1 /*- 2 * SPDX-License-Identifier: BSD-3-Clause 3 * 4 * Copyright (c) 1982, 1986, 1993, 1994, 1995 5 * The Regents of the University of California. All rights reserved. 6 * 7 * Redistribution and use in source and binary forms, with or without 8 * modification, are permitted provided that the following conditions 9 * are met: 10 * 1. Redistributions of source code must retain the above copyright 11 * notice, this list of conditions and the following disclaimer. 12 * 2. Redistributions in binary form must reproduce the above copyright 13 * notice, this list of conditions and the following disclaimer in the 14 * documentation and/or other materials provided with the distribution. 15 * 3. Neither the name of the University nor the names of its contributors 16 * may be used to endorse or promote products derived from this software 17 * without specific prior written permission. 18 * 19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 22 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 29 * SUCH DAMAGE. 30 * 31 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 32 * $FreeBSD$ 33 */ 34 35 #ifndef _NETINET_TCP_VAR_H_ 36 #define _NETINET_TCP_VAR_H_ 37 38 #include <netinet/tcp.h> 39 #include <netinet/tcp_fsm.h> 40 41 #ifdef _KERNEL 42 #include <net/vnet.h> 43 #include <sys/mbuf.h> 44 #endif 45 46 #if defined(_KERNEL) || defined(_WANT_TCPCB) 47 /* TCP segment queue entry */ 48 struct tseg_qent { 49 TAILQ_ENTRY(tseg_qent) tqe_q; 50 struct mbuf *tqe_m; /* mbuf contains packet */ 51 struct mbuf *tqe_last; /* last mbuf in chain */ 52 tcp_seq tqe_start; /* TCP Sequence number start */ 53 int tqe_len; /* TCP segment data length */ 54 uint32_t tqe_flags; /* The flags from the th->th_flags */ 55 uint32_t tqe_mbuf_cnt; /* Count of mbuf overhead */ 56 }; 57 TAILQ_HEAD(tsegqe_head, tseg_qent); 58 59 struct sackblk { 60 tcp_seq start; /* start seq no. of sack block */ 61 tcp_seq end; /* end seq no. */ 62 }; 63 64 struct sackhole { 65 tcp_seq start; /* start seq no. of hole */ 66 tcp_seq end; /* end seq no. */ 67 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 68 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 69 }; 70 71 struct sackhint { 72 struct sackhole *nexthole; 73 int sack_bytes_rexmit; 74 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 75 76 int ispare; /* explicit pad for 64bit alignment */ 77 int sacked_bytes; /* 78 * Total sacked bytes reported by the 79 * receiver via sack option 80 */ 81 uint32_t _pad1[1]; /* TBD */ 82 uint64_t _pad[1]; /* TBD */ 83 }; 84 85 #define SEGQ_EMPTY(tp) TAILQ_EMPTY(&(tp)->t_segq) 86 87 STAILQ_HEAD(tcp_log_stailq, tcp_log_mem); 88 89 /* 90 * Tcp control block, one per tcp; fields: 91 * Organized for 64 byte cacheline efficiency based 92 * on common tcp_input/tcp_output processing. 93 */ 94 struct tcpcb { 95 /* Cache line 1 */ 96 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 97 struct tcp_function_block *t_fb;/* TCP function call block */ 98 void *t_fb_ptr; /* Pointer to t_fb specific data */ 99 uint32_t t_maxseg:24, /* maximum segment size */ 100 t_logstate:8; /* State of "black box" logging */ 101 uint32_t t_port:16, /* Tunneling (over udp) port */ 102 t_state:4, /* state of this connection */ 103 t_idle_reduce : 1, 104 t_delayed_ack: 7, /* Delayed ack variable */ 105 t_fin_is_rst: 1, /* Are fin's treated as resets */ 106 t_log_state_set: 1, 107 bits_spare : 2; 108 u_int t_flags; 109 tcp_seq snd_una; /* sent but unacknowledged */ 110 tcp_seq snd_max; /* highest sequence number sent; 111 * used to recognize retransmits 112 */ 113 tcp_seq snd_nxt; /* send next */ 114 tcp_seq snd_up; /* send urgent pointer */ 115 uint32_t snd_wnd; /* send window */ 116 uint32_t snd_cwnd; /* congestion-controlled window */ 117 uint32_t t_peakrate_thr; /* pre-calculated peak rate threshold */ 118 /* Cache line 2 */ 119 u_int32_t ts_offset; /* our timestamp offset */ 120 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 121 int rcv_numsacks; /* # distinct sack blks present */ 122 u_int t_tsomax; /* TSO total