bsd/netinet/tcp_var.h

   1 /*
   2  * Copyright (c) 2000-2015 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/types.h>
  67 #include <sys/appleapiopts.h>
  68 #include <sys/queue.h>
  69 #include <netinet/in_pcb.h>
  70 #include <netinet/tcp.h>
  71 #include <netinet/tcp_timer.h>
  72
  73 #if defined(__LP64__)
  74 #define _TCPCB_PTR(x)                   u_int32_t
  75 #define _TCPCB_LIST_HEAD(name, type)    \
  76 struct name {                           \
  77         u_int32_t       lh_first;       \
  78 };
  79 #else
  80 #define _TCPCB_PTR(x)                   x
  81 #define _TCPCB_LIST_HEAD(name, type)    LIST_HEAD(name, type)
  82 #endif
  83
  84 #ifdef KERNEL_PRIVATE
  85
  86 #define TCP_RETRANSHZ   1000    /* granularity of TCP timestamps, 1ms */
  87 /* Minimum time quantum within which the timers are coalesced */
  88 #define TCP_TIMER_10MS_QUANTUM  (TCP_RETRANSHZ/100) /* every 10ms */
  89 #define TCP_TIMER_100MS_QUANTUM   (TCP_RETRANSHZ/10) /* every 100ms */
  90 #define TCP_TIMER_500MS_QUANTUM   (TCP_RETRANSHZ/2) /* every 500ms */
  91
  92 #define TCP_RETRANSHZ_TO_USEC 1000
  93
  94 #define N_TIME_WAIT_SLOTS   128         /* must be power of 2 */
  95
  96 /* Base RTT is stored for N_MIN_RTT_HISTORY slots. This is used to
  97  * estimate expected minimum RTT for delay based congestion control
  98  * algorithms.
  99  */
 100 #define N_RTT_BASE      5
 101
 102 /* Always allow at least 4 packets worth of recv window when adjusting
 103  * recv window using inter-packet arrival jitter.
 104  */
 105 #define MIN_IAJ_WIN 4
 106
 107 /* A variation in delay of this many milliseconds is tolerable. This limit has to
 108  * be low but greater than zero. We also use standard deviation on jitter to adjust
 109  * this limit for different link and connection types.
 110  */
 111 #define ALLOWED_IAJ 5
 112
 113 /* Ignore the first few packets on a connection until the ACK clock gets going
 114  */
 115 #define IAJ_IGNORE_PKTCNT 40
 116
 117 /* Let the accumulated IAJ value increase by this threshold at most. This limit
 118  * will control how many ALLOWED_IAJ measurements a receiver will have to see
 119  * before opening the receive window
 120  */
 121 #define ACC_IAJ_HIGH_THRESH 100
 122
 123 /* When accumulated IAJ reaches this value, the receiver starts to react by
 124  * closing the window
 125  */
 126 #define ACC_IAJ_REACT_LIMIT 200
 127
 128 /* If the number of small packets (smaller than IAJ packet size) seen on a
 129  * connection is more than this threshold, reset the size and learn it again.
 130  * This is needed because the sender might send smaller segments after PMTU
 131  * discovery and the receiver has to learn the new size.
 132  */
 133 #define RESET_IAJ_SIZE_THRESH 20
 134
 135 /*
 136  * Adaptive timeout is a read/write timeout specified by the application to
 137  * get a socket event when the transport layer detects a stall in data
 138  * transfer. The value specified is the number of probes that can be sent
 139  * to the peer before generating an event. Since it is not specified as
 140  * a time value, the timeout will adjust based on the RTT seen on the link.
 141  * The timeout will start only when there is an indication that the read/write
 142  * operation is not making progress.
 143  *
 144  * If a write operation stalls, the probe will be retransmission of data.
 145  * If a read operation stalls, the probe will be a keep-alive packet.
 146  *
 147  * The maximum value of adaptive timeout is set to 10 which will allow
 148  * transmission of enough number of probes to the peer.
 149  */
 150 #define TCP_ADAPTIVE_TIMEOUT_MAX        10
 151
 152 #define TCP_CONNECTIVITY_PROBES_MAX     5
 153
 154 /*
 155  * Kernel variables for tcp.
 156  */
 157
 158 /* TCP segment queue entry */
 159 struct tseg_qent {
 160         LIST_ENTRY(tseg_qent) tqe_q;
 161         int     tqe_len;                /* TCP segment data length */
 162         struct  tcphdr *tqe_th;         /* a pointer to tcp header */
 163         struct  mbuf    *tqe_m;         /* mbuf contains packet */
 164 };
 165 LIST_HEAD(tsegqe_head, tseg_qent);
 166
 167 struct sackblk {
 168         tcp_seq start;          /* start seq no. of sack block */
 169         tcp_seq end;            /* end seq no. */
 170 };
 171
 172 struct sackhole {
 173         tcp_seq start;          /* start seq no. of hole */
 174         tcp_seq end;            /* end seq no. */
 175         tcp_seq rxmit;          /* next seq. no in hole to be retransmitted */
 176         u_int32_t rxmit_start;  /* timestamp of first retransmission */
 177         TAILQ_ENTRY(sackhole) scblink;  /* scoreboard linkage */
 178 };
 179
 180 struct sackhint {
 181         struct sackhole *nexthole;
 182         int     sack_bytes_rexmit;
 183 };
 184
 185 struct tcp_rxt_seg {
 186         tcp_seq rx_start;
 187         tcp_seq rx_end;
 188         u_int16_t rx_count;
 189         u_int16_t rx_flags;
 190 #define TCP_RXT_SPURIOUS        0x1     /* received DSACK notification */
 191 #define TCP_RXT_DSACK_FOR_TLP   0x2
 192         SLIST_ENTRY(tcp_rxt_seg) rx_link;
 193 };
 194
 195 struct tcptemp {
 196         u_char  tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
 197         struct  tcphdr tt_t;
 198 };
 199
 200 struct bwmeas {
 201         tcp_seq bw_start;               /* start of bw measurement */
 202         uint32_t bw_ts;         /* timestamp when bw measurement started */
 203         uint32_t bw_size;               /* burst size in bytes for this bw measurement */
 204         uint32_t bw_minsizepkts;        /* Min burst size as segments */
 205         uint32_t bw_maxsizepkts;        /* Max burst size as segments */
 206         uint32_t bw_minsize;    /* Min size in bytes */
 207         uint32_t bw_maxsize;    /* Max size in bytes */
 208         uint32_t bw_sndbw;              /* Measured send bw */
 209 };
 210
 211 /* MPTCP Data sequence map entry */
 212 struct mpt_dsn_map {
 213         uint64_t                mpt_dsn;        /* data seq num recvd */
 214         uint32_t                mpt_sseq;       /* relative subflow # */
 215         uint16_t                mpt_len;        /* length of mapping */
 216         uint16_t                mpt_csum;       /* checksum value if on */
 217 };
 218 #define tcp6cb          tcpcb  /* for KAME src sync over BSD*'s */
 219
 220 struct tcp_ccstate {
 221         union {
 222                 struct tcp_cubic_state {
 223                         u_int32_t tc_last_max;  /* cwnd at last loss */
 224                         u_int32_t tc_epoch_start; /* TS of last loss */
 225                         u_int32_t tc_origin_point; /* window at the start of an epoch */
 226                         u_int32_t tc_tcp_win; /* computed tcp win */
 227                         u_int32_t tc_tcp_bytes_acked; /* bytes acked */
 228                         u_int32_t tc_target_win; /* cubic target win */
 229                         u_int32_t tc_avg_lastmax; /* Average of last max */
 230                         u_int32_t tc_mean_deviation; /* Mean absolute deviation */
 231                         float     tc_epoch_period; /* K parameter */
 232                 } _cubic_state_;
 233 #define cub_last_max __u__._cubic_state_.tc_last_max
 234 #define cub_epoch_start __u__._cubic_state_.tc_epoch_start
 235 #define cub_origin_point __u__._cubic_state_.tc_origin_point
 236 #define cub_tcp_win __u__._cubic_state_.tc_tcp_win
 237 #define cub_tcp_bytes_acked __u__._cubic_state_.tc_tcp_bytes_acked
 238 #define cub_epoch_period __u__._cubic_state_.tc_epoch_period
 239 #define cub_target_win __u__._cubic_state_.tc_target_win
 240 #define cub_avg_lastmax __u__._cubic_state_.tc_avg_lastmax
 241 #define cub_mean_dev __u__._cubic_state_.tc_mean_deviation
 242         } __u__;
 243 };
 244
 245 /*
 246  * Tcp control block, one per tcp; fields:
 247  * Organized for 16 byte cacheline efficiency.
