| 1 | /* |
| 2 | * Copyright (c) 2000-2017 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 | /* Always allow at least 16 packets worth of recv window when adjusting |
| 97 | * recv window using inter-packet arrival jitter. |
| 98 | */ |
| 99 | #define MIN_IAJ_WIN 16 |
| 100 | |
| 101 | /* A variation in delay of this many milliseconds is tolerable. This limit has to |
| 102 | * be low but greater than zero. We also use standard deviation on jitter to adjust |
| 103 | * this limit for different link and connection types. |
| 104 | */ |
| 105 | #define ALLOWED_IAJ 5 |
| 106 | |
| 107 | /* Ignore the first few packets on a connection until the ACK clock gets going |
| 108 | */ |
| 109 | #define IAJ_IGNORE_PKTCNT 40 |
| 110 | |
| 111 | /* Let the accumulated IAJ value increase by this threshold at most. This limit |
| 112 | * will control how many ALLOWED_IAJ measurements a receiver will have to see |
| 113 | * before opening the receive window |
| 114 | */ |
| 115 | #define ACC_IAJ_HIGH_THRESH 100 |
| 116 | |
| 117 | /* When accumulated IAJ reaches this value, the receiver starts to react by |
| 118 | * closing the window |
| 119 | */ |
| 120 | #define ACC_IAJ_REACT_LIMIT 200 |
| 121 | |
| 122 | /* If the number of small packets (smaller than IAJ packet size) seen on a |
| 123 | * connection is more than this threshold, reset the size and learn it again. |
| 124 | * This is needed because the sender might send smaller segments after PMTU |
| 125 | * discovery and the receiver has to learn the new size. |
| 126 | */ |
| 127 | #define RESET_IAJ_SIZE_THRESH 20 |
| 128 | |
| 129 | /* |
| 130 | * Adaptive timeout is a read/write timeout specified by the application to |
| 131 | * get a socket event when the transport layer detects a stall in data |
| 132 | * transfer. The value specified is the number of probes that can be sent |
| 133 | * to the peer before generating an event. Since it is not specified as |
| 134 | * a time value, the timeout will adjust based on the RTT seen on the link. |
| 135 | * The timeout will start only when there is an indication that the read/write |
| 136 | * operation is not making progress. |
| 137 | * |
| 138 | * If a write operation stalls, the probe will be retransmission of data. |
| 139 | * If a read operation stalls, the probe will be a keep-alive packet. |
| 140 | * |
| 141 | * The maximum value of adaptive timeout is set to 10 which will allow |
| 142 | * transmission of enough number of probes to the peer. |
| 143 | */ |
| 144 | #define TCP_ADAPTIVE_TIMEOUT_MAX 10 |
| 145 | |
| 146 | #define TCP_CONNECTIVITY_PROBES_MAX 5 |
| 147 | |
| 148 | /* |
| 149 | * Kernel variables for tcp. |
| 150 | */ |
| 151 | |
| 152 | /* TCP segment queue entry */ |
| 153 | struct tseg_qent { |
| 154 | LIST_ENTRY(tseg_qent) tqe_q; |
| 155 | int tqe_len; /* TCP segment data length */ |
| 156 | struct tcphdr *tqe_th; /* a pointer to tcp header */ |
| 157 | struct mbuf *tqe_m; /* mbuf contains packet */ |
| 158 | }; |
| 159 | LIST_HEAD(tsegqe_head, tseg_qent); |
| 160 | |
| 161 | struct sackblk { |
| 162 | tcp_seq start; /* start seq no. of sack block */ |
| 163 | tcp_seq end; /* end seq no. */ |
| 164 | }; |
| 165 | |
| 166 | struct sackhole { |
| 167 | tcp_seq start; /* start seq no. of hole */ |
| 168 | tcp_seq end; /* end seq no. */ |
| 169 | tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ |
| 170 | u_int32_t rxmit_start; /* timestamp of first retransmission */ |
| 171 | TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ |
| 172 | }; |
| 173 | |
| 174 | struct sackhint { |
| 175 | struct sackhole *nexthole; |
| 176 | int sack_bytes_rexmit; |
| 177 | }; |
| 178 | |
| 179 | struct tcp_rxt_seg { |
| 180 | tcp_seq rx_start; |
| 181 | tcp_seq rx_end; |
| 182 | u_int16_t rx_count; |
| 183 | u_int16_t rx_flags; |
| 184 | #define TCP_RXT_SPURIOUS 0x1 /* received DSACK notification */ |
| 185 | #define TCP_RXT_DSACK_FOR_TLP 0x2 |
| 186 | SLIST_ENTRY(tcp_rxt_seg) rx_link; |
| 187 | }; |
| 188 | |
| 189 | struct tcp_notify_ack_marker { |
| 190 | tcp_seq notify_snd_una; /* Notify when snd_una crosses this seq */ |
| 191 | tcp_notify_ack_id_t notify_id; |
| 192 | SLIST_ENTRY(tcp_notify_ack_marker) notify_next; |
| 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 bandwidth */ |
| 209 | uint32_t bw_sndbw_max; /* Max measured bandwidth */ |
| 210 | uint32_t bw_rcvbw_max; /* Max receive bandwidth measured */ |
| 211 | }; |
| 212 | |
| 213 | /* MPTCP Data sequence map entry */ |
| 214 | struct mpt_dsn_map { |
| 215 | uint64_t mpt_dsn; /* data seq num recvd */ |
| 216 | uint32_t mpt_sseq; /* relative subflow # */ |
| 217 | uint16_t mpt_len; /* length of mapping */ |
| 218 | uint16_t mpt_csum; /* checksum value if on */ |
| 219 | uint8_t mpt_dfin; /* It's a DATA_FIN */ |
| 220 | }; |
| 221 | #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ |
| 222 | |
| 223 | struct tcp_ccstate { |
| 224 | union { |
| 225 | struct tcp_cubic_state { |
| 226 | u_int32_t tc_last_max; /* cwnd at last loss */ |
| 227 | u_int32_t tc_epoch_start; /* TS of last loss */ |
| 228 | u_int32_t tc_origin_point; /* window at the start of an epoch */ |
| 229 | u_int32_t tc_tcp_win; /* computed tcp win */ |
| 230 | u_int32_t tc_tcp_bytes_acked; /* bytes acked */ |
| 231 | u_int32_t tc_target_win; /* cubic target win */ |
| 232 | u_int32_t tc_avg_lastmax; /* Average of last max */ |
| 233 | u_int32_t tc_mean_deviation; /* Mean absolute deviation */ |
| 234 | float tc_epoch_period; /* K parameter */ |
| 235 | } _cubic_state_; |
| 236 | #define cub_last_max __u__._cubic_state_.tc_last_max |
| 237 | #define cub_epoch_start __u__._cubic_state_.tc_epoch_start |
| 238 | #define cub_origin_point __u__._cubic_state_.tc_origin_point |
| 239 | #define cub_tcp_win __u__._cubic_state_.tc_tcp_win |
| 240 | #define cub_tcp_bytes_acked __u__._cubic_state_.tc_tcp_bytes_acked |
| 241 | #define cub_epoch_period __u__._cubic_state_.tc_epoch_period |
| 242 | #define cub_target_win __u__._cubic_state_.tc_target_win |
| 243 | #define cub_avg_lastmax __u__._cubic_state_.tc_avg_lastmax |
| 244 | #define cub_mean_dev __u__._cubic_state_.tc_mean_deviation |
| 245 | } __u__; |
| 246 | }; |
| 247 | |
| 248 | /* |
| 249 | * Tcp control block, one per tcp; fields: |
| 250 | * Organized for 16 byte cacheline efficiency. |
| 251 | */ |
| 252 | struct tcpcb { |
| 253 | struct tsegqe_head t_segq; |
| 254 | int t_dupacks; /* consecutive dup acks recd */ |
| 255 | int t_state; /* state of this connection */ |
| 256 | uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */ |
| 257 | struct tcptimerentry tentry; /* entry in timer list */ |
| 258 | |
| 259 | struct inpcb *t_inpcb; /* back pointer to internet pcb */ |
| 260 | uint32_t t_flags; |
| 261 | #define TF_ACKNOW 0x00001 /* ack peer immediately */ |
| 262 | #define TF_DELACK 0x00002 /* ack, but try to delay it */ |
| 263 | #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ |
| 264 | #define TF_NOOPT 0x00008 /* don't use tcp options */ |
| 265 | #define TF_SENTFIN 0x00010 /* have sent FIN */ |
| 266 | #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ |
| 267 | #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ |
| 268 | #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ |
| 269 | #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ |
| 270 | #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ |
| 271 | #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */ |
| 272 | #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ |
| 273 | #define TF_NOPUSH 0x01000 /* don't push */ |
| 274 | #define TF_REQ_CC 0x02000 /* have/will request CC */ |
| 275 | #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ |
| 276 | #define TF_SENDCCNEW 0x08000 /* Unused */ |
| 277 | #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ |
| 278 | #define TF_LOCAL 0x20000 /* connection to a host on local link */ |
| 279 | #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ |
| 280 | #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ |
| 281 | #define TF_LASTIDLE 0x100000 /* connection was previously idle */ |
| 282 | #define TF_FASTRECOVERY 0x200000 /* in NewReno Fast Recovery */ |
| 283 | #define TF_WASFRECOVERY 0x400000 /* was in NewReno Fast Recovery */ |
| 284 | #define TF_SIGNATURE 0x800000 /* require MD5 digests (RFC2385) */ |
| 285 | #define TF_MAXSEGSNT 0x1000000 /* last segment sent was a full segment */ |
| 286 | #define TF_STREAMING_ON 0x2000000 /* Receiver detected streaming */ |
| 287 | #define TF_PMTUD 0x4000000 /* Perform Path MTU Discovery for this connection */ |
| 288 | #define TF_CLOSING 0x8000000 /* pending tcp close */ |
| 289 | #define TF_TSO 0x10000000 /* TCP Segment Offloading is enable on this connection */ |
| 290 | #define TF_BLACKHOLE 0x20000000 /* Path MTU Discovery Black Hole detection */ |
| 291 | #define TF_TIMER_ONLIST 0x40000000 /* pcb is on tcp_timer_list */ |
| 292 | #define TF_STRETCHACK 0x80000000 /* receiver is going to delay acks */ |
| 293 | |
| 294 | tcp_seq snd_una; /* send unacknowledged */ |
| 295 | tcp_seq snd_max; /* highest sequence number sent; |
| 296 | * used to recognize retransmits |
| 297 | */ |
| 298 | tcp_seq snd_nxt; /* send next */ |
| 299 | tcp_seq snd_up; /* send urgent pointer */ |
| 300 | |
| 301 | tcp_seq snd_wl1; /* window update seg seq number */ |
| 302 | tcp_seq snd_wl2; /* window update seg ack number */ |
| 303 | tcp_seq iss; /* initial send sequence number */ |
| 304 | tcp_seq irs; /* initial receive sequence number */ |
| 305 | |
| 306 | tcp_seq rcv_nxt; /* receive next */ |
| 307 | tcp_seq rcv_adv; /* advertised window */ |
| 308 | u_int32_t rcv_wnd; /* receive window */ |
| 309 | tcp_seq rcv_up; /* receive urgent pointer */ |
| 310 | |
| 311 | u_int32_t snd_wnd; /* send window */ |
| 312 | u_int32_t snd_cwnd; /* congestion-controlled window */ |
| 313 | u_int32_t snd_ssthresh; /* snd_cwnd size threshold for |
| 314 | * for slow start exponential to |
| 315 | * linear switch |
| 316 | */ |
| 317 | tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ |
| 318 | |
| 319 | u_int32_t t_maxopd; /* mss plus options */ |
| 320 | u_int32_t t_rcvtime; /* time at which a packet was received */ |
| 321 | u_int32_t t_sndtime; /* time at which we last sent new data */ |
| 322 | u_int32_t t_starttime; /* time connection was established */ |
| 323 | int t_rtttime; /* tcp clock when rtt calculation was started */ |
| 324 | tcp_seq t_rtseq; /* sequence number being timed */ |
| 325 | |
| 326 | u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ |
| 327 | u_int32_t rfbuf_cnt; /* recv buffer autoscaling byte count */ |
| 328 | |