burst length limit in bytes */ 123 u_int t_tsomaxsegcount; /* TSO maximum segment count */ 124 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 125 tcp_seq rcv_nxt; /* receive next */ 126 tcp_seq rcv_adv; /* advertised window */ 127 uint32_t rcv_wnd; /* receive window */ 128 u_int t_flags2; /* More tcpcb flags storage */ 129 int t_srtt; /* smoothed round-trip time */ 130 int t_rttvar; /* variance in round-trip time */ 131 u_int32_t ts_recent; /* timestamp echo data */ 132 u_char snd_scale; /* window scaling for send window */ 133 u_char rcv_scale; /* window scaling for recv window */ 134 u_char snd_limited; /* segments limited transmitted */ 135 u_char request_r_scale; /* pending window scaling */ 136 tcp_seq last_ack_sent; 137 u_int t_rcvtime; /* inactivity time */ 138 /* Cache line 3 */ 139 tcp_seq rcv_up; /* receive urgent pointer */ 140 int t_segqlen; /* segment reassembly queue length */ 141 uint32_t t_segqmbuflen; /* Count of bytes mbufs on all entries */ 142 struct tsegqe_head t_segq; /* segment reassembly queue */ 143 struct mbuf *t_in_pkt; 144 struct mbuf *t_tail_pkt; 145 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 146 struct vnet *t_vnet; /* back pointer to parent vnet */ 147 uint32_t snd_ssthresh; /* snd_cwnd size threshold for 148 * for slow start exponential to 149 * linear switch 150 */ 151 tcp_seq snd_wl1; /* window update seg seq number */ 152 /* Cache line 4 */ 153 tcp_seq snd_wl2; /* window update seg ack number */ 154 155 tcp_seq irs; /* initial receive sequence number */ 156 tcp_seq iss; /* initial send sequence number */ 157 u_int t_acktime; 158 u_int ts_recent_age; /* when last updated */ 159 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 160 uint16_t cl4_spare; /* Spare to adjust CL 4 */ 161 char t_oobflags; /* have some */ 162 char t_iobc; /* input character */ 163 int t_rxtcur; /* current retransmit value (ticks) */ 164 165 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 166 u_int t_rtttime; /* RTT measurement start time */ 167 168 tcp_seq t_rtseq; /* sequence number being timed */ 169 u_int t_starttime; /* time connection was established */ 170 171 u_int t_pmtud_saved_maxseg; /* pre-blackhole MSS */ 172 u_int t_rttmin; /* minimum rtt allowed */ 173 174 u_int t_rttbest; /* best rtt we've seen */ 175 176 int t_softerror; /* possible error not yet reported */ 177 uint32_t max_sndwnd; /* largest window peer has offered */ 178 /* Cache line 5 */ 179 uint32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 180 uint32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 181 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 182 int t_sndzerowin; /* zero-window updates sent */ 183 u_long t_rttupdated; /* number of times rtt sampled */ 184 int snd_numholes; /* number of holes seen by sender */ 185 u_int t_badrxtwin; /* window for retransmit recovery */ 186 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 187 /* SACK scoreboard (sorted) */ 188 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 189 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 190 struct sackhint sackhint; /* SACK scoreboard hint */ 191 int t_rttlow; /* smallest observerved RTT */ 192 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 193 struct toedev *tod; /* toedev handling this connection */ 194 int t_sndrexmitpack; /* retransmit packets sent */ 195 int t_rcvoopack; /* out-of-order packets received */ 196 void *t_toe; /* TOE pcb pointer */ 197 struct cc_algo *cc_algo; /* congestion control algorithm */ 198 struct cc_var *ccv; /* congestion control specific vars */ 199 struct osd *osd; /* storage for Khelp module data */ 200 int t_bytes_acked; /* # bytes acked during current RTT */ 201 u_int t_maxunacktime; 202 u_int t_keepinit; /* time to establish connection */ 203 u_int t_keepidle; /* time before keepalive probes begin */ 204 u_int t_keepintvl; /* interval between keepalives */ 205 u_int t_keepcnt; /* number of keepalives before close */ 206 int t_dupacks; /* consecutive dup acks recd */ 207 int t_lognum; /* Number