1 /* 2 * Copyright (c) 2000-2021 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 #if !KERNEL 73 #include <TargetConditionals.h> 74 #endif 75 76 #if defined(__LP64__) 77 #define _TCPCB_PTR(x) u_int32_t 78 #define _TCPCB_LIST_HEAD(name, type) \ 79 struct name { \ 80 u_int32_t lh_first; \ 81 } 82 #else 83 #define _TCPCB_PTR(x) x 84 #define _TCPCB_LIST_HEAD(name, type) LIST_HEAD(name, type) 85 #endif 86 87 #ifdef KERNEL_PRIVATE 88 89 #define TCP_RETRANSHZ 1000 /* granularity of TCP timestamps, 1ms */ 90 /* Minimum time quantum within which the timers are coalesced */ 91 #define TCP_TIMER_10MS_QUANTUM (TCP_RETRANSHZ/100) /* every 10ms */ 92 #define TCP_TIMER_100MS_QUANTUM (TCP_RETRANSHZ/10) /* every 100ms */ 93 #define TCP_TIMER_500MS_QUANTUM (TCP_RETRANSHZ/2) /* every 500ms */ 94 95 #define TCP_RETRANSHZ_TO_USEC 1000 96 97 #define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */ 98 99 /* Always allow at least 16 packets worth of recv window when adjusting 100 * recv window using inter-packet arrival jitter. 101 */ 102 #define MIN_IAJ_WIN 16 103 104 /* A variation in delay of this many milliseconds is tolerable. This limit has to 105 * be low but greater than zero. We also use standard deviation on jitter to adjust 106 * this limit for different link and connection types. 107 */ 108 #define ALLOWED_IAJ 5 109 110 /* Ignore the first few packets on a connection until the ACK clock gets going 111 */ 112 #define IAJ_IGNORE_PKTCNT 40 113 114 /* Let the accumulated IAJ value increase by this threshold at most. This limit 115 * will control how many ALLOWED_IAJ measurements a receiver will have to see 116 * before opening the receive window 117 */ 118 #define ACC_IAJ_HIGH_THRESH 100 119 120 /* When accumulated IAJ reaches this value, the receiver starts to react by 121 * closing the window 122 */ 123 #define ACC_IAJ_REACT_LIMIT 200 124 125 /* If the number of small packets (smaller than IAJ packet size) seen on a 126 * connection is more than this threshold, reset the size and learn it again. 127 * This is needed because the sender might send smaller segments after PMTU 128 * discovery and the receiver has to learn the new size. 129 */ 130 #define RESET_IAJ_SIZE_THRESH 20 131 132 /* 133 * Adaptive timeout is a read/write timeout specified by the application to 134 * get a socket event when the transport layer detects a stall in data 135 * transfer. The value specified is the number of probes that can be sent 136 * to the peer before generating an event. Since it is not specified as 137 * a time value, the timeout will adjust based on the RTT seen on the link. 138 * The timeout will start only when there is an indication that the read/write 139 * operation is not making progress. 140 * 141 * If a write operation stalls, the probe will be retransmission of data. 142 * If a read operation stalls, the probe will be a keep-alive packet. 143 * 144 * The maximum value of adaptive timeout is set to 10 which will allow 145 * transmission of enough number of probes to the peer. 146 */ 147 #define TCP_ADAPTIVE_TIMEOUT_MAX 10 148 149 #define TCP_CONNECTIVITY_PROBES_MAX 5 150 151 /* 152 * Kernel variables for tcp. 153 */ 154 155 /* TCP segment queue entry */ 156 struct tseg_qent { 157 LIST_ENTRY(tseg_qent) tqe_q; 158 int tqe_len; /* TCP segment data length */ 159 struct tcphdr *tqe_th; /* a pointer to tcp header */ 160 struct mbuf *tqe_m; /* mbuf contains packet */ 161 }; 162 LIST_HEAD(tsegqe_head, tseg_qent); 163 164 struct sackblk { 165 tcp_seq start; /* start seq no. of sack block */ 166 tcp_seq end; /* end seq no. */ 167 }; 168 169 struct sackhole { 170 tcp_seq start; /* start seq no. of hole */ 171 tcp_seq end; /* end seq no. */ 172 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 173 u_int32_t rxmit_start; /* timestamp of first retransmission */ 174 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 175 }; 176 177 struct sackhint { 178 struct sackhole *nexthole; 179 int sack_bytes_rexmit; 180 int sack_bytes_acked; 181 }; 182 183 struct tcp_rxt_seg { 184 tcp_seq rx_start; 185 tcp_seq rx_end; 186 u_int16_t rx_count; 187 u_int16_t rx_flags; 188 #define TCP_RXT_SPURIOUS 0x1 /* received DSACK notification */ 189 #define TCP_RXT_DSACK_FOR_TLP 0x2 190 SLIST_ENTRY(tcp_rxt_seg) rx_link; 191 }; 192 193 struct tcp_notify_ack_marker { 194 tcp_seq notify_snd_una; /* Notify when snd_una crosses this seq */ 195 tcp_notify_ack_id_t notify_id; 196 SLIST_ENTRY(tcp_notify_ack_marker) notify_next; 197 }; 198 199 struct tcptemp { 200 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 201 struct tcphdr tt_t; 202 }; 203 204 struct bwmeas { 205 tcp_seq bw_start; /* start of bw measurement */ 206 uint32_t bw_ts; /* timestamp when bw measurement started */ 207 uint32_t bw_size; /* burst size in bytes for this bw measurement */ 208 uint32_t bw_minsizepkts; /* Min burst size as segments */ 209 uint32_t bw_maxsizepkts; /* Max burst size as segments */ 210 uint32_t bw_minsize; /* Min size in bytes */ 211 uint32_t bw_maxsize; /* Max size in bytes */ 212 uint32_t bw_sndbw; /* Measured send bandwidth */ 213 uint32_t bw_sndbw_max; /* Max measured bandwidth */ 214 uint32_t bw_rcvbw_max; /* Max receive bandwidth measured */ 215 }; 216 217 /* MPTCP Data sequence map entry */ 218 struct mpt_dsn_map { 219 uint64_t mpt_dsn; /* data seq num recvd */ 220 uint32_t mpt_sseq; /* relative subflow # */ 221 uint16_t mpt_len; /* length of mapping */ 222 uint16_t mpt_csum; /* checksum value if on */ 223 uint8_t mpt_dfin; /* It's a DATA_FIN */ 224 }; 225 #define tcp6cb tcpcb /* for KAME src sync over BSD*'s */ 226 227 struct tcp_ccstate { 228 union { 229 struct tcp_cubic_state { 230 u_int32_t tc_last_max; /* cwnd at last loss */ 231 u_int32_t tc_epoch_start; /* TS of last loss */ 232 u_int32_t tc_origin_point; /* window at the start of an epoch */ 233 u_int32_t tc_tcp_win; /* computed tcp win */ 234 u_int32_t tc_tcp_bytes_acked; /* bytes acked */ 235 u_int32_t tc_avg_lastmax; /* Average of last max */ 236 u_int32_t tc_mean_deviation; /* Mean absolute deviation */ 237 float tc_epoch_period; /* K parameter */ 238 } _cubic_state_; 239 #define cub_last_max __u__._cubic_state_.tc_last_max 240 #define cub_epoch_start __u__._cubic_state_.tc_epoch_start 241 #define cub_origin_point __u__._cubic_state_.tc_origin_point 242 #define cub_tcp_win __u__._cubic_state_.tc_tcp_win 243 #define cub_tcp_bytes_acked __u__._cubic_state_.tc_tcp_bytes_acked 244 #define cub_epoch_period __u__._cubic_state_.tc_epoch_period 245 #define cub_avg_lastmax __u__._cubic_state_.tc_avg_lastmax 246 #define cub_mean_dev __u__._cubic_state_.tc_mean_deviation 247 struct tcp_ledbat_state { 248 uint32_t num_slowdown_events; 249 uint32_t slowdown_ts; 250 uint32_t slowdown_begin; 251 uint32_t md_bytes_acked; 252 } _ledbat_state_; 253 #define ledbat_slowdown_events __u__._ledbat_state_.num_slowdown_events 254 #define ledbat_slowdown_ts __u__._ledbat_state_.slowdown_ts 255 #define ledbat_slowdown_begin __u__._ledbat_state_.slowdown_begin 256 #define ledbat_md_bytes_acked __u__._ledbat_state_.md_bytes_acked 257 } __u__; 258 }; 259 260 struct tcp_rledbat_state { 261 uint32_t num_slowdown_events; /* Number of slowdown events until reset */ 262 uint32_t slowdown_ts; /* Next slowdown timestamp */ 263 uint32_t slowdown_begin; /* Slowdown begin time */ 264 uint32_t reduction_end; /* Used to track rl_win reduction once per RTT */ 265 uint32_t rcvd_bytes; /* bytes received used for byte counting */ 266 uint32_t md_rcvd_bytes; /* bytes received during MD used for byte counting */ 267 uint32_t win; /* receive Ledbat window size */ 268 uint32_t ssthresh; /* receive Ledbat ssthresh */ 269 uint32_t drained_bytes; /* bytes drained from the flight-size */ 270 uint32_t win_ws; /* receive Ledbat window after avoiding window shrinking */ 271 }; 272 273 /* 274 * Tcp control block, one per tcp; fields: 275 * Organized for 16 byte cacheline efficiency. 276 */ 277 struct tcpcb { 278 struct tsegqe_head t_segq; 279 uint32_t t_dupacks; /* consecutive dup acks recd */ 280 int t_state; /* state of this connection */ 281 uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */ 282 struct tcptimerentry tentry; /* entry in timer list */ 283 284 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 285 uint32_t t_flags; 286 #define TF_ACKNOW 0x00001 /* ack peer immediately */ 287 #define TF_DELACK 0x00002 /* ack, but try to delay it */ 288 #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ 289 #define TF_NOOPT 0x00008 /* don't use tcp options */ 290 #define TF_SENTFIN 0x00010 /* have sent FIN */ 291 #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ 292 #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ 293 #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ 294 #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ 295 #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ 296 #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) - unused but needed for backwards compatibility */ 297 #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ 298 #define TF_NOPUSH 0x01000 /* don't push */ 299 #define TF_REQ_CC 0x02000 /* have/will request CC */ 300 #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ 301 #define TF_SENDCCNEW 0x08000 /* Unused */ 302 #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ 303 #define TF_LOCAL 0x20000 /* connection to a host on local link */ 304 #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ 305 #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ 306 #define TF_LASTIDLE 0x100000 /* connection was previously idle */ 307 #define TF_FASTRECOVERY 0x200000 /* in NewReno Fast Recovery */ 308 #define TF_WASFRECOVERY 0x400000 /* was in NewReno Fast Recovery */ 309 #define TF_SIGNATURE 0x800000 /* require MD5 digests (RFC2385) */ 310 #define TF_MAXSEGSNT 0x1000000 /* last segment sent was a full segment */ 311 #define TF_STREAMING_ON 0x2000000 /* Receiver detected streaming */ 312 #define TF_PMTUD 0x4000000 /* Perform Path MTU Discovery for this connection */ 313 #define TF_CLOSING 0x8000000 /* pending tcp close */ 314 #define TF_TSO 0x10000000 /* TCP Segment Offloading is enable on this connection */ 315 #define TF_BLACKHOLE 0x20000000 /* Path MTU Discovery Black Hole detection */ 316 #define TF_TIMER_ONLIST 0x40000000 /* pcb is on tcp_timer_list */ 317 #define TF_STRETCHACK 0x80000000 /* receiver is going to delay acks */ 318 319 tcp_seq snd_una; /* send unacknowledged */ 320 tcp_seq snd_max; /* highest sequence number sent; 321 * used to recognize retransmits 322 */ 323 tcp_seq snd_nxt; /* send next */ 324 tcp_seq snd_up; /* send urgent pointer */ 325 326 tcp_seq snd_wl1; /* window update seg seq number */ 327 tcp_seq snd_wl2; /* window update seg ack number */ 328 tcp_seq iss; /* initial send sequence number */ 329 tcp_seq irs; /* initial receive sequence number */ 330 331 tcp_seq rcv_nxt; /* receive next */ 332 tcp_seq rcv_adv; /* advertised window */ 333 uint32_t rcv_wnd; /* receive window */ 334 uint32_t t_last_recwin; 335 tcp_seq rcv_up; /* receive urgent pointer */ 336 337 uint32_t snd_wnd; /* send window */ 338 uint32_t snd_cwnd; /* congestion-controlled window */ 339 uint32_t snd_ssthresh; /* snd_cwnd size threshold for 340 * for slow start exponential to 341 * linear switch 342 */ 343 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 344 345 uint32_t t_maxopd; /* mss plus options */ 346 uint32_t t_rcvtime; /* time at which a packet was received */ 347 uint32_t t_sndtime; /* time at which we last sent new data */ 348 uint32_t t_starttime; /* time connection was established */ 349 int t_rtttime; /* tcp clock when rtt calculation was started */ 350 tcp_seq t_rtseq; /* sequence number being timed */ 351 352 uint32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 353 uint32_t rfbuf_cnt; /* recv buffer autoscaling byte count */ 354 uint32_t rfbuf_space; /* Current "ideal" estimate of the space */ 355 356 int t_rxtcur; /* current retransmit value (ticks) */ 357 unsigned int t_maxseg; /* maximum segment size */ 358 int t_srtt; /* smoothed round-trip time */ 359 int t_rttvar; /* variance in round-trip time */ 360 361 uint64_t t_accsleep_ms; /* accumulated sleep time since last boot */ 362 uint32_t t_reassqlen; /* length of reassembly queue */ 363 uint32_t t_reassq_mbcnt; /* amount in bytes of mbuf space used */ 364 uint16_t t_rxtshift; /* log(2) of rexmt exp. backoff */ 365 uint32_t t_rttmin; /* minimum rtt allowed */ 366 uint32_t t_rttbest; /* best rtt we've seen */ 367 uint32_t t_rttcur; /* most recent value of rtt */ 368 uint32_t t_rttupdated; /* number of times rtt sampled */ 369 uint32_t t_rxt_conndroptime; /* retxmt conn gets dropped after this time, when set */ 370 uint32_t t_rxtstart; /* time at which retransmission started */ 371 uint32_t max_sndwnd; /* largest window peer has offered */ 372 373 int t_softerror; /* possible error not yet reported */ 374 /* out-of-band data */ 375 char t_oobflags; /* have some */ 376 char t_iobc; /* input character */ 377 #define TCPOOB_HAVEDATA 0x01 378 #define TCPOOB_HADDATA 0x02 379 /* RFC 1323 variables */ 380 u_int8_t snd_scale; /* window scaling for send window */ 381 u_int8_t rcv_scale; /* window scaling for recv window */ 382 u_int8_t request_r_scale; /* pending window scaling */ 383 u_int8_t requested_s_scale; 384 u_int8_t tcp_cc_index; /* index of congestion control algorithm */ 385 u_int8_t t_adaptive_rtimo; /* Read timeout used as a multiple of RTT */ 386 u_int8_t t_adaptive_wtimo; /* Write timeout used as a multiple of RTT */ 387 u_int8_t t_stretchack_delayed; /* stretch ack delayed */ 388 389 /* State for limiting early retransmits when SACK is not enabled */ 390 u_int16_t t_early_rexmt_count; /* count of early rexmts */ 391 u_int32_t t_early_rexmt_win; /* window for limiting early rexmts */ 392 393 u_int32_t ts_recent; /* timestamp echo data */ 394 395 u_int32_t ts_recent_age; /* when last updated */ 396 tcp_seq last_ack_sent; 397 /* RFC 3465 variables */ 398 u_int32_t t_bytes_acked; /* ABC "bytes_acked" parameter */ 399 400 int t_lastchain; /* amount of packets chained last time around */ 401 uint16_t t_unacksegs; /* received but unacked segments for delaying acks */ 402 403 /* 404 * Pretty arbitrary value ;-) 405 * Goal is to make sure that some ACKs are being sent more frequently 406 * to allow the other side to ramp-up. 407 */ 408 #define TCP_FORCED_ACKS_COUNT 16 409 uint16_t t_forced_acks; /* count of pure ACKs that need to be forced out */ 410 uint8_t t_rexmtthresh; /* duplicate ack threshold for entering fast recovery */ 411 uint8_t t_rtimo_probes; /* number of adaptive rtimo probes sent */ 412 uint32_t t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */ 413 uint32_t t_persist_stop; /* persistence limit deadline if triggered by ZWP */ 414 uint32_t t_notsent_lowat; /* Low water for not sent data */ 415 416 /* Receiver state for stretch-ack algorithm */ 417 u_int32_t rcv_unackwin; /* to measure win for stretching acks */ 418 u_int32_t rcv_by_unackwin; /* bytes seen during the last ack-stretching win */ 419 u_int32_t rcv_by_unackhalfwin; 420 u_int32_t rcv_nostrack_ts; /* timestamp when stretch ack was disabled automatically */ 421 u_int32_t rcv_nostrack_pkts; /* pkts received since strech ack was disabled */ 422 u_int16_t rcv_waitforss; /* wait for packets during slow-start */ 423 424 /* ECN stats */ 425 u_int16_t ecn_flags; 426 #define TE_SETUPSENT 0x0001 /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */ 427 #define TE_SETUPRECEIVED 0x0002 /* Indicate we have received ECN-SETUP SYN or SYN-ACK */ 428 #define TE_SENDIPECT 0x0004 /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */ 429 #define TE_SENDCWR 0x0008 /* Indicate that the next non-retransmit should have the TCP CWR flag set */ 430 #define TE_SENDECE 0x0010 /* Indicate that the next packet should have the TCP ECE flag set */ 431 #define TE_INRECOVERY 0x0020 /* connection entered recovery after receiving ECE */ 432 #define TE_RECV_ECN_CE 0x0040 /* Received IPTOS_ECN_CE marking atleast once */ 433 #define TE_RECV_ECN_ECE 0x0080 /* Received ECE marking atleast once */ 434 #define TE_LOST_SYN 0x0100 /* Lost SYN with ECN setup */ 435 #define TE_LOST_SYNACK 0x0200 /* Lost SYN-ACK with ECN setup */ 436 #define TE_ECN_MODE_ENABLE 0x0400 /* Option ECN mode set to enable */ 437 #define TE_ECN_MODE_DISABLE 0x0800 /* Option ECN mode set to disable */ 438 #define TE_ENABLE_ECN 0x1000 /* Enable negotiation of ECN */ 439 #define TE_ECN_ON (TE_SETUPSENT | TE_SETUPRECEIVED) /* Indicate ECN was successfully negotiated on a connection) */ 440 #define TE_CEHEURI_SET 0x2000 /* We did our CE-probing at the beginning */ 441 #define TE_CLIENT_SETUP 0x4000 /* setup from client side */ 442 #define TE_RCVD_SYN_RST 0x8000 /* Received RST to the first ECN enabled SYN */ 443 444 u_int32_t t_ecn_recv_ce; /* Received CE from the network */ 445 u_int32_t t_ecn_recv_cwr; /* Packets received with CWR */ 446 447 /* state for bad retransmit recovery */ 448 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 449 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 450 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 451 int t_srtt_prev; /* srtt prior to retransmit */ 452 int t_rttvar_prev; /* rttvar prior to retransmit */ 453 u_int32_t t_badrexmt_time; /* bad rexmt detection time */ 454 455 /* Packet reordering metric */ 456 u_int32_t t_reorderwin; /* Reordering late time offset */ 457 458 /* SACK related state */ 459 int16_t snd_numholes; /* number of holes seen by sender */ 460 tcp_seq sack_newdata; /* New data xmitted in this recovery 461 * episode starts at this seq number */ 462 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 463 /* SACK scoreboard (sorted) */ 464 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 465 int rcv_numsacks; /* # distinct sack blks present */ 466 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 467 struct sackhint sackhint; /* SACK scoreboard hint */ 468 tcp_seq send_highest_sack; /* Sequence number of fresh data sent after the most recent fast-retransmit */ 469 int t_new_dupacks; /* Dupacks received above send_highest_sack */ 470 471 struct mbuf *t_pktlist_head; /* First packet in transmit chain */ 472 struct mbuf *t_pktlist_tail; /* Last packet in transmit chain */ 473 u_int32_t t_pktlist_sentlen; /* total bytes in transmit chain */ 474 475 u_int32_t t_keepidle; /* keepalive idle timer (override global if > 0) */ 476 u_int32_t t_keepinit; /* connection timeout, i.e. idle time 477 * in SYN_SENT or SYN_RECV state */ 478 u_int32_t t_keepintvl; /* interval between keepalives */ 479 u_int32_t t_keepcnt; /* number of keepalives before close */ 480 481 u_int32_t tso_max_segment_size; /* TSO maximum segment unit for NIC */ 482 u_int16_t t_pmtud_lastseg_size; /* size of the last sent segment */ 483 u_int32_t t_pmtud_saved_maxopd; /* MSS saved before performing PMTU-D BlackHole detection */ 484 u_int32_t t_pmtud_start_ts; /* Time of PMTUD blackhole detection */ 485 486 struct{ 487 u_int32_t rxduplicatebytes; 488 u_int32_t rxoutoforderbytes; 489 u_int32_t txretransmitbytes; 490 u_int16_t synrxtshift; 491 u_int16_t rxmitsyns; 492 u_int16_t unused_pad_to_8; 493 u_int32_t rxmitpkts; 494 uint32_t delayed_acks_sent; 495 uint32_t acks_delayed; 496 } t_stat; 497 u_int8_t t_notify_ack_count; 498 u_int8_t t_ecn_recv_ce_pkt; /* Received packet with CE-bit set (independent from last_ack_sent) */ 499 u_int32_t t_cached_maxopd; /* default for MSS adjustment using link status report */ 500 501 uint32_t bg_ssthresh; /* Slow start threshold until delay increases */ 502 uint32_t t_flagsext; /* Another field to accommodate more flags */ 503 #define TF_RXTFINDROP 0x1 /* Drop conn after retransmitting FIN 3 times */ 504 #define TF_RCVUNACK_WAITSS 0x2 /* set when the receiver should not stretch acks */ 505 #define TF_BWMEAS_INPROGRESS 0x4 /* Indicate BW meas is happening */ 506 #define TF_MEASURESNDBW 0x8 /* Measure send bw on this connection */ 507 #define TF_LAST_IS_PSH 0x10 /* Indicates whether the last packet in the rcv socket buffer had the PUSH-flag set */ 508 #define