   0x40000         /* sent a receiver win 0 in response */
 416 #define TF_SLOWLINK     0x80000         /* route is a on a modem speed link */
 417
 418         int     t_force;                /* 1 if forcing out a byte */
 419
 420         tcp_seq snd_una;                /* send unacknowledged */
 421         tcp_seq snd_max;                /* highest sequence number sent;
 422                                          * used to recognize retransmits
 423                                          */
 424         tcp_seq snd_nxt;                /* send next */
 425         tcp_seq snd_up;                 /* send urgent pointer */
 426
 427         tcp_seq snd_wl1;                /* window update seg seq number */
 428         tcp_seq snd_wl2;                /* window update seg ack number */
 429         tcp_seq iss;                    /* initial send sequence number */
 430         tcp_seq irs;                    /* initial receive sequence number */
 431
 432         tcp_seq rcv_nxt;                /* receive next */
 433         tcp_seq rcv_adv;                /* advertised window */
 434         u_int32_t rcv_wnd;              /* receive window */
 435         tcp_seq rcv_up;                 /* receive urgent pointer */
 436
 437         u_int32_t snd_wnd;              /* send window */
 438         u_int32_t snd_cwnd;             /* congestion-controlled window */
 439         u_int32_t snd_ssthresh;         /* snd_cwnd size threshold for
 440                                          * for slow start exponential to
 441                                          * linear switch
 442                                          */
 443         u_int   t_maxopd;               /* mss plus options */
 444
 445         u_int32_t t_rcvtime;            /* inactivity time */
 446         u_int32_t t_starttime;          /* time connection was established */
 447         int     t_rtttime;              /* round trip time */
 448         tcp_seq t_rtseq;                /* sequence number being timed */
 449
 450         int     t_rxtcur;               /* current retransmit value (ticks) */
 451         u_int   t_maxseg;               /* maximum segment size */
 452         int     t_srtt;                 /* smoothed round-trip time */
 453         int     t_rttvar;               /* variance in round-trip time */
 454
 455         int     t_rxtshift;             /* log(2) of rexmt exp. backoff */
 456         u_int   t_rttmin;               /* minimum rtt allowed */
 457         u_int32_t t_rttupdated;         /* number of times rtt sampled */
 458         u_int32_t max_sndwnd;           /* largest window peer has offered */
 459
 460         int     t_softerror;            /* possible error not yet reported */
 461 /* out-of-band data */
 462         char    t_oobflags;             /* have some */
 463         char    t_iobc;                 /* input character */
 464 #define TCPOOB_HAVEDATA 0x01
 465 #define TCPOOB_HADDATA  0x02
 466 /* RFC 1323 variables */
 467         u_char  snd_scale;              /* window scaling for send window */
 468         u_char  rcv_scale;              /* window scaling for recv window */
 469         u_char  request_r_scale;        /* pending window scaling */
 470         u_char  requested_s_scale;
 471         u_int32_t ts_recent;            /* timestamp echo data */
 472
 473         u_int32_t ts_recent_age;        /* when last updated */
 474         tcp_seq last_ack_sent;
 475 /* RFC 1644 variables */
 476         tcp_cc  cc_send;                /* send connection count */
 477         tcp_cc  cc_recv;                /* receive connection count */
 478         tcp_seq snd_recover;            /* for use in fast recovery */
 479 /* experimental */
 480         u_int32_t snd_cwnd_prev;        /* cwnd prior to retransmit */
 481         u_int32_t snd_ssthresh_prev;    /* ssthresh prior to retransmit */
 482         u_int32_t t_badrxtwin;          /* window for retransmit recovery */
 483 };
 484
 485 /*
 486  * TCP statistics.
 487  * Many of these should be kept per connection,
 488  * but that's inconvenient at the moment.