of log entries */ 208 int t_loglimit; /* Maximum number of log entries */ 209 struct tcp_log_stailq t_logs; /* Log buffer */ 210 struct tcp_log_id_node *t_lin; 211 struct tcp_log_id_bucket *t_lib; 212 const char *t_output_caller; /* Function that called tcp_output */ 213 struct statsblob *t_stats; /* Per-connection stats */ 214 uint32_t t_logsn; /* Log "serial number" */ 215 uint32_t gput_ts; /* Time goodput measurement started */ 216 tcp_seq gput_seq; /* Outbound measurement seq */ 217 tcp_seq gput_ack; /* Inbound measurement ack */ 218 int32_t t_stats_gput_prev; /* XXXLAS: Prev gput measurement */ 219 uint8_t t_tfo_client_cookie_len; /* TCP Fast Open client cookie length */ 220 unsigned int *t_tfo_pending; /* TCP Fast Open server pending counter */ 221 union { 222 uint8_t client[TCP_FASTOPEN_MAX_COOKIE_LEN]; 223 uint64_t server; 224 } t_tfo_cookie; /* TCP Fast Open cookie to send */ 225 #ifdef TCPPCAP 226 struct mbufq t_inpkts; /* List of saved input packets. */ 227 struct mbufq t_outpkts; /* List of saved output packets. */ 228 #endif 229 }; 230 #endif /* _KERNEL || _WANT_TCPCB */ 231 232 #ifdef _KERNEL 233 struct tcptemp { 234 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 235 struct tcphdr tt_t; 236 }; 237 238 /* Minimum map entries limit value, if set */ 239 #define TCP_MIN_MAP_ENTRIES_LIMIT 128 240 241 /* 242 * TODO: We yet need to brave plowing in 243 * to tcp_input() and the pru_usrreq() block. 244 * Right now these go to the old standards which 245 * are somewhat ok, but in the long term may 246 * need to be changed. If we do tackle tcp_input() 247 * then we need to get rid of the tcp_do_segment() 248 * function below. 249 */ 250 /* Flags for tcp functions */ 251 #define TCP_FUNC_BEING_REMOVED 0x01 /* Can no longer be referenced */ 252 253 /* 254 * If defining the optional tcp_timers, in the 255 * tfb_tcp_timer_stop call you must use the 256 * callout_async_drain() function with the 257 * tcp_timer_discard callback. You should check 258 * the return of callout_async_drain() and if 0 259 * increment tt_draincnt. Since the timer sub-system 260 * does not know your callbacks you must provide a 261 * stop_all function that loops through and calls 262 * tcp_timer_stop() with each of your defined timers. 263 * Adding a tfb_tcp_handoff_ok function allows the socket 264 * option to change stacks to query you even if the 265 * connection is in a later stage. You return 0 to 266 * say you can take over and run your stack, you return 267 * non-zero (an error number) to say no you can't. 268 * If the function is undefined you can only change 269 * in the early states (before connect or listen). 270 * tfb_tcp_fb_fini is changed to add a flag to tell 271 * the old stack if the tcb is being destroyed or 272 * not. A one in the flag means the TCB is being 273 * destroyed, a zero indicates its transitioning to 274 * another stack (via socket option). 275 */ 276 struct tcp_function_block { 277 char tfb_tcp_block_name[TCP_FUNCTION_NAME_LEN_MAX]; 278 int (*tfb_tcp_output)(struct tcpcb *); 279 int (*tfb_tcp_output_wtime)(struct tcpcb *, const struct timeval *); 280 void (*tfb_tcp_do_segment)(struct mbuf *, struct tcphdr *, 281 struct socket *, struct tcpcb *, 282 int, int, uint8_t); 283 int (*tfb_do_queued_segments)(struct socket *, struct tcpcb *, int); 284 int (*tfb_do_segment_nounlock)(struct mbuf *, struct tcphdr *, 285 struct socket *, struct tcpcb *, 286 int, int, uint8_t, 287 int, struct timeval *); 288 void (*tfb_tcp_hpts_do_segment)(struct mbuf *, struct tcphdr *, 289 struct socket *, struct tcpcb *, 290 int, int, uint8_t, 291 int, struct timeval *); 292 int (*tfb_tcp_ctloutput)(struct socket *so, struct sockopt *sopt, 293 struct inpcb *inp, struct tcpcb *tp); 294 /* Optional memory allocation/free routine */ 295 int (*tfb_tcp_fb_init)(struct tcpcb *); 296 void (*tfb_tcp_fb_fini)(struct tcpcb *, int); 297 /* Optional timers, must define all if you define one */ 298 int (*tfb_tcp_timer_stop_all)(struct tcpcb *); 299 void (*tfb_tcp_timer_