TF_SACK_ENABLE 0x20 /* SACK is enabled */ 509 #define TF_RECOMPUTE_RTT 0x40 /* recompute RTT after spurious retransmit */ 510 #define TF_DETECT_READSTALL 0x80 /* Used to detect a stall during read operation */ 511 #define TF_RECV_THROTTLE 0x100 /* Input throttling active */ 512 #define TF_NOSTRETCHACK 0x200 /* ack every other packet */ 513 #define TF_NOTIMEWAIT 0x800 /* Avoid going into time-wait */ 514 #define TF_SENT_TLPROBE 0x1000 /* Sent data in PTO */ 515 #define TF_PKTS_REORDERED 0x2000 /* Detected reordering */ 516 #define TF_DELAY_RECOVERY 0x4000 /* delay fast recovery */ 517 #define TF_FORCE 0x8000 /* force 1 byte out */ 518 #define TF_DISABLE_STRETCHACK 0x10000 /* auto-disable stretch ack */ 519 #define TF_NOBLACKHOLE_DETECTION 0x20000 /* Disable PMTU blackhole detection */ 520 #define TF_RESCUE_RXT 0x80000 /* SACK rescue retransmit */ 521 #define TF_CWND_NONVALIDATED 0x100000 /* cwnd non validated */ 522 #define TF_IF_PROBING 0x200000 /* Trigger interface probe timeout */ 523 #define TF_FASTOPEN 0x400000 /* TCP Fastopen is enabled */ 524 #define TF_REASS_INPROG 0x800000 /* Reassembly is in progress */ 525 #define TF_FASTOPEN_FORCE_ENABLE 0x1000000 /* Force-enable TCP Fastopen */ 526 #define TF_LOGGED_CONN_SUMMARY 0x2000000 /* Connection summary was logged */ 527 528 #if TRAFFIC_MGT 529 /* Inter-arrival jitter related state */ 530 uint32_t iaj_rcv_ts; /* tcp clock when the first packet was received */ 531 int iaj_size; /* Size of packet for iaj measurement */ 532 uint8_t iaj_small_pkt; /* Count of packets smaller than iaj_size */ 533 uint8_t t_pipeack_ind; /* index for next pipeack sample */ 534 uint16_t iaj_pktcnt; /* packet count, to avoid throttling initially */ 535 uint32_t acc_iaj; /* Accumulated iaj */ 536 uint32_t avg_iaj; /* Mean */ 537 uint32_t std_dev_iaj; /* Standard deviation */ 538 #endif /* TRAFFIC_MGT */ 539 struct bwmeas *t_bwmeas; /* State for bandwidth measurement */ 540 tcp_seq t_idleat; /* rcv_nxt at idle time */ 541 TAILQ_ENTRY(tcpcb) t_twentry; /* link for time wait queue */ 542 struct tcp_ccstate *t_ccstate; /* congestion control related state */ 543 /* Tail loss probe related state */ 544 tcp_seq t_tlphighrxt; /* snd_nxt after PTO */ 545 u_int32_t t_tlpstart; /* timestamp at PTO */ 546 /* DSACK data receiver state */ 547 tcp_seq t_dsack_lseq; /* DSACK left sequence */ 548 tcp_seq t_dsack_rseq; /* DSACK right sequence */ 549 /* DSACK data sender state */ 550 SLIST_HEAD(tcp_rxt_seghead, tcp_rxt_seg) t_rxt_segments; 551 uint32_t t_rxt_seg_count; 552 uint32_t t_rxt_seg_drop; 553 tcp_seq t_dsack_lastuna; /* snd_una when last recovery episode started */ 554 /* state for congestion window validation (draft-ietf-tcpm-newcwv-07) */ 555 #define TCP_PIPEACK_SAMPLE_COUNT 3 556 u_int32_t t_pipeack_sample[TCP_PIPEACK_SAMPLE_COUNT]; /* pipeack, bytes acked within RTT */ 557 tcp_seq t_pipeack_lastuna; /* una when pipeack measurement started */ 558 u_int32_t t_pipeack; 559 u_int32_t t_lossflightsize; 560 561 #if MPTCP 562 u_int32_t t_mpflags; /* flags for multipath TCP */ 563 564 #define TMPF_PREESTABLISHED 0x00000001 /* conn in pre-established state */ 565 #define TMPF_SND_KEYS 0x00000002 /* indicates that keys should be send */ 566 #define TMPF_MPTCP_TRUE 0x00000004 /* negotiated MPTCP successfully */ 567 #define TMPF_MPTCP_RCVD_KEY 0x00000008 /* state for 3-way handshake */ 568 #define TMPF_SND_MPPRIO 0x00000010 /* send priority of subflow */ 569 #define TMPF_SND_REM_ADDR 0x00000020 /* initiate address removal */ 570 #define TMPF_RCVD_DACK 0x00000040 /* received a data-ack */ 571 #define TMPF_JOINED_FLOW 0x00000080 /* Indicates additional flow */ 572 #define TMPF_BACKUP_PATH 0x00000100 /* Indicates backup path */ 573 #define TMPF_MPTCP_ACKNOW 0x00000200 /* Send Data ACK */ 574 #define TMPF_SEND_DSN 0x00000400 /* Send DSN mapping */ 575 #define TMPF_SEND_DFIN 0x00000800 /* Send Data FIN */ 576 #define TMPF_RECV_DFIN 0x00001000 /* Recv Data FIN */ 577 #define TMPF_SENT_JOIN 0x00002000 /* Sent Join */ 578 #define TMPF_RECVD_JOIN 0x00004000 /* Received Join */ 579 #define TMPF_RESET 0x00008000 /* Send RST */ 580 #define TMPF_TCP_FALLBACK 0x00010000 /* Fallback to TCP */ 581 #define TMPF_FASTCLOSERCV 0x00020000 /* Received Fastclose option */ 582 #define TMPF_EMBED_DSN 0x00040000 /* tp has DSN mapping */ 583 #define TMPF_MPTCP_READY 0x00080000 /* Can send DSS options on data */ 584 #define TMPF_INFIN_SENT 0x00100000 /* Sent infinite mapping */ 585 #define TMPF_SND_MPFAIL 0x00200000 /* Received mapping csum failure */ 586 #define TMPF_SND_JACK 0x00400000 /* Send a Join-ACK */ 587 #define TMPF_TFO_REQUEST 0x00800000 /* TFO Requested */ 588 #define TMPF_MPTCP_ECHO_ADDR 0x01000000 /* MPTCP echoes add_addr */ 589 590 #define TMPF_MPTCP_SIGNALS (TMPF_SND_MPPRIO | TMPF_SND_REM_ADDR | TMPF_SND_MPFAIL | TMPF_SND_KEYS | TMPF_SND_JACK) 591 592 tcp_seq t_mpuna; /* unacknowledged sequence */ 593 struct mptcb *t_mptcb; /* pointer to MPTCP TCB */ 594 struct mptsub *t_mpsub; /* pointer to the MPTCP subflow */ 595 struct mpt_dsn_map t_rcv_map; /* Receive mapping list */ 596 u_int8_t t_local_aid; /* Addr Id for authentication */ 597 u_int8_t t_rem_aid; /* Addr ID of another subflow */ 598 u_int8_t t_mprxtshift; /* join retransmission */ 599 #endif /* MPTCP */ 600 601 #define TFO_F_OFFER_COOKIE 0x01 /* We will offer a cookie */ 602 #define TFO_F_COOKIE_VALID 0x02 /* The received cookie is valid */ 603 #define TFO_F_COOKIE_REQ 0x04 /* Client requested a new cookie */ 604 #define TFO_F_COOKIE_SENT 0x08 /* Client did send a cookie in the SYN */ 605 #define TFO_F_SYN_LOSS 0x10 /* A SYN-loss triggered a fallback to regular TCP on the client-side */ 606 #define TFO_F_NO_SNDPROBING 0x20 /* This network is guaranteed to support TFO in the upstream direction */ 607 #define TFO_F_HEURISTIC_DONE 0x40 /* We have already marked this network as bad */ 608 u_int8_t t_tfo_flags; 609 #define TFO_S_SYNDATA_RCV 0x01 /* SYN+data has been received */ 610 #define TFO_S_COOKIEREQ_RECV 0x02 /* TFO-cookie request received */ 611 #define TFO_S_COOKIE_SENT 0x04 /* TFO-cookie announced in SYN/ACK */ 612 #define TFO_S_COOKIE_INVALID 0x08 /* Received TFO-cookie is invalid */ 613 #define TFO_S_COOKIE_REQ 0x10 /* TFO-cookie requested within the SYN */ 614 #define TFO_S_COOKIE_RCV 0x20 /* TFO-cookie received in SYN/ACK */ 615 #define TFO_S_SYN_DATA_SENT 0x40 /* SYN+data sent */ 616 #define TFO_S_SYN_DATA_ACKED 0x80 /* SYN+data has been acknowledged in SYN/ACK */ 617 #define TFO_S_SYN_LOSS 0x0100 /* SYN+TFO has been lost - fallback to regular TCP */ 618 #define TFO_S_COOKIE_WRONG 0x0200 /* Cookie we sent in the SYN was wrong */ 619 #define TFO_S_NO_COOKIE_RCV 0x0400 /* We asked for a cookie but didn't get one */ 620 #define TFO_S_HEURISTICS_DISABLE 0x0800 /* TFO-heuristics disabled it for this connection */ 621 #define TFO_S_SEND_BLACKHOLE 0x1000 /* TFO got blackholed in the send direction */ 622 #define TFO_S_RECV_BLACKHOLE 0x2000 /* TFO got blackholed in the recv direction */ 623 #define TFO_S_ONE_BYTE_PROXY 0x4000 /* TFO failed because of a proxy acknowledging just one byte */ 624 u_int16_t t_tfo_stats; 625 626 u_int8_t t_tfo_probes; /* TFO-probes we did send */ 627 /* 628 * This here is the TFO-probing state-machine. Transitions are as follows: 629 * 630 * Current state: PROBE_NONE 631 * Event: SYN+DATA acknowledged 632 * Action: Transition to PROBE_PROBING and set keepalive-timer 633 * 634 * Current state: PROBE_PROBING (initial state) 635 * Event: Receive data 636 * Action: Transition to PROBE_NONE and cancel keepalive-timer 637 * Event: Receive ACK that does not indicate a hole 638 * Action: Transition to PROBE_NONE and cancel keepalive-timer 639 * Event: Receive ACK that indicates a hole 640 * Action: Transition to PROBE_WAIT_DATA and set a short timer 641 * to wait for the final segment. 642 * Event: Keepalive-timeout (did not receive any segment) 643 * Action: Signal ETIMEDOUT as with regular keepalive-timers 644 * 645 * Current state: PROBE_WAIT_DATA 646 * Event: Receive data 647 * Action: Transition to PROBE_NONE and cancel keepalive-timer 648 * Event: Data-timeout (did not receive the expected data) 649 * Action: Signal ENODATA up to the app and close everything. 650 */ 651 #define TFO_PROBE_NONE 0 /* Not probing now */ 652 #define TFO_PROBE_PROBING 1 /* Sending out TCP-keepalives waiting for reply */ 653 #define TFO_PROBE_WAIT_DATA 2 /* Received reply, waiting for data */ 654 u_int8_t t_tfo_probe_state; 655 656 u_int32_t t_rcvoopack; /* out-of-order packets received */ 657 u_int32_t t_pawsdrop; /* segments dropped due to PAWS */ 658 u_int32_t t_sack_recovery_episode; /* SACK recovery episodes */ 659 u_int32_t t_reordered_pkts; /* packets reorderd */ 660 u_int32_t t_dsack_sent; /* Sent DSACK notification */ 661 u_int32_t t_dsack_recvd; /* Received a valid DSACK option */ 662 SLIST_HEAD(, tcp_notify_ack_marker) t_notify_ack; /* state for notifying data acknowledgements */ 663 u_int32_t t_recv_throttle_ts; /* TS for start of recv throttle */ 664 u_int32_t t_rxt_minimum_timeout; /* minimum retransmit timeout in ms */ 665 uint32_t t_challengeack_last; /* last time challenge ACK was sent per sec */ 666 uint32_t t_challengeack_count; /* # of challenge ACKs already sent per sec */ 667 668 u_int32_t t_log_flags; /* TCP logging flags*/ 669 u_int32_t t_connect_time; /* time when the connection started */ 670 671 uint32_t t_comp_gencnt; /* Current compression generation-count */ 672 uint32_t t_comp_lastinc; /* Last time the gen-count was changed - should change every TCP_COMP_CHANGE_RATE ms */ 673 #define TCP_COMP_CHANGE_RATE 5 /* Intervals at which we change the gencnt. Means that worst-case we send one ACK every TCP_COMP_CHANGE_RATE ms */ 674 675 uint32_t t_ts_offset; /* Randomized timestamp offset to hide on-the-wire timestamp */ 676 677 #define NCURR_RTT_HIST 4 /* Number of current RTT samples (k) */ 678 uint32_t curr_rtt_hist[NCURR_RTT_HIST]; /* last k current RTT samples */ 679 uint32_t curr_rtt_min; /* Minimum current RTT from last k samples */ 680 uint32_t curr_rtt_index; /* Index for current RTT samples */ 681 682 tcp_seq rcv_high; /* highest sequence number received */ 683 uint32_t tsv_high; /* timestamp value of highest received sequence number */ 684 struct tcp_rledbat_state t_rlstate; /* State used by rLedbat */ 685 686 uint32_t rcv_srtt; /* receiver's SRTT, coarse when Timestamp not supported */ 687 uint32_t rcv_rtt_est_ts; /* start of measurement for estimating RTT when Timestamp not supported */ 688 uint32_t rcv_rtt_est_seq; /* expected sequence number for completion of 1RTT when Timestamp not supported */ 689 690 uuid_t t_fsw_uuid; 691 uuid_t t_flow_uuid; 692 }; 693 694 #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) 695 #define SACK_ENABLED(tp) (tp->t_flagsext & TF_SACK_ENABLE) 696 697 /* 698 * If the connection is in a throttled state due to advisory feedback from 699 * the interface output queue, reset that state. We do this in favor 700 * of entering recovery because the data transfer during recovery 701 * should be just a trickle and it will help to improve performance. 