| 329 | int t_rxtcur; /* current retransmit value (ticks) */ |
| 330 | u_int t_maxseg; /* maximum segment size */ |
| 331 | int t_srtt; /* smoothed round-trip time */ |
| 332 | int t_rttvar; /* variance in round-trip time */ |
| 333 | |
| 334 | u_int64_t t_accsleep_ms; /* accumulated sleep time since last boot */ |
| 335 | u_int16_t t_reassqlen; /* length of reassembly queue */ |
| 336 | u_int16_t t_rxtshift; /* log(2) of rexmt exp. backoff */ |
| 337 | u_int32_t t_rttmin; /* minimum rtt allowed */ |
| 338 | u_int32_t t_rttbest; /* best rtt we've seen */ |
| 339 | u_int32_t t_rttcur; /* most recent value of rtt */ |
| 340 | u_int32_t t_rttupdated; /* number of times rtt sampled */ |
| 341 | u_int32_t t_rxt_conndroptime; /* retxmt conn gets dropped after this time, when set */ |
| 342 | u_int32_t t_rxtstart; /* time at which retransmission started */ |
| 343 | u_int32_t max_sndwnd; /* largest window peer has offered */ |
| 344 | |
| 345 | int t_softerror; /* possible error not yet reported */ |
| 346 | /* out-of-band data */ |
| 347 | char t_oobflags; /* have some */ |
| 348 | char t_iobc; /* input character */ |
| 349 | #define TCPOOB_HAVEDATA 0x01 |
| 350 | #define TCPOOB_HADDATA 0x02 |
| 351 | /* RFC 1323 variables */ |
| 352 | u_int8_t snd_scale; /* window scaling for send window */ |
| 353 | u_int8_t rcv_scale; /* window scaling for recv window */ |
| 354 | u_int8_t request_r_scale; /* pending window scaling */ |
| 355 | u_int8_t requested_s_scale; |
| 356 | u_int8_t tcp_cc_index; /* index of congestion control algorithm */ |
| 357 | u_int8_t t_adaptive_rtimo; /* Read timeout used as a multiple of RTT */ |
| 358 | u_int8_t t_adaptive_wtimo; /* Write timeout used as a multiple of RTT */ |
| 359 | u_int8_t t_stretchack_delayed; /* stretch ack delayed */ |
| 360 | |
| 361 | /* State for limiting early retransmits when SACK is not enabled */ |
| 362 | u_int16_t t_early_rexmt_count; /* count of early rexmts */ |
| 363 | u_int32_t t_early_rexmt_win; /* window for limiting early rexmts */ |
| 364 | |
| 365 | u_int32_t ts_recent; /* timestamp echo data */ |
| 366 | |
| 367 | u_int32_t ts_recent_age; /* when last updated */ |
| 368 | tcp_seq last_ack_sent; |
| 369 | /* RFC 3465 variables */ |
| 370 | u_int32_t t_bytes_acked; /* ABC "bytes_acked" parameter */ |
| 371 | |
| 372 | int t_lastchain; /* amount of packets chained last time around */ |
| 373 | u_int16_t t_unacksegs; /* received but unacked segments for delaying acks */ |
| 374 | u_int8_t t_rexmtthresh; /* duplicate ack threshold for entering fast recovery */ |
| 375 | u_int8_t t_rtimo_probes; /* number of adaptive rtimo probes sent */ |
| 376 | u_int32_t t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */ |
| 377 | u_int32_t t_persist_stop; /* persistence limit deadline if triggered by ZWP */ |
| 378 | u_int32_t t_notsent_lowat; /* Low water for not sent data */ |
| 379 | |
| 380 | /* Receiver state for stretch-ack algorithm */ |
| 381 | u_int32_t rcv_unackwin; /* to measure win for stretching acks */ |
| 382 | u_int32_t rcv_by_unackwin; /* bytes seen during the last ack-stretching win */ |
| 383 | u_int32_t rcv_nostrack_ts; /* timestamp when stretch ack was disabled automatically */ |
| 384 | u_int32_t rcv_nostrack_pkts; /* pkts received since strech ack was disabled */ |
| 385 | u_int16_t ; /* wait for packets during slow-start */ |
| 386 | |
| 387 | /* ECN stats */ |
| 388 | u_int16_t ecn_flags; |
| 389 | #define TE_SETUPSENT 0x0001 /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */ |
| 390 | #define TE_SETUPRECEIVED 0x0002 /* Indicate we have received ECN-SETUP SYN or SYN-ACK */ |
| 391 | #define TE_SENDIPECT 0x0004 /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */ |
| 392 | #define TE_SENDCWR 0x0008 /* Indicate that the next non-retransmit should have the TCP CWR flag set */ |
| 393 | #define TE_SENDECE 0x0010 /* Indicate that the next packet should have the TCP ECE flag set */ |
| 394 | #define TE_INRECOVERY 0x0020 /* connection entered recovery after receiving ECE */ |
| 395 | #define TE_RECV_ECN_CE 0x0040 /* Received IPTOS_ECN_CE marking atleast once */ |
| 396 | #define TE_RECV_ECN_ECE 0x0080 /* Received ECE marking atleast once */ |
| 397 | #define TE_LOST_SYN 0x0100 /* Lost SYN with ECN setup */ |
| 398 | #define TE_LOST_SYNACK 0x0200 /* Lost SYN-ACK with ECN setup */ |
| 399 | #define TE_ECN_MODE_ENABLE 0x0400 /* Option ECN mode set to enable */ |
| 400 | #define TE_ECN_MODE_DISABLE 0x0800 /* Option ECN mode set to disable */ |
| 401 | #define TE_ENABLE_ECN 0x1000 /* Enable negotiation of ECN */ |
| 402 | #define TE_ECN_ON (TE_SETUPSENT | TE_SETUPRECEIVED) /* Indicate ECN was successfully negotiated on a connection) */ |
| 403 | #define TE_CEHEURI_SET 0x2000 /* We did our CE-probing at the beginning */ |
| 404 | #define TE_CLIENT_SETUP 0x4000 /* setup from client side */ |
| 405 | #define TE_RCVD_SYN_RST 0x8000 /* Received RST to the first ECN enabled SYN */ |
| 406 | |
| 407 | u_int32_t t_ecn_recv_ce; /* Received CE from the network */ |
| 408 | u_int32_t t_ecn_recv_cwr; /* Packets received with CWR */ |
| 409 | |
| 410 | /* state for bad retransmit recovery */ |
| 411 | u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ |
| 412 | u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ |
| 413 | tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ |
| 414 | int t_srtt_prev; /* srtt prior to retransmit */ |
| 415 | int t_rttvar_prev; /* rttvar prior to retransmit */ |
| 416 | u_int32_t t_badrexmt_time; /* bad rexmt detection time */ |
| 417 | |
| 418 | /* Packet reordering metric */ |
| 419 | u_int16_t t_reorderwin; /* Reordering late time offset */ |
| 420 | |
| 421 | /* SACK related state */ |
| 422 | int16_t snd_numholes; /* number of holes seen by sender */ |
| 423 | tcp_seq sack_newdata; /* New data xmitted in this recovery |
| 424 | episode starts at this seq number */ |
| 425 | TAILQ_HEAD(sackhole_head, sackhole) snd_holes; |
| 426 | /* SACK scoreboard (sorted) */ |
| 427 | tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ |
| 428 | int rcv_numsacks; /* # distinct sack blks present */ |
| 429 | struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ |
| 430 | struct sackhint sackhint; /* SACK scoreboard hint */ |
| 431 | |
| 432 | struct mbuf *t_pktlist_head; /* First packet in transmit chain */ |
| 433 | struct mbuf *t_pktlist_tail; /* Last packet in transmit chain */ |
| 434 | u_int32_t t_pktlist_sentlen; /* total bytes in transmit chain */ |
| 435 | |
| 436 | u_int32_t t_keepidle; /* keepalive idle timer (override global if > 0) */ |
| 437 | u_int32_t t_keepinit; /* connection timeout, i.e. idle time |
| 438 | in SYN_SENT or SYN_RECV state */ |
| 439 | u_int32_t t_keepintvl; /* interval between keepalives */ |
| 440 | u_int32_t t_keepcnt; /* number of keepalives before close */ |
| 441 | |
| 442 | u_int32_t tso_max_segment_size; /* TSO maximum segment unit for NIC */ |
| 443 | u_int16_t t_pmtud_lastseg_size; /* size of the last sent segment */ |
| 444 | u_int16_t t_pmtud_saved_maxopd; /* MSS saved before performing PMTU-D BlackHole detection */ |
| 445 | u_int32_t t_pmtud_start_ts; /* Time of PMTUD blackhole detection */ |
| 446 | |
| 447 | struct |
| 448 | { |
| 449 | u_int32_t rxduplicatebytes; |
| 450 | u_int32_t rxoutoforderbytes; |
| 451 | u_int32_t txretransmitbytes; |
| 452 | u_int8_t synrxtshift; |
| 453 | u_int8_t unused; |
| 454 | u_int16_t unused_pad_to_8; |
| 455 | u_int32_t rxmitpkts; |
| 456 | } t_stat; |
| 457 | u_int8_t t_notify_ack_count; |
| 458 | u_int8_t t_ecn_recv_ce_pkt; /* Received packet with CE-bit set (independent from last_ack_sent) */ |
| 459 | u_int16_t t_cached_maxopd; /* default for MSS adjustment using link status report */ |
| 460 | |
| 461 | uint32_t bg_ssthresh; /* Slow start threshold until delay increases */ |
| 462 | uint32_t t_flagsext; /* Another field to accommodate more flags */ |
| 463 | #define TF_RXTFINDROP 0x1 /* Drop conn after retransmitting FIN 3 times */ |
| 464 | #define TF_RCVUNACK_WAITSS 0x2 /* set when the receiver should not stretch acks */ |
| 465 | #define TF_BWMEAS_INPROGRESS 0x4 /* Indicate BW meas is happening */ |
| 466 | #define TF_MEASURESNDBW 0x8 /* Measure send bw on this connection */ |
| 467 | #define TF_LRO_OFFLOADED 0x10 /* Connection LRO offloaded */ |
| 468 | #define TF_SACK_ENABLE 0x20 /* SACK is enabled */ |
| 469 | #define TF_RECOMPUTE_RTT 0x40 /* recompute RTT after spurious retransmit */ |
| 470 | #define TF_DETECT_READSTALL 0x80 /* Used to detect a stall during read operation */ |
| 471 | #define TF_RECV_THROTTLE 0x100 /* Input throttling active */ |
| 472 | #define TF_NOSTRETCHACK 0x200 /* ack every other packet */ |
| 473 | #define TF_NOTIMEWAIT 0x800 /* Avoid going into time-wait */ |
| 474 | #define TF_SENT_TLPROBE 0x1000 /* Sent data in PTO */ |
| 475 | #define TF_PKTS_REORDERED 0x2000 /* Detected reordering */ |
| 476 | #define TF_DELAY_RECOVERY 0x4000 /* delay fast recovery */ |
| 477 | #define TF_FORCE 0x8000 /* force 1 byte out */ |
| 478 | #define TF_DISABLE_STRETCHACK 0x10000 /* auto-disable stretch ack */ |
| 479 | #define TF_NOBLACKHOLE_DETECTION 0x20000 /* Disable PMTU blackhole detection */ |
| 480 | #define TF_DISABLE_DSACK 0x40000 /* Ignore DSACK due to n/w duplication */ |
| 481 | #define TF_RESCUE_RXT 0x80000 /* SACK rescue retransmit */ |
| 482 | #define TF_CWND_NONVALIDATED 0x100000 /* cwnd non validated */ |
| 483 | #define TF_PROBING 0x200000 /* Trigger probe timeout */ |
| 484 | #define TF_FASTOPEN 0x400000 /* TCP Fastopen is enabled */ |
| 485 | #define TF_REASS_INPROG 0x800000 /* Reassembly is in progress */ |
| 486 | #define TF_FASTOPEN_HEUR 0x1000000 /* Make sure that heuristics get never skipped */ |
| 487 | |
| 488 | #if TRAFFIC_MGT |
| 489 | /* Inter-arrival jitter related state */ |
| 490 | uint32_t iaj_rcv_ts; /* tcp clock when the first packet was received */ |
| 491 | uint16_t iaj_size; /* Size of packet for iaj measurement */ |
| 492 | uint8_t iaj_small_pkt; /* Count of packets smaller than iaj_size */ |
| 493 | uint8_t t_pipeack_ind; /* index for next pipeack sample */ |
| 494 | uint16_t iaj_pktcnt; /* packet count, to avoid throttling initially */ |
| 495 | uint16_t acc_iaj; /* Accumulated iaj */ |
| 496 | uint32_t avg_iaj; /* Mean */ |
| 497 | uint32_t std_dev_iaj; /* Standard deviation */ |
| 498 | #endif /* TRAFFIC_MGT */ |
| 499 | struct bwmeas *t_bwmeas; /* State for bandwidth measurement */ |
| 500 | uint32_t t_lropktlen; /* Bytes in a LRO frame */ |
| 501 | tcp_seq t_idleat; /* rcv_nxt at idle time */ |
| 502 | TAILQ_ENTRY(tcpcb) t_twentry; /* link for time wait queue */ |