 248  */
 249 struct tcpcb {
 250         struct  tsegqe_head t_segq;
 251         int     t_dupacks;              /* consecutive dup acks recd */
 252         int     t_state;                /* state of this connection */
 253         uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */
 254         struct tcptimerentry tentry;    /* entry in timer list */
 255
 256         struct  inpcb *t_inpcb;         /* back pointer to internet pcb */
 257         uint32_t        t_flags;
 258 #define TF_ACKNOW       0x00001         /* ack peer immediately */
 259 #define TF_DELACK       0x00002         /* ack, but try to delay it */
 260 #define TF_NODELAY      0x00004         /* don't delay packets to coalesce */
 261 #define TF_NOOPT        0x00008         /* don't use tcp options */
 262 #define TF_SENTFIN      0x00010         /* have sent FIN */
 263 #define TF_REQ_SCALE    0x00020         /* have/will request window scaling */
 264 #define TF_RCVD_SCALE   0x00040         /* other side has requested scaling */
 265 #define TF_REQ_TSTMP    0x00080         /* have/will request timestamps */
 266 #define TF_RCVD_TSTMP   0x00100         /* a timestamp was received in SYN */
 267 #define TF_SACK_PERMIT  0x00200         /* other side said I could SACK */
 268 #define TF_NEEDSYN      0x00400         /* send SYN (implicit state) */
 269 #define TF_NEEDFIN      0x00800         /* send FIN (implicit state) */
 270 #define TF_NOPUSH       0x01000         /* don't push */
 271 #define TF_REQ_CC       0x02000         /* have/will request CC */
 272 #define TF_RCVD_CC      0x04000         /* a CC was received in SYN */
 273 #define TF_SENDCCNEW    0x08000         /* send CCnew instead of CC in SYN */
 274 #define TF_MORETOCOME   0x10000         /* More data to be appended to sock */
 275 #define TF_LOCAL        0x20000         /* connection to a host on local link */
 276 #define TF_RXWIN0SENT   0x40000         /* sent a receiver win 0 in response */
 277 #define TF_SLOWLINK     0x80000         /* route is a on a modem speed link */
 278 #define TF_LASTIDLE     0x100000        /* connection was previously idle */
 279 #define TF_FASTRECOVERY 0x200000        /* in NewReno Fast Recovery */
 280 #define TF_WASFRECOVERY 0x400000        /* was in NewReno Fast Recovery */
 281 #define TF_SIGNATURE    0x800000        /* require MD5 digests (RFC2385) */
 282 #define TF_MAXSEGSNT    0x1000000       /* last segment sent was a full segment */
 283 #define TF_PMTUD        0x4000000       /* Perform Path MTU Discovery for this connection */
 284 #define TF_CLOSING      0x8000000       /* pending tcp close */
 285 #define TF_TSO          0x10000000      /* TCP Segment Offloading is enable on this connection */
 286 #define TF_BLACKHOLE    0x20000000      /* Path MTU Discovery Black Hole detection */
 287 #define TF_TIMER_ONLIST 0x40000000      /* pcb is on tcp_timer_list */
 288 #define TF_STRETCHACK   0x80000000      /* receiver is going to delay acks */
 289
 290         tcp_seq snd_una;                /* send unacknowledged */
 291         tcp_seq snd_max;                /* highest sequence number sent;
 292                                          * used to recognize retransmits
 293                                          */
 294         tcp_seq snd_nxt;                /* send next */
 295         tcp_seq snd_up;                 /* send urgent pointer */
 296
 297         tcp_seq snd_wl1;                /* window update seg seq number */
 298         tcp_seq snd_wl2;                /* window update seg ack number */
 299         tcp_seq iss;                    /* initial send sequence number */
 300         tcp_seq irs;                    /* initial receive sequence number */
 301
 302         tcp_seq rcv_nxt;                /* receive next */
 303         tcp_seq rcv_adv;                /* advertised window */
 304         u_int32_t       rcv_wnd;                /* receive window */
 305         tcp_seq rcv_up;                 /* receive urgent pointer */
 306
 307         u_int32_t       snd_wnd;        /* send window */
 308         u_int32_t       snd_cwnd;       /* congestion-controlled window */
 309         u_int32_t       snd_ssthresh;   /* snd_cwnd size threshold for
 310                                          * for slow start exponential to
 311                                          * linear switch
 312                                          */
 313         tcp_seq snd_recover;            /* for use in NewReno Fast Recovery */
 314
 315         u_int   t_maxopd;               /* mss plus options */
 316
 317         u_int32_t       t_rcvtime;      /* time at which a packet was received */
 318         u_int32_t       t_starttime;    /* time connection was established */
 319         int     t_rtttime;              /* tcp clock when rtt calculation was started */
 320         tcp_seq t_rtseq;                /* sequence number being timed */
 321
 322         u_int32_t rfbuf_ts;             /* recv buffer autoscaling timestamp */
 323         u_int32_t rfbuf_cnt;            /* recv buffer autoscaling byte count */
 324
 325         int     t_rxtcur;               /* current retransmit value (ticks) */
 326         u_int   t_maxseg;               /* maximum segment size */
 327         int     t_srtt;                 /* smoothed round-trip time */
 328         int     t_rttvar;               /* variance in round-trip time */
 329
 330         u_int64_t       t_accsleep_ms;  /* accumulated sleep time since last boot */
 331         u_int16_t       t_reassqlen;    /* length of reassembly queue */
 332         u_int16_t       t_rxtshift;     /* log(2) of rexmt exp. backoff */
 333         u_int32_t       t_rttmin;       /* minimum rtt allowed */
 334         u_int32_t       t_rttbest;      /* best rtt we've seen */
 335         u_int32_t       t_rttcur;       /* most recent value of rtt */
 336         u_int32_t       t_rttupdated;   /* number of times rtt sampled */
 337         u_int32_t       t_rxt_conndroptime;     /* retxmt conn gets dropped after this time, when set */
 338         u_int32_t       t_rxtstart;     /* time at which retransmission started */
 339         u_int32_t       max_sndwnd;     /* largest window peer has offered */
 340
 341         int     t_softerror;            /* possible error not yet reported */
 342 /* out-of-band data */
 343         char    t_oobflags;             /* have some */
 344         char    t_iobc;                 /* input character */
 345 #define TCPOOB_HAVEDATA 0x01
 346 #define TCPOOB_HADDATA  0x02
 347 /* RFC 1323 variables */
 348         u_int8_t        snd_scale;      /* window scaling for send window */
 349         u_int8_t        rcv_scale;      /* window scaling for recv window */
 350         u_int8_t        request_r_scale; /* pending window scaling */
 351         u_int8_t        requested_s_scale;
 352         u_int8_t        tcp_cc_index;   /* index of congestion control algorithm */
 353         u_int8_t        t_adaptive_rtimo;       /* Read timeout used as a multiple of RTT */
 354         u_int8_t        t_adaptive_wtimo;       /* Write timeout used as a multiple of RTT */
 355         u_int8_t        t_stretchack_delayed;   /* stretch ack delayed */
 356
 357 /* State for limiting early retransmits when SACK is not enabled */
 358         u_int16_t       t_early_rexmt_count; /* count of early rexmts */
 359         u_int32_t       t_early_rexmt_win; /* window for limiting early rexmts */
 360
 361         u_int32_t       ts_recent;      /* timestamp echo data */
 362
 363         u_int32_t       ts_recent_age;  /* when last updated */
 364         tcp_seq last_ack_sent;
 365 /* RFC 1644 variables */
 366         tcp_cc  cc_send;                /* send connection count */
 367         tcp_cc  cc_recv;                /* receive connection count */
 368 /* RFC 3465 variables */
 369         u_int32_t       t_bytes_acked;  /* ABC "bytes_acked" parameter */
 370
 371         int     t_lastchain;            /* amount of packets chained last time around */
 372         u_int16_t       t_unacksegs;    /* received but unacked segments for delaying acks */
 373         u_int8_t        t_rexmtthresh;  /* duplicate ack threshold for entering fast recovery */
 374         u_int8_t        t_rtimo_probes; /* number of adaptive rtimo probes sent */
 375         u_int32_t       t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */
 376         u_int32_t       t_persist_stop;         /* persistence limit deadline if triggered by ZWP */
 377         u_int32_t       t_notsent_lowat;        /* Low water for not sent data */
 378
 379 /* Receiver state for stretch-ack algorithm */
 380         u_int32_t       rcv_unackwin;   /* to measure win for stretching acks */
 381         u_int32_t       rcv_by_unackwin; /* bytes seen during the last ack-stretching win */
 382         u_int32_t       rcv_nostrack_ts; /* timestamp when stretch ack was disabled automatically */
 383         u_int16_t       rcv_waitforss;  /* wait for packets during slow-start */
 384
 385 /* ECN stats */
 386         u_int16_t       ecn_flags;
 387 #define TE_SETUPSENT            0x0001  /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */
 388 #define TE_SETUPRECEIVED        0x0002  /* Indicate we have received ECN-SETUP SYN or SYN-ACK */
 389 #define TE_SENDIPECT            0x0004  /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */
 390 #define TE_SENDCWR              0x0008  /* Indicate that the next non-retransmit should have the TCP CWR flag set */
 391 #define TE_SENDECE              0x0010  /* Indicate that the next packet should have the TCP ECE flag set */
 392 #define TE_INRECOVERY           0x0020  /* connection entered recovery after receiving ECE */
 393 #define TE_RECV_ECN_CE          0x0040  /* Received IPTOS_ECN_CE marking atleast once */
 394 #define TE_RECV_ECN_ECE         0x0080  /* Received ECE marking atleast once */
 395 #define TE_LOST_SYN             0x0100  /* Lost SYN with ECN setup */
 396 #define TE_LOST_SYNACK          0x0200  /* Lost SYN-ACK with ECN setup */
 397 #define TE_ECN_MODE_ENABLE      0x0400  /* Option ECN mode set to enable */
 398 #define TE_ECN_MODE_DISABLE     0x0800  /* Option ECN mode set to disable */
 399 #define TE_ENABLE_ECN           0x1000  /* Enable negotiation of ECN */
 400 #define TE_ECN_ON               (TE_SETUPSENT | TE_SETUPRECEIVED) /* Indicate ECN was successfully negotiated on a connection) */
 401 #define TE_CEHEURI_SET          0x2000 /* We did our CE-probing at the beginning */
 402 #define TE_CLIENT_SETUP         0x4000  /* setup from client side */
 403
 404         u_int32_t       t_ecn_recv_ce;  /* Received CE from the network */
 405         u_int32_t       t_ecn_recv_cwr; /* Packets received with CWR */
 406         u_int8_t        t_ecn_recv_ce_pkt; /* Received packet with CE-bit set (independent from last_ack_sent) */
 407
 408 /* state for bad retransmit recovery */
 409         u_int32_t       snd_cwnd_prev;  /* cwnd prior to retransmit */
 410         u_int32_t       snd_ssthresh_prev; /* ssthresh prior to retransmit */
 411         tcp_seq snd_recover_prev;       /* snd_recover prior to retransmit */
 412         int     t_srtt_prev;            /* srtt prior to retransmit */
 413         int     t_rttvar_prev;          /* rttvar prior to retransmit */
 414         u_int32_t       t_badrexmt_time; /* bad rexmt detection time */
 415
 416 /* Packet reordering metric */
 417         u_int16_t       t_reorderwin; /* Reordering late time offset */
 418
 419 /* SACK related state */
 420         int16_t snd_numholes;           /* number of holes seen by sender */
 421         tcp_seq sack_newdata;           /* New data xmitted in this recovery
 422                                            episode starts at this seq number */
 423         TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
 424                                                 /* SACK scoreboard (sorted) */