702 * We also do not want to back off twice in the same RTT. 703 */ 704 #define ENTER_FASTRECOVERY(_tp_) do { \ 705 (_tp_)->t_flags |= TF_FASTRECOVERY; \ 706 if (INP_IS_FLOW_CONTROLLED((_tp_)->t_inpcb)) \ 707 inp_reset_fc_state((_tp_)->t_inpcb); \ 708 if (!SLIST_EMPTY(&tp->t_rxt_segments)) \ 709 tcp_rxtseg_clean(tp); \ 710 (_tp_)->t_new_dupacks = 0; \ 711 } while(0) 712 713 #define EXIT_FASTRECOVERY(_tp_) do { \ 714 (_tp_)->t_flags &= ~TF_FASTRECOVERY; \ 715 (_tp_)->t_dupacks = 0; \ 716 (_tp_)->t_new_dupacks = 0; \ 717 (_tp_)->t_rexmtthresh = (uint8_t)tcprexmtthresh; \ 718 (_tp_)->t_bytes_acked = 0; \ 719 (_tp_)->ecn_flags &= ~TE_INRECOVERY; \ 720 (_tp_)->t_timer[TCPT_PTO] = 0; \ 721 (_tp_)->t_flagsext &= ~TF_RESCUE_RXT; \ 722 (_tp_)->t_lossflightsize = 0; \ 723 (_tp_)->sackhint.sack_bytes_acked = 0; \ 724 } while(0) 725 726 /* 727 * When the number of duplicate acks received is less than 728 * the retransmit threshold, use Limited Transmit algorithm 729 */ 730 extern uint8_t tcprexmtthresh; 731 #define ALLOW_LIMITED_TRANSMIT(_tp_) \ 732 ((_tp_)->t_dupacks > 0 && \ 733 (_tp_)->t_dupacks < (_tp_)->t_rexmtthresh && \ 734 ((_tp_)->t_flagsext & (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) \ 735 != (TF_PKTS_REORDERED|TF_DELAY_RECOVERY)) 736 737 /* 738 * This condition is true is timestamp option is supported 739 * on a connection. 740 */ 741 #define TSTMP_SUPPORTED(_tp_) \ 742 (((_tp_)->t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) == \ 743 (TF_REQ_TSTMP|TF_RCVD_TSTMP)) 744 745 /* 746 * This condition is true if window scale option is supported 747 * on a connection 748 */ 749 #define TCP_WINDOW_SCALE_ENABLED(_tp_) \ 750 (((_tp_)->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) == \ 751 (TF_RCVD_SCALE|TF_REQ_SCALE)) 752 753 /* Is ECN enabled end-to-end */ 754 #define TCP_ECN_ENABLED(_tp_) \ 755 (((_tp_)->ecn_flags & (TE_ECN_ON)) == (TE_ECN_ON)) 756 757 /* 758 * Gives number of bytes acked by this ack 759 */ 760 #define BYTES_ACKED(_th_, _tp_) \ 761 ((_th_)->th_ack - (_tp_)->snd_una) 762 763 /* Returns true if a DSACK option should be sent */ 764 #define TCP_SEND_DSACK_OPT(_tp_) \ 765 ((_tp_)->t_dsack_lseq > 0 && (_tp_)->t_dsack_rseq > 0) 766 767 /* 768 * Returns true if a DSACK sequence is within the max send window that will 769 * be accepted. In order to set a window to validate sequence numbers, the 770 * max send window within which a DSACK option is processed is limited. 771 * 772 * We need to choose a maximum window to check if the sequence number is 773 * within the window. One arbitrary choice is 256 * MSS because if the 774 * window is as large as 256 segments it might be big enough to ignore the 775 * DSACK option. Choosing a much larger limit means that the memory for 776 * retransmit segments can be held for a longer time. 777 */ 778 #define TCP_DSACK_MAX_SEND_WINDOW(_tp_) (MIN((_tp_)->snd_wnd, tcp_autosndbuf_max)) 779 #define TCP_DSACK_SEQ_IN_WINDOW(_tp_, _seq_, _una_) \ 780 (SEQ_LEQ((_seq_), (_tp_)->snd_max) && \ 781 SEQ_GEQ((_seq_), ((_una_) - TCP_DSACK_MAX_SEND_WINDOW(_tp_)))) 782 783 #define TCP_RESET_REXMT_STATE(_tp_) do { \ 784 (_tp_)->t_rxtshift = 0; \ 785 (_tp_)->t_rxtstart = 0; \ 786 mptcp_reset_rexmit_state((_tp_)); \ 787 } while(0); 788 789 #define TCP_IF_STATE_CHANGED(tp, probe_if_index) \ 790 (probe_if_index > 0 && tp->t_inpcb->inp_last_outifp != NULL && \ 791 probe_if_index == tp->t_inpcb->inp_last_outifp->if_index) 792 793 #define TCP_RLEDBAT_ENABLED(_tp) \ 794 (tcp_rledbat == 1 && TSTMP_SUPPORTED(_tp)) 795 796 #define TCP_RECV_BG(_so) \ 797 (tcp_recv_bg == 1 || IS_TCP_RECV_BG(_so)) 798 799 #define TCP_USE_RLEDBAT(_tp, _so) \ 800 TCP_RLEDBAT_ENABLED(_tp) && TCP_RECV_BG(_so) 801 802 /* 803 * Structure to hold TCP options that are only used during segment 804 * processing (in tcp_input), but not held in the tcpcb. 805 * It's basically used to reduce the number of parameters 806 * to tcp_dooptions. 807 */ 808 struct tcpopt { 809 uint32_t to_flags; /* which options are present */ 810 #define TOF_TS 0x0001 /* timestamp */ 811 #define TOF_MSS 0x0010 812 #define TOF_SCALE 0x0020 813 #define TOF_SIGNATURE 0x0040 /* signature option present */ 814 #define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */ 815 #define TOF_SACK 0x0100 /* Peer sent SACK option */ 816 #define TOF_MPTCP 0x0200 /* MPTCP options to be dropped */ 817 #define TOF_TFO 0x0400 /* TFO cookie option present */ 818 #define TOF_TFOREQ 0x0800 /* TFO cookie request present */ 819 uint32_t to_tsval; 820 uint32_t to_tsecr; 821 uint16_t to_mss; 822 uint8_t to_requested_s_scale; 823 uint8_t to_nsacks; /* number of SACK blocks */ 824 u_char *to_sacks; /* pointer to the first SACK blocks */ 825 u_char *to_tfo; /* pointer to the TFO cookie */ 826 }; 827 828 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 829 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 830 831 /* TFO-specific defines */ 832 #define TFO_COOKIE_LEN_MIN 4 833 #define TFO_COOKIE_LEN_DEFAULT 8 834 #define TFO_COOKIE_LEN_MAX 16 835 836 /* 837 * The initial retransmission should happen at rtt + 4 * rttvar. 838 * Because of the way we do the smoothing, srtt and rttvar 839 * will each average +1/2 tick of bias. When we compute 840 * the retransmit timer, we want 1/2 tick of rounding and 841 * 1 extra tick because of +-1/2 tick uncertainty in the 842 * firing of the timer. The bias will give us exactly the 843 * 1.5 tick we need. But, because the bias is 844 * statistical, we have to test that we don't drop below 845 * the minimum feasible timer (which is 2 ticks). 846 * This version of the macro adapted from a paper by Lawrence 847 * Brakmo and Larry Peterson which outlines a problem caused 848 * by insufficient precision in the original implementation, 849 * which results in inappropriately large RTO values for very 850 * fast networks. 851 */ 852 #define TCP_REXMTVAL(tp) \ 853 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 854 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 855 856 /* 857 * Jaguar compatible TCP control block, for xtcpcb 858 * Does not have the old fields 859 */ 860 struct otcpcb { 861 #else 862 struct tseg_qent; 863 _TCPCB_LIST_HEAD(tsegqe_head, tseg_qent); 864 865 struct tcpcb { 866 #endif /* KERNEL_PRIVATE */ 867 #if defined(KERNEL_PRIVATE) 868 u_int32_t t_segq; 869 #else 870 struct tsegqe_head t_segq; 871 #endif /* KERNEL_PRIVATE */ 872 int t_dupacks; /* consecutive dup acks recd */ 873 u_int32_t unused; /* unused now: was t_template */ 874 875 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 876 877 _TCPCB_PTR(struct inpcb *) t_inpcb; /* back pointer to internet pcb */ 878 int t_state; /* state of this connection */ 879 u_int t_flags; 880 #define TF_ACKNOW 0x00001 /* ack peer immediately */ 881 #define TF_DELACK 0x00002 /* ack, but try to delay it */ 882 #define TF_NODELAY 0x00004 /* don't delay packets to coalesce */ 883 #define TF_NOOPT 0x00008 /* don't use tcp options */ 884 #define TF_SENTFIN 0x00010 /* have sent FIN */ 885 #define TF_REQ_SCALE 0x00020 /* have/will request window scaling */ 886 #define TF_RCVD_SCALE 0x00040 /* other side has requested scaling */ 887 #define TF_REQ_TSTMP 0x00080 /* have/will request timestamps */ 888 #define TF_RCVD_TSTMP 0x00100 /* a timestamp was received in SYN */ 889 #define TF_SACK_PERMIT 0x00200 /* other side said I could SACK */ 890 #define TF_NEEDSYN 0x00400 /* send SYN (implicit state) - unused but needed for backwards compatibility */ 891 #define TF_NEEDFIN 0x00800 /* send FIN (implicit state) */ 892 #define TF_NOPUSH 0x01000 /* don't push */ 893 #define TF_REQ_CC 0x02000 /* have/will request CC */ 894 #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ 895 #define TF_SENDCCNEW 0x08000 /* Not implemented */ 896 #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ 897 #define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */ 898 #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ 899 #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ 900 901 int t_force; /* 1 if forcing out a byte */ 902 903 tcp_seq snd_una; /* send unacknowledged */ 904 tcp_seq snd_max; /* highest sequence number sent; 905 * used to recognize retransmits 906 */ 907 tcp_seq snd_nxt; /* send next */ 908 tcp_seq snd_up; /* send urgent pointer */ 909 910 tcp_seq snd_wl1; /* window update seg seq number */ 911 tcp_seq snd_wl2; /* window update seg ack number */ 912 tcp_seq iss; /* initial send sequence number */ 913 tcp_seq irs; /* initial receive sequence number */ 914 915 tcp_seq rcv_nxt; /* receive next */ 916 tcp_seq rcv_adv; /* advertised window */ 917 u_int32_t rcv_wnd; /* receive window */ 918 tcp_seq rcv_up; /* receive urgent pointer */ 919 920 u_int32_t snd_wnd; /* send window */ 921 u_int32_t snd_cwnd; /* congestion-controlled window */ 922 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 923 * for slow start exponential to 924 * linear switch 925 */ 926 u_int t_maxopd; /* mss plus options */ 927 928 u_int32_t