| 503 | struct tcp_ccstate *t_ccstate; /* congestion control related state */ |
| 504 | /* Tail loss probe related state */ |
| 505 | tcp_seq t_tlphighrxt; /* snd_nxt after PTO */ |
| 506 | u_int32_t t_tlpstart; /* timestamp at PTO */ |
| 507 | /* DSACK data receiver state */ |
| 508 | tcp_seq t_dsack_lseq; /* DSACK left sequence */ |
| 509 | tcp_seq t_dsack_rseq; /* DSACK right sequence */ |
| 510 | /* DSACK data sender state */ |
| 511 | SLIST_HEAD(tcp_rxt_seghead, tcp_rxt_seg) t_rxt_segments; |
| 512 | tcp_seq t_dsack_lastuna; /* snd_una when last recovery episode started */ |
| 513 | /* state for congestion window validation (draft-ietf-tcpm-newcwv-07) */ |
| 514 | #define TCP_PIPEACK_SAMPLE_COUNT 3 |
| 515 | u_int32_t t_pipeack_sample[TCP_PIPEACK_SAMPLE_COUNT]; /* pipeack, bytes acked within RTT */ |
| 516 | tcp_seq t_pipeack_lastuna; /* una when pipeack measurement started */ |
| 517 | u_int32_t t_pipeack; |
| 518 | u_int32_t t_lossflightsize; |
| 519 | |
| 520 | #if MPTCP |
| 521 | u_int32_t t_mpflags; /* flags for multipath TCP */ |
| 522 | |
| 523 | #define TMPF_PREESTABLISHED 0x00000001 /* conn in pre-established state */ |
| 524 | #define TMPF_SND_KEYS 0x00000002 /* indicates that keys should be send */ |
| 525 | #define TMPF_MPTCP_TRUE 0x00000004 /* negotiated MPTCP successfully */ |
| 526 | #define TMPF_MPTCP_RCVD_KEY 0x00000008 /* state for 3-way handshake */ |
| 527 | #define TMPF_SND_MPPRIO 0x00000010 /* send priority of subflow */ |
| 528 | #define TMPF_SND_REM_ADDR 0x00000020 /* initiate address removal */ |
| 529 | #define TMPF_RCVD_DACK 0x00000040 /* received a data-ack */ |
| 530 | #define TMPF_JOINED_FLOW 0x00000080 /* Indicates additional flow */ |
| 531 | #define TMPF_BACKUP_PATH 0x00000100 /* Indicates backup path */ |
| 532 | #define TMPF_MPTCP_ACKNOW 0x00000200 /* Send Data ACK */ |
| 533 | #define TMPF_SEND_DSN 0x00000400 /* Send DSN mapping */ |
| 534 | #define TMPF_SEND_DFIN 0x00000800 /* Send Data FIN */ |
| 535 | #define TMPF_RECV_DFIN 0x00001000 /* Recv Data FIN */ |
| 536 | #define TMPF_SENT_JOIN 0x00002000 /* Sent Join */ |
| 537 | #define TMPF_RECVD_JOIN 0x00004000 /* Received Join */ |
| 538 | #define TMPF_RESET 0x00008000 /* Send RST */ |
| 539 | #define TMPF_TCP_FALLBACK 0x00010000 /* Fallback to TCP */ |
| 540 | #define TMPF_FASTCLOSERCV 0x00020000 /* Received Fastclose option */ |
| 541 | #define TMPF_EMBED_DSN 0x00040000 /* tp has DSN mapping */ |
| 542 | #define TMPF_MPTCP_READY 0x00080000 /* Can send DSS options on data */ |
| 543 | #define TMPF_INFIN_SENT 0x00100000 /* Sent infinite mapping */ |
| 544 | #define TMPF_SND_MPFAIL 0x00200000 /* Received mapping csum failure */ |
| 545 | #define TMPF_SND_JACK 0x00400000 /* Send a Join-ACK */ |
| 546 | #define TMPF_TFO_REQUEST 0x00800000 /* TFO Requested */ |
| 547 | |
| 548 | #define TMPF_MPTCP_SIGNALS (TMPF_SND_MPPRIO | TMPF_SND_REM_ADDR | TMPF_SND_MPFAIL | TMPF_SND_KEYS | TMPF_SND_JACK) |
| 549 | |
| 550 | tcp_seq t_mpuna; /* unacknowledged sequence */ |
| 551 | struct mptcb *t_mptcb; /* pointer to MPTCP TCB */ |
| 552 | struct mptsub *t_mpsub; /* pointer to the MPTCP subflow */ |
| 553 | struct mpt_dsn_map t_rcv_map; /* Receive mapping list */ |
| 554 | u_int8_t t_local_aid; /* Addr Id for authentication */ |
| 555 | u_int8_t t_rem_aid; /* Addr ID of another subflow */ |
| 556 | u_int8_t t_mprxtshift; /* join retransmission */ |
| 557 | #endif /* MPTCP */ |
| 558 | |
| 559 | #define TFO_F_OFFER_COOKIE 0x01 /* We will offer a cookie */ |
| 560 | #define TFO_F_COOKIE_VALID 0x02 /* The received cookie is valid */ |
| 561 | #define TFO_F_COOKIE_REQ 0x04 /* Client requested a new cookie */ |
| 562 | #define TFO_F_COOKIE_SENT 0x08 /* Client did send a cookie in the SYN */ |
| 563 | #define TFO_F_SYN_LOSS 0x10 /* A SYN-loss triggered a fallback to regular TCP on the client-side */ |
| 564 | #define TFO_F_NO_SNDPROBING 0x20 /* This network is guaranteed to support TFO in the upstream direction */ |
| 565 | #define TFO_F_HEURISTIC_DONE 0x40 /* We have already marked this network as bad */ |
| 566 | u_int8_t t_tfo_flags; |
| 567 | #define TFO_S_SYNDATA_RCV 0x01 /* SYN+data has been received */ |
| 568 | #define TFO_S_COOKIEREQ_RECV 0x02 /* TFO-cookie request received */ |
| 569 | #define TFO_S_COOKIE_SENT 0x04 /* TFO-cookie announced in SYN/ACK */ |
| 570 | #define TFO_S_COOKIE_INVALID 0x08 /* Received TFO-cookie is invalid */ |
| 571 | #define TFO_S_COOKIE_REQ 0x10 /* TFO-cookie requested within the SYN */ |
| 572 | #define TFO_S_COOKIE_RCV 0x20 /* TFO-cookie received in SYN/ACK */ |
| 573 | #define TFO_S_SYN_DATA_SENT 0x40 /* SYN+data sent */ |
| 574 | #define TFO_S_SYN_DATA_ACKED 0x80 /* SYN+data has been acknowledged in SYN/ACK */ |
| 575 | #define TFO_S_SYN_LOSS 0x0100 /* SYN+TFO has been lost - fallback to regular TCP */ |
| 576 | #define TFO_S_COOKIE_WRONG 0x0200 /* Cookie we sent in the SYN was wrong */ |
| 577 | #define TFO_S_NO_COOKIE_RCV 0x0400 /* We asked for a cookie but didn't get one */ |
| 578 | #define TFO_S_HEURISTICS_DISABLE 0x0800 /* TFO-heuristics disabled it for this connection */ |
| 579 | #define TFO_S_SEND_BLACKHOLE 0x1000 /* TFO got blackholed in the send direction */ |
| 580 | #define TFO_S_RECV_BLACKHOLE 0x2000 /* TFO got blackholed in the recv direction */ |
| 581 | #define TFO_S_ONE_BYTE_PROXY 0x4000 /* TFO failed because of a proxy acknowledging just one byte */ |
| 582 | u_int16_t t_tfo_stats; |
| 583 | |
| 584 | u_int8_t t_tfo_probes; /* TFO-probes we did send */ |
| 585 | /* |
| 586 | * This here is the TFO-probing state-machine. Transitions are as follows: |
| 587 | * |
| 588 | * Current state: PROBE_NONE |
| 589 | * Event: SYN+DATA acknowledged |
| 590 | * Action: Transition to PROBE_PROBING and set keepalive-timer |
| 591 | * |
| 592 | * Current state: PROBE_PROBING (initial state) |
| 593 | * Event: Receive data |
| 594 | * Action: Transition to PROBE_NONE and cancel keepalive-timer |
| 595 | * Event: Receive ACK that does not indicate a hole |
| 596 | * Action: Transition to PROBE_NONE and cancel keepalive-timer |
| 597 | * Event: Receive ACK that indicates a hole |
| 598 | * Action: Transition to PROBE_WAIT_DATA and set a short timer |
| 599 | * to wait for the final segment. |
| 600 | * Event: Keepalive-timeout (did not receive any segment) |
| 601 | * Action: Signal ETIMEDOUT as with regular keepalive-timers |
| 602 | * |
| 603 | * Current state: PROBE_WAIT_DATA |
| 604 | * Event: Receive data |
| 605 | * Action: Transition to PROBE_NONE and cancel keepalive-timer |
| 606 | * Event: Data-timeout (did not receive the expected data) |
| 607 | * Action: Signal ENODATA up to the app and close everything. |
| 608 | */ |
| 609 | #define TFO_PROBE_NONE 0 /* Not probing now */ |
| 610 | #define TFO_PROBE_PROBING 1 /* Sending out TCP-keepalives waiting for reply */ |
| 611 | #define TFO_PROBE_WAIT_DATA 2 /* Received reply, waiting for data */ |
| 612 | u_int8_t t_tfo_probe_state; |
| 613 | |
| 614 | u_int32_t t_rcvoopack; /* out-of-order packets received */ |
| 615 | u_int32_t t_pawsdrop; /* segments dropped due to PAWS */ |
| 616 | u_int32_t t_sack_recovery_episode; /* SACK recovery episodes */ |
| 617 | u_int32_t t_reordered_pkts; /* packets reorderd */ |
| 618 | u_int32_t t_dsack_sent; /* Sent DSACK notification */ |
| 619 | u_int32_t t_dsack_recvd; /* Received a valid DSACK option */ |
| 620 | SLIST_HEAD(,tcp_notify_ack_marker) t_notify_ack; /* state for notifying data acknowledgements */ |
| 621 | u_int32_t t_recv_throttle_ts; /* TS for start of recv throttle */ |
| 622 | u_int32_t t_rxt_minimum_timeout; /* minimum retransmit timeout in ms */ |
| 623 | uint32_t t_challengeack_last; /* last time challenge ACK was sent per sec */ |
| 624 | uint32_t t_challengeack_count; /* # of challenge ACKs already sent per sec */ |
| 625 | }; |
| 626 | |
| 627 | #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) |
| 628 | #define SACK_ENABLED(tp) (tp->t_flagsext & TF_SACK_ENABLE) |
| 629 | |
| 630 | /* |
| 631 | * If the connection is in a throttled state due to advisory feedback from |
| 632 | * the interface output queue, reset that state. We do this in favor |
| 633 | * of entering recovery because the data transfer during recovery |
| 634 | * should be just a trickle and it will help to improve performance. |
| 635 | * We also do not want to back off twice in the same RTT. |
| 636 | */ |
| 637 | #define ENTER_FASTRECOVERY(_tp_) do { \ |
| 638 | (_tp_)->t_flags |= TF_FASTRECOVERY; \ |
| 639 | if (INP_IS_FLOW_CONTROLLED((_tp_)->t_inpcb)) \ |
| 640 | inp_reset_fc_state((_tp_)->t_inpcb); \ |
| 641 | if (!SLIST_EMPTY(&tp->t_rxt_segments)) \ |
| 642 | tcp_rxtseg_clean(tp); \ |
| 643 | } while(0) |
| 644 | |
| 645 | #define EXIT_FASTRECOVERY(_tp_) do { \ |
| 646 | (_tp_)->t_flags &= ~TF_FASTRECOVERY; \ |
| 647 | (_tp_)->t_dupacks = 0; \ |
| 648 | (_tp_)->t_rexmtthresh = tcprexmtthresh; \ |
| 649 | (_tp_)->t_bytes_acked = 0; \ |
| 650 | (_tp_)->ecn_flags &= ~TE_INRECOVERY; \ |
| 651 | (_tp_)->t_timer[TCPT_PTO] = 0; \ |
| 652 | (_tp_)->t_flagsext &= ~TF_RESCUE_RXT; \ |
| 653 | (_tp_)->t_lossflightsize = 0; \ |
| 654 | } while(0) |
| 655 | |
| 656 | /* |
| 657 | * When the number of duplicate acks received is less than |
| 658 | * the retransmit threshold, use Limited Transmit algorithm |
| 659 | */ |
| 660 | extern int tcprexmtthresh; |
| 661 | #define ALLOW_LIMITED_TRANSMIT(_tp_) \ |
| 662 | ((_tp_)->t_dupacks > 0 && \ |
| 663 | (_tp_)->t_dupacks < (_tp_)->t_rexmtthresh && \ |
| 664 | ((_tp_)->t_flagsext & (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) \ |
| 665 | != (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) |
| 666 | |
| 667 | /* |
| 668 | * This condition is true is timestamp option is supported |
| 669 | * on a connection. |
| 670 | */ |
| 671 | #define TSTMP_SUPPORTED(_tp_) \ |
| 672 | (((_tp_)->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) == \ |
| 673 | (TF_REQ_TSTMP|TF_RCVD_TSTMP)) |
| 674 | |
| 675 | /* |
| 676 | * This condition is true if window scale option is supported |
| 677 | * on a connection |
| 678 | */ |
| 679 | #define TCP_WINDOW_SCALE_ENABLED(_tp_) \ |
| 680 | (((_tp_)->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == \ |
| 681 | (TF_RCVD_SCALE|TF_REQ_SCALE)) |
| 682 | |
| 683 | /* Is ECN enabled end-to-end */ |
| 684 | #define TCP_ECN_ENABLED(_tp_) \ |
| 685 | (((_tp_)->ecn_flags & (TE_ECN_ON)) == (TE_ECN_ON)) |
| 686 | |
| 687 | /* |
| 688 | * Gives number of bytes acked by this ack |
| 689 | */ |
| 690 | #define BYTES_ACKED(_th_, _tp_) \ |
| 691 | ((_th_)->th_ack - (_tp_)->snd_una) |
| 692 | |
| 693 | /* Returns true if a DSACK option should be sent */ |