 425         tcp_seq snd_fack;               /* last seq number(+1) sack'd by rcv'r*/
 426         int     rcv_numsacks;           /* # distinct sack blks present */
 427         struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
 428         struct sackhint sackhint;       /* SACK scoreboard hint */
 429
 430         struct mbuf     *t_pktlist_head; /* First packet in transmit chain */
 431         struct mbuf     *t_pktlist_tail; /* Last packet in transmit chain */
 432         u_int32_t       t_pktlist_sentlen; /* total bytes in transmit chain */
 433
 434         u_int32_t       t_keepidle;     /* keepalive idle timer (override global if > 0) */
 435         u_int32_t       t_keepinit;     /* connection timeout, i.e. idle time
 436                                         in SYN_SENT or SYN_RECV state */
 437         u_int32_t       t_keepintvl;    /* interval between keepalives */
 438         u_int32_t       t_keepcnt;      /* number of keepalives before close */
 439
 440         u_int32_t       tso_max_segment_size;   /* TSO maximum segment unit for NIC */
 441         u_int16_t       t_pmtud_lastseg_size;   /* size of the last sent segment */
 442         u_int16_t       t_pmtud_saved_maxopd;   /* MSS saved before performing PMTU-D BlackHole detection */
 443         u_int32_t       t_pmtud_start_ts;       /* Time of PMTUD blackhole detection */
 444
 445         struct
 446         {
 447                 u_int32_t       rxduplicatebytes;
 448                 u_int32_t       rxoutoforderbytes;
 449                 u_int32_t       txretransmitbytes;
 450                 u_int8_t        synrxtshift;
 451                 u_int8_t        unused;
 452                 u_int16_t       unused_pad_to_8;
 453         } t_stat;
 454
 455         /* Background congestion related state */
 456         uint32_t        rtt_hist[N_RTT_BASE];   /* history of minimum RTT */
 457         uint32_t        rtt_count;              /* Number of RTT samples in recent base history */
 458         uint32_t        bg_ssthresh;            /* Slow start threshold until delay increases */
 459         uint32_t        t_flagsext;             /* Another field to accommodate more flags */
 460 #define TF_RXTFINDROP           0x1                     /* Drop conn after retransmitting FIN 3 times */
 461 #define TF_RCVUNACK_WAITSS      0x2             /* set when the receiver should not stretch acks */
 462 #define TF_BWMEAS_INPROGRESS    0x4             /* Indicate BW meas is happening */
 463 #define TF_MEASURESNDBW         0x8             /* Measure send bw on this connection */
 464 #define TF_LRO_OFFLOADED        0x10            /* Connection LRO offloaded */
 465 #define TF_SACK_ENABLE          0x20            /* SACK is enabled */
 466 #define TF_RECOMPUTE_RTT        0x40            /* recompute RTT after spurious retransmit */
 467 #define TF_DETECT_READSTALL     0x80            /* Used to detect a stall during read operation */
 468 #define TF_RECV_THROTTLE        0x100           /* Input throttling active */
 469 #define TF_NOSTRETCHACK         0x200           /* ack every other packet */
 470 #define TF_STREAMEOW            0x400           /* Last packet was small indicating end of write */
 471 #define TF_NOTIMEWAIT           0x800           /* Avoid going into time-wait */
 472 #define TF_SENT_TLPROBE         0x1000          /* Sent data in PTO */
 473 #define TF_PKTS_REORDERED       0x2000          /* Detected reordering */
 474 #define TF_DELAY_RECOVERY       0x4000          /* delay fast recovery */
 475 #define TF_FORCE                0x8000          /* force 1 byte out */
 476 #define TF_DISABLE_STRETCHACK   0x10000         /* auto-disable stretch ack */
 477 #define TF_NOBLACKHOLE_DETECTION 0x20000        /* Disable PMTU blackhole detection */
 478 #define TF_DISABLE_DSACK        0x40000         /* Ignore DSACK due to n/w duplication */
 479 #define TF_RESCUE_RXT           0x80000         /* SACK rescue retransmit */
 480 #define TF_CWND_NONVALIDATED    0x100000        /* cwnd non validated */
 481 #define TF_PROBING              0x200000        /* Trigger probe timeout */
 482 #define TF_FASTOPEN             0x400000        /* TCP Fastopen is enabled */
 483
 484 #if TRAFFIC_MGT
 485         /* Inter-arrival jitter related state */
 486         uint32_t        iaj_rcv_ts;             /* tcp clock when the first packet was received */
 487         uint16_t        iaj_size;               /* Size of packet for iaj measurement */
 488         uint8_t         iaj_small_pkt;          /* Count of packets smaller than iaj_size */
 489         uint8_t         t_pipeack_ind;          /* index for next pipeack sample */
 490         uint16_t        iaj_pktcnt;             /* packet count, to avoid throttling initially */
 491         uint16_t        acc_iaj;                /* Accumulated iaj */
 492         tcp_seq         iaj_rwintop;            /* recent max advertised window */
 493         uint32_t        avg_iaj;                /* Mean */
 494         uint32_t        std_dev_iaj;            /* Standard deviation */
 495 #endif /* TRAFFIC_MGT */
 496         struct bwmeas   *t_bwmeas;              /* State for bandwidth measurement */
 497         uint32_t        t_lropktlen;            /* Bytes in a LRO frame */
 498         tcp_seq         t_idleat;               /* rcv_nxt at idle time */
 499         TAILQ_ENTRY(tcpcb) t_twentry;           /* link for time wait queue */
 500         struct tcp_ccstate      *t_ccstate;     /* congestion control related state */
 501 /* Tail loss probe related state */
 502         tcp_seq         t_tlphighrxt;           /* snd_nxt after PTO */
 503         u_int32_t       t_tlpstart;             /* timestamp at PTO */
 504 /* DSACK data receiver state */
 505         tcp_seq         t_dsack_lseq;           /* DSACK left sequence */
 506         tcp_seq         t_dsack_rseq;           /* DSACK right sequence */
 507 /* DSACK data sender state */
 508         SLIST_HEAD(tcp_rxt_seghead, tcp_rxt_seg) t_rxt_segments;
 509         tcp_seq         t_dsack_lastuna;        /* snd_una when last recovery episode started */
 510 /* state for congestion window validation (draft-ietf-tcpm-newcwv-07) */
 511 #define TCP_PIPEACK_SAMPLE_COUNT        3
 512         u_int32_t       t_pipeack_sample[TCP_PIPEACK_SAMPLE_COUNT];     /* pipeack, bytes acked within RTT */
 513         tcp_seq         t_pipeack_lastuna; /* una when pipeack measurement started */
 514         u_int32_t       t_pipeack;
 515         u_int32_t       t_lossflightsize;
 516
 517 #if MPTCP
 518         u_int32_t       t_mpflags;              /* flags for multipath TCP */
 519
 520 #define TMPF_PREESTABLISHED     0x00000001 /* conn in pre-established state */
 521 #define TMPF_SENT_KEYS          0x00000002 /* indicates that keys were sent */
 522 #define TMPF_MPTCP_TRUE         0x00000004 /* negotiated MPTCP successfully */
 523 #define TMPF_MPTCP_RCVD_KEY     0x00000008 /* state for 3-way handshake */
 524 #define TMPF_SND_MPPRIO         0x00000010 /* send priority of subflow */
 525 #define TMPF_SND_REM_ADDR       0x00000020 /* initiate address removal */
 526 #define TMPF_UNUSED             0x00000040 /* address addition acked by peer */
 527 #define TMPF_JOINED_FLOW        0x00000080 /* Indicates additional flow */
 528 #define TMPF_BACKUP_PATH        0x00000100 /* Indicates backup path */
 529 #define TMPF_MPTCP_ACKNOW       0x00000200 /* Send Data ACK */
 530 #define TMPF_SEND_DSN           0x00000400 /* Send DSN mapping */
 531 #define TMPF_SEND_DFIN          0x00000800 /* Send Data FIN */
 532 #define TMPF_RECV_DFIN          0x00001000 /* Recv Data FIN */
 533 #define TMPF_SENT_JOIN          0x00002000 /* Sent Join */
 534 #define TMPF_RECVD_JOIN         0x00004000 /* Received Join */
 535 #define TMPF_RESET              0x00008000 /* Send RST */
 536 #define TMPF_TCP_FALLBACK       0x00010000 /* Fallback to TCP */
 537 #define TMPF_FASTCLOSE          0x00020000 /* Send Fastclose option */
 538 #define TMPF_EMBED_DSN          0x00040000 /* tp has DSN mapping */
 539 #define TMPF_MPTCP_READY        0x00080000 /* Can send DSS options on data */
 540 #define TMPF_INFIN_SENT         0x00100000 /* Sent infinite mapping */
 541 #define TMPF_SND_MPFAIL         0x00200000 /* Received mapping csum failure */
 542 #define TMPF_FASTJOIN_SEND      0x00400000 /* Fast join early data send */
 543 #define TMPF_FASTJOINBY2_SEND   0x00800000 /* Fast join send after 3 WHS */
 544 #define TMPF_MPCAP_RETRANSMIT   0x01000000 /* Retransmission of 3rd ACK */
 545
 546         tcp_seq                 t_mpuna;        /* unacknowledged sequence */
 547         void                    *t_mptcb;       /* pointer to MPTCP TCB */
 548         struct mpt_dsn_map      t_rcv_map;      /* Receive mapping list */
 549         u_int8_t                t_local_aid;    /* Addr Id for authentication */
 550         u_int8_t                t_rem_aid;      /* Addr ID of another subflow */
 551         u_int8_t                t_mprxtshift;   /* join retransmission */
 552 #endif /* MPTCP */
 553
 554 #define TFO_F_OFFER_COOKIE      0x01 /* We will offer a cookie */
 555 #define TFO_F_COOKIE_VALID      0x02 /* The received cookie is valid */
 556 #define TFO_F_COOKIE_REQ        0x04 /* Client requested a new cookie */
 557 #define TFO_F_COOKIE_SENT       0x08 /* Client did send a cookie in the SYN */
 558 #define TFO_F_SYN_LOSS          0x10 /* A SYN-loss triggered a fallback to regular TCP on the client-side */
 559 #define TFO_F_NO_RCVPROBING     0x20 /* This network is guaranteed to support TFO in the downstream direction */
 560 #define TFO_F_NO_SNDPROBING     0x40 /* This network is guaranteed to support TFO in the upstream direction */
 561         u_int8_t                t_tfo_flags;
 562 #define TFO_S_SYNDATA_RCV       0x01 /* SYN+data has been received */
 563 #define TFO_S_COOKIEREQ_RECV    0x02 /* TFO-cookie request received */
 564 #define TFO_S_COOKIE_SENT       0x04 /* TFO-cookie announced in SYN/ACK */
 565 #define TFO_S_COOKIE_INVALID    0x08 /* Received TFO-cookie is invalid */
 566 #define TFO_S_COOKIE_REQ        0x10 /* TFO-cookie requested within the SYN */
 567 #define TFO_S_COOKIE_RCV        0x20 /* TFO-cookie received in SYN/ACK */
 568 #define TFO_S_SYN_DATA_SENT     0x40 /* SYN+data sent */
 569 #define TFO_S_SYN_DATA_ACKED    0x80 /* SYN+data has been acknowledged in SYN/ACK */
 570 #define TFO_S_SYN_LOSS          0x0100 /* SYN+TFO has been lost - fallback to regular TCP */
 571         u_int16_t               t_tfo_stats;
 572
 573         u_int8_t                t_tfo_probes; /* TFO-probes we did send */
 574 /*
 575  * This here is the TFO-probing state-machine. Transitions are as follows:
 576  *
 577  * Current state: PROBE_NONE
 578  *                Event: SYN+DATA acknowledged
 579  *                       Action: Transition to PROBE_PROBING and set keepalive-timer
 580  *
 581  * Current state: PROBE_PROBING (initial state)
 582  *                Event: Receive data
 583  *                       Action: Transition to PROBE_NONE and cancel keepalive-timer
 584  *                Event: Receive ACK that does not indicate a hole
 585  *                       Action: Transition to PROBE_NONE and cancel keepalive-timer
 586  *                Event: Receive ACK that indicates a hole
 587  *                       Action: Transition to PROBE_WAIT_DATA and set a short timer
 588  *                               to wait for the final segment.
 589  *                Event: Keepalive-timeout (did not receive any segment)
 590  *                       Action: Signal ETIMEDOUT as with regular keepalive-timers
 591  *
 592  * Current state: PROBE_WAIT_DATA
 593  *                Event: Receive data
 594  *                       Action: Transition to PROBE_NONE and cancel keepalive-timer
 595  *                Event: Data-timeout (did not receive the expected data)
 596  *                       Action: Signal ENODATA up to the app and close everything.