 489  */
 490 struct  tcpstat {
 491         u_int32_t       tcps_connattempt;       /* connections initiated */
 492         u_int32_t       tcps_accepts;           /* connections accepted */
 493         u_int32_t       tcps_connects;          /* connections established */
 494         u_int32_t       tcps_drops;             /* connections dropped */
 495         u_int32_t       tcps_conndrops;         /* embryonic connections dropped */
 496         u_int32_t       tcps_closed;            /* conn. closed (includes drops) */
 497         u_int32_t       tcps_segstimed;         /* segs where we tried to get rtt */
 498         u_int32_t       tcps_rttupdated;        /* times we succeeded */
 499         u_int32_t       tcps_delack;            /* delayed acks sent */
 500         u_int32_t       tcps_timeoutdrop;       /* conn. dropped in rxmt timeout */
 501         u_int32_t       tcps_rexmttimeo;        /* retransmit timeouts */
 502         u_int32_t       tcps_persisttimeo;      /* persist timeouts */
 503         u_int32_t       tcps_keeptimeo;         /* keepalive timeouts */
 504         u_int32_t       tcps_keepprobe;         /* keepalive probes sent */
 505         u_int32_t       tcps_keepdrops;         /* connections dropped in keepalive */
 506
 507         u_int32_t       tcps_sndtotal;          /* total packets sent */
 508         u_int32_t       tcps_sndpack;           /* data packets sent */
 509         u_int32_t       tcps_sndbyte;           /* data bytes sent */
 510         u_int32_t       tcps_sndrexmitpack;     /* data packets retransmitted */
 511         u_int32_t       tcps_sndrexmitbyte;     /* data bytes retransmitted */
 512         u_int32_t       tcps_sndacks;           /* ack-only packets sent */
 513         u_int32_t       tcps_sndprobe;          /* window probes sent */
 514         u_int32_t       tcps_sndurg;            /* packets sent with URG only */
 515         u_int32_t       tcps_sndwinup;          /* window update-only packets sent */
 516         u_int32_t       tcps_sndctrl;           /* control (SYN|FIN|RST) packets sent */
 517
 518         u_int32_t       tcps_rcvtotal;          /* total packets received */
 519         u_int32_t       tcps_rcvpack;           /* packets received in sequence */
 520         u_int32_t       tcps_rcvbyte;           /* bytes received in sequence */
 521         u_int32_t       tcps_rcvbadsum;         /* packets received with ccksum errs */
 522         u_int32_t       tcps_rcvbadoff;         /* packets received with bad offset */
 523         u_int32_t       tcps_rcvmemdrop;        /* packets dropped for lack of memory */
 524         u_int32_t       tcps_rcvshort;          /* packets received too short */
 525         u_int32_t       tcps_rcvduppack;        /* duplicate-only packets received */
 526         u_int32_t       tcps_rcvdupbyte;        /* duplicate-only bytes received */
 527         u_int32_t       tcps_rcvpartduppack;    /* packets with some duplicate data */
 528         u_int32_t       tcps_rcvpartdupbyte;    /* dup. bytes in part-dup. packets */
 529         u_int32_t       tcps_rcvoopack;         /* out-of-order packets received */
 530         u_int32_t       tcps_rcvoobyte;         /* out-of-order bytes received */
 531         u_int32_t       tcps_rcvpackafterwin;   /* packets with data after window */
 532         u_int32_t       tcps_rcvbyteafterwin;   /* bytes rcvd after window */
 533         u_int32_t       tcps_rcvafterclose;     /* packets rcvd after "close" */
 534         u_int32_t       tcps_rcvwinprobe;       /* rcvd window probe packets */
 535         u_int32_t       tcps_rcvdupack;         /* rcvd duplicate acks */
 536         u_int32_t       tcps_rcvacktoomuch;     /* rcvd acks for unsent data */
 537         u_int32_t       tcps_rcvackpack;        /* rcvd ack packets */
 538         u_int32_t       tcps_rcvackbyte;        /* bytes acked by rcvd acks */
 539         u_int32_t       tcps_rcvwinupd;         /* rcvd window update packets */
 540         u_int32_t       tcps_pawsdrop;          /* segments dropped due to PAWS */
 541         u_int32_t       tcps_predack;           /* times hdr predict ok for acks */
 542         u_int32_t       tcps_preddat;           /* times hdr predict ok for data pkts */
 543         u_int32_t       tcps_pcbcachemiss;
 544         u_int32_t       tcps_cachedrtt;         /* times cached RTT in route updated */
 545         u_int32_t       tcps_cachedrttvar;      /* times cached rttvar updated */