t_rcvtime; /* time at which a packet was received */ 929 u_int32_t t_starttime; /* time connection was established */ 930 int t_rtttime; /* round trip time */ 931 tcp_seq t_rtseq; /* sequence number being timed */ 932 933 int t_rxtcur; /* current retransmit value (ticks) */ 934 u_int t_maxseg; /* maximum segment size */ 935 int t_srtt; /* smoothed round-trip time */ 936 int t_rttvar; /* variance in round-trip time */ 937 938 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 939 u_int t_rttmin; /* minimum rtt allowed */ 940 u_int32_t t_rttupdated; /* number of times rtt sampled */ 941 u_int32_t max_sndwnd; /* largest window peer has offered */ 942 943 int t_softerror; /* possible error not yet reported */ 944 /* out-of-band data */ 945 char t_oobflags; /* have some */ 946 char t_iobc; /* input character */ 947 #define TCPOOB_HAVEDATA 0x01 948 #define TCPOOB_HADDATA 0x02 949 /* RFC 1323 variables */ 950 u_char snd_scale; /* window scaling for send window */ 951 u_char rcv_scale; /* window scaling for recv window */ 952 u_char request_r_scale; /* pending window scaling */ 953 u_char requested_s_scale; 954 u_int32_t ts_recent; /* timestamp echo data */ 955 956 u_int32_t ts_recent_age; /* when last updated */ 957 tcp_seq last_ack_sent; 958 /* RFC 1644 variables */ 959 tcp_cc cc_send; /* send connection count */ 960 tcp_cc cc_recv; /* receive connection count */ 961 tcp_seq snd_recover; /* for use in fast recovery */ 962 /* experimental */ 963 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 964 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 965 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 966 }; 967 968 #define tcps_ecn_setup tcps_ecn_client_success 969 #define tcps_sent_cwr tcps_ecn_recv_ece 970 #define tcps_sent_ece tcps_ecn_sent_ece 971 972 /* 973 * TCP statistics. 974 * Many of these should be kept per connection, 975 * but that's inconvenient at the moment. 976 */ 977 struct tcpstat { 978 u_int32_t tcps_connattempt; /* connections initiated */ 979 u_int32_t tcps_accepts; /* connections accepted */ 980 u_int32_t tcps_connects; /* connections established */ 981 u_int32_t tcps_drops; /* connections dropped */ 982 u_int32_t tcps_conndrops; /* embryonic connections dropped */ 983 u_int32_t tcps_closed; /* conn. closed (includes drops) */ 984 u_int32_t tcps_segstimed; /* segs where we tried to get rtt */ 985 u_int32_t tcps_rttupdated; /* times we succeeded */ 986 u_int32_t tcps_delack; /* delayed acks sent */ 987 u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 988 u_int32_t tcps_rexmttimeo; /* retransmit timeouts */ 989 u_int32_t tcps_persisttimeo; /* persist timeouts */ 990 u_int32_t tcps_keeptimeo; /* keepalive timeouts */ 991 u_int32_t tcps_keepprobe; /* keepalive probes sent */ 992 u_int32_t tcps_keepdrops; /* connections dropped in keepalive */ 993 994 u_int32_t tcps_sndtotal; /* total packets sent */ 995 u_int32_t tcps_sndpack; /* data packets sent */ 996 u_int32_t tcps_sndbyte; /* data bytes sent */ 997 u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */ 998 u_int32_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 999 u_int32_t tcps_sndacks; /* ack-only packets sent */ 1000 u_int32_t tcps_sndprobe; /* window probes sent */ 1001 u_int32_t tcps_sndurg; /* packets sent with URG only */ 1002 u_int32_t tcps_sndwinup; /* window update-only packets sent */ 1003 u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 1004 1005 u_int32_t tcps_rcvtotal; /* total packets received */ 1006 u_int32_t tcps_rcvpack; /* packets received in sequence */ 1007 u_int32_t tcps_rcvbyte; /* bytes received in sequence */ 1008 u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */ 1009 u_int32_t tcps_rcvbadoff; /* packets received with bad offset */ 1010 u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ 1011 u_int32_t tcps_rcvshort; /* packets received too short */ 1012 u_int32_t tcps_rcvduppack; /* duplicate-only packets received */ 1013 u_int32_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 1014 u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */ 1015 u_int32_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 1016 u_int32_t tcps_rcvoopack; /* out-of-order packets received */ 1017 u_int32_t tcps_rcvoobyte; /* out-of-order bytes received */ 1018 u_int32_t tcps_rcvpackafterwin; /* packets with data after window */ 1019 u_int32_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 1020 u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */ 1021 u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */ 1022 u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */ 1023 u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 1024 u_int32_t tcps_rcvackpack; /* rcvd ack packets */ 1025 u_int32_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 1026 u_int32_t tcps_rcvwinupd; /* rcvd window update packets */ 1027 u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */ 1028 u_int32_t tcps_predack; /* times hdr predict ok for acks */ 1029 u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */ 1030 u_int32_t tcps_pcbcachemiss; 1031 u_int32_t tcps_cachedrtt; /* times cached RTT in route updated */ 1032 u_int32_t tcps_cachedrttvar; /* times cached rttvar updated */ 1033 u_int32_t tcps_cachedssthresh; /* times cached ssthresh updated */ 1034 u_int32_t tcps_usedrtt; /* times RTT initialized from route */ 1035 u_int32_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 1036 u_int32_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 1037 u_int32_t tcps_persistdrop; /* timeout in persist state */ 1038 u_int32_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 1039 u_int32_t tcps_mturesent; /* resends due to MTU discovery */ 1040 u_int32_t tcps_listendrop; /* listen queue overflows */ 1041 u_int32_t tcps_synchallenge; /* challenge ACK due to bad SYN */ 1042 u_int32_t tcps_rstchallenge; /* challenge ACK due to bad RST */ 1043 1044 /* new stats from FreeBSD 5.4 sync up */ 1045 u_int32_t tcps_minmssdrops; /* average minmss too low drops */ 1046 1047 u_int32_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 1048 u_int32_t tcps_badrst; /* ignored RSTs in the window */ 1049 1050 u_int32_t tcps_sc_added; /* entry added to syncache */ 1051 u_int32_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 1052 u_int32_t tcps_sc_dupsyn; /* duplicate SYN packet */ 1053 u_int32_t tcps_sc_dropped; /* could not reply to packet */ 1054 u_int32_t tcps_sc_completed; /* successful extraction of entry */ 1055 u_int32_t tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ 1056 u_int32_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 1057 u_int32_t tcps_sc_reset; /* RST removed entry from syncache */ 1058 u_int32_t tcps_sc_stale; /* timed out or listen socket gone */ 1059 u_int32_t tcps_sc_aborted; /* syncache entry aborted */ 1060 u_int32_t tcps_sc_badack; /* removed due to bad ACK */ 1061 u_int32_t tcps_sc_unreach; /* ICMP unreachable received */ 1062 u_int32_t tcps_sc_zonefail; /* zalloc() failed */ 1063 u_int32_t tcps_sc_sendcookie; /* SYN cookie sent */ 1064 u_int32_t tcps_sc_recvcookie; /* SYN cookie received */ 1065 1066 u_int32_t tcps_hc_added; /* entry added to hostcache */ 1067 u_int32_t tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ 1068 1069 /* SACK related stats */ 1070 u_int32_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 1071 u_int32_t tcps_sack_rexmits; /* SACK rexmit segments */ 1072 u_int32_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 1073 u_int32_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 1074 u_int32_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 1075 u_int32_t tcps_sack_sboverflow; /* SACK sendblock overflow */ 1076 1077 u_int32_t tcps_bg_rcvtotal; /* total background packets received */ 1078 u_int32_t tcps_rxtfindrop; /* drop conn after retransmitting FIN */ 1079 u_int32_t tcps_fcholdpacket; /* packets withheld because of flow control */ 1080 1081 u_int32_t tcps_limited_txt; /* Limited transmit used */ 1082 u_int32_t tcps_early_rexmt; /* Early retransmit used */ 1083 u_int32_t tcps_sack_ackadv; /* Cumulative ack advanced along with sack */ 1084 1085 /* Checksum related stats */ 1086 u_int32_t tcps_rcv_swcsum; /* tcp swcksum (inbound), packets */ 1087 u_int32_t tcps_rcv_swcsum_bytes; /* tcp swcksum (inbound), bytes */ 1088 u_int32_t tcps_rcv6_swcsum; /* tcp6 swcksum (inbound), packets */ 1089 u_int32_t tcps_rcv6_swcsum_bytes; /* tcp6 swcksum (inbound), bytes */ 1090 u_int32_t tcps_snd_swcsum; /* tcp swcksum (outbound), packets */ 1091 u_int32_t tcps_snd_swcsum_bytes; /* tcp swcksum (outbound), bytes */ 1092 u_int32_t tcps_snd6_swcsum; /* tcp6 swcksum (outbound), packets */ 1093 u_int32_t tcps_snd6_swcsum_bytes; /* tcp6 swcksum (outbound), bytes */ 1094 u_int32_t tcps_unused_1; 1095 u_int32_t tcps_unused_2; 1096 u_int32_t tcps_unused_3; 1097 1098 /* MPTCP Related stats */ 1099 u_int32_t tcps_invalid_mpcap; /* Invalid MPTCP capable opts */ 1100 u_int32_t tcps_invalid_joins; /* Invalid MPTCP joins */ 1101 u_int32_t tcps_mpcap_fallback; /* TCP fallback in primary */ 1102 u_int32_t tcps_join_fallback; /* No MPTCP in secondary */ 1103 u_int32_t tcps_estab_fallback; /* DSS option dropped */ 1104 u_int32_t tcps_invalid_opt; /* Catchall error stat */ 1105 u_int32_t tcps_mp_outofwin; /* Packet lies outside the 1106 * shared recv window */ 1107 u_int32_t tcps_mp_reducedwin; /* Reduced subflow window */ 1108 u_int32_t tcps_mp_badcsum; /* Bad DSS csum */ 1109 u_int32_t tcps_mp_oodata; /* Out of order data */ 1110 u_int32_t tcps_mp_switches; /* number of subflow switch */ 1111 u_int32_t tcps_mp_rcvtotal; /* number of rcvd packets */ 1112 u_int32_t tcps_mp_rcvbytes; /* number of bytes received */ 1113 u_int32_t tcps_mp_sndpacks; /* number of data packs sent */ 1114 u_int32_t tcps_mp_sndbytes; /* number of bytes sent */ 1115 u_int32_t tcps_join_rxmts; /* join ack retransmits */ 1116 u_int32_t tcps_tailloss_rto; /* RTO due to tail loss */ 1117 u_int32_t tcps_reordered_pkts; /* packets reorderd */ 1118 u_int32_t tcps_recovered_pkts; /* recovered after