 597  */
 598 #define TFO_PROBE_NONE          0 /* Not probing now */
 599 #define TFO_PROBE_PROBING       1 /* Sending out TCP-keepalives waiting for reply */
 600 #define TFO_PROBE_WAIT_DATA     2 /* Received reply, waiting for data */
 601         u_int8_t                t_tfo_probe_state;
 602
 603         u_int32_t       t_rcvoopack;            /* out-of-order packets received */
 604         u_int32_t       t_pawsdrop;             /* segments dropped due to PAWS */
 605         u_int32_t       t_sack_recovery_episode; /* SACK recovery episodes */
 606         u_int32_t       t_reordered_pkts;       /* packets reorderd */
 607         u_int32_t       t_dsack_sent;           /* Sent DSACK notification */
 608         u_int32_t       t_dsack_recvd;          /* Received a valid DSACK option */
 609 };
 610
 611 #define IN_FASTRECOVERY(tp)     (tp->t_flags & TF_FASTRECOVERY)
 612 #define SACK_ENABLED(tp)        (tp->t_flagsext & TF_SACK_ENABLE)
 613
 614 /*
 615  * If the connection is in a throttled state due to advisory feedback from
 616  * the interface output queue, reset that state. We do this in favor
 617  * of entering recovery because the data transfer during recovery
 618  * should be just a trickle and it will help to improve performance.
 619  * We also do not want to back off twice in the same RTT.
 620  */
 621 #define ENTER_FASTRECOVERY(_tp_) do {                           \
 622         (_tp_)->t_flags |= TF_FASTRECOVERY;                     \
 623         if (INP_IS_FLOW_CONTROLLED((_tp_)->t_inpcb))            \
 624                 inp_reset_fc_state((_tp_)->t_inpcb);            \
 625         if (!SLIST_EMPTY(&tp->t_rxt_segments))                  \
 626                 tcp_rxtseg_clean(tp);                           \
 627 } while(0)
 628
 629 #define EXIT_FASTRECOVERY(_tp_) do {            \
 630         (_tp_)->t_flags &= ~TF_FASTRECOVERY;    \
 631         (_tp_)->t_dupacks = 0;                  \
 632         (_tp_)->t_rexmtthresh = tcprexmtthresh; \
 633         (_tp_)->t_bytes_acked = 0;              \
 634         (_tp_)->ecn_flags &= ~TE_INRECOVERY;    \
 635         (_tp_)->t_timer[TCPT_PTO] = 0;          \
 636         (_tp_)->t_flagsext &= ~TF_RESCUE_RXT;   \
 637         (_tp_)->t_lossflightsize = 0;           \
 638 } while(0)
 639
 640 /*
 641  * When the number of duplicate acks received is less than
 642  * the retransmit threshold, use Limited Transmit algorithm
 643  */
 644 extern int tcprexmtthresh;
 645 #define ALLOW_LIMITED_TRANSMIT(_tp_) \
 646         ((_tp_)->t_dupacks > 0 && \
 647         (_tp_)->t_dupacks < (_tp_)->t_rexmtthresh && \
 648         ((_tp_)->t_flagsext & (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) \
 649             != (TF_PKTS_REORDERED|TF_DELAY_RECOVERY))
 650
 651 /*
 652  * This condition is true is timestamp option is supported
 653  * on a connection.
 654  */
 655 #define TSTMP_SUPPORTED(_tp_) \
 656         (((_tp_)->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) == \
 657                 (TF_REQ_TSTMP|TF_RCVD_TSTMP))
 658
 659 /*
 660  * This condition is true if window scale option is supported
 661  * on a connection
 662  */
 663 #define TCP_WINDOW_SCALE_ENABLED(_tp_) \
 664         (((_tp_)->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == \
 665                 (TF_RCVD_SCALE|TF_REQ_SCALE))
 666
 667 /* Is ECN enabled end-to-end */
 668 #define TCP_ECN_ENABLED(_tp_) \
 669         (((_tp_)->ecn_flags & (TE_ECN_ON)) == (TE_ECN_ON))
 670
 671 /*
 672  * Gives number of bytes acked by this ack
 673  */
 674 #define BYTES_ACKED(_th_, _tp_) \
 675         ((_th_)->th_ack - (_tp_)->snd_una)
 676
 677 /* Returns true if a DSACK option should be sent */
 678 #define TCP_SEND_DSACK_OPT(_tp_) \
 679         ((_tp_)->t_dsack_lseq > 0 && (_tp_)->t_dsack_rseq > 0)
 680
 681 /* Check if DSACK option should be processed */
 682 #define TCP_DSACK_ENABLED(tp)   (tcp_dsack_enable == 1 && \
 683     !(tp->t_flagsext & TF_DISABLE_DSACK))
 684
 685 /*
 686  * Returns true if a DSACK sequence is within the max send window that will
 687  * be accepted. In order to set a window to validate sequence numbers, the
 688  * max send window within which a DSACK option is processed is limited.
 689  *
 690  * We need to choose a maximum window to check if the sequence number is
 691  * within the window. One arbitrary choice is 256 * MSS because if the
 692  * window is as large as 256 segments it might be big enough to ignore the
 693  * DSACK option. Choosing a much larger limit means that the memory for
 694  * retransmit segments can be held for a longer time.
 695  */
 696 #define TCP_DSACK_MAX_SEND_WINDOW(_tp_) ((_tp_)->t_maxseg << 8)
 697 #define TCP_DSACK_SEQ_IN_WINDOW(_tp_, _seq_, _una_) \
 698     (SEQ_LEQ((_seq_), (_tp_)->snd_max) && \
 699     SEQ_GEQ((_seq_), ((_una_) - TCP_DSACK_MAX_SEND_WINDOW(_tp_))))
 700
 701
 702 enum tcp_cc_event {
 703         TCP_CC_CWND_INIT,       /* 0 */
 704         TCP_CC_INSEQ_ACK_RCVD,  /* 1 */
 705         TCP_CC_ACK_RCVD,        /* 2 */
 706         TCP_CC_ENTER_FASTRECOVERY, /* 3 */
 707         TCP_CC_IN_FASTRECOVERY, /* 4 */
 708         TCP_CC_EXIT_FASTRECOVERY,  /* 5 */
 709         TCP_CC_PARTIAL_ACK,     /* 6 */
 710         TCP_CC_IDLE_TIMEOUT,    /* 7 */
 711         TCP_CC_REXMT_TIMEOUT,   /* 8 */
 712         TCP_CC_ECN_RCVD,        /* 9 */
 713         TCP_CC_BAD_REXMT_RECOVERY, /* 10 */
 714         TCP_CC_OUTPUT_ERROR,    /* 11 */
 715         TCP_CC_CHANGE_ALGO,     /* 12 */
 716         TCP_CC_FLOW_CONTROL,    /* 13 */
 717         TCP_CC_SUSPEND,         /* 14 */
 718         TCP_CC_LIMITED_TRANSMIT, /* 15 */
 719         TCP_CC_EARLY_RETRANSMIT, /* 16 */
 720         TCP_CC_TLP_RECOVERY,    /* 17 */
 721         TCP_CC_TLP_RECOVER_LASTPACKET, /* 18 */
 722         TCP_CC_DELAY_FASTRECOVERY, /* 19 */
 723         TCP_CC_TLP_IN_FASTRECOVERY, /* 20 */
 724         TCP_CC_DSACK_BAD_REXMT  /* 21 */
 725 };
 726
 727 /*
 728  * Structure to hold TCP options that are only used during segment
 729  * processing (in tcp_input), but not held in the tcpcb.
 730  * It's basically used to reduce the number of parameters
 731  * to tcp_dooptions.
 732  */
 733 struct tcpopt {
 734         u_int32_t       to_flags;               /* which options are present */
 735 #define TOF_TS          0x0001          /* timestamp */
 736 #define TOF_MSS         0x0010
 737 #define TOF_SCALE       0x0020
 738 #define TOF_SIGNATURE   0x0040  /* signature option present */
 739 #define TOF_SIGLEN      0x0080  /* signature length valid (RFC2385) */
 740 #define TOF_SACK        0x0100          /* Peer sent SACK option */
 741 #define TOF_MPTCP       0x0200  /* MPTCP options to be dropped */
 742 #define TOF_TFO         0x0400  /* TFO cookie option present */
 743 #define TOF_TFOREQ      0x0800  /* TFO cookie request present */
 744         u_int32_t               to_tsval;
 745         u_int32_t               to_tsecr;
 746         u_int16_t       to_mss;
 747         u_int8_t        to_requested_s_scale;
 748         u_int8_t        to_nsacks;      /* number of SACK blocks */
 749         u_char          *to_sacks;      /* pointer to the first SACK blocks */
 750         u_char          *to_tfo;        /* pointer to the TFO cookie */
 751 };
 752
 753 /*
 754  * The TAO cache entry which is stored in the protocol family specific
 755  * portion of the route metrics.
 756  */
 757 struct rmxp_tao {
 758         tcp_cc  tao_cc;                 /* latest CC in valid SYN */
 759         tcp_cc  tao_ccsent;             /* latest CC sent to peer */
 760         u_short tao_mssopt;             /* peer's cached MSS */
 761 #ifdef notyet
 762         u_short tao_flags;              /* cache status flags */
 763 #define TAOF_DONT       0x0001          /* peer doesn't understand rfc1644 */
 764 #define TAOF_OK         0x0002          /* peer does understand rfc1644 */
 765 #define TAOF_UNDEF      0               /* we don't know yet */
 766 #endif /* notyet */
 767 };
 768 #define rmx_taop(r)     ((struct rmxp_tao *)(r).rmx_filler)
 769
 770 #define intotcpcb(ip)   ((struct tcpcb *)(ip)->inp_ppcb)
 771 #define sototcpcb(so)   (intotcpcb(sotoinpcb(so)))
 772
 773 /*
 774  * The rtt measured is in milliseconds as the timestamp granularity is
 775  * a millisecond. The smoothed round-trip time and estimated variance
 776  * are stored as fixed point numbers scaled by the values below.
 777  * For convenience, these scales are also used in smoothing the average
 778  * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
 779  * With these scales, srtt has 5 bits to the right of the binary point,
 780  * and thus an "ALPHA" of 0.875.  rttvar has 4 bits to the right of the
 781  * binary point, and is smoothed with an ALPHA of 0.75.
 782  */
 783 #define TCP_RTT_SCALE           32      /* multiplier for srtt; 3 bits frac. */
 784 #define TCP_RTT_SHIFT           5       /* shift for srtt; 5 bits frac. */
 785 #define TCP_RTTVAR_SCALE        16      /* multiplier for rttvar; 4 bits */
 786 #define TCP_RTTVAR_SHIFT        4       /* shift for rttvar; 4 bits */
 787 #define TCP_DELTA_SHIFT         2       /* see tcp_input.c */
 788
 789 /* TFO-specific defines */
 790 #define TFO_COOKIE_LEN_MIN      4
 791 #define TFO_COOKIE_LEN_DEFAULT  8
 792 #define TFO_COOKIE_LEN_MAX      16
 793
 794 /*
 795  * The initial retransmission should happen at rtt + 4 * rttvar.
 796  * Because of the way we do the smoothing, srtt and rttvar
 797  * will each average +1/2 tick of bias.  When we compute
 798  * the retransmit timer, we want 1/2 tick of rounding and
 799  * 1 extra tick because of +-1/2 tick uncertainty in the
 800  * firing of the timer.  The bias will give us exactly the
 801  * 1.5 tick we need.  But, because the bias is
 802  * statistical, we have to test that we don't drop below
 803  * the minimum feasible timer (which is 2 ticks).