| 694 | #define TCP_SEND_DSACK_OPT(_tp_) \ |
| 695 | ((_tp_)->t_dsack_lseq > 0 && (_tp_)->t_dsack_rseq > 0) |
| 696 | |
| 697 | /* Check if DSACK option should be processed */ |
| 698 | #define TCP_DSACK_ENABLED(tp) (tcp_dsack_enable == 1 && \ |
| 699 | !(tp->t_flagsext & TF_DISABLE_DSACK)) |
| 700 | |
| 701 | /* |
| 702 | * Returns true if a DSACK sequence is within the max send window that will |
| 703 | * be accepted. In order to set a window to validate sequence numbers, the |
| 704 | * max send window within which a DSACK option is processed is limited. |
| 705 | * |
| 706 | * We need to choose a maximum window to check if the sequence number is |
| 707 | * within the window. One arbitrary choice is 256 * MSS because if the |
| 708 | * window is as large as 256 segments it might be big enough to ignore the |
| 709 | * DSACK option. Choosing a much larger limit means that the memory for |
| 710 | * retransmit segments can be held for a longer time. |
| 711 | */ |
| 712 | #define TCP_DSACK_MAX_SEND_WINDOW(_tp_) ((_tp_)->t_maxseg << 8) |
| 713 | #define TCP_DSACK_SEQ_IN_WINDOW(_tp_, _seq_, _una_) \ |
| 714 | (SEQ_LEQ((_seq_), (_tp_)->snd_max) && \ |
| 715 | SEQ_GEQ((_seq_), ((_una_) - TCP_DSACK_MAX_SEND_WINDOW(_tp_)))) |
| 716 | |
| 717 | #define TCP_RESET_REXMT_STATE(_tp_) do { \ |
| 718 | (_tp_)->t_rxtshift = 0; \ |
| 719 | (_tp_)->t_rxtstart = 0; \ |
| 720 | mptcp_reset_rexmit_state((_tp_)); \ |
| 721 | } while(0); |
| 722 | |
| 723 | #define TCP_AUTORCVBUF_MAX(_ifp_) (((_ifp_) != NULL && (IFNET_IS_CELLULAR((_ifp_))) && ((_ifp_)->if_eflags & IFEF_3CA)) ? \ |
| 724 | (tcp_autorcvbuf_max << 1) : tcp_autorcvbuf_max) |
| 725 | |
| 726 | enum tcp_cc_event { |
| 727 | TCP_CC_CWND_INIT, /* 0 */ |
| 728 | TCP_CC_INSEQ_ACK_RCVD, /* 1 */ |
| 729 | TCP_CC_ACK_RCVD, /* 2 */ |
| 730 | TCP_CC_ENTER_FASTRECOVERY, /* 3 */ |
| 731 | TCP_CC_IN_FASTRECOVERY, /* 4 */ |
| 732 | TCP_CC_EXIT_FASTRECOVERY, /* 5 */ |
| 733 | TCP_CC_PARTIAL_ACK, /* 6 */ |
| 734 | TCP_CC_IDLE_TIMEOUT, /* 7 */ |
| 735 | TCP_CC_REXMT_TIMEOUT, /* 8 */ |
| 736 | TCP_CC_ECN_RCVD, /* 9 */ |
| 737 | TCP_CC_BAD_REXMT_RECOVERY, /* 10 */ |
| 738 | TCP_CC_OUTPUT_ERROR, /* 11 */ |
| 739 | TCP_CC_CHANGE_ALGO, /* 12 */ |
| 740 | TCP_CC_FLOW_CONTROL, /* 13 */ |
| 741 | TCP_CC_SUSPEND, /* 14 */ |
| 742 | TCP_CC_LIMITED_TRANSMIT, /* 15 */ |
| 743 | TCP_CC_EARLY_RETRANSMIT, /* 16 */ |
| 744 | TCP_CC_TLP_RECOVERY, /* 17 */ |
| 745 | TCP_CC_TLP_RECOVER_LASTPACKET, /* 18 */ |
| 746 | TCP_CC_DELAY_FASTRECOVERY, /* 19 */ |
| 747 | TCP_CC_TLP_IN_FASTRECOVERY, /* 20 */ |
| 748 | TCP_CC_DSACK_BAD_REXMT /* 21 */ |
| 749 | }; |
| 750 | |
| 751 | /* |
| 752 | * Structure to hold TCP options that are only used during segment |
| 753 | * processing (in tcp_input), but not held in the tcpcb. |
| 754 | * It's basically used to reduce the number of parameters |
| 755 | * to tcp_dooptions. |
| 756 | */ |
| 757 | struct tcpopt { |
| 758 | u_int32_t to_flags; /* which options are present */ |
| 759 | #define TOF_TS 0x0001 /* timestamp */ |
| 760 | #define TOF_MSS 0x0010 |
| 761 | #define TOF_SCALE 0x0020 |
| 762 | #define TOF_SIGNATURE 0x0040 /* signature option present */ |
| 763 | #define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */ |
| 764 | #define TOF_SACK 0x0100 /* Peer sent SACK option */ |
| 765 | #define TOF_MPTCP 0x0200 /* MPTCP options to be dropped */ |
| 766 | #define TOF_TFO 0x0400 /* TFO cookie option present */ |
| 767 | #define TOF_TFOREQ 0x0800 /* TFO cookie request present */ |
| 768 | u_int32_t to_tsval; |
| 769 | u_int32_t to_tsecr; |
| 770 | u_int16_t to_mss; |
| 771 | u_int8_t to_requested_s_scale; |
| 772 | u_int8_t to_nsacks; /* number of SACK blocks */ |
| 773 | u_char *to_sacks; /* pointer to the first SACK blocks */ |
| 774 | u_char *to_tfo; /* pointer to the TFO cookie */ |
| 775 | }; |
| 776 | |
| 777 | /* |
| 778 | * The TAO cache entry which is stored in the protocol family specific |
| 779 | * portion of the route metrics. |
| 780 | */ |
| 781 | struct rmxp_tao { |
| 782 | tcp_cc tao_cc; /* latest CC in valid SYN */ |
| 783 | tcp_cc tao_ccsent; /* latest CC sent to peer */ |
| 784 | u_short tao_mssopt; /* peer's cached MSS */ |
| 785 | #ifdef notyet |
| 786 | u_short tao_flags; /* cache status flags */ |
| 787 | #define TAOF_DONT 0x0001 /* peer doesn't understand rfc1644 */ |
| 788 | #define TAOF_OK 0x0002 /* peer does understand rfc1644 */ |
| 789 | #define TAOF_UNDEF 0 /* we don't know yet */ |
| 790 | #endif /* notyet */ |
| 791 | }; |
| 792 | #define rmx_taop(r) ((struct rmxp_tao *)(r).rmx_filler) |
| 793 | |
| 794 | #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) |
| 795 | #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) |
| 796 | |
| 797 | /* TFO-specific defines */ |
| 798 | #define TFO_COOKIE_LEN_MIN 4 |
| 799 | #define TFO_COOKIE_LEN_DEFAULT 8 |
| 800 | #define TFO_COOKIE_LEN_MAX 16 |
| 801 | |
| 802 | /* |
| 803 | * The initial retransmission should happen at rtt + 4 * rttvar. |
| 804 | * Because of the way we do the smoothing, srtt and rttvar |
| 805 | * will each average +1/2 tick of bias. When we compute |
| 806 | * the retransmit timer, we want 1/2 tick of rounding and |
| 807 | * 1 extra tick because of +-1/2 tick uncertainty in the |
| 808 | * firing of the timer. The bias will give us exactly the |
| 809 | * 1.5 tick we need. But, because the bias is |
| 810 | * statistical, we have to test that we don't drop below |
| 811 | * the minimum feasible timer (which is 2 ticks). |
| 812 | * This version of the macro adapted from a paper by Lawrence |
| 813 | * Brakmo and Larry Peterson which outlines a problem caused |
| 814 | * by insufficient precision in the original implementation, |
| 815 | * which results in inappropriately large RTO values for very |
| 816 | * fast networks. |
| 817 | */ |
| 818 | #define TCP_REXMTVAL(tp) \ |
| 819 | max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ |
| 820 | + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) |
| 821 | |
| 822 | /* |
| 823 | * Jaguar compatible TCP control block, for xtcpcb |
| 824 | * Does not have the old fields |
| 825 | */ |
| 826 | struct otcpcb { |
| 827 | #else |
| 828 | struct tseg_qent; |
| 829 | _TCPCB_LIST_HEAD(tsegqe_head, tseg_qent); |
| 830 | |
| 831 | struct tcpcb { |
| 832 | #endif /* KERNEL_PRIVATE */ |
| 833 | #if defined(KERNEL_PRIVATE) |
| 834 | u_int32_t t_segq; |
| 835 | #else |
| 836 | struct tsegqe_head t_segq; |
| 837 | #endif /* KERNEL_PRIVATE */ |
| 838 | int t_dupacks; /* consecutive dup acks recd */ |
| 839 | u_int32_t unused; /* unused now: was t_template */ |
| 840 | |
| 841 | int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ |
| 842 | |
| 843 | _TCPCB_PTR(struct inpcb *) t_inpcb; /* back pointer to internet pcb */ |
| 844 | int t_state; /* state of this connection */ |
| 845 | u_int t_flags; |
| 846 | #define TF_ACKNOW 0x00001 /* ack peer immediately */ |
| 847 | #define TF_DELACK 0x00002 /* ack, but try to delay it */ |
| 848 | #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ |
| 849 | #define TF_NOOPT 0x00008 /* don't use tcp options */ |
| 850 | #define TF_SENTFIN 0x00010 /* have sent FIN */ |
| 851 | #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ |
| 852 | #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ |
| 853 | #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ |
| 854 | #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ |
| 855 | #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ |
| 856 | #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) */ |
| 857 | #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ |
| 858 | #define TF_NOPUSH 0x01000 /* don't push */ |
| 859 | #define TF_REQ_CC 0x02000 /* have/will request CC */ |
| 860 | #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ |
| 861 | #define TF_SENDCCNEW 0x08000 /* Not implemented */ |
| 862 | #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ |
| 863 | #define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */ |
| 864 | #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ |
| 865 | #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ |
| 866 | |
| 867 | int t_force; /* 1 if forcing out a byte */ |
| 868 | |
| 869 | tcp_seq snd_una; /* send unacknowledged */ |
| 870 | tcp_seq snd_max; /* highest sequence number sent; |
| 871 | * used to recognize retransmits |
| 872 | */ |
| 873 | tcp_seq snd_nxt; /* send next */ |
| 874 | tcp_seq snd_up; /* send urgent pointer */ |
| 875 | |
| 876 | tcp_seq snd_wl1; /* window update seg seq number */ |
| 877 | tcp_seq snd_wl2; /* window update seg ack number */ |
| 878 | tcp_seq iss; /* initial send sequence number */ |
| 879 | tcp_seq irs; /* initial receive sequence number */ |
| 880 | |
| 881 | tcp_seq rcv_nxt; /* receive next */ |
| 882 | tcp_seq rcv_adv; /* advertised window */ |
| 883 | u_int32_t rcv_wnd; /* receive window */ |
| 884 | tcp_seq rcv_up; /* receive urgent pointer */ |
| 885 | |
| 886 | u_int32_t snd_wnd; /* send window */ |
| 887 | u_int32_t snd_cwnd; /* congestion-controlled window */ |
| 888 | u_int32_t snd_ssthresh; /* snd_cwnd size threshold for |
| 889 | * for slow start exponential to |
| 890 | * linear switch |
| 891 | */ |
| 892 | u_int t_maxopd; /* mss plus options */ |
| 893 | |
| 894 | u_int32_t t_rcvtime; /* time at which a packet was received */ |
| 895 | u_int32_t t_starttime; /* time connection was established */ |
| 896 | int t_rtttime; /* round trip time */ |
| 897 | tcp_seq t_rtseq; /* sequence number being timed */ |
| 898 | |
| 899 | int t_rxtcur; /* current retransmit value (ticks) */ |
| 900 | u_int t_maxseg; /* maximum segment size */ |
| 901 | int t_srtt; /* smoothed round-trip time */ |
| 902 | int t_rttvar; /* variance in round-trip time */ |
| 903 | |
| 904 | int t_rxtshift; /* log(2) of rexmt exp. backoff */ |
| 905 | u_int t_rttmin; /* minimum rtt allowed */ |
| 906 | u_int32_t t_rttupdated; /* number of times rtt sampled */ |
| 907 | u_int32_t max_sndwnd; /* largest window peer has offered */ |
| 908 | |
| 909 | int t_softerror; /* possible error not yet reported */ |
| 910 | /* out-of-band data */ |
| 911 | char t_oobflags; /* have some */ |
| 912 | char t_iobc; /* input character */ |
| 913 | #define TCPOOB_HAVEDATA 0x01 |
| 914 | #define TCPOOB_HADDATA 0x02 |
| 915 | /* RFC 1323 variables */ |
| 916 | u_char snd_scale; /* window scaling for send window */ |
| 917 | u_char rcv_scale; /* window scaling for recv window */ |
| 918 | u_char request_r_scale; /* pending window scaling */ |
| 919 | u_char requested_s_scale; |
| 920 | u_int32_t ts_recent; /* timestamp echo data */ |
| 921 | |