loss */ 1119 u_int32_t tcps_pto; /* probe timeout */ 1120 u_int32_t tcps_rto_after_pto; /* RTO after a probe */ 1121 u_int32_t tcps_tlp_recovery; /* TLP induced fast recovery */ 1122 u_int32_t tcps_tlp_recoverlastpkt; /* TLP recoverd last pkt */ 1123 u_int32_t tcps_ecn_client_success; /* client-side connection negotiated ECN */ 1124 u_int32_t tcps_ecn_recv_ece; /* ECE received, sent CWR */ 1125 u_int32_t tcps_ecn_sent_ece; /* Sent ECE notification */ 1126 u_int32_t tcps_detect_reordering; /* Detect pkt reordering */ 1127 u_int32_t tcps_delay_recovery; /* Delay fast recovery */ 1128 u_int32_t tcps_avoid_rxmt; /* Retransmission was avoided */ 1129 u_int32_t tcps_unnecessary_rxmt; /* Retransmission was not needed */ 1130 u_int32_t tcps_nostretchack; /* disabled stretch ack algorithm on a connection */ 1131 u_int32_t tcps_rescue_rxmt; /* SACK rescue retransmit */ 1132 u_int32_t tcps_pto_in_recovery; /* rescue retransmit in fast recovery */ 1133 u_int32_t tcps_pmtudbh_reverted; /* PMTU Blackhole detection, segment size reverted */ 1134 1135 /* DSACK related statistics */ 1136 u_int32_t tcps_dsack_disable; /* DSACK disabled due to n/w duplication */ 1137 u_int32_t tcps_dsack_ackloss; /* ignore DSACK due to ack loss */ 1138 u_int32_t tcps_dsack_badrexmt; /* DSACK based bad rexmt recovery */ 1139 u_int32_t tcps_dsack_sent; /* Sent DSACK notification */ 1140 u_int32_t tcps_dsack_recvd; /* Received a valid DSACK option */ 1141 u_int32_t tcps_dsack_recvd_old; /* Received an out of window DSACK option */ 1142 1143 /* MPTCP Subflow selection stats */ 1144 u_int32_t tcps_mp_sel_symtomsd; /* By symptomsd */ 1145 u_int32_t tcps_mp_sel_rtt; /* By RTT comparison */ 1146 u_int32_t tcps_mp_sel_rto; /* By RTO comparison */ 1147 u_int32_t tcps_mp_sel_peer; /* By peer's output pattern */ 1148 u_int32_t tcps_mp_num_probes; /* Number of probes sent */ 1149 u_int32_t tcps_mp_verdowngrade; /* MPTCP version downgrade */ 1150 u_int32_t tcps_drop_after_sleep; /* drop after long AP sleep */ 1151 u_int32_t tcps_probe_if; /* probe packets after interface availability */ 1152 u_int32_t tcps_probe_if_conflict; /* Can't send probe packets for interface */ 1153 1154 u_int32_t tcps_ecn_client_setup; /* Attempted ECN setup from client side */ 1155 u_int32_t tcps_ecn_server_setup; /* Attempted ECN setup from server side */ 1156 u_int32_t tcps_ecn_server_success; /* server-side connection negotiated ECN */ 1157 u_int32_t tcps_ecn_lost_synack; /* Lost SYN-ACK with ECN setup */ 1158 u_int32_t tcps_ecn_lost_syn; /* Lost SYN with ECN setup */ 1159 u_int32_t tcps_ecn_not_supported; /* Server did not support ECN setup */ 1160 u_int32_t tcps_ecn_recv_ce; /* Received CE from the network */ 1161 u_int32_t tcps_ecn_conn_recv_ce; /* Number of connections received CE atleast once */ 1162 u_int32_t tcps_ecn_conn_recv_ece; /* Number of connections received ECE atleast once */ 1163 u_int32_t tcps_ecn_conn_plnoce; /* Number of connections that received no CE and sufferred packet loss */ 1164 u_int32_t tcps_ecn_conn_pl_ce; /* Number of connections that received CE and sufferred packet loss */ 1165 u_int32_t tcps_ecn_conn_nopl_ce; /* Number of connections that received CE and sufferred no packet loss */ 1166 u_int32_t tcps_ecn_fallback_synloss; /* Number of times we did fall back due to SYN-Loss */ 1167 u_int32_t tcps_ecn_fallback_reorder; /* Number of times we fallback because we detected the PAWS-issue */ 1168 u_int32_t tcps_ecn_fallback_ce; /* Number of times we fallback because we received too many CEs */ 1169 1170 /* TFO-related statistics */ 1171 u_int32_t tcps_tfo_syn_data_rcv; /* Received a SYN+data with valid cookie */ 1172 u_int32_t tcps_tfo_cookie_req_rcv;/* Received a TFO cookie-request */ 1173 u_int32_t tcps_tfo_cookie_sent; /* Offered a TFO-cookie to the client */ 1174 u_int32_t tcps_tfo_cookie_invalid;/* Received an invalid TFO-cookie */ 1175 u_int32_t tcps_tfo_cookie_req; /* Cookie requested with the SYN */ 1176 u_int32_t tcps_tfo_cookie_rcv; /* Cookie received in a SYN/ACK */ 1177 u_int32_t tcps_tfo_syn_data_sent; /* SYN+data+cookie sent */ 1178 u_int32_t tcps_tfo_syn_data_acked;/* SYN+data has been acknowledged */ 1179 u_int32_t tcps_tfo_syn_loss; /* SYN+TFO has been lost and we fallback */ 1180 u_int32_t tcps_tfo_blackhole; /* TFO got blackholed by a middlebox. */ 1181 u_int32_t tcps_tfo_cookie_wrong; /* TFO-cookie we sent was wrong */ 1182 u_int32_t tcps_tfo_no_cookie_rcv; /* We asked for a cookie but didn't get one */ 1183 u_int32_t tcps_tfo_heuristics_disable; /* TFO got disabled due to heuristics */ 1184 u_int32_t tcps_tfo_sndblackhole; /* TFO got blackholed in the sending direction */ 1185 u_int32_t tcps_mss_to_default; /* Change MSS to default using link status report */ 1186 u_int32_t tcps_mss_to_medium; /* Change MSS to medium using link status report */ 1187 u_int32_t tcps_mss_to_low; /* Change MSS to low using link status report */ 1188 u_int32_t tcps_ecn_fallback_droprst; /* ECN fallback caused by connection drop due to RST */ 1189 u_int32_t tcps_ecn_fallback_droprxmt; /* ECN fallback due to drop after multiple retransmits */ 1190 u_int32_t tcps_ecn_fallback_synrst; /* ECN fallback due to rst after syn */ 1191 1192 u_int32_t tcps_mptcp_rcvmemdrop; /* MPTCP packets dropped for lack of memory */ 1193 u_int32_t tcps_mptcp_rcvduppack; /* MPTCP duplicate-only packets received */ 1194 u_int32_t tcps_mptcp_rcvpackafterwin; /* MPTCP packets with data after window */ 1195 1196 /* TCP timer statistics */ 1197 u_int32_t tcps_timer_drift_le_1_ms; /* Timer drift less or equal to 1 ms */ 1198 u_int32_t tcps_timer_drift_le_10_ms; /* Timer drift less or equal to 10 ms */ 1199 u_int32_t tcps_timer_drift_le_20_ms; /* Timer drift less or equal to 20 ms */ 1200 u_int32_t tcps_timer_drift_le_50_ms; /* Timer drift less or equal to 50 ms */ 1201 u_int32_t tcps_timer_drift_le_100_ms; /* Timer drift less or equal to 100 ms */ 1202 u_int32_t tcps_timer_drift_le_200_ms; /* Timer drift less or equal to 200 ms */ 1203 u_int32_t tcps_timer_drift_le_500_ms; /* Timer drift less or equal to 500 ms */ 1204 u_int32_t tcps_timer_drift_le_1000_ms; /* Timer drift less or equal to 1000 ms */ 1205 u_int32_t tcps_timer_drift_gt_1000_ms; /* Timer drift greater than 1000 ms */ 1206 1207 u_int32_t tcps_mptcp_handover_attempt; /* Total number of MPTCP-attempts using handover mode */ 1208 u_int32_t tcps_mptcp_interactive_attempt; /* Total number of MPTCP-attempts using interactive mode */ 1209 u_int32_t tcps_mptcp_aggregate_attempt; /* Total number of MPTCP-attempts using aggregate mode */ 1210 u_int32_t tcps_mptcp_fp_handover_attempt; /* Same as previous three but only for first-party apps */ 1211 u_int32_t tcps_mptcp_fp_interactive_attempt; 1212 u_int32_t tcps_mptcp_fp_aggregate_attempt; 1213 u_int32_t tcps_mptcp_heuristic_fallback; /* Total number of MPTCP-connections that fell back due to heuristics */ 1214 u_int32_t tcps_mptcp_fp_heuristic_fallback; /* Same as previous but for first-party apps */ 1215 u_int32_t tcps_mptcp_handover_success_wifi; /* Total number of successfull handover-mode connections that *started* on WiFi */ 1216 u_int32_t tcps_mptcp_handover_success_cell; /* Total number of successfull handover-mode connections that *started* on Cell */ 1217 u_int32_t tcps_mptcp_interactive_success; /* Total number of interactive-mode connections that negotiated MPTCP */ 1218 u_int32_t tcps_mptcp_aggregate_success; /* Same as previous but for aggregate */ 1219 u_int32_t tcps_mptcp_fp_handover_success_wifi; /* Same as previous four, but for first-party apps */ 1220 u_int32_t tcps_mptcp_fp_handover_success_cell; 1221 u_int32_t tcps_mptcp_fp_interactive_success; 1222 u_int32_t tcps_mptcp_fp_aggregate_success; 1223 u_int32_t tcps_mptcp_handover_cell_from_wifi; /* Total number of connections that use cell in handover-mode (coming from WiFi) */ 1224 u_int32_t tcps_mptcp_handover_wifi_from_cell; /* Total number of connections that use WiFi in handover-mode (coming from cell) */ 1225 u_int32_t tcps_mptcp_interactive_cell_from_wifi; /* Total number of connections that use cell in interactive mode (coming from WiFi) */ 1226 u_int64_t tcps_mptcp_handover_cell_bytes; /* Total number of bytes sent on cell in handover-mode (on new subflows, ignoring initial one) */ 1227 u_int64_t tcps_mptcp_interactive_cell_bytes; /* Same as previous but for interactive */ 1228 u_int64_t tcps_mptcp_aggregate_cell_bytes; 1229 u_int64_t tcps_mptcp_handover_all_bytes; /* Total number of bytes sent in handover */ 1230 u_int64_t tcps_mptcp_interactive_all_bytes; 1231 u_int64_t tcps_mptcp_aggregate_all_bytes; 1232 u_int32_t tcps_mptcp_back_to_wifi; /* Total number of connections that succeed to move traffic away from cell (when starting on cell) */ 1233 u_int32_t tcps_mptcp_wifi_proxy; /* Total number of new subflows that fell back to regular TCP on cell */ 1234 u_int32_t tcps_mptcp_cell_proxy; /* Total number of new subflows that fell back to regular TCP on WiFi */ 1235 1236 /* TCP offload statistics */ 1237 u_int32_t tcps_ka_offload_drops; /* Keep alive drops for timeout reported by firmware */ 1238 1239 u_int32_t tcps_mptcp_triggered_cell; /* Total number of times an MPTCP-connection triggered cell bringup */ 1240 1241 u_int32_t tcps_fin_timeout_drops; 1242 }; 1243 1244 1245 struct tcpstat_local { 1246 u_int64_t badformat; 1247 u_int64_t unspecv6; 1248 u_int64_t synfin; 1249 u_int64_t badformatipsec; 1250 u_int64_t noconnnolist; 1251 u_int64_t noconnlist; 1252 u_int64_t listbadsyn; 1253 u_int64_t icmp6unreach; 1254 u_int64_t deprecate6; 1255 u_int64_t ooopacket; 1256 u_int64_t rstinsynrcv; 1257 u_int64_t dospacket; 1258 u_int64_t cleanup; 1259 u_int64_t synwindow; 1260 }; 1261 1262 #pragma pack(4) 1263 1264 /* 1265 * TCB structure exported to user-land via sysctl(3). 1266 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 1267 * included. Not all of our clients do. 1268 */ 1269 1270 struct xtcpcb { 1271 u_int32_t xt_len; 1272 #ifdef KERNEL_PRIVATE 1273 struct inpcb_compat xt_inp; 1274 #else 1275 struct inpcb xt_inp; 1276 #endif 1277 #ifdef KERNEL_PRIVATE 1278 struct otcpcb xt_tp; 1279 #else 1280 struct tcpcb xt_tp; 1281 #endif 1282 struct xsocket xt_socket; 1283 u_quad_t xt_alignment_hack; 1284 }; 1285 1286 #if XNU_TARGET_OS_OSX || KERNEL || !