 804  * This version of the macro adapted from a paper by Lawrence
 805  * Brakmo and Larry Peterson which outlines a problem caused
 806  * by insufficient precision in the original implementation,
 807  * which results in inappropriately large RTO values for very
 808  * fast networks.
 809  */
 810 #define TCP_REXMTVAL(tp) \
 811         max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT))  \
 812           + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
 813
 814 /*
 815  * Jaguar compatible TCP control block, for xtcpcb
 816  * Does not have the old fields
 817  */
 818 struct otcpcb {
 819 #else
 820 struct tseg_qent;
 821 _TCPCB_LIST_HEAD(tsegqe_head, tseg_qent);
 822
 823 struct tcpcb {
 824 #endif /* KERNEL_PRIVATE */
 825 #if defined(KERNEL_PRIVATE)
 826         u_int32_t t_segq;
 827 #else
 828         struct  tsegqe_head t_segq;
 829 #endif /* KERNEL_PRIVATE */
 830         int     t_dupacks;              /* consecutive dup acks recd */
 831         u_int32_t unused;               /* unused now: was t_template */
 832
 833         int     t_timer[TCPT_NTIMERS_EXT];      /* tcp timers */
 834
 835         _TCPCB_PTR(struct inpcb *) t_inpcb;     /* back pointer to internet pcb */
 836         int     t_state;                /* state of this connection */
 837         u_int   t_flags;
 838 #define TF_ACKNOW       0x00001         /* ack peer immediately */
 839 #define TF_DELACK       0x00002         /* ack, but try to delay it */
 840 #define TF_NODELAY      0x00004         /* don't delay packets to coalesce */
 841 #define TF_NOOPT        0x00008         /* don't use tcp options */
 842 #define TF_SENTFIN      0x00010         /* have sent FIN */
 843 #define TF_REQ_SCALE    0x00020         /* have/will request window scaling */
 844 #define TF_RCVD_SCALE   0x00040         /* other side has requested scaling */
 845 #define TF_REQ_TSTMP    0x00080         /* have/will request timestamps */
 846 #define TF_RCVD_TSTMP   0x00100         /* a timestamp was received in SYN */
 847 #define TF_SACK_PERMIT  0x00200         /* other side said I could SACK */
 848 #define TF_NEEDSYN      0x00400         /* send SYN (implicit state) */
 849 #define TF_NEEDFIN      0x00800         /* send FIN (implicit state) */
 850 #define TF_NOPUSH       0x01000         /* don't push */
 851 #define TF_REQ_CC       0x02000         /* have/will request CC */
 852 #define TF_RCVD_CC      0x04000         /* a CC was received in SYN */
 853 #define TF_SENDCCNEW    0x08000         /* send CCnew instead of CC in SYN */
 854 #define TF_MORETOCOME   0x10000         /* More data to be appended to sock */
 855 #define TF_LQ_OVERFLOW  0x20000         /* listen queue overflow */
 856 #define TF_RXWIN0SENT   0x40000         /* sent a receiver win 0 in response */
 857 #define TF_SLOWLINK     0x80000         /* route is a on a modem speed link */
 858
 859         int     t_force;                /* 1 if forcing out a byte */
 860
 861         tcp_seq snd_una;                /* send unacknowledged */
 862         tcp_seq snd_max;                /* highest sequence number sent;
 863                                          * used to recognize retransmits
 864                                          */
 865         tcp_seq snd_nxt;                /* send next */
 866         tcp_seq snd_up;                 /* send urgent pointer */
 867
 868         tcp_seq snd_wl1;                /* window update seg seq number */
 869         tcp_seq snd_wl2;                /* window update seg ack number */
 870         tcp_seq iss;                    /* initial send sequence number */
 871         tcp_seq irs;                    /* initial receive sequence number */
 872
 873         tcp_seq rcv_nxt;                /* receive next */
 874         tcp_seq rcv_adv;                /* advertised window */
 875         u_int32_t rcv_wnd;              /* receive window */
 876         tcp_seq rcv_up;                 /* receive urgent pointer */
 877
 878         u_int32_t snd_wnd;              /* send window */
 879         u_int32_t snd_cwnd;             /* congestion-controlled window */
 880         u_int32_t snd_ssthresh;         /* snd_cwnd size threshold for
 881                                          * for slow start exponential to
 882                                          * linear switch
 883                                          */
 884         u_int   t_maxopd;               /* mss plus options */
 885
 886         u_int32_t t_rcvtime;            /* time at which a packet was received */
 887         u_int32_t t_starttime;          /* time connection was established */
 888         int     t_rtttime;              /* round trip time */
 889         tcp_seq t_rtseq;                /* sequence number being timed */
 890
 891         int     t_rxtcur;               /* current retransmit value (ticks) */
 892         u_int   t_maxseg;               /* maximum segment size */
 893         int     t_srtt;                 /* smoothed round-trip time */
 894         int     t_rttvar;               /* variance in round-trip time */
 895
 896         int     t_rxtshift;             /* log(2) of rexmt exp. backoff */
 897         u_int   t_rttmin;               /* minimum rtt allowed */
 898         u_int32_t t_rttupdated;         /* number of times rtt sampled */
 899         u_int32_t max_sndwnd;           /* largest window peer has offered */
 900
 901         int     t_softerror;            /* possible error not yet reported */
 902 /* out-of-band data */
 903         char    t_oobflags;             /* have some */
 904         char    t_iobc;                 /* input character */
 905 #define TCPOOB_HAVEDATA 0x01
 906 #define TCPOOB_HADDATA  0x02
 907 /* RFC 1323 variables */
 908         u_char  snd_scale;              /* window scaling for send window */
 909         u_char  rcv_scale;              /* window scaling for recv window */
 910         u_char  request_r_scale;        /* pending window scaling */
 911         u_char  requested_s_scale;
 912         u_int32_t ts_recent;            /* timestamp echo data */
 913
 914         u_int32_t ts_recent_age;        /* when last updated */
 915         tcp_seq last_ack_sent;
 916 /* RFC 1644 variables */
 917         tcp_cc  cc_send;                /* send connection count */
 918         tcp_cc  cc_recv;                /* receive connection count */
 919         tcp_seq snd_recover;            /* for use in fast recovery */
 920 /* experimental */
 921         u_int32_t snd_cwnd_prev;        /* cwnd prior to retransmit */
 922         u_int32_t snd_ssthresh_prev;    /* ssthresh prior to retransmit */
 923         u_int32_t t_badrxtwin;          /* window for retransmit recovery */
 924 };
 925
 926 #define        tcps_ecn_setup  tcps_ecn_client_success
 927 #define        tcps_sent_cwr   tcps_ecn_recv_ece
 928 #define        tcps_sent_ece   tcps_ecn_sent_ece
 929
 930 /*
 931  * TCP statistics.
 932  * Many of these should be kept per connection,
 933  * but that's inconvenient at the moment.
 934  */
 935 struct  tcpstat {
 936         u_int32_t       tcps_connattempt;       /* connections initiated */
 937         u_int32_t       tcps_accepts;           /* connections accepted */
 938         u_int32_t       tcps_connects;          /* connections established */
 939         u_int32_t       tcps_drops;             /* connections dropped */
 940         u_int32_t       tcps_conndrops;         /* embryonic connections dropped */
 941         u_int32_t       tcps_closed;            /* conn. closed (includes drops) */
 942         u_int32_t       tcps_segstimed;         /* segs where we tried to get rtt */
 943         u_int32_t       tcps_rttupdated;        /* times we succeeded */
 944         u_int32_t       tcps_delack;            /* delayed acks sent */
 945         u_int32_t       tcps_timeoutdrop;       /* conn. dropped in rxmt timeout */
 946         u_int32_t       tcps_rexmttimeo;        /* retransmit timeouts */
 947         u_int32_t       tcps_persisttimeo;      /* persist timeouts */
 948         u_int32_t       tcps_keeptimeo;         /* keepalive timeouts */
 949         u_int32_t       tcps_keepprobe;         /* keepalive probes sent */
 950         u_int32_t       tcps_keepdrops;         /* connections dropped in keepalive */
 951
 952         u_int32_t       tcps_sndtotal;          /* total packets sent */
 953         u_int32_t       tcps_sndpack;           /* data packets sent */
 954         u_int32_t       tcps_sndbyte;           /* data bytes sent */
 955         u_int32_t       tcps_sndrexmitpack;     /* data packets retransmitted */
 956         u_int32_t       tcps_sndrexmitbyte;     /* data bytes retransmitted */
 957         u_int32_t       tcps_sndacks;           /* ack-only packets sent */
 958         u_int32_t       tcps_sndprobe;          /* window probes sent */
 959         u_int32_t       tcps_sndurg;            /* packets sent with URG only */
 960         u_int32_t       tcps_sndwinup;          /* window update-only packets sent */
 961         u_int32_t       tcps_sndctrl;           /* control (SYN|FIN|RST) packets sent */
 962
 963         u_int32_t       tcps_rcvtotal;          /* total packets received */
 964         u_int32_t       tcps_rcvpack;           /* packets received in sequence */
 965         u_int32_t       tcps_rcvbyte;           /* bytes received in sequence */
 966         u_int32_t       tcps_rcvbadsum;         /* packets received with ccksum errs */
 967         u_int32_t       tcps_rcvbadoff;         /* packets received with bad offset */
 968         u_int32_t       tcps_rcvmemdrop;        /* packets dropped for lack of memory */
 969         u_int32_t       tcps_rcvshort;          /* packets received too short */
 970         u_int32_t       tcps_rcvduppack;        /* duplicate-only packets received */
 971         u_int32_t       tcps_rcvdupbyte;        /* duplicate-only bytes received */
 972         u_int32_t       tcps_rcvpartduppack;    /* packets with some duplicate data */
 973         u_int32_t       tcps_rcvpartdupbyte;    /* dup. bytes in part-dup. packets */
 974         u_int32_t       tcps_rcvoopack;         /* out-of-order packets received */
 975         u_int32_t       tcps_rcvoobyte;         /* out-of-order bytes received */
 976         u_int32_t       tcps_rcvpackafterwin;   /* packets with data after window */
 977         u_int32_t       tcps_rcvbyteafterwin;   /* bytes rcvd after window */
 978         u_int32_t       tcps_rcvafterclose;     /* packets rcvd after "close" */
 979         u_int32_t       tcps_rcvwinprobe;       /* rcvd window probe packets */
 980         u_int32_t       tcps_rcvdupack;         /* rcvd duplicate acks */
 981         u_int32_t       tcps_rcvacktoomuch;     /* rcvd acks for unsent data */
 982         u_int32_t       tcps_rcvackpack;        /* rcvd ack packets */
 983         u_int32_t       tcps_rcvackbyte;        /* bytes acked by rcvd acks */
 984         u_int32_t       tcps_rcvwinupd;         /* rcvd window update packets */
 985         u_int32_t       tcps_pawsdrop;          /* segments dropped due to PAWS */
 986         u_int32_t       tcps_predack;           /* times hdr predict ok for acks */
 987         u_int32_t       tcps_preddat;           /* times hdr predict ok for data pkts */
 988         u_int32_t       tcps_pcbcachemiss;
 989         u_int32_t       tcps_cachedrtt;         /* times cached RTT in route updated */
 990         u_int32_t       tcps_cachedrttvar;      /* times cached rttvar updated */
 991         u_int32_t       tcps_cachedssthresh;    /* times cached ssthresh updated */
 992         u_int32_t       tcps_usedrtt;           /* times RTT initialized from route */
 993         u_int32_t       tcps_usedrttvar;        /* times RTTVAR initialized from rt */
 994         u_int32_t       tcps_usedssthresh;      /* times ssthresh initialized from rt*/
 995         u_int32_t       tcps_persistdrop;       /* timeout in persist state */
 996         u_int32_t       tcps_badsyn;            /* bogus SYN, e.g. premature ACK */