| 922 | u_int32_t ts_recent_age; /* when last updated */ |
| 923 | tcp_seq last_ack_sent; |
| 924 | /* RFC 1644 variables */ |
| 925 | tcp_cc cc_send; /* send connection count */ |
| 926 | tcp_cc cc_recv; /* receive connection count */ |
| 927 | tcp_seq snd_recover; /* for use in fast recovery */ |
| 928 | /* experimental */ |
| 929 | u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ |
| 930 | u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ |
| 931 | u_int32_t t_badrxtwin; /* window for retransmit recovery */ |
| 932 | }; |
| 933 | |
| 934 | #define tcps_ecn_setup tcps_ecn_client_success |
| 935 | #define tcps_sent_cwr tcps_ecn_recv_ece |
| 936 | #define tcps_sent_ece tcps_ecn_sent_ece |
| 937 | |
| 938 | /* |
| 939 | * TCP statistics. |
| 940 | * Many of these should be kept per connection, |
| 941 | * but that's inconvenient at the moment. |
| 942 | */ |
| 943 | struct tcpstat { |
| 944 | u_int32_t tcps_connattempt; /* connections initiated */ |
| 945 | u_int32_t tcps_accepts; /* connections accepted */ |
| 946 | u_int32_t tcps_connects; /* connections established */ |
| 947 | u_int32_t tcps_drops; /* connections dropped */ |
| 948 | u_int32_t tcps_conndrops; /* embryonic connections dropped */ |
| 949 | u_int32_t tcps_closed; /* conn. closed (includes drops) */ |
| 950 | u_int32_t tcps_segstimed; /* segs where we tried to get rtt */ |
| 951 | u_int32_t tcps_rttupdated; /* times we succeeded */ |
| 952 | u_int32_t tcps_delack; /* delayed acks sent */ |
| 953 | u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ |
| 954 | u_int32_t tcps_rexmttimeo; /* retransmit timeouts */ |
| 955 | u_int32_t tcps_persisttimeo; /* persist timeouts */ |
| 956 | u_int32_t tcps_keeptimeo; /* keepalive timeouts */ |
| 957 | u_int32_t tcps_keepprobe; /* keepalive probes sent */ |
| 958 | u_int32_t tcps_keepdrops; /* connections dropped in keepalive */ |
| 959 | |
| 960 | u_int32_t tcps_sndtotal; /* total packets sent */ |
| 961 | u_int32_t tcps_sndpack; /* data packets sent */ |
| 962 | u_int32_t tcps_sndbyte; /* data bytes sent */ |
| 963 | u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */ |
| 964 | u_int32_t tcps_sndrexmitbyte; /* data bytes retransmitted */ |
| 965 | u_int32_t tcps_sndacks; /* ack-only packets sent */ |
| 966 | u_int32_t tcps_sndprobe; /* window probes sent */ |
| 967 | u_int32_t tcps_sndurg; /* packets sent with URG only */ |
| 968 | u_int32_t tcps_sndwinup; /* window update-only packets sent */ |
| 969 | u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ |
| 970 | |
| 971 | u_int32_t tcps_rcvtotal; /* total packets received */ |
| 972 | u_int32_t tcps_rcvpack; /* packets received in sequence */ |
| 973 | u_int32_t tcps_rcvbyte; /* bytes received in sequence */ |
| 974 | u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */ |
| 975 | u_int32_t tcps_rcvbadoff; /* packets received with bad offset */ |
| 976 | u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ |
| 977 | u_int32_t tcps_rcvshort; /* packets received too short */ |
| 978 | u_int32_t tcps_rcvduppack; /* duplicate-only packets received */ |
| 979 | u_int32_t tcps_rcvdupbyte; /* duplicate-only bytes received */ |
| 980 | u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */ |
| 981 | u_int32_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ |
| 982 | u_int32_t tcps_rcvoopack; /* out-of-order packets received */ |
| 983 | u_int32_t tcps_rcvoobyte; /* out-of-order bytes received */ |
| 984 | u_int32_t tcps_rcvpackafterwin; /* packets with data after window */ |
| 985 | u_int32_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ |
| 986 | u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */ |
| 987 | u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */ |
| 988 | u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */ |
| 989 | u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ |
| 990 | u_int32_t tcps_rcvackpack; /* rcvd ack packets */ |
| 991 | u_int32_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ |
| 992 | u_int32_t tcps_rcvwinupd; /* rcvd window update packets */ |
| 993 | u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */ |
| 994 | u_int32_t tcps_predack; /* times hdr predict ok for acks */ |
| 995 | u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */ |
| 996 | u_int32_t tcps_pcbcachemiss; |
| 997 | u_int32_t tcps_cachedrtt; /* times cached RTT in route updated */ |
| 998 | u_int32_t tcps_cachedrttvar; /* times cached rttvar updated */ |
| 999 | u_int32_t tcps_cachedssthresh; /* times cached ssthresh updated */ |
| 1000 | u_int32_t tcps_usedrtt; /* times RTT initialized from route */ |
| 1001 | u_int32_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ |
| 1002 | u_int32_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ |
| 1003 | u_int32_t tcps_persistdrop; /* timeout in persist state */ |
| 1004 | u_int32_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ |
| 1005 | u_int32_t tcps_mturesent; /* resends due to MTU discovery */ |
| 1006 | u_int32_t tcps_listendrop; /* listen queue overflows */ |
| 1007 | u_int32_t tcps_synchallenge; /* challenge ACK due to bad SYN */ |
| 1008 | u_int32_t tcps_rstchallenge; /* challenge ACK due to bad RST */ |
| 1009 | |
| 1010 | /* new stats from FreeBSD 5.4 sync up */ |
| 1011 | u_int32_t tcps_minmssdrops; /* average minmss too low drops */ |
| 1012 | |
| 1013 | u_int32_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ |
| 1014 | u_int32_t tcps_badrst; /* ignored RSTs in the window */ |
| 1015 | |
| 1016 | u_int32_t tcps_sc_added; /* entry added to syncache */ |
| 1017 | u_int32_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ |
| 1018 | u_int32_t tcps_sc_dupsyn; /* duplicate SYN packet */ |
| 1019 | u_int32_t tcps_sc_dropped; /* could not reply to packet */ |
| 1020 | u_int32_t tcps_sc_completed; /* successful extraction of entry */ |
| 1021 | u_int32_t tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ |
| 1022 | u_int32_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ |
| 1023 | u_int32_t tcps_sc_reset; /* RST removed entry from syncache */ |
| 1024 | u_int32_t tcps_sc_stale; /* timed out or listen socket gone */ |
| 1025 | u_int32_t tcps_sc_aborted; /* syncache entry aborted */ |
| 1026 | u_int32_t tcps_sc_badack; /* removed due to bad ACK */ |
| 1027 | u_int32_t tcps_sc_unreach; /* ICMP unreachable received */ |
| 1028 | u_int32_t tcps_sc_zonefail; /* zalloc() failed */ |
| 1029 | u_int32_t tcps_sc_sendcookie; /* SYN cookie sent */ |
| 1030 | u_int32_t tcps_sc_recvcookie; /* SYN cookie received */ |
| 1031 | |
| 1032 | u_int32_t tcps_hc_added; /* entry added to hostcache */ |
| 1033 | u_int32_t tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ |
| 1034 | |
| 1035 | /* SACK related stats */ |
| 1036 | u_int32_t tcps_sack_recovery_episode; /* SACK recovery episodes */ |
| 1037 | u_int32_t tcps_sack_rexmits; /* SACK rexmit segments */ |
| 1038 | u_int32_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ |
| 1039 | u_int32_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ |
| 1040 | u_int32_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ |
| 1041 | u_int32_t tcps_sack_sboverflow; /* SACK sendblock overflow */ |
| 1042 | |
| 1043 | u_int32_t tcps_bg_rcvtotal; /* total background packets received */ |
| 1044 | u_int32_t tcps_rxtfindrop; /* drop conn after retransmitting FIN */ |
| 1045 | u_int32_t tcps_fcholdpacket; /* packets withheld because of flow control */ |
| 1046 | |
| 1047 | /* LRO related stats */ |
| 1048 | u_int32_t tcps_coalesced_pack; /* number of coalesced packets */ |
| 1049 | u_int32_t tcps_flowtbl_full; /* times flow table was full */ |
| 1050 | u_int32_t tcps_flowtbl_collision; /* collisions in flow tbl */ |
| 1051 | u_int32_t tcps_lro_twopack; /* 2 packets coalesced */ |
| 1052 | u_int32_t tcps_lro_multpack; /* 3 or 4 pkts coalesced */ |
| 1053 | u_int32_t tcps_lro_largepack; /* 5 or more pkts coalesced */ |
| 1054 | |
| 1055 | u_int32_t tcps_limited_txt; /* Limited transmit used */ |
| 1056 | u_int32_t tcps_early_rexmt; /* Early retransmit used */ |
| 1057 | u_int32_t tcps_sack_ackadv; /* Cumulative ack advanced along with sack */ |
| 1058 | |
| 1059 | /* Checksum related stats */ |
| 1060 | u_int32_t tcps_rcv_swcsum; /* tcp swcksum (inbound), packets */ |
| 1061 | u_int32_t tcps_rcv_swcsum_bytes; /* tcp swcksum (inbound), bytes */ |
| 1062 | u_int32_t tcps_rcv6_swcsum; /* tcp6 swcksum (inbound), packets */ |
| 1063 | u_int32_t tcps_rcv6_swcsum_bytes; /* tcp6 swcksum (inbound), bytes */ |
| 1064 | u_int32_t tcps_snd_swcsum; /* tcp swcksum (outbound), packets */ |
| 1065 | u_int32_t tcps_snd_swcsum_bytes; /* tcp swcksum (outbound), bytes */ |
| 1066 | u_int32_t tcps_snd6_swcsum; /* tcp6 swcksum (outbound), packets */ |
| 1067 | u_int32_t tcps_snd6_swcsum_bytes; /* tcp6 swcksum (outbound), bytes */ |
| 1068 | u_int32_t tcps_msg_unopkts; /* unordered packet on TCP msg stream */ |
| 1069 | u_int32_t tcps_msg_unoappendfail; /* failed to append unordered pkt */ |
| 1070 | u_int32_t tcps_msg_sndwaithipri; /* send is waiting for high priority data */ |
| 1071 | |
| 1072 | /* MPTCP Related stats */ |
| 1073 | u_int32_t tcps_invalid_mpcap; /* Invalid MPTCP capable opts */ |
| 1074 | u_int32_t tcps_invalid_joins; /* Invalid MPTCP joins */ |
| 1075 | u_int32_t tcps_mpcap_fallback; /* TCP fallback in primary */ |
| 1076 | u_int32_t tcps_join_fallback; /* No MPTCP in secondary */ |
| 1077 | u_int32_t tcps_estab_fallback; /* DSS option dropped */ |
| 1078 | u_int32_t tcps_invalid_opt; /* Catchall error stat */ |
| 1079 | u_int32_t tcps_mp_outofwin; /* Packet lies outside the |
| 1080 | shared recv window */ |
| 1081 | u_int32_t tcps_mp_reducedwin; /* Reduced subflow window */ |
| 1082 | u_int32_t tcps_mp_badcsum; /* Bad DSS csum */ |
| 1083 | u_int32_t tcps_mp_oodata; /* Out of order data */ |
| 1084 | u_int32_t tcps_mp_switches; /* number of subflow switch */ |
| 1085 | u_int32_t tcps_mp_rcvtotal; /* number of rcvd packets */ |
| 1086 | u_int32_t tcps_mp_rcvbytes; /* number of bytes received */ |
| 1087 | u_int32_t tcps_mp_sndpacks; /* number of data packs sent */ |
| 1088 | u_int32_t tcps_mp_sndbytes; /* number of bytes sent */ |
| 1089 | u_int32_t tcps_join_rxmts; /* join ack retransmits */ |
| 1090 | u_int32_t tcps_tailloss_rto; /* RTO due to tail loss */ |
| 1091 | u_int32_t tcps_reordered_pkts; /* packets reorderd */ |
| 1092 | u_int32_t tcps_recovered_pkts; /* recovered after loss */ |
| 1093 | u_int32_t tcps_pto; /* probe timeout */ |
| 1094 | u_int32_t tcps_rto_after_pto; /* RTO after a probe */ |
| 1095 | u_int32_t tcps_tlp_recovery; /* TLP induced fast recovery */ |
| 1096 | u_int32_t tcps_tlp_recoverlastpkt; /* TLP recoverd last pkt */ |