(TARGET_OS_IPHONE && !TARGET_OS_SIMULATOR) 1287 1288 struct xtcpcb64 { 1289 u_int32_t xt_len; 1290 struct xinpcb64 xt_inpcb; 1291 1292 u_int64_t t_segq; 1293 int t_dupacks; /* consecutive dup acks recd */ 1294 1295 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 1296 1297 int t_state; /* state of this connection */ 1298 u_int t_flags; 1299 1300 int t_force; /* 1 if forcing out a byte */ 1301 1302 tcp_seq snd_una; /* send unacknowledged */ 1303 tcp_seq snd_max; /* highest sequence number sent; 1304 * used to recognize retransmits 1305 */ 1306 tcp_seq snd_nxt; /* send next */ 1307 tcp_seq snd_up; /* send urgent pointer */ 1308 1309 tcp_seq snd_wl1; /* window update seg seq number */ 1310 tcp_seq snd_wl2; /* window update seg ack number */ 1311 tcp_seq iss; /* initial send sequence number */ 1312 tcp_seq irs; /* initial receive sequence number */ 1313 1314 tcp_seq rcv_nxt; /* receive next */ 1315 tcp_seq rcv_adv; /* advertised window */ 1316 u_int32_t rcv_wnd; /* receive window */ 1317 tcp_seq rcv_up; /* receive urgent pointer */ 1318 1319 u_int32_t snd_wnd; /* send window */ 1320 u_int32_t snd_cwnd; /* congestion-controlled window */ 1321 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 1322 * for slow start exponential to 1323 * linear switch 1324 */ 1325 u_int t_maxopd; /* mss plus options */ 1326 1327 u_int32_t t_rcvtime; /* time at which a packet was received */ 1328 u_int32_t t_starttime; /* time connection was established */ 1329 int t_rtttime; /* round trip time */ 1330 tcp_seq t_rtseq; /* sequence number being timed */ 1331 1332 int t_rxtcur; /* current retransmit value (ticks) */ 1333 u_int t_maxseg; /* maximum segment size */ 1334 int t_srtt; /* smoothed round-trip time */ 1335 int t_rttvar; /* variance in round-trip time */ 1336 1337 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 1338 u_int t_rttmin; /* minimum rtt allowed */ 1339 u_int32_t t_rttupdated; /* number of times rtt sampled */ 1340 u_int32_t max_sndwnd; /* largest window peer has offered */ 1341 1342 int t_softerror; /* possible error not yet reported */ 1343 /* out-of-band data */ 1344 char t_oobflags; /* have some */ 1345 char t_iobc; /* input character */ 1346 /* RFC 1323 variables */ 1347 u_char snd_scale; /* window scaling for send window */ 1348 u_char rcv_scale; /* window scaling for recv window */ 1349 u_char request_r_scale; /* pending window scaling */ 1350 u_char requested_s_scale; 1351 u_int32_t ts_recent; /* timestamp echo data */ 1352 1353 u_int32_t ts_recent_age; /* when last updated */ 1354 tcp_seq last_ack_sent; 1355 /* RFC 1644 variables */ 1356 tcp_cc cc_send; /* send connection count */ 1357 tcp_cc cc_recv; /* receive connection count */ 1358 tcp_seq snd_recover; /* for use in fast recovery */ 1359 /* experimental */ 1360 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 1361 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 1362 u_int32_t t_badrxtwin; /* window for retransmit recovery */ 1363 1364 u_quad_t xt_alignment_hack; 1365 }; 1366 1367 #endif /* XNU_TARGET_OS_OSX || KERNEL || !(TARGET_OS_IPHONE && !TARGET_OS_SIMULATOR) */ 1368 1369 #ifdef PRIVATE 1370 1371 struct xtcpcb_n { 1372 u_int32_t xt_len; 1373 u_int32_t xt_kind; /* XSO_TCPCB */ 1374 1375 u_int64_t t_segq; 1376 int t_dupacks; /* consecutive dup acks recd */ 1377 1378 int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ 1379 1380 int t_state; /* state of this connection */ 1381 u_int t_flags; 1382 1383 int t_force; /* 1 if forcing out a byte */ 1384 1385 tcp_seq snd_una; /* send unacknowledged */ 1386 tcp_seq snd_max; /* highest sequence number sent; 1387 * used to recognize retransmits 1388 */ 1389 tcp_seq snd_nxt; /* send next */ 1390 tcp_seq snd_up; /* send urgent pointer */ 1391 1392 tcp_seq snd_wl1; /* window update seg seq number */ 1393 tcp_seq snd_wl2; /* window update seg ack number */ 1394 tcp_seq iss; /* initial send sequence number */ 1395 tcp_seq irs; /* initial receive sequence number */ 1396 1397 tcp_seq rcv_nxt; /* receive next */ 1398 tcp_seq rcv_adv; /* advertised window */ 1399 u_int32_t rcv_wnd; /* receive window */ 1400 tcp_seq rcv_up; /* receive urgent pointer */ 1401 1402 u_int32_t snd_wnd; /* send window */ 1403 u_int32_t snd_cwnd; /* congestion-controlled window */ 1404 u_int32_t snd_ssthresh; /* snd_cwnd size threshold for 1405 * for slow start exponential to 1406 * linear switch 1407 */ 1408 u_int t_maxopd; /* mss plus options */ 1409 1410 u_int32_t t_rcvtime; /* time at which a packet was received */ 1411 u_int32_t t_starttime; /* time connection was established */ 1412 int t_rtttime; /* round trip time */ 1413 tcp_seq t_rtseq; /* sequence number being timed */ 1414 1415 int t_rxtcur; /* current retransmit value (ticks) */ 1416 u_int t_maxseg; /* maximum segment size */ 1417 int t_srtt; /* smoothed round-trip time */ 1418 int t_rttvar; /* variance in round-trip time */ 1419 1420 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 1421 u_int t_rttmin; /* minimum rtt allowed */ 1422 u_int32_t t_rttupdated; /* number of times rtt sampled */ 1423 u_int32_t max_sndwnd; /* largest window peer has offered */ 1424 1425 int t_softerror; /* possible error not yet reported */ 1426 /* out-of-band data */ 1427 char t_oobflags; /* have some */ 1428 char t_iobc; /* input character */ 1429 /* RFC 1323 variables */ 1430 u_char snd_scale; /* window scaling for send window */ 1431 u_char rcv_scale; /* window scaling for recv window */ 1432 u_char request_r_scale; /* pending window scaling */ 1433 u_char requested_s_scale; 1434 u_int32_t ts_recent; /* timestamp echo data */ 1435 1436 u_int32_t ts_recent_age; /* when last updated */ 1437 tcp_seq last_ack_sent; 1438 /* RFC 1644 variables */ 1439 tcp_cc cc_send; /* send connection count */ 1440 tcp_cc cc_recv; /* receive connection count */ 1441 tcp_seq snd_recover; /* for use in fast recovery */ 1442 /* experimental */ 1443 u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 1444 u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 1445 }; 1446 1447 /* 1448 * The rtt measured is in milliseconds as the timestamp granularity is 1449 * a millisecond. The smoothed round-trip time and estimated variance 1450 * are stored as fixed point numbers scaled by the values below. 1451 * For convenience, these scales are also used in smoothing the average 1452 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 1453 * With these scales, srtt has 5 bits to the right of the binary point, 1454 * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the 1455 * binary point, and is smoothed with an ALPHA of 0.75. 1456 */ 1457 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 1458 #define TCP_RTT_SHIFT 5 /* shift for srtt; 5 bits frac. */ 1459 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 4 bits */ 1460 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 4 bits */ 1461 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 1462 1463 1464 /* 1465 * TCP structure with information that gives insight into forward progress on an interface, 1466 * exported to user-land via sysctl(3). 1467 */ 1468 struct xtcpprogress_indicators { 1469 u_int32_t xp_numflows; /* Total number of flows */ 1470 u_int32_t xp_conn_probe_fails; /* Count of connection failures */ 1471 u_int32_t xp_read_probe_fails; /* Count of read probe failures */ 1472 u_int32_t xp_write_probe_fails; /* Count of write failures */ 1473 u_int32_t xp_recentflows; /* Total of "recent" flows */ 1474 u_int32_t xp_recentflows_unacked; /* Total of "recent" flows with unacknowledged data */ 1475 u_int64_t xp_recentflows_rxbytes; /* Total of "recent" flows received bytes */ 1476 u_int64_t xp_recentflows_txbytes; /* Total of "recent" flows transmitted bytes */ 1477 u_int64_t xp_recentflows_rxooo; /* Total of "recent" flows received out of order bytes */ 1478 u_int64_t xp_recentflows_rxdup; /* Total of "recent" flows received duplicate bytes */ 1479 u_int64_t xp_recentflows_retx; /* Total of "recent" flows retransmitted bytes */ 1480 u_int64_t xp_reserved1; /* Expansion */ 1481 u_int64_t xp_reserved2; /* Expansion */ 1482 u_int64_t xp_reserved3; /* Expansion */ 1483 u_int64_t xp_reserved4; /* Expansion */ 1484 }; 1485 1486 struct tcpprogressreq { 1487 u_int64_t ifindex; /* Interface index for progress indicators */ 1488 u_int64_t recentflow_maxduration; /* In mach_absolute_time, max duration for flow to be counted as "recent" */ 1489 u_int64_t filter_flags; /* Optional additional filtering, values are interface properties per ntstat.h */ 1490 u_int64_t xp_reserved2; /* Expansion */ 1491 }; 1492 1493 struct tcpprobereq { 1494 u_int64_t ifindex; /* Optional interface index for TCP keep-alive probing */ 1495 u_int64_t enable; /* Flag to enable or disable probing (1=on, 0=off)*/ 1496 u_int64_t filter_flags; /* Optional flags for filtering interfaces per ntstat.h (NSTAT_IFNET_IS_*) */ 1497 u_int32_t reserved; /* Expansion */ 1498 u_int32_t reserved2; /* Expansion */ 1499 }; 1500 1501 #endif /* PRIVATE */ 1502 1503 #pragma pack() 1504 1505 /* 1506 * Names for TCP sysctl objects 1507 */ 1508 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 1509 #define TCPCTL_DO_RFC1644 2 /* use RFC-1644 extensions */ 1510 #define TCPCTL_MSSDFLT 3 /* MSS default */ 1511 #define TCPCTL_STATS 4 /* statistics (read-only) */ 1512 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 1513 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 1514 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 1515 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 1516 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 1517 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 1518 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 1519 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 1520 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 