 997         u_int32_t       tcps_mturesent;         /* resends due to MTU discovery */
 998         u_int32_t       tcps_listendrop;        /* listen queue overflows */
 999
1000         /* new stats from FreeBSD 5.4 sync up */
1001         u_int32_t       tcps_minmssdrops;       /* average minmss too low drops */
1002         u_int32_t       tcps_sndrexmitbad;      /* unnecessary packet retransmissions */
1003         u_int32_t       tcps_badrst;            /* ignored RSTs in the window */
1004
1005         u_int32_t       tcps_sc_added;          /* entry added to syncache */
1006         u_int32_t       tcps_sc_retransmitted;  /* syncache entry was retransmitted */
1007         u_int32_t       tcps_sc_dupsyn;         /* duplicate SYN packet */
1008         u_int32_t       tcps_sc_dropped;        /* could not reply to packet */
1009         u_int32_t       tcps_sc_completed;      /* successful extraction of entry */
1010         u_int32_t       tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
1011         u_int32_t       tcps_sc_cacheoverflow;  /* syncache cache limit hit */
1012         u_int32_t       tcps_sc_reset;          /* RST removed entry from syncache */
1013         u_int32_t       tcps_sc_stale;          /* timed out or listen socket gone */
1014         u_int32_t       tcps_sc_aborted;        /* syncache entry aborted */
1015         u_int32_t       tcps_sc_badack;         /* removed due to bad ACK */
1016         u_int32_t       tcps_sc_unreach;        /* ICMP unreachable received */
1017         u_int32_t       tcps_sc_zonefail;       /* zalloc() failed */
1018         u_int32_t       tcps_sc_sendcookie;     /* SYN cookie sent */
1019         u_int32_t       tcps_sc_recvcookie;     /* SYN cookie received */
1020
1021         u_int32_t       tcps_hc_added;          /* entry added to hostcache */
1022         u_int32_t       tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
1023
1024         /* SACK related stats */
1025         u_int32_t       tcps_sack_recovery_episode; /* SACK recovery episodes */
1026         u_int32_t       tcps_sack_rexmits;          /* SACK rexmit segments   */
1027         u_int32_t       tcps_sack_rexmit_bytes;     /* SACK rexmit bytes      */
1028         u_int32_t       tcps_sack_rcv_blocks;       /* SACK blocks (options) received */
1029         u_int32_t       tcps_sack_send_blocks;      /* SACK blocks (options) sent     */
1030         u_int32_t       tcps_sack_sboverflow;       /* SACK sendblock overflow   */
1031
1032         u_int32_t       tcps_bg_rcvtotal;       /* total background packets received */
1033         u_int32_t       tcps_rxtfindrop;        /* drop conn after retransmitting FIN */
1034         u_int32_t       tcps_fcholdpacket;      /* packets withheld because of flow control */
1035
1036         /* LRO related stats */
1037         u_int32_t       tcps_coalesced_pack;    /* number of coalesced packets */
1038         u_int32_t       tcps_flowtbl_full;      /* times flow table was full */
1039         u_int32_t       tcps_flowtbl_collision; /* collisions in flow tbl */
1040         u_int32_t       tcps_lro_twopack;       /* 2 packets coalesced */
1041         u_int32_t       tcps_lro_multpack;      /* 3 or 4 pkts coalesced */
1042         u_int32_t       tcps_lro_largepack;     /* 5 or more pkts coalesced */
1043
1044         u_int32_t       tcps_limited_txt;       /* Limited transmit used */
1045         u_int32_t       tcps_early_rexmt;       /* Early retransmit used */
1046         u_int32_t       tcps_sack_ackadv;       /* Cumulative ack advanced along with sack */
1047
1048         /* Checksum related stats */
1049         u_int32_t       tcps_rcv_swcsum;        /* tcp swcksum (inbound), packets */
1050         u_int32_t       tcps_rcv_swcsum_bytes;  /* tcp swcksum (inbound), bytes */
1051         u_int32_t       tcps_rcv6_swcsum;       /* tcp6 swcksum (inbound), packets */
1052         u_int32_t       tcps_rcv6_swcsum_bytes; /* tcp6 swcksum (inbound), bytes */
1053         u_int32_t       tcps_snd_swcsum;        /* tcp swcksum (outbound), packets */
1054         u_int32_t       tcps_snd_swcsum_bytes;  /* tcp swcksum (outbound), bytes */
1055         u_int32_t       tcps_snd6_swcsum;       /* tcp6 swcksum (outbound), packets */
1056         u_int32_t       tcps_snd6_swcsum_bytes; /* tcp6 swcksum (outbound), bytes */
1057         u_int32_t       tcps_msg_unopkts;       /* unordered packet on TCP msg stream */
1058         u_int32_t       tcps_msg_unoappendfail; /* failed to append unordered pkt */
1059         u_int32_t       tcps_msg_sndwaithipri; /* send is waiting for high priority data */
1060
1061         /* MPTCP Related stats */
1062         u_int32_t       tcps_invalid_mpcap;     /* Invalid MPTCP capable opts */
1063         u_int32_t       tcps_invalid_joins;     /* Invalid MPTCP joins */
1064         u_int32_t       tcps_mpcap_fallback;    /* TCP fallback in primary */
1065         u_int32_t       tcps_join_fallback;     /* No MPTCP in secondary */
1066         u_int32_t       tcps_estab_fallback;    /* DSS option dropped */
1067         u_int32_t       tcps_invalid_opt;       /* Catchall error stat */
1068         u_int32_t       tcps_mp_outofwin;       /* Packet lies outside the
1069                                                    shared recv window */
1070         u_int32_t       tcps_mp_reducedwin;     /* Reduced subflow window */
1071         u_int32_t       tcps_mp_badcsum;        /* Bad DSS csum */
1072         u_int32_t       tcps_mp_oodata;         /* Out of order data */
1073         u_int32_t       tcps_mp_switches;       /* number of subflow switch */
1074         u_int32_t       tcps_mp_rcvtotal;       /* number of rcvd packets */
1075         u_int32_t       tcps_mp_rcvbytes;       /* number of bytes received */
1076         u_int32_t       tcps_mp_sndpacks;       /* number of data packs sent */
1077         u_int32_t       tcps_mp_sndbytes;       /* number of bytes sent */
1078         u_int32_t       tcps_join_rxmts;        /* join ack retransmits */
1079         u_int32_t       tcps_tailloss_rto;      /* RTO due to tail loss */
1080         u_int32_t       tcps_reordered_pkts;    /* packets reorderd */
1081         u_int32_t       tcps_recovered_pkts;    /* recovered after loss */
1082         u_int32_t       tcps_pto;               /* probe timeout */
1083         u_int32_t       tcps_rto_after_pto;     /* RTO after a probe */
1084         u_int32_t       tcps_tlp_recovery;      /* TLP induced fast recovery */
1085         u_int32_t       tcps_tlp_recoverlastpkt; /* TLP recoverd last pkt */
1086         u_int32_t       tcps_ecn_client_success; /* client-side connection negotiated ECN */
1087         u_int32_t       tcps_ecn_recv_ece;      /* ECE received, sent CWR */
1088         u_int32_t       tcps_ecn_sent_ece;      /* Sent ECE notification */
1089         u_int32_t       tcps_detect_reordering; /* Detect pkt reordering */
1090         u_int32_t       tcps_delay_recovery;    /* Delay fast recovery */
1091         u_int32_t       tcps_avoid_rxmt;        /* Retransmission was avoided */
1092         u_int32_t       tcps_unnecessary_rxmt;  /* Retransmission was not needed */
1093         u_int32_t       tcps_nostretchack;      /* disabled stretch ack algorithm on a connection */
1094         u_int32_t       tcps_rescue_rxmt;       /* SACK rescue retransmit */
1095         u_int32_t       tcps_pto_in_recovery;   /* rescue retransmit in fast recovery */
1096         u_int32_t       tcps_pmtudbh_reverted;  /* PMTU Blackhole detection, segment size reverted */
1097
1098         /* DSACK related statistics */
1099         u_int32_t       tcps_dsack_disable;     /* DSACK disabled due to n/w duplication */
1100         u_int32_t       tcps_dsack_ackloss;     /* ignore DSACK due to ack loss */
1101         u_int32_t       tcps_dsack_badrexmt;    /* DSACK based bad rexmt recovery */
1102         u_int32_t       tcps_dsack_sent;        /* Sent DSACK notification */
1103         u_int32_t       tcps_dsack_recvd;       /* Received a valid DSACK option */
1104         u_int32_t       tcps_dsack_recvd_old;   /* Received an out of window DSACK option */
1105
1106         /* MPTCP Subflow selection stats */
1107         u_int32_t       tcps_mp_sel_symtomsd;   /* By symptomsd */
1108         u_int32_t       tcps_mp_sel_rtt;        /* By RTT comparison */
1109         u_int32_t       tcps_mp_sel_rto;        /* By RTO comparison */
1110         u_int32_t       tcps_mp_sel_peer;       /* By peer's output pattern */
1111         u_int32_t       tcps_mp_num_probes;     /* Number of probes sent */
1112         u_int32_t       tcps_mp_verdowngrade;   /* MPTCP version downgrade */
1113         u_int32_t       tcps_drop_after_sleep;  /* drop after long AP sleep */
1114         u_int32_t       tcps_probe_if;          /* probe packets after interface availability */
1115         u_int32_t       tcps_probe_if_conflict; /* Can't send probe packets for interface */
1116
1117         u_int32_t       tcps_ecn_client_setup;  /* Attempted ECN setup from client side */
1118         u_int32_t       tcps_ecn_server_setup;  /* Attempted ECN setup from server side */
1119         u_int32_t       tcps_ecn_server_success; /* server-side connection negotiated ECN */
1120         u_int32_t       tcps_ecn_lost_synack;   /* Lost SYN-ACK with ECN setup */
1121         u_int32_t       tcps_ecn_lost_syn;      /* Lost SYN with ECN setup */
1122         u_int32_t       tcps_ecn_not_supported; /* Server did not support ECN setup */
1123         u_int32_t       tcps_ecn_recv_ce;       /* Received CE from the network */
1124         u_int32_t       tcps_ecn_conn_recv_ce;  /* Number of connections received CE atleast once */
1125         u_int32_t       tcps_ecn_conn_recv_ece; /* Number of connections received ECE atleast once */
1126         u_int32_t       tcps_ecn_conn_plnoce;   /* Number of connections that received no CE and sufferred packet loss */
1127         u_int32_t       tcps_ecn_conn_pl_ce;    /* Number of connections that received CE and sufferred packet loss */
1128         u_int32_t       tcps_ecn_conn_nopl_ce;  /* Number of connections that received CE and sufferred no packet loss */
1129         u_int32_t       tcps_ecn_fallback_synloss; /* Number of times we did fall back due to SYN-Loss */
1130         u_int32_t       tcps_ecn_fallback_reorder; /* Number of times we fallback because we detected the PAWS-issue */
1131         u_int32_t       tcps_ecn_fallback_ce;   /* Number of times we fallback because we received too many CEs */
1132
1133         /* TFO-related statistics */
1134         u_int32_t       tcps_tfo_syn_data_rcv;  /* Received a SYN+data with valid cookie */
1135         u_int32_t       tcps_tfo_cookie_req_rcv;/* Received a TFO cookie-request */
1136         u_int32_t       tcps_tfo_cookie_sent;   /* Offered a TFO-cookie to the client */
1137         u_int32_t       tcps_tfo_cookie_invalid;/* Received an invalid TFO-cookie */
1138         u_int32_t       tcps_tfo_cookie_req;    /* Cookie requested with the SYN */
1139         u_int32_t       tcps_tfo_cookie_rcv;    /* Cookie received in a SYN/ACK */
1140         u_int32_t       tcps_tfo_syn_data_sent; /* SYN+data+cookie sent */
1141         u_int32_t       tcps_tfo_syn_data_acked;/* SYN+data has been acknowledged */
1142         u_int32_t       tcps_tfo_syn_loss;      /* SYN+TFO has been lost and we fallback */
1143         u_int32_t       tcps_tfo_blackhole;     /* TFO got blackholed by a middlebox. */
1144 };
1145
1146
1147 struct tcpstat_local {
1148         u_int64_t badformat;
1149         u_int64_t unspecv6;
1150         u_int64_t synfin;
1151         u_int64_t badformatipsec;
1152         u_int64_t noconnnolist;
1153         u_int64_t noconnlist;
1154         u_int64_t listbadsyn;
1155         u_int64_t icmp6unreach;
1156         u_int64_t deprecate6;
1157         u_int64_t ooopacket;
1158         u_int64_t rstinsynrcv;
1159         u_int64_t dospacket;
1160         u_int64_t cleanup;
1161         u_int64_t synwindow;
1162 };
1163
1164 #pragma pack(4)
1165
1166 /*
1167  * TCB structure exported to user-land via sysctl(3).