 546         u_int32_t       tcps_cachedssthresh;    /* times cached ssthresh updated */
 547         u_int32_t       tcps_usedrtt;           /* times RTT initialized from route */
 548         u_int32_t       tcps_usedrttvar;        /* times RTTVAR initialized from rt */
 549         u_int32_t       tcps_usedssthresh;      /* times ssthresh initialized from rt*/
 550         u_int32_t       tcps_persistdrop;       /* timeout in persist state */
 551         u_int32_t       tcps_badsyn;            /* bogus SYN, e.g. premature ACK */
 552         u_int32_t       tcps_mturesent;         /* resends due to MTU discovery */
 553         u_int32_t       tcps_listendrop;        /* listen queue overflows */
 554
 555         /* new stats from FreeBSD 5.4 sync up */
 556         u_int32_t       tcps_minmssdrops;       /* average minmss too low drops */
 557         u_int32_t       tcps_sndrexmitbad;      /* unnecessary packet retransmissions */
 558         u_int32_t       tcps_badrst;            /* ignored RSTs in the window */
 559
 560         u_int32_t       tcps_sc_added;          /* entry added to syncache */
 561         u_int32_t       tcps_sc_retransmitted;  /* syncache entry was retransmitted */
 562         u_int32_t       tcps_sc_dupsyn;         /* duplicate SYN packet */
 563         u_int32_t       tcps_sc_dropped;        /* could not reply to packet */
 564         u_int32_t       tcps_sc_completed;      /* successful extraction of entry */
 565         u_int32_t       tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
 566         u_int32_t       tcps_sc_cacheoverflow;  /* syncache cache limit hit */
 567         u_int32_t       tcps_sc_reset;          /* RST removed entry from syncache */
 568         u_int32_t       tcps_sc_stale;          /* timed out or listen socket gone */
 569         u_int32_t       tcps_sc_aborted;        /* syncache entry aborted */
 570         u_int32_t       tcps_sc_badack;         /* removed due to bad ACK */
 571         u_int32_t       tcps_sc_unreach;        /* ICMP unreachable received */
 572         u_int32_t       tcps_sc_zonefail;       /* zalloc() failed */
 573         u_int32_t       tcps_sc_sendcookie;     /* SYN cookie sent */
 574         u_int32_t       tcps_sc_recvcookie;     /* SYN cookie received */
 575
 576         u_int32_t       tcps_hc_added;          /* entry added to hostcache */
 577         u_int32_t       tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
 578
 579         /* SACK related stats */
 580         u_int32_t       tcps_sack_recovery_episode; /* SACK recovery episodes */
 581         u_int32_t  tcps_sack_rexmits;       /* SACK rexmit segments   */
 582         u_int32_t  tcps_sack_rexmit_bytes;          /* SACK rexmit bytes      */
 583         u_int32_t  tcps_sack_rcv_blocks;            /* SACK blocks (options) received */
 584         u_int32_t  tcps_sack_send_blocks;           /* SACK blocks (options) sent     */
 585         u_int32_t  tcps_sack_sboverflow;            /* SACK sendblock overflow   */
 586
 587 #if TRAFFIC_MGT
 588         u_int32_t       tcps_bg_rcvtotal;       /* total background packets received */
 589 #endif /* TRAFFIC_MGT */
 590 };
 591
 592 #pragma pack(4)
 593
 594 /*
 595  * TCB structure exported to user-land via sysctl(3).
 596  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
 597  * included.  Not all of our clients do.
 598  */
 599 struct  xtcpcb {
 600         u_int32_t       xt_len;
 601 #ifdef KERNEL_PRIVATE
 602         struct  inpcb_compat    xt_inp;
 603 #else
 604         struct  inpcb   xt_inp;
 605 #endif
 606 #ifdef KERNEL_PRIVATE
 607         struct  otcpcb  xt_tp;
 608 #else
 609         struct  tcpcb   xt_tp;
 610 #endif
 611         struct  xsocket xt_socket;
 612         u_quad_t        xt_alignment_hack;
 613 };
 614
 615 #pragma pack()
 616
 617 /*
 618  * Names for TCP sysctl objects
 619  */
 620 #define TCPCTL_DO_RFC1323       1       /* use RFC-1323 extensions */
 621 #define TCPCTL_DO_RFC1644       2       /* use RFC-1644 extensions */
 622 #define TCPCTL_MSSDFLT          3       /* MSS default */
 623 #define TCPCTL_STATS            4       /* statistics (read-only) */
 624 #define TCPCTL_RTTDFLT          5       /* default RTT estimate */
 625 #define TCPCTL_KEEPIDLE         6       /* keepalive idle timer */
 626 #define TCPCTL_KEEPINTVL        7       /* interval to send keepalives */
 627 #define TCPCTL_SENDSPACE        8       /* send buffer space */