1521 #define TCPCTL_MAXID 14 1522 1523 #ifdef BSD_KERNEL_PRIVATE 1524 #include <sys/bitstring.h> 1525 1526 #define TCP_PKTLIST_CLEAR(tp) { \ 1527 (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL; \ 1528 (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0; \ 1529 } 1530 1531 extern int tcp_TCPTV_MIN; 1532 1533 #ifdef SYSCTL_DECL 1534 SYSCTL_DECL(_net_inet_tcp); 1535 #endif /* SYSCTL_DECL */ 1536 1537 extern struct inpcbhead tcb; /* head of queue of active tcpcb's */ 1538 extern struct inpcbinfo tcbinfo; 1539 extern struct tcpstat tcpstat; /* tcp statistics */ 1540 extern int tcp_mssdflt; /* XXX */ 1541 extern int tcp_minmss; 1542 #define TCP_FASTOPEN_SERVER 0x01 1543 #define TCP_FASTOPEN_CLIENT 0x02 1544 1545 extern int tcp_tfo_halfcnt; 1546 extern int tcp_tfo_backlog; 1547 extern int tcp_fastopen; 1548 extern int ss_fltsz_local; 1549 extern int target_qdelay; 1550 extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ 1551 extern struct timeval tcp_uptime; 1552 extern lck_spin_t tcp_uptime_lock; 1553 extern int tcp_delack_enabled; 1554 extern int maxseg_unacked; 1555 extern int tcp_use_newreno; 1556 extern int tcp_use_ledbat; 1557 extern struct zone *tcp_reass_zone; 1558 extern struct zone *tcp_rxt_seg_zone; 1559 extern int tcp_ecn_outbound; 1560 extern int tcp_ecn_inbound; 1561 extern uint32_t tcp_do_autorcvbuf; 1562 extern uint32_t tcp_autorcvbuf_max; 1563 extern int tcp_recv_bg; 1564 extern int tcp_do_ack_compression; 1565 extern int tcp_randomize_timestamps; 1566 extern int tcp_rledbat; 1567 extern int tcp_use_min_curr_rtt; 1568 /* 1569 * Dummy value used for when there is no flow and we want to ensure that compression 1570 * can happen. 1571 */ 1572 #define TCP_ACK_COMPRESSION_DUMMY 1 1573 1574 extern int tcp_do_better_lr; 1575 extern int tcp_cubic_minor_fixes; 1576 extern int tcp_cubic_rfc_compliant; 1577 extern int tcp_flow_control_response; 1578 1579 extern int tcp_reass_total_qlen; 1580 1581 struct protosw; 1582 struct domain; 1583 1584 struct tcp_respond_args { 1585 unsigned int ifscope; 1586 unsigned int nocell:1, 1587 noexpensive:1, 1588 awdl_unrestricted:1, 1589 intcoproc_allowed:1, 1590 keep_alive:1, 1591 noconstrained:1; 1592 }; 1593 1594 void tcp_canceltimers(struct tcpcb *); 1595 struct tcpcb * 1596 tcp_close(struct tcpcb *); 1597 void tcp_ctlinput(int, struct sockaddr *, void *, struct ifnet *); 1598 int tcp_ctloutput(struct socket *, struct sockopt *); 1599 struct tcpcb * 1600 tcp_drop(struct tcpcb *, int); 1601 void tcp_drain(void); 1602 void tcp_getrt_rtt(struct tcpcb *tp, struct rtentry *rt); 1603 void tcp_init(struct protosw *, struct domain *); 1604 void tcp_input(struct mbuf *, int); 1605 void tcp_mss(struct tcpcb *, int, unsigned int); 1606 uint32_t tcp_ceil(double a); 1607 uint32_t tcp_round_to(uint32_t val, uint32_t round); 1608 uint32_t tcp_round_up(uint32_t val, uint32_t base); 1609 int tcp_mssopt(struct tcpcb *); 1610 void tcp_drop_syn_sent(struct inpcb *, int); 1611 void tcp_mtudisc(struct inpcb *, int); 1612 struct tcpcb * 1613 tcp_newtcpcb(struct inpcb *); 1614 int tcp_output(struct tcpcb *); 1615 void tcp_respond(struct tcpcb *, void *, struct tcphdr *, struct mbuf *, 1616 tcp_seq, tcp_seq, uint8_t, struct tcp_respond_args *); 1617 struct rtentry * 1618 tcp_rtlookup(struct inpcb *, unsigned int); 1619 void tcp_setpersist(struct tcpcb *); 1620 void tcp_gc(struct inpcbinfo *); 1621 void tcp_itimer(struct inpcbinfo *ipi); 1622 void tcp_check_timer_state(struct tcpcb *tp); 1623 void tcp_run_timerlist(void *arg1, void *arg2); 1624 void tcp_sched_timers(struct tcpcb *tp); 1625 1626 struct tcptemp *tcp_maketemplate(struct tcpcb *); 1627 void tcp_fillheaders(struct mbuf *, struct tcpcb *, void *, void *); 1628 struct tcpcb *tcp_timers(struct tcpcb *, int); 1629 void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int); 1630 1631 void tcp_fill_info(struct tcpcb *, struct tcp_info *); 1632 void tcp_sack_doack(struct tcpcb *, struct tcpopt *, struct tcphdr *, 1633 u_int32_t *, uint32_t *); 1634 extern boolean_t tcp_sack_process_dsack(struct tcpcb *, struct tcpopt *, 1635 struct tcphdr *); 1636 int tcp_detect_bad_rexmt(struct tcpcb *, struct tcphdr *, struct tcpopt *, 1637 u_int32_t rxtime); 1638 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 1639 void tcp_clean_sackreport(struct tcpcb *tp); 1640 void tcp_sack_adjust(struct tcpcb *tp); 1641 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 1642 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 1643 void tcp_free_sackholes(struct tcpcb *tp); 1644 void tcp_sack_lost_rexmit(struct tcpcb *tp); 1645 int32_t tcp_sbspace(struct tcpcb *tp); 1646 void tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp); 1647 void tcp_set_ecn(struct tcpcb *tp, struct ifnet *ifp); 1648 int tcp_flight_size(struct tcpcb *tp); 1649 void tcp_reset_stretch_ack(struct tcpcb *tp); 1650 extern void tcp_get_ports_used(ifnet_t ifp, int, u_int32_t, bitstr_t *); 1651 uint32_t tcp_count_opportunistic(unsigned int ifindex, u_int32_t flags); 1652 uint32_t tcp_find_anypcb_byaddr(struct ifaddr *ifa); 1653 void tcp_set_max_rwinscale(struct tcpcb *tp, struct socket *so); 1654 struct bwmeas* tcp_bwmeas_alloc(struct tcpcb *tp); 1655 void tcp_bwmeas_free(struct tcpcb *tp); 1656 extern int32_t timer_diff(uint32_t t1, uint32_t toff1, uint32_t t2, uint32_t toff2); 1657 1658 extern void tcp_set_background_cc(struct socket *); 1659 extern void tcp_set_foreground_cc(struct socket *); 1660 extern void tcp_set_recv_bg(struct socket *); 1661 extern void tcp_clear_recv_bg(struct socket *); 1662 extern boolean_t tcp_sack_byte_islost(struct tcpcb *tp); 1663 #define IS_TCP_RECV_BG(_so) \ 1664 ((_so)->so_flags1 & SOF1_TRAFFIC_MGT_TCP_RECVBG) 1665 1666 #if TRAFFIC_MGT 1667 #define CLEAR_IAJ_STATE(_tp_) (_tp_)->iaj_rcv_ts = 0 1668 void reset_acc_iaj(struct tcpcb *tp); 1669 #endif /* TRAFFIC_MGT */ 1670 1671 int tcp_lock(struct socket *, int, void *); 1672 int tcp_unlock(struct socket *, int, void *); 1673 void calculate_tcp_clock(void); 1674 1675 extern void tcp_keepalive_reset(struct tcpcb *); 1676 extern uint32_t get_base_rtt(struct tcpcb *tp); 1677 1678 #ifdef _KERN_LOCKS_H_ 1679 lck_mtx_t * tcp_getlock(struct socket *, int); 1680 #else 1681 void * tcp_getlock(struct socket *, int); 1682 #endif 1683 1684 extern struct pr_usrreqs tcp_usrreqs; 1685 extern u_int32_t tcp_sendspace; 1686 extern u_int32_t tcp_recvspace; 1687 tcp_seq tcp_new_isn(struct tcpcb *); 1688 1689 extern int tcp_input_checksum(int, struct mbuf *, struct tcphdr *, int, int); 1690 extern void tcp_getconninfo(struct socket *, struct conninfo_tcp *); 1691 extern void add_to_time_wait(struct tcpcb *, uint32_t delay); 1692 extern void tcp_pmtud_revert_segment_size(struct tcpcb *tp); 1693 extern void tcp_rxtseg_insert(struct tcpcb *, tcp_seq, tcp_seq); 1694 extern struct tcp_rxt_seg *tcp_rxtseg_find(struct tcpcb *, tcp_seq, tcp_seq); 1695 extern void tcp_rxtseg_set_spurious(struct tcpcb *tp, tcp_seq start, tcp_seq end); 1696 extern void tcp_rxtseg_clean(struct tcpcb *); 1697 extern boolean_t tcp_rxtseg_detect_bad_rexmt(struct tcpcb *, tcp_seq); 1698 extern boolean_t tcp_rxtseg_dsack_for_tlp(struct tcpcb *); 1699 extern u_int32_t tcp_rxtseg_total_size(struct tcpcb *tp); 1700 extern void tcp_rexmt_save_state(struct tcpcb *tp); 1701 extern void tcp_interface_send_probe(u_int16_t if_index_available); 1702 extern void tcp_probe_connectivity(struct ifnet *ifp, u_int32_t enable); 1703 extern void tcp_get_connectivity_status(struct tcpcb *, 1704 struct tcp_conn_status *); 1705 1706 extern void tcp_clear_keep_alive_offload(struct socket *so); 1707 extern void tcp_fill_keepalive_offload_frames(struct ifnet *, 1708 struct ifnet_keepalive_offload_frame *, u_int32_t, size_t, u_int32_t *); 1709 extern int tcp_notify_kao_timeout(ifnet_t ifp, 1710 struct ifnet_keepalive_offload_frame *frame); 1711 1712 extern boolean_t tfo_enabled(const struct tcpcb *tp); 1713 extern void tcp_disable_tfo(struct tcpcb *tp); 1714 extern void tcp_tfo_gen_cookie(struct inpcb *inp, u_char *out, size_t blk_size); 1715 #define TCP_FASTOPEN_KEYLEN 16 1716 extern int tcp_freeq(struct tcpcb *tp); 1717 extern errno_t tcp_notify_ack_id_valid(struct tcpcb *, struct socket *, u_int32_t); 1718 extern errno_t tcp_add_notify_ack_marker(struct tcpcb *, u_int32_t); 1719 extern void tcp_notify_ack_free(struct tcpcb *); 1720 extern void tcp_notify_acknowledgement(struct tcpcb *, struct socket *); 1721 extern void tcp_get_notify_ack_count(struct tcpcb *, 1722 struct tcp_notify_ack_complete *); 1723 extern void tcp_get_notify_ack_ids(struct tcpcb *tp, 1724 struct tcp_notify_ack_complete *); 1725 extern void tcp_update_mss_locked(struct socket *, struct ifnet *); 1726 1727 extern int get_tcp_inp_list(struct inpcb **, int, inp_gen_t); 1728 extern bool tcp_notify_ack_active(struct socket *so); 1729 extern void tcp_set_finwait_timeout(struct tcpcb *); 1730 1731 #if MPTCP 1732 extern int mptcp_input_preproc(struct tcpcb *tp, struct mbuf *m, 1733 struct tcphdr *th, int drop_hdrlen); 1734 extern uint16_t mptcp_output_csum(struct mbuf *m, uint64_t dss_val, 1735 uint32_t sseq, uint16_t dlen); 1736 extern int mptcp_adj_mss(struct tcpcb *, boolean_t); 1737 extern void mptcp_insert_rmap(struct tcpcb *tp, struct mbuf *m, struct tcphdr *th); 1738 extern int dump_mptcp_reass_qlen(char *, int); 1739 #endif 1740 1741 extern int dump_tcp_reass_qlen(char *, int); 1742 extern uint32_t tcp_reass_qlen_space(struct socket *); 1743 1744 __private_extern__ void tcp_update_stats_per_flow( 1745 struct ifnet_stats_per_flow *, struct ifnet *); 1746 1747 #define TCP_ACK_STRATEGY_LEGACY 0 1748 #define TCP_ACK_STRATEGY_MODERN 1 1749 1750 extern int tcp_ack_strategy; 1751 1752 #if defined(SKYWALK) 1753 void tcp_add_fsw_flow(struct tcpcb *, struct ifnet *); 1754 void tcp_del_fsw_flow(struct tcpcb *); 1755 #else 1756 #define tcp_add_fsw_flow(...) 1757 #define tcp_del_fsw_flow(...) 1758 #endif 1759 1760 #endif /* BSD_KERNEL_PRIVATE */ 1761 1762 #endif /* _NETINET_TCP_VAR_H_ */ 1763