1168  * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
1169  * included.  Not all of our clients do.
1170  */
1171
1172 struct  xtcpcb {
1173         u_int32_t       xt_len;
1174 #ifdef KERNEL_PRIVATE
1175         struct  inpcb_compat    xt_inp;
1176 #else
1177         struct  inpcb   xt_inp;
1178 #endif
1179 #ifdef KERNEL_PRIVATE
1180         struct  otcpcb  xt_tp;
1181 #else
1182         struct  tcpcb   xt_tp;
1183 #endif
1184         struct  xsocket xt_socket;
1185         u_quad_t        xt_alignment_hack;
1186 };
1187
1188
1189 struct  xtcpcb64 {
1190         u_int32_t               xt_len;
1191         struct xinpcb64         xt_inpcb;
1192
1193         u_int64_t t_segq;
1194         int     t_dupacks;              /* consecutive dup acks recd */
1195
1196         int t_timer[TCPT_NTIMERS_EXT];  /* tcp timers */
1197
1198         int     t_state;                /* state of this connection */
1199         u_int   t_flags;
1200
1201         int     t_force;                /* 1 if forcing out a byte */
1202
1203         tcp_seq snd_una;                /* send unacknowledged */
1204         tcp_seq snd_max;                /* highest sequence number sent;
1205                                          * used to recognize retransmits
1206                                          */
1207         tcp_seq snd_nxt;                /* send next */
1208         tcp_seq snd_up;                 /* send urgent pointer */
1209
1210         tcp_seq snd_wl1;                /* window update seg seq number */
1211         tcp_seq snd_wl2;                /* window update seg ack number */
1212         tcp_seq iss;                    /* initial send sequence number */
1213         tcp_seq irs;                    /* initial receive sequence number */
1214
1215         tcp_seq rcv_nxt;                /* receive next */
1216         tcp_seq rcv_adv;                /* advertised window */
1217         u_int32_t rcv_wnd;              /* receive window */
1218         tcp_seq rcv_up;                 /* receive urgent pointer */
1219
1220         u_int32_t snd_wnd;              /* send window */
1221         u_int32_t snd_cwnd;             /* congestion-controlled window */
1222         u_int32_t snd_ssthresh;         /* snd_cwnd size threshold for
1223                                          * for slow start exponential to
1224                                          * linear switch
1225                                          */
1226         u_int   t_maxopd;               /* mss plus options */
1227
1228         u_int32_t t_rcvtime;            /* time at which a packet was received */
1229         u_int32_t t_starttime;          /* time connection was established */
1230         int     t_rtttime;              /* round trip time */
1231         tcp_seq t_rtseq;                /* sequence number being timed */
1232
1233         int     t_rxtcur;               /* current retransmit value (ticks) */
1234         u_int   t_maxseg;               /* maximum segment size */
1235         int     t_srtt;                 /* smoothed round-trip time */
1236         int     t_rttvar;               /* variance in round-trip time */
1237
1238         int     t_rxtshift;             /* log(2) of rexmt exp. backoff */
1239         u_int   t_rttmin;               /* minimum rtt allowed */
1240         u_int32_t t_rttupdated;         /* number of times rtt sampled */
1241         u_int32_t max_sndwnd;           /* largest window peer has offered */
1242
1243         int     t_softerror;            /* possible error not yet reported */
1244 /* out-of-band data */
1245         char    t_oobflags;             /* have some */
1246         char    t_iobc;                 /* input character */
1247 /* RFC 1323 variables */
1248         u_char  snd_scale;              /* window scaling for send window */
1249         u_char  rcv_scale;              /* window scaling for recv window */
1250         u_char  request_r_scale;        /* pending window scaling */
1251         u_char  requested_s_scale;
1252         u_int32_t ts_recent;            /* timestamp echo data */
1253
1254         u_int32_t ts_recent_age;        /* when last updated */
1255         tcp_seq last_ack_sent;
1256 /* RFC 1644 variables */
1257         tcp_cc  cc_send;                /* send connection count */
1258         tcp_cc  cc_recv;                /* receive connection count */
1259         tcp_seq snd_recover;            /* for use in fast recovery */
1260 /* experimental */
1261         u_int32_t snd_cwnd_prev;        /* cwnd prior to retransmit */
1262         u_int32_t snd_ssthresh_prev;    /* ssthresh prior to retransmit */
1263         u_int32_t t_badrxtwin;          /* window for retransmit recovery */
1264
1265         u_quad_t                xt_alignment_hack;
1266 };
1267
1268
1269 #ifdef PRIVATE
1270
1271 struct  xtcpcb_n {
1272         u_int32_t               xt_len;
1273         u_int32_t                       xt_kind;                /* XSO_TCPCB */
1274
1275         u_int64_t t_segq;
1276         int     t_dupacks;              /* consecutive dup acks recd */
1277
1278         int t_timer[TCPT_NTIMERS_EXT];  /* tcp timers */
1279
1280         int     t_state;                /* state of this connection */
1281         u_int   t_flags;
1282
1283         int     t_force;                /* 1 if forcing out a byte */
1284
1285         tcp_seq snd_una;                /* send unacknowledged */
1286         tcp_seq snd_max;                /* highest sequence number sent;
1287                                                                          * used to recognize retransmits
1288                                                                          */
1289         tcp_seq snd_nxt;                /* send next */
1290         tcp_seq snd_up;                 /* send urgent pointer */
1291
1292         tcp_seq snd_wl1;                /* window update seg seq number */
1293         tcp_seq snd_wl2;                /* window update seg ack number */
1294         tcp_seq iss;                    /* initial send sequence number */
1295         tcp_seq irs;                    /* initial receive sequence number */
1296
1297         tcp_seq rcv_nxt;                /* receive next */
1298         tcp_seq rcv_adv;                /* advertised window */
1299         u_int32_t rcv_wnd;              /* receive window */
1300         tcp_seq rcv_up;                 /* receive urgent pointer */
1301
1302         u_int32_t snd_wnd;              /* send window */
1303         u_int32_t snd_cwnd;             /* congestion-controlled window */
1304         u_int32_t snd_ssthresh;         /* snd_cwnd size threshold for
1305                                                                          * for slow start exponential to
1306                                                                          * linear switch
1307                                                                          */
1308         u_int   t_maxopd;               /* mss plus options */
1309
1310         u_int32_t t_rcvtime;            /* time at which a packet was received */
1311         u_int32_t t_starttime;          /* time connection was established */
1312         int     t_rtttime;              /* round trip time */
1313         tcp_seq t_rtseq;                /* sequence number being timed */
1314
1315         int     t_rxtcur;               /* current retransmit value (ticks) */
1316         u_int   t_maxseg;               /* maximum segment size */
1317         int     t_srtt;                 /* smoothed round-trip time */
1318         int     t_rttvar;               /* variance in round-trip time */
1319
1320         int     t_rxtshift;             /* log(2) of rexmt exp. backoff */
1321         u_int   t_rttmin;               /* minimum rtt allowed */
1322         u_int32_t t_rttupdated;         /* number of times rtt sampled */
1323         u_int32_t max_sndwnd;           /* largest window peer has offered */
1324
1325         int     t_softerror;            /* possible error not yet reported */
1326         /* out-of-band data */
1327         char    t_oobflags;             /* have some */
1328         char    t_iobc;                 /* input character */
1329         /* RFC 1323 variables */
1330         u_char  snd_scale;              /* window scaling for send window */
1331         u_char  rcv_scale;              /* window scaling for recv window */
1332         u_char  request_r_scale;        /* pending window scaling */
1333         u_char  requested_s_scale;
1334         u_int32_t ts_recent;            /* timestamp echo data */
1335
1336         u_int32_t ts_recent_age;        /* when last updated */
1337         tcp_seq last_ack_sent;
1338         /* RFC 1644 variables */
1339         tcp_cc  cc_send;                /* send connection count */
1340         tcp_cc  cc_recv;                /* receive connection count */
1341         tcp_seq snd_recover;            /* for use in fast recovery */
1342         /* experimental */
1343         u_int32_t snd_cwnd_prev;        /* cwnd prior to retransmit */
1344         u_int32_t snd_ssthresh_prev;    /* ssthresh prior to retransmit */
1345 };
1346
1347 #endif /* PRIVATE */
1348
1349 #pragma pack()
1350
1351 /*
1352  * Names for TCP sysctl objects
1353  */
1354 #define TCPCTL_DO_RFC1323       1       /* use RFC-1323 extensions */