 628 #define TCPCTL_RECVSPACE        9       /* receive buffer space */
 629 #define TCPCTL_KEEPINIT         10      /* timeout for establishing syn */
 630 #define TCPCTL_PCBLIST          11      /* list of all outstanding PCBs */
 631 #define TCPCTL_DELACKTIME       12      /* time before sending delayed ACK */
 632 #define TCPCTL_V6MSSDFLT        13      /* MSS default for IPv6 */
 633 #define TCPCTL_MAXID            14
 634
 635 #ifdef KERNEL_PRIVATE
 636 #define TCP_PKTLIST_CLEAR(tp) {                                         \
 637         (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL;             \
 638         (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0;                \
 639 }
 640
 641 #define TCPCTL_NAMES { \
 642         { 0, 0 }, \
 643         { "rfc1323", CTLTYPE_INT }, \
 644         { "rfc1644", CTLTYPE_INT }, \
 645         { "mssdflt", CTLTYPE_INT }, \
 646         { "stats", CTLTYPE_STRUCT }, \
 647         { "rttdflt", CTLTYPE_INT }, \
 648         { "keepidle", CTLTYPE_INT }, \
 649         { "keepintvl", CTLTYPE_INT }, \
 650         { "sendspace", CTLTYPE_INT }, \
 651         { "recvspace", CTLTYPE_INT }, \
 652         { "keepinit", CTLTYPE_INT }, \
 653         { "pcblist", CTLTYPE_STRUCT }, \
 654         { "delacktime", CTLTYPE_INT }, \
 655         { "v6mssdflt", CTLTYPE_INT }, \
 656 }
 657
 658 #ifdef SYSCTL_DECL
 659 SYSCTL_DECL(_net_inet_tcp);
 660 #endif /* SYSCTL_DECL */
 661
 662 extern  struct inpcbhead tcb;           /* head of queue of active tcpcb's */
 663 extern  struct inpcbinfo tcbinfo;
 664 extern  struct tcpstat tcpstat; /* tcp statistics */
 665 extern  int tcp_mssdflt;        /* XXX */
 666 extern  int tcp_minmss;
 667 extern  int tcp_minmssoverload;
 668 extern  int tcp_do_newreno;
 669 extern  int ss_fltsz;
 670 extern  int ss_fltsz_local;
 671 #ifdef __APPLE__
 672 extern  u_long tcp_now;         /* for RFC 1323 timestamps */
 673 extern  int tcp_delack_enabled;
 674 #endif /* __APPLE__ */
 675
 676 extern  int tcp_do_sack;        /* SACK enabled/disabled */
 677
 678 void     tcp_canceltimers(struct tcpcb *);
 679 struct tcpcb *
 680          tcp_close(struct tcpcb *);
 681 void     tcp_ctlinput(int, struct sockaddr *, void *);
 682 int      tcp_ctloutput(struct socket *, struct sockopt *);
 683 struct tcpcb *
 684          tcp_drop(struct tcpcb *, int);
 685 void     tcp_drain(void);
 686 void     tcp_fasttimo(void);
 687 struct rmxp_tao *
 688          tcp_gettaocache(struct inpcb *);
 689 void     tcp_init(void) __attribute__((section("__TEXT, initcode")));
 690 void     tcp_input(struct mbuf *, int);
 691 void     tcp_mss(struct tcpcb *, int);
 692 int      tcp_mssopt(struct tcpcb *);
 693 void     tcp_drop_syn_sent(struct inpcb *, int);
 694 void     tcp_mtudisc(struct inpcb *, int);
 695 struct tcpcb *
 696          tcp_newtcpcb(struct inpcb *);
 697 int      tcp_output(struct tcpcb *);
 698 void     tcp_quench(struct inpcb *, int);
 699 void     tcp_respond(struct tcpcb *, void *,
 700             struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int, ifnet_t);
 701 struct rtentry *
 702          tcp_rtlookup(struct inpcb *);
 703 void     tcp_setpersist(struct tcpcb *);
 704 void     tcp_slowtimo(void);
 705 struct tcptemp *
 706          tcp_maketemplate(struct tcpcb *);
 707 void     tcp_fillheaders(struct tcpcb *, void *, void *);
 708 struct tcpcb *
 709          tcp_timers(struct tcpcb *, int);
 710 void     tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
 711
 712 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
 713 void     tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
 714 void     tcp_clean_sackreport(struct tcpcb *tp);
 715 void     tcp_sack_adjust(struct tcpcb *tp);
 716 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
 717 void     tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
 718 void     tcp_free_sackholes(struct tcpcb *tp);
 719 long     tcp_sbspace(struct tcpcb *tp);
 720
 721
 722 int      tcp_lock (struct socket *, int, int);
 723 int      tcp_unlock (struct socket *, int, int);
 724 #ifdef _KERN_LOCKS_H_
 725 lck_mtx_t *      tcp_getlock (struct socket *, int);
 726 #else
 727 void *   tcp_getlock (struct socket *, int);
 728 #endif
 729
 730
 731 extern  struct pr_usrreqs tcp_usrreqs;
 732 extern  u_long tcp_sendspace;
 733 extern  u_long tcp_recvspace;
 734 tcp_seq tcp_new_isn(struct tcpcb *);
 735
 736 #endif /* KERNEL_RPIVATE */
 737
 738 #endif /* _NETINET_TCP_VAR_H_ */