| 1097 | u_int32_t tcps_ecn_client_success; /* client-side connection negotiated ECN */ |
| 1098 | u_int32_t tcps_ecn_recv_ece; /* ECE received, sent CWR */ |
| 1099 | u_int32_t tcps_ecn_sent_ece; /* Sent ECE notification */ |
| 1100 | u_int32_t tcps_detect_reordering; /* Detect pkt reordering */ |
| 1101 | u_int32_t tcps_delay_recovery; /* Delay fast recovery */ |
| 1102 | u_int32_t tcps_avoid_rxmt; /* Retransmission was avoided */ |
| 1103 | u_int32_t tcps_unnecessary_rxmt; /* Retransmission was not needed */ |
| 1104 | u_int32_t tcps_nostretchack; /* disabled stretch ack algorithm on a connection */ |
| 1105 | u_int32_t tcps_rescue_rxmt; /* SACK rescue retransmit */ |
| 1106 | u_int32_t tcps_pto_in_recovery; /* rescue retransmit in fast recovery */ |
| 1107 | u_int32_t tcps_pmtudbh_reverted; /* PMTU Blackhole detection, segment size reverted */ |
| 1108 | |
| 1109 | /* DSACK related statistics */ |
| 1110 | u_int32_t tcps_dsack_disable; /* DSACK disabled due to n/w duplication */ |
| 1111 | u_int32_t tcps_dsack_ackloss; /* ignore DSACK due to ack loss */ |
| 1112 | u_int32_t tcps_dsack_badrexmt; /* DSACK based bad rexmt recovery */ |
| 1113 | u_int32_t tcps_dsack_sent; /* Sent DSACK notification */ |
| 1114 | u_int32_t tcps_dsack_recvd; /* Received a valid DSACK option */ |
| 1115 | u_int32_t tcps_dsack_recvd_old; /* Received an out of window DSACK option */ |
| 1116 | |
| 1117 | /* MPTCP Subflow selection stats */ |
| 1118 | u_int32_t tcps_mp_sel_symtomsd; /* By symptomsd */ |
| 1119 | u_int32_t tcps_mp_sel_rtt; /* By RTT comparison */ |
| 1120 | u_int32_t tcps_mp_sel_rto; /* By RTO comparison */ |
| 1121 | u_int32_t tcps_mp_sel_peer; /* By peer's output pattern */ |
| 1122 | u_int32_t tcps_mp_num_probes; /* Number of probes sent */ |
| 1123 | u_int32_t tcps_mp_verdowngrade; /* MPTCP version downgrade */ |
| 1124 | u_int32_t tcps_drop_after_sleep; /* drop after long AP sleep */ |
| 1125 | u_int32_t tcps_probe_if; /* probe packets after interface availability */ |
| 1126 | u_int32_t tcps_probe_if_conflict; /* Can't send probe packets for interface */ |
| 1127 | |
| 1128 | u_int32_t tcps_ecn_client_setup; /* Attempted ECN setup from client side */ |
| 1129 | u_int32_t tcps_ecn_server_setup; /* Attempted ECN setup from server side */ |
| 1130 | u_int32_t tcps_ecn_server_success; /* server-side connection negotiated ECN */ |
| 1131 | u_int32_t tcps_ecn_lost_synack; /* Lost SYN-ACK with ECN setup */ |
| 1132 | u_int32_t tcps_ecn_lost_syn; /* Lost SYN with ECN setup */ |
| 1133 | u_int32_t tcps_ecn_not_supported; /* Server did not support ECN setup */ |
| 1134 | u_int32_t tcps_ecn_recv_ce; /* Received CE from the network */ |
| 1135 | u_int32_t tcps_ecn_conn_recv_ce; /* Number of connections received CE atleast once */ |
| 1136 | u_int32_t tcps_ecn_conn_recv_ece; /* Number of connections received ECE atleast once */ |
| 1137 | u_int32_t tcps_ecn_conn_plnoce; /* Number of connections that received no CE and sufferred packet loss */ |
| 1138 | u_int32_t tcps_ecn_conn_pl_ce; /* Number of connections that received CE and sufferred packet loss */ |
| 1139 | u_int32_t tcps_ecn_conn_nopl_ce; /* Number of connections that received CE and sufferred no packet loss */ |
| 1140 | u_int32_t tcps_ecn_fallback_synloss; /* Number of times we did fall back due to SYN-Loss */ |
| 1141 | u_int32_t tcps_ecn_fallback_reorder; /* Number of times we fallback because we detected the PAWS-issue */ |
| 1142 | u_int32_t tcps_ecn_fallback_ce; /* Number of times we fallback because we received too many CEs */ |
| 1143 | |
| 1144 | /* TFO-related statistics */ |
| 1145 | u_int32_t tcps_tfo_syn_data_rcv; /* Received a SYN+data with valid cookie */ |
| 1146 | u_int32_t tcps_tfo_cookie_req_rcv;/* Received a TFO cookie-request */ |
| 1147 | u_int32_t tcps_tfo_cookie_sent; /* Offered a TFO-cookie to the client */ |
| 1148 | u_int32_t tcps_tfo_cookie_invalid;/* Received an invalid TFO-cookie */ |
| 1149 | u_int32_t tcps_tfo_cookie_req; /* Cookie requested with the SYN */ |
| 1150 | u_int32_t tcps_tfo_cookie_rcv; /* Cookie received in a SYN/ACK */ |
| 1151 | u_int32_t tcps_tfo_syn_data_sent; /* SYN+data+cookie sent */ |
| 1152 | u_int32_t tcps_tfo_syn_data_acked;/* SYN+data has been acknowledged */ |
| 1153 | u_int32_t tcps_tfo_syn_loss; /* SYN+TFO has been lost and we fallback */ |
| 1154 | u_int32_t tcps_tfo_blackhole; /* TFO got blackholed by a middlebox. */ |
| 1155 | u_int32_t tcps_tfo_cookie_wrong; /* TFO-cookie we sent was wrong */ |
| 1156 | u_int32_t tcps_tfo_no_cookie_rcv; /* We asked for a cookie but didn't get one */ |
| 1157 | u_int32_t tcps_tfo_heuristics_disable; /* TFO got disabled due to heuristics */ |
| 1158 | u_int32_t tcps_tfo_sndblackhole; /* TFO got blackholed in the sending direction */ |
| 1159 | u_int32_t tcps_mss_to_default; /* Change MSS to default using link status report */ |
| 1160 | u_int32_t tcps_mss_to_medium; /* Change MSS to medium using link status report */ |
| 1161 | u_int32_t tcps_mss_to_low; /* Change MSS to low using link status report */ |
| 1162 | u_int32_t tcps_ecn_fallback_droprst; /* ECN fallback caused by connection drop due to RST */ |
| 1163 | u_int32_t tcps_ecn_fallback_droprxmt; /* ECN fallback due to drop after multiple retransmits */ |
| 1164 | u_int32_t tcps_ecn_fallback_synrst; /* ECN fallback due to rst after syn */ |
| 1165 | |
| 1166 | u_int32_t tcps_mptcp_rcvmemdrop; /* MPTCP packets dropped for lack of memory */ |
| 1167 | u_int32_t tcps_mptcp_rcvduppack; /* MPTCP duplicate-only packets received */ |
| 1168 | u_int32_t tcps_mptcp_rcvpackafterwin; /* MPTCP packets with data after window */ |
| 1169 | |
| 1170 | /* TCP timer statistics */ |
| 1171 | u_int32_t tcps_timer_drift_le_1_ms; /* Timer drift less or equal to 1 ms */ |
| 1172 | u_int32_t tcps_timer_drift_le_10_ms; /* Timer drift less or equal to 10 ms */ |
| 1173 | u_int32_t tcps_timer_drift_le_20_ms; /* Timer drift less or equal to 20 ms */ |
| 1174 | u_int32_t tcps_timer_drift_le_50_ms; /* Timer drift less or equal to 50 ms */ |
| 1175 | u_int32_t tcps_timer_drift_le_100_ms; /* Timer drift less or equal to 100 ms */ |
| 1176 | u_int32_t tcps_timer_drift_le_200_ms; /* Timer drift less or equal to 200 ms */ |
| 1177 | u_int32_t tcps_timer_drift_le_500_ms; /* Timer drift less or equal to 500 ms */ |
| 1178 | u_int32_t tcps_timer_drift_le_1000_ms; /* Timer drift less or equal to 1000 ms */ |
| 1179 | u_int32_t tcps_timer_drift_gt_1000_ms; /* Timer drift greater than 1000 ms */ |
| 1180 | |
| 1181 | u_int32_t tcps_mptcp_handover_attempt; /* Total number of MPTCP-attempts using handover mode */ |
| 1182 | u_int32_t tcps_mptcp_interactive_attempt; /* Total number of MPTCP-attempts using interactive mode */ |
| 1183 | u_int32_t tcps_mptcp_aggregate_attempt; /* Total number of MPTCP-attempts using aggregate mode */ |
| 1184 | u_int32_t tcps_mptcp_fp_handover_attempt; /* Same as previous three but only for first-party apps */ |
| 1185 | u_int32_t tcps_mptcp_fp_interactive_attempt; |
| 1186 | u_int32_t tcps_mptcp_fp_aggregate_attempt; |
| 1187 | u_int32_t tcps_mptcp_heuristic_fallback; /* Total number of MPTCP-connections that fell back due to heuristics */ |
| 1188 | u_int32_t tcps_mptcp_fp_heuristic_fallback; /* Same as previous but for first-party apps */ |
| 1189 | u_int32_t tcps_mptcp_handover_success_wifi; /* Total number of successfull handover-mode connections that *started* on WiFi */ |
| 1190 | u_int32_t tcps_mptcp_handover_success_cell; /* Total number of successfull handover-mode connections that *started* on Cell */ |
| 1191 | u_int32_t tcps_mptcp_interactive_success; /* Total number of interactive-mode connections that negotiated MPTCP */ |
| 1192 | u_int32_t tcps_mptcp_aggregate_success; /* Same as previous but for aggregate */ |
| 1193 | u_int32_t tcps_mptcp_fp_handover_success_wifi; /* Same as previous four, but for first-party apps */ |
| 1194 | u_int32_t tcps_mptcp_fp_handover_success_cell; |
| 1195 | u_int32_t tcps_mptcp_fp_interactive_success; |
| 1196 | u_int32_t tcps_mptcp_fp_aggregate_success; |
| 1197 | u_int32_t tcps_mptcp_handover_cell_from_wifi; /* Total number of connections that use cell in handover-mode (coming from WiFi) */ |
| 1198 | u_int32_t tcps_mptcp_handover_wifi_from_cell; /* Total number of connections that use WiFi in handover-mode (coming from cell) */ |
| 1199 | u_int32_t tcps_mptcp_interactive_cell_from_wifi; /* Total number of connections that use cell in interactive mode (coming from WiFi) */ |
| 1200 | u_int64_t tcps_mptcp_handover_cell_bytes; /* Total number of bytes sent on cell in handover-mode (on new subflows, ignoring initial one) */ |
| 1201 | u_int64_t tcps_mptcp_interactive_cell_bytes; /* Same as previous but for interactive */ |
| 1202 | u_int64_t tcps_mptcp_aggregate_cell_bytes; |
| 1203 | u_int64_t tcps_mptcp_handover_all_bytes; /* Total number of bytes sent in handover */ |
| 1204 | u_int64_t tcps_mptcp_interactive_all_bytes; |
| 1205 | u_int64_t tcps_mptcp_aggregate_all_bytes; |
| 1206 | u_int32_t tcps_mptcp_back_to_wifi; /* Total number of connections that succeed to move traffic away from cell (when starting on cell) */ |
| 1207 | u_int32_t tcps_mptcp_wifi_proxy; /* Total number of new subflows that fell back to regular TCP on cell */ |
| 1208 | u_int32_t tcps_mptcp_cell_proxy; /* Total number of new subflows that fell back to regular TCP on WiFi */ |
| 1209 | u_int32_t tcps_mptcp_triggered_cell; /* Total number of times an MPTCP-connection triggered cell bringup */ |
| 1210 | }; |
| 1211 | |
| 1212 | |
| 1213 | struct tcpstat_local { |
| 1214 | u_int64_t badformat; |
| 1215 | u_int64_t unspecv6; |
| 1216 | u_int64_t synfin; |
| 1217 | u_int64_t badformatipsec; |
| 1218 | u_int64_t noconnnolist; |
| 1219 | u_int64_t noconnlist; |
| 1220 | u_int64_t listbadsyn; |
| 1221 | u_int64_t icmp6unreach; |
| 1222 | u_int64_t deprecate6; |
| 1223 | u_int64_t ooopacket; |
| 1224 | u_int64_t rstinsynrcv; |
| 1225 | u_int64_t dospacket; |
| 1226 | u_int64_t cleanup; |
| 1227 | u_int64_t synwindow; |
| 1228 | }; |
| 1229 | |
| 1230 | #pragma pack(4) |
| 1231 | |
| 1232 | /* |
| 1233 | * TCB structure exported to user-land via sysctl(3). |
| 1234 | * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been |
| 1235 | * included. Not all of our clients do. |
| 1236 | */ |
| 1237 | |
| 1238 | struct xtcpcb { |
| 1239 | u_int32_t xt_len; |
| 1240 | #ifdef KERNEL_PRIVATE |
| 1241 | struct inpcb_compat xt_inp; |
| 1242 | #else |
| 1243 | struct inpcb xt_inp; |
| 1244 | #endif |
| 1245 | #ifdef KERNEL_PRIVATE |
| 1246 | struct otcpcb xt_tp; |
| 1247 | #else |
| 1248 | struct tcpcb xt_tp; |
| 1249 | #endif |
| 1250 | struct xsocket xt_socket; |
| 1251 | u_quad_t xt_alignment_hack; |
| 1252 | }; |
| 1253 | |
| 1254 | #if !CONFIG_EMBEDDED |
| 1255 | |
| 1256 | struct xtcpcb64 { |
| 1257 | u_int32_t xt_len; |
| 1258 | struct xinpcb64 xt_inpcb; |
| 1259 | |
| 1260 | u_int64_t t_segq; |
| 1261 | int t_dupacks; /* consecutive dup acks recd */ |
| 1262 | |
| 1263 | int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ |
| 1264 | |
| 1265 | int t_state; /* state of this connection */ |
| 1266 | u_int t_flags; |
| 1267 | |
| 1268 | int t_force; /* 1 if forcing out a byte */ |
| 1269 | |
| 1270 | tcp_seq snd_una; /* send unacknowledged */ |
| 1271 | tcp_seq snd_max; /* highest sequence number sent; |
| 1272 | * used to recognize retransmits |
| 1273 | */ |
| 1274 | tcp_seq snd_nxt; /* send next */ |
| 1275 | tcp_seq snd_up; /* send urgent pointer */ |
| 1276 | |
| 1277 | tcp_seq snd_wl1; /* window update seg seq number */ |
| 1278 | tcp_seq snd_wl2; /* window update seg ack number */ |
| 1279 | tcp_seq iss; /* initial send sequence number */ |
| 1280 | tcp_seq irs; /* initial receive sequence number */ |
| 1281 | |
| 1282 | tcp_seq rcv_nxt; /* receive next */ |
| 1283 | tcp_seq rcv_adv; /* advertised window */ |
| 1284 | u_int32_t rcv_wnd; /* receive window */ |
| 1285 | tcp_seq rcv_up; /* receive urgent pointer */ |
| 1286 | |
| 1287 | u_int32_t snd_wnd; /* send window */ |
| 1288 | u_int32_t snd_cwnd; /* congestion-controlled window */ |
| 1289 | u_int32_t snd_ssthresh; /* snd_cwnd size threshold for |
| 1290 | * for slow start exponential to |
| 1291 | * linear switch |
| 1292 | */ |
| 1293 | u_int t_maxopd; /* mss plus options */ |
| 1294 | |
| 1295 | u_int32_t t_rcvtime; /* time at which a packet was received */ |
| 1296 | u_int32_t t_starttime; /* time connection was established */ |
| 1297 | int t_rtttime; /* round trip time */ |
| 1298 | tcp_seq t_rtseq; /* sequence number being timed */ |
| 1299 | |
| 1300 | int t_rxtcur; /* current retransmit value (ticks) */ |
| 1301 | u_int t_maxseg; /* maximum segment size */ |
| 1302 | int t_srtt; /* smoothed round-trip time */ |
| 1303 | int t_rttvar; /* variance in round-trip time */ |
| 1304 | |
| 1305 | int t_rxtshift; /* log(2) of rexmt exp. backoff */ |
| 1306 | u_int t_rttmin; /* minimum rtt allowed */ |
| 1307 | u_int32_t t_rttupdated; /* number of times rtt sampled */ |
| 1308 | u_int32_t max_sndwnd; /* largest window peer has offered */ |
| 1309 | |
| 1310 | int t_softerror; /* possible error not yet reported */ |
| 1311 | /* out-of-band data */ |
| 1312 | char t_oobflags; /* have some */ |
| 1313 | char t_iobc; /* input character */ |
| 1314 | /* RFC 1323 variables */ |
| 1315 | u_char snd_scale; /* window scaling for send window */ |
| 1316 | u_char rcv_scale; /* window scaling for recv window */ |
| 1317 | u_char request_r_scale; /* pending window scaling */ |
| 1318 | u_char requested_s_scale; |
| 1319 | u_int32_t ts_recent; /* timestamp echo data */ |
| 1320 | |
| 1321 | u_int32_t ts_recent_age; /* when last updated */ |
| 1322 | tcp_seq last_ack_sent; |
| 1323 | /* RFC 1644 variables */ |
| 1324 | tcp_cc cc_send; /* send connection count */ |
| 1325 | tcp_cc cc_recv; /* receive connection count */ |
| 1326 | tcp_seq snd_recover; /* for use in fast recovery */ |
| 1327 | /* experimental */ |
| 1328 | u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ |
| 1329 | u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ |
| 1330 | u_int32_t t_badrxtwin; /* window for retransmit recovery */ |
| 1331 | |
| 1332 | u_quad_t xt_alignment_hack; |
| 1333 | }; |
| 1334 | |
| 1335 | #endif /* !CONFIG_EMBEDDED */ |
| 1336 | |
| 1337 | #ifdef PRIVATE |
| 1338 | |
| 1339 | struct xtcpcb_n { |
| 1340 | u_int32_t xt_len; |
| 1341 | u_int32_t xt_kind; /* XSO_TCPCB */ |
| 1342 | |
| 1343 | u_int64_t t_segq; |
| 1344 | int t_dupacks; /* consecutive dup acks recd */ |
| 1345 | |
| 1346 | int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ |
| 1347 | |
| 1348 | int t_state; /* state of this connection */ |
| 1349 | u_int t_flags; |
| 1350 | |
| 1351 | int t_force; /* 1 if forcing out a byte */ |
| 1352 | |
| 1353 | tcp_seq snd_una; /* send unacknowledged */ |
| 1354 | tcp_seq snd_max; /* highest sequence number sent; |
| 1355 | * used to recognize retransmits |
| 1356 | */ |
| 1357 | tcp_seq snd_nxt; /* send next */ |
| 1358 | tcp_seq snd_up; /* send urgent pointer */ |
| 1359 | |
| 1360 | tcp_seq snd_wl1; /* window update seg seq number */ |
| 1361 | tcp_seq snd_wl2; /* window update seg ack number */ |
| 1362 | tcp_seq iss; /* initial send sequence number */ |
| 1363 | tcp_seq irs; /* initial receive sequence number */ |
| 1364 | |
| 1365 | tcp_seq rcv_nxt; /* receive next */ |
| 1366 | tcp_seq rcv_adv; /* advertised window */ |
| 1367 | u_int32_t rcv_wnd; /* receive window */ |
| 1368 | tcp_seq rcv_up; /* receive urgent pointer */ |
| 1369 | |
| 1370 | u_int32_t snd_wnd; /* send window */ |
| 1371 | u_int32_t snd_cwnd; /* congestion-controlled window */ |
| 1372 | u_int32_t snd_ssthresh; /* snd_cwnd size threshold for |
| 1373 | * for slow start exponential to |
| 1374 | * linear switch |
| 1375 | */ |
| 1376 | u_int t_maxopd; /* mss plus options */ |
| 1377 | |
| 1378 | u_int32_t t_rcvtime; /* time at which a packet was received */ |
| 1379 | u_int32_t t_starttime; /* time connection was established */ |
| 1380 | int t_rtttime; /* round trip time */ |
| 1381 | tcp_seq t_rtseq; /* sequence number being timed */ |
| 1382 | |
| 1383 | int t_rxtcur; /* current retransmit value (ticks) */ |
| 1384 | u_int t_maxseg; /* maximum segment size */ |
| 1385 | int t_srtt; /* smoothed round-trip time */ |
| 1386 | int t_rttvar; /* variance in round-trip time */ |
| 1387 | |
| 1388 | int t_rxtshift; /* log(2) of rexmt exp. backoff */ |
| 1389 | u_int t_rttmin; /* minimum rtt allowed */ |
| 1390 | u_int32_t t_rttupdated; /* number of times rtt sampled */ |
| 1391 | u_int32_t max_sndwnd; /* largest window peer has offered */ |
| 1392 | |
| 1393 | int t_softerror; /* possible error not yet reported */ |
| 1394 | /* out-of-band data */ |
| 1395 | char t_oobflags; /* have some */ |
| 1396 | char t_iobc; /* input character */ |
| 1397 | /* RFC 1323 variables */ |
| 1398 | u_char snd_scale; /* window scaling for send window */ |
| 1399 | u_char rcv_scale; /* window scaling for recv window */ |
| 1400 | u_char request_r_scale; /* pending window scaling */ |
| 1401 | u_char requested_s_scale; |
| 1402 | u_int32_t ts_recent; /* timestamp echo data */ |
| 1403 | |
| 1404 | u_int32_t ts_recent_age; /* when last updated */ |
| 1405 | tcp_seq last_ack_sent; |
| 1406 | /* RFC 1644 variables */ |
| 1407 | tcp_cc cc_send; /* send connection count */ |
| 1408 | tcp_cc cc_recv; /* receive connection count */ |
| 1409 | tcp_seq snd_recover; /* for use in fast recovery */ |
| 1410 | /* experimental */ |
| 1411 | u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ |
| 1412 | u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ |
| 1413 | }; |
| 1414 | |
| 1415 | /* |
| 1416 | * The rtt measured is in milliseconds as the timestamp granularity is |
| 1417 | * a millisecond. The smoothed round-trip time and estimated variance |
| 1418 | * are stored as fixed point numbers scaled by the values below. |
| 1419 | * For convenience, these scales are also used in smoothing the average |
| 1420 | * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). |
| 1421 | * With these scales, srtt has 5 bits to the right of the binary point, |
| 1422 | * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the |
| 1423 | * binary point, and is smoothed with an ALPHA of 0.75. |
| 1424 | */ |
| 1425 | #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ |
| 1426 | #define TCP_RTT_SHIFT 5 /* shift for srtt; 5 bits frac. */ |
| 1427 | #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 4 bits */ |
| 1428 | #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 4 bits */ |
| 1429 | #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ |
| 1430 | |
| 1431 | |
| 1432 | /* |
| 1433 | * TCP structure with information that gives insight into forward progress on an interface, |
| 1434 | * exported to user-land via sysctl(3). |
| 1435 | */ |
| 1436 | struct xtcpprogress_indicators { |
| 1437 | u_int32_t xp_numflows; /* Total number of flows */ |
| 1438 | u_int32_t xp_conn_probe_fails; /* Count of connection failures */ |
| 1439 | u_int32_t xp_read_probe_fails; /* Count of read probe failures */ |
| 1440 | u_int32_t xp_write_probe_fails; /* Count of write failures */ |
| 1441 | u_int32_t xp_recentflows; /* Total of "recent" flows */ |
| 1442 | u_int32_t xp_recentflows_unacked; /* Total of "recent" flows with unacknowledged data */ |
| 1443 | u_int64_t xp_recentflows_rxbytes; /* Total of "recent" flows received bytes */ |
| 1444 | u_int64_t xp_recentflows_txbytes; /* Total of "recent" flows transmitted bytes */ |
| 1445 | u_int64_t xp_recentflows_rxooo; /* Total of "recent" flows received out of order bytes */ |
| 1446 | u_int64_t xp_recentflows_rxdup; /* Total of "recent" flows received duplicate bytes */ |
| 1447 | u_int64_t xp_recentflows_retx; /* Total of "recent" flows retransmitted bytes */ |
| 1448 | u_int64_t xp_reserved1; /* Expansion */ |
| 1449 | u_int64_t xp_reserved2; /* Expansion */ |
| 1450 | u_int64_t xp_reserved3; /* Expansion */ |
| 1451 | u_int64_t xp_reserved4; /* Expansion */ |
| 1452 | }; |
| 1453 | |
| 1454 | struct tcpprogressreq { |
| 1455 | u_int64_t ifindex; /* Interface index for progress indicators */ |
| 1456 | u_int64_t recentflow_maxduration; /* In mach_absolute_time, max duration for flow to be counted as "recent" */ |
| 1457 | u_int64_t xp_reserved1; /* Expansion */ |
| 1458 | u_int64_t xp_reserved2; /* Expansion */ |
| 1459 | }; |
| 1460 | |
| 1461 | #endif /* PRIVATE */ |
| 1462 | |
| 1463 | #pragma pack() |
| 1464 | |
| 1465 | /* |
| 1466 | * Names for TCP sysctl objects |
| 1467 | */ |
| 1468 | #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ |
| 1469 | #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ |
| 1470 | #define TCPCTL_MSSDFLT 3 /* MSS default */ |
| 1471 | #define TCPCTL_STATS 4 /* statistics (read-only) */ |
| 1472 | #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ |
| 1473 | #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ |
| 1474 | #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ |
| 1475 | #define TCPCTL_SENDSPACE 8 /* send buffer space */ |
| 1476 | #define TCPCTL_RECVSPACE 9 /* receive buffer space */ |
| 1477 | #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ |
| 1478 | #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ |
| 1479 | #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ |
| 1480 | #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ |
| 1481 | #define TCPCTL_MAXID 14 |
| 1482 | |
| 1483 | #ifdef BSD_KERNEL_PRIVATE |
| 1484 | #include <sys/bitstring.h> |
| 1485 | |
| 1486 | #define TCP_PKTLIST_CLEAR(tp) { \ |
| 1487 | (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL; \ |
| 1488 | (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0; \ |
| 1489 | } |
| 1490 | |
| 1491 | #define TCPCTL_NAMES { \ |
| 1492 | { 0, 0 }, \ |
| 1493 | { "rfc1323", CTLTYPE_INT }, \ |
| 1494 | { "rfc1644", CTLTYPE_INT }, \ |
| 1495 | { "mssdflt", CTLTYPE_INT }, \ |
| 1496 | { "stats", CTLTYPE_STRUCT }, \ |
| 1497 | { "rttdflt", CTLTYPE_INT }, \ |
| 1498 | { "keepidle", CTLTYPE_INT }, \ |
| 1499 | { "keepintvl", CTLTYPE_INT }, \ |
| 1500 | { "sendspace", CTLTYPE_INT }, \ |
| 1501 | { "recvspace", CTLTYPE_INT }, \ |
| 1502 | { "keepinit", CTLTYPE_INT }, \ |
| 1503 | { "pcblist", CTLTYPE_STRUCT }, \ |
| 1504 | { "delacktime", CTLTYPE_INT }, \ |
| 1505 | { "v6mssdflt", CTLTYPE_INT }, \ |
| 1506 | } |
| 1507 | |
| 1508 | #ifdef SYSCTL_DECL |
| 1509 | SYSCTL_DECL(_net_inet_tcp); |
| 1510 | #endif /* SYSCTL_DECL */ |
| 1511 | |
| 1512 | extern struct inpcbhead tcb; /* head of queue of active tcpcb's */ |
| 1513 | extern struct inpcbinfo tcbinfo; |
| 1514 | extern struct tcpstat tcpstat; /* tcp statistics */ |
| 1515 | extern int tcp_mssdflt; /* XXX */ |
| 1516 | extern int tcp_minmss; |
| 1517 | #define TCP_FASTOPEN_SERVER 0x01 |
| 1518 | #define TCP_FASTOPEN_CLIENT 0x02 |
| 1519 | |
| 1520 | extern int tcp_tfo_halfcnt; |
| 1521 | extern int tcp_tfo_backlog; |
| 1522 | extern int tcp_fastopen; |
| 1523 | extern int ss_fltsz; |
| 1524 | extern int ss_fltsz_local; |
| 1525 | extern int tcp_do_rfc3390; /* Calculate ss_fltsz according to RFC 3390 */ |
| 1526 | extern int tcp_do_rfc1323; |
| 1527 | extern int target_qdelay; |
| 1528 | extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ |
| 1529 | extern struct timeval tcp_uptime; |
| 1530 | extern lck_spin_t *tcp_uptime_lock; |
| 1531 | extern int tcp_delack_enabled; |
| 1532 | extern int tcp_do_sack; /* SACK enabled/disabled */ |
| 1533 | extern int tcp_do_rfc3465; |
| 1534 | extern int tcp_do_rfc3465_lim2; |
| 1535 | extern int maxseg_unacked; |
| 1536 | extern int tcp_use_newreno; |
| 1537 | extern struct zone *tcp_reass_zone; |
| 1538 | extern struct zone *tcp_rxt_seg_zone; |
| 1539 | extern int tcp_ecn_outbound; |
| 1540 | extern int tcp_ecn_inbound; |
| 1541 | extern u_int32_t tcp_do_autorcvbuf; |
| 1542 | extern u_int32_t tcp_autorcvbuf_max; |
| 1543 | extern u_int32_t tcp_autorcvbuf_inc_shift; |
| 1544 | extern int tcp_recv_bg; |
| 1545 | |
| 1546 | struct protosw; |
| 1547 | struct domain; |
| 1548 | |
| 1549 | struct tcp_respond_args { |
| 1550 | unsigned int ifscope; |
| 1551 | unsigned int nocell:1, |
| 1552 | noexpensive:1, |
| 1553 | awdl_unrestricted:1, |
| 1554 | intcoproc_allowed:1; |
| 1555 | }; |
| 1556 | |
| 1557 | void tcp_canceltimers(struct tcpcb *); |
| 1558 | struct tcpcb * |
| 1559 | tcp_close(struct tcpcb *); |
| 1560 | void tcp_ctlinput(int, struct sockaddr *, void *, struct ifnet *); |
| 1561 | int tcp_ctloutput(struct socket *, struct sockopt *); |
| 1562 | struct tcpcb * |
| 1563 | tcp_drop(struct tcpcb *, int); |
| 1564 | void tcp_drain(void); |
| 1565 | void tcp_getrt_rtt(struct tcpcb *tp, struct rtentry *rt); |
| 1566 | struct rmxp_tao * |
| 1567 | tcp_gettaocache(struct inpcb *); |
| 1568 | void tcp_init(struct protosw *, struct domain *); |
| 1569 | void tcp_input(struct mbuf *, int); |
| 1570 | void tcp_mss(struct tcpcb *, int, unsigned int); |
| 1571 | int tcp_mssopt(struct tcpcb *); |
| 1572 | void tcp_drop_syn_sent(struct inpcb *, int); |
| 1573 | void tcp_mtudisc(struct inpcb *, int); |
| 1574 | struct tcpcb * |
| 1575 | tcp_newtcpcb(struct inpcb *); |
| 1576 | int tcp_output(struct tcpcb *); |
| 1577 | void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *, |
| 1578 | tcp_seq, tcp_seq, int, struct tcp_respond_args *); |
| 1579 | struct rtentry * |
| 1580 | tcp_rtlookup(struct inpcb *, unsigned int); |
| 1581 | void tcp_setpersist(struct tcpcb *); |
| 1582 | void tcp_gc(struct inpcbinfo *); |
| 1583 | void tcp_itimer(struct inpcbinfo *ipi); |
| 1584 | void tcp_check_timer_state(struct tcpcb *tp); |
| 1585 | void tcp_run_timerlist(void *arg1, void *arg2); |
| 1586 | |
| 1587 | struct tcptemp *tcp_maketemplate(struct tcpcb *); |
| 1588 | void (struct tcpcb *, void *, void *); |
| 1589 | struct tcpcb *tcp_timers(struct tcpcb *, int); |
| 1590 | void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int); |
| 1591 | |
| 1592 | void tcp_fill_info(struct tcpcb *, struct tcp_info *); |
| 1593 | void tcp_sack_doack(struct tcpcb *, struct tcpopt *, struct tcphdr *, |
| 1594 | u_int32_t *); |
| 1595 | extern boolean_t tcp_sack_process_dsack(struct tcpcb *, struct tcpopt *, |
| 1596 | struct tcphdr *); |
| 1597 | int tcp_detect_bad_rexmt(struct tcpcb *, struct tcphdr *, struct tcpopt *, |
| 1598 | u_int32_t rxtime); |
| 1599 | void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); |
| 1600 | void tcp_clean_sackreport(struct tcpcb *tp); |
| 1601 | void tcp_sack_adjust(struct tcpcb *tp); |
| 1602 | struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); |
| 1603 | void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); |
| 1604 | void tcp_free_sackholes(struct tcpcb *tp); |
| 1605 | int32_t tcp_sbspace(struct tcpcb *tp); |
| 1606 | void tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp); |
| 1607 | void tcp_set_ecn(struct tcpcb *tp, struct ifnet *ifp); |
| 1608 | void tcp_reset_stretch_ack(struct tcpcb *tp); |
| 1609 | extern void tcp_get_ports_used(u_int32_t, int, u_int32_t, bitstr_t *); |
| 1610 | uint32_t tcp_count_opportunistic(unsigned int ifindex, u_int32_t flags); |
| 1611 | uint32_t tcp_find_anypcb_byaddr(struct ifaddr *ifa); |
| 1612 | void tcp_set_max_rwinscale(struct tcpcb *tp, struct socket *so, struct ifnet *ifp); |
| 1613 | struct bwmeas* tcp_bwmeas_alloc(struct tcpcb *tp); |
| 1614 | void tcp_bwmeas_free(struct tcpcb *tp); |
| 1615 | extern int32_t timer_diff(uint32_t t1, uint32_t toff1, uint32_t t2, uint32_t toff2); |
| 1616 | |
| 1617 | extern void tcp_set_background_cc(struct socket *); |
| 1618 | extern void tcp_set_foreground_cc(struct socket *); |
| 1619 | extern void tcp_set_recv_bg(struct socket *); |
| 1620 | extern void tcp_clear_recv_bg(struct socket *); |
| 1621 | extern boolean_t tcp_sack_byte_islost(struct tcpcb *tp); |
| 1622 | #define IS_TCP_RECV_BG(_so) \ |
| 1623 | ((_so)->so_flags1 & SOF1_TRAFFIC_MGT_TCP_RECVBG) |
| 1624 | |
| 1625 | #if TRAFFIC_MGT |
| 1626 | #define CLEAR_IAJ_STATE(_tp_) (_tp_)->iaj_rcv_ts = 0 |
| 1627 | void reset_acc_iaj(struct tcpcb *tp); |
| 1628 | #endif /* TRAFFIC_MGT */ |
| 1629 | |
| 1630 | int tcp_lock (struct socket *, int, void *); |
| 1631 | int tcp_unlock (struct socket *, int, void *); |
| 1632 | void calculate_tcp_clock(void); |
| 1633 | |
| 1634 | extern void tcp_keepalive_reset(struct tcpcb *); |
| 1635 | extern uint32_t get_base_rtt(struct tcpcb *tp); |
| 1636 | |
| 1637 | #ifdef _KERN_LOCKS_H_ |
| 1638 | lck_mtx_t * tcp_getlock (struct socket *, int); |
| 1639 | #else |
| 1640 | void * tcp_getlock (struct socket *, int); |
| 1641 | #endif |
| 1642 | |
| 1643 | extern struct pr_usrreqs tcp_usrreqs; |
| 1644 | extern u_int32_t tcp_sendspace; |
| 1645 | extern u_int32_t tcp_recvspace; |
| 1646 | tcp_seq tcp_new_isn(struct tcpcb *); |
| 1647 | |
| 1648 | extern int tcp_input_checksum(int, struct mbuf *, struct tcphdr *, int, int); |
| 1649 | extern void tcp_getconninfo(struct socket *, struct conninfo_tcp *); |
| 1650 | extern void add_to_time_wait(struct tcpcb *, uint32_t delay); |
| 1651 | extern void tcp_pmtud_revert_segment_size(struct tcpcb *tp); |
| 1652 | extern void tcp_rxtseg_insert(struct tcpcb *, tcp_seq, tcp_seq); |
| 1653 | extern struct tcp_rxt_seg *tcp_rxtseg_find(struct tcpcb *, tcp_seq, tcp_seq); |
| 1654 | extern void tcp_rxtseg_clean(struct tcpcb *); |
| 1655 | extern boolean_t tcp_rxtseg_detect_bad_rexmt(struct tcpcb *, tcp_seq); |
| 1656 | extern boolean_t tcp_rxtseg_dsack_for_tlp(struct tcpcb *); |
| 1657 | extern u_int32_t tcp_rxtseg_total_size(struct tcpcb *tp); |
| 1658 | extern void tcp_rexmt_save_state(struct tcpcb *tp); |
| 1659 | extern void tcp_interface_send_probe(u_int16_t if_index_available); |
| 1660 | extern void tcp_probe_connectivity(struct ifnet *ifp, u_int32_t enable); |
| 1661 | extern void tcp_get_connectivity_status(struct tcpcb *, |
| 1662 | struct tcp_conn_status *); |
| 1663 | |
| 1664 | extern void tcp_fill_keepalive_offload_frames(struct ifnet *, |
| 1665 | struct ifnet_keepalive_offload_frame *, u_int32_t, size_t, u_int32_t *); |
| 1666 | |
| 1667 | extern boolean_t tfo_enabled(const struct tcpcb *tp); |
| 1668 | extern void tcp_disable_tfo(struct tcpcb *tp); |
| 1669 | extern void tcp_tfo_gen_cookie(struct inpcb *inp, u_char *out, size_t blk_size); |
| 1670 | #define TCP_FASTOPEN_KEYLEN 16 |
| 1671 | extern int tcp_freeq(struct tcpcb *tp); |
| 1672 | extern errno_t tcp_notify_ack_id_valid(struct tcpcb *, struct socket *, u_int32_t); |
| 1673 | extern errno_t tcp_add_notify_ack_marker(struct tcpcb *, u_int32_t); |
| 1674 | extern void tcp_notify_ack_free(struct tcpcb *); |
| 1675 | extern void tcp_notify_acknowledgement(struct tcpcb *, struct socket *); |
| 1676 | extern void tcp_get_notify_ack_count(struct tcpcb *, |
| 1677 | struct tcp_notify_ack_complete *); |
| 1678 | extern void tcp_get_notify_ack_ids(struct tcpcb *tp, |
| 1679 | struct tcp_notify_ack_complete *); |
| 1680 | extern void tcp_update_mss_locked(struct socket *, struct ifnet *); |
| 1681 | |
| 1682 | extern int get_tcp_inp_list(struct inpcb **, int, inp_gen_t); |
| 1683 | extern bool tcp_notify_ack_active(struct socket *so); |
| 1684 | |
| 1685 | #if MPTCP |
| 1686 | extern int mptcp_input_preproc(struct tcpcb *tp, struct mbuf *m, |
| 1687 | struct tcphdr *th, int drop_hdrlen); |
| 1688 | extern uint32_t mptcp_output_csum(struct mbuf *m, uint64_t dss_val, |
| 1689 | uint32_t sseq, uint16_t dlen); |
| 1690 | extern int mptcp_adj_mss(struct tcpcb *, boolean_t); |
| 1691 | extern void mptcp_insert_rmap(struct tcpcb *tp, struct mbuf *m, struct tcphdr *th); |
| 1692 | #endif |
| 1693 | |
| 1694 | __private_extern__ void tcp_update_stats_per_flow( |
| 1695 | struct ifnet_stats_per_flow *, struct ifnet *); |
| 1696 | |
| 1697 | #endif /* BSD_KERNEL_RPIVATE */ |
| 1698 | |
| 1699 | #endif /* _NETINET_TCP_VAR_H_ */ |
| 1700 | |