1355 #define TCPCTL_DO_RFC1644       2       /* use RFC-1644 extensions */
1356 #define TCPCTL_MSSDFLT          3       /* MSS default */
1357 #define TCPCTL_STATS            4       /* statistics (read-only) */
1358 #define TCPCTL_RTTDFLT          5       /* default RTT estimate */
1359 #define TCPCTL_KEEPIDLE         6       /* keepalive idle timer */
1360 #define TCPCTL_KEEPINTVL        7       /* interval to send keepalives */
1361 #define TCPCTL_SENDSPACE        8       /* send buffer space */
1362 #define TCPCTL_RECVSPACE        9       /* receive buffer space */
1363 #define TCPCTL_KEEPINIT         10      /* timeout for establishing syn */
1364 #define TCPCTL_PCBLIST          11      /* list of all outstanding PCBs */
1365 #define TCPCTL_DELACKTIME       12      /* time before sending delayed ACK */
1366 #define TCPCTL_V6MSSDFLT        13      /* MSS default for IPv6 */
1367 #define TCPCTL_MAXID            14
1368
1369 #ifdef BSD_KERNEL_PRIVATE
1370 #include <sys/bitstring.h>
1371
1372 #define TCP_PKTLIST_CLEAR(tp) {                                         \
1373         (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL;             \
1374         (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0;                \
1375 }
1376
1377 #define TCPCTL_NAMES { \
1378         { 0, 0 }, \
1379         { "rfc1323", CTLTYPE_INT }, \
1380         { "rfc1644", CTLTYPE_INT }, \
1381         { "mssdflt", CTLTYPE_INT }, \
1382         { "stats", CTLTYPE_STRUCT }, \
1383         { "rttdflt", CTLTYPE_INT }, \
1384         { "keepidle", CTLTYPE_INT }, \
1385         { "keepintvl", CTLTYPE_INT }, \
1386         { "sendspace", CTLTYPE_INT }, \
1387         { "recvspace", CTLTYPE_INT }, \
1388         { "keepinit", CTLTYPE_INT }, \
1389         { "pcblist", CTLTYPE_STRUCT }, \
1390         { "delacktime", CTLTYPE_INT }, \
1391         { "v6mssdflt", CTLTYPE_INT }, \
1392 }
1393
1394 #ifdef SYSCTL_DECL
1395 SYSCTL_DECL(_net_inet_tcp);
1396 #endif /* SYSCTL_DECL */
1397
1398 extern  struct inpcbhead tcb;           /* head of queue of active tcpcb's */
1399 extern  struct inpcbinfo tcbinfo;
1400 extern  struct tcpstat tcpstat; /* tcp statistics */
1401 extern  int tcp_mssdflt;        /* XXX */
1402 extern  int tcp_minmss;
1403 #define TCP_FASTOPEN_SERVER 0x01
1404 #define TCP_FASTOPEN_CLIENT 0x02
1405
1406 extern  int tcp_tfo_halfcnt;
1407 extern  int tcp_tfo_backlog;
1408 extern  int tcp_fastopen;
1409 extern  int tcp_tfo_fallback_min;
1410 extern  int ss_fltsz;
1411 extern  int ss_fltsz_local;
1412 extern  int tcp_do_rfc3390;             /* Calculate ss_fltsz according to RFC 3390 */
1413 extern int tcp_do_rfc1323;
1414 extern int target_qdelay;
1415 extern  u_int32_t tcp_now;              /* for RFC 1323 timestamps */
1416 extern struct timeval tcp_uptime;
1417 extern lck_spin_t *tcp_uptime_lock;
1418 extern  int tcp_delack_enabled;
1419 extern  int tcp_do_sack;        /* SACK enabled/disabled */
1420 extern int tcp_do_rfc3465;
1421 extern int tcp_do_rfc3465_lim2;
1422 extern int maxseg_unacked;
1423 extern int tcp_use_newreno;
1424 extern struct zone *tcp_reass_zone;
1425 extern struct zone *tcp_rxt_seg_zone;
1426 extern int tcp_ecn_outbound;
1427 extern int tcp_ecn_inbound;
1428
1429
1430 #if CONFIG_IFEF_NOWINDOWSCALE
1431 extern int tcp_obey_ifef_nowindowscale;
1432 #endif
1433
1434 struct protosw;
1435 struct domain;
1436
1437 struct tcp_respond_args {
1438         unsigned int ifscope;
1439         unsigned int nocell:1,
1440                      noexpensive:1,
1441                      awdl_unrestricted:1;
1442 };
1443
1444 void     tcp_canceltimers(struct tcpcb *);
1445 struct tcpcb *
1446          tcp_close(struct tcpcb *);
1447 void     tcp_ctlinput(int, struct sockaddr *, void *);
1448 int      tcp_ctloutput(struct socket *, struct sockopt *);
1449 struct tcpcb *
1450          tcp_drop(struct tcpcb *, int);
1451 void     tcp_drain(void);
1452 void     tcp_getrt_rtt(struct tcpcb *tp, struct rtentry *rt);
1453 struct rmxp_tao *
1454          tcp_gettaocache(struct inpcb *);
1455 void     tcp_init(struct protosw *, struct domain *);
1456 void     tcp_input(struct mbuf *, int);
1457 void     tcp_mss(struct tcpcb *, int, unsigned int);
1458 int      tcp_mssopt(struct tcpcb *);
1459 void     tcp_drop_syn_sent(struct inpcb *, int);
1460 void     tcp_mtudisc(struct inpcb *, int);
1461 struct tcpcb *
1462          tcp_newtcpcb(struct inpcb *);
1463 int      tcp_output(struct tcpcb *);
1464 void     tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *,
1465     tcp_seq, tcp_seq, int, struct tcp_respond_args *);
1466 struct rtentry *
1467          tcp_rtlookup(struct inpcb *, unsigned int);
1468 void     tcp_setpersist(struct tcpcb *);
1469 void     tcp_gc(struct inpcbinfo *);
1470 void     tcp_itimer(struct inpcbinfo *ipi);
1471 void     tcp_check_timer_state(struct tcpcb *tp);
1472 void     tcp_run_timerlist(void *arg1, void *arg2);
1473
1474 struct tcptemp *tcp_maketemplate(struct tcpcb *);
1475 void     tcp_fillheaders(struct tcpcb *, void *, void *);
1476 struct tcpcb *tcp_timers(struct tcpcb *, int);
1477 void     tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int);
1478
1479 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, struct tcphdr *,
1480     u_int32_t *);
1481 extern boolean_t tcp_sack_process_dsack(struct tcpcb *, struct tcpopt *,
1482     struct tcphdr *);
1483 int tcp_detect_bad_rexmt(struct tcpcb *, struct tcphdr *, struct tcpopt *,
1484     u_int32_t rxtime);
1485 void     tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
1486 void     tcp_clean_sackreport(struct tcpcb *tp);
1487 void     tcp_sack_adjust(struct tcpcb *tp);
1488 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
1489 void     tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
1490 void     tcp_free_sackholes(struct tcpcb *tp);
1491 int32_t  tcp_sbspace(struct tcpcb *tp);
1492 void     tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp);
1493 void     tcp_set_ecn(struct tcpcb *tp, struct ifnet *ifp);
1494 void     tcp_reset_stretch_ack(struct tcpcb *tp);
1495 extern void tcp_get_ports_used(u_int32_t, int, u_int32_t, bitstr_t *);
1496 uint32_t tcp_count_opportunistic(unsigned int ifindex, u_int32_t flags);
1497 uint32_t tcp_find_anypcb_byaddr(struct ifaddr *ifa);
1498 void     tcp_set_max_rwinscale(struct tcpcb *tp, struct socket *so);
1499 struct bwmeas* tcp_bwmeas_alloc(struct tcpcb *tp);
1500 void tcp_bwmeas_free(struct tcpcb *tp);
1501 extern int32_t timer_diff(uint32_t t1, uint32_t toff1, uint32_t t2, uint32_t toff2);
1502
1503 extern void tcp_set_background_cc(struct socket *);
1504 extern void tcp_set_foreground_cc(struct socket *);
1505 extern void tcp_set_recv_bg(struct socket *);
1506 extern void tcp_clear_recv_bg(struct socket *);
1507 extern boolean_t tcp_sack_byte_islost(struct tcpcb *tp);
1508 #define IS_TCP_RECV_BG(_so)     \
1509         ((_so)->so_traffic_mgt_flags & TRAFFIC_MGT_TCP_RECVBG)
1510
1511 #if TRAFFIC_MGT
1512 #define CLEAR_IAJ_STATE(_tp_) (_tp_)->iaj_rcv_ts = 0
1513 void     reset_acc_iaj(struct tcpcb *tp);
1514 #endif /* TRAFFIC_MGT */
1515
1516 int      tcp_lock (struct socket *, int, void *);
1517 int      tcp_unlock (struct socket *, int, void *);
1518 void     calculate_tcp_clock(void);
1519
1520 extern void tcp_keepalive_reset(struct tcpcb *);
1521 extern uint32_t get_base_rtt(struct tcpcb *tp);
1522
1523 #ifdef _KERN_LOCKS_H_
1524 lck_mtx_t *      tcp_getlock (struct socket *, int);
1525 #else
1526 void *   tcp_getlock (struct socket *, int);
1527 #endif
1528
1529
1530 extern  struct pr_usrreqs tcp_usrreqs;
1531 extern  u_int32_t tcp_sendspace;
1532 extern  u_int32_t tcp_recvspace;
1533 tcp_seq tcp_new_isn(struct tcpcb *);
1534
1535 extern int tcp_input_checksum(int, struct mbuf *, struct tcphdr *, int, int);
1536 extern void tcp_getconninfo(struct socket *, struct conninfo_tcp *);
1537 extern void add_to_time_wait(struct tcpcb *, uint32_t delay);
1538 extern void tcp_pmtud_revert_segment_size(struct tcpcb *tp);
1539 extern void tcp_rxtseg_insert(struct tcpcb *, tcp_seq, tcp_seq);
1540 extern struct tcp_rxt_seg *tcp_rxtseg_find(struct tcpcb *, tcp_seq, tcp_seq);
1541 extern void tcp_rxtseg_clean(struct tcpcb *);
1542 extern boolean_t tcp_rxtseg_detect_bad_rexmt(struct tcpcb *, tcp_seq);
1543 extern boolean_t tcp_rxtseg_dsack_for_tlp(struct tcpcb *);
1544 extern u_int32_t tcp_rxtseg_total_size(struct tcpcb *tp);
1545 extern void tcp_rexmt_save_state(struct tcpcb *tp);
1546 extern void tcp_interface_send_probe(u_int16_t if_index_available);
1547 extern void tcp_probe_connectivity(struct ifnet *ifp, u_int32_t enable);
1548 extern void tcp_get_connectivity_status(struct tcpcb *,
1549     struct tcp_conn_status *);
1550
1551 extern boolean_t tfo_enabled(const struct tcpcb *tp);
1552 extern void tcp_disable_tfo(struct tcpcb *tp);
1553 extern void tcp_tfo_gen_cookie(struct inpcb *inp, u_char *out, size_t blk_size);
1554 #define TCP_FASTOPEN_KEYLEN 16
1555
1556 #if MPTCP
1557 extern int mptcp_input_preproc(struct tcpcb *, struct mbuf *, int);
1558 extern void mptcp_output_csum(struct tcpcb *, struct mbuf *, int32_t, unsigned,
1559     u_int64_t, u_int32_t *);
1560 extern int mptcp_adj_mss(struct tcpcb *, boolean_t);
1561 extern void mptcp_insert_rmap(struct tcpcb *, struct mbuf *);
1562 #endif
1563 #endif /* BSD_KERNEL_RPIVATE */
1564
1565 #endif /* _NETINET_TCP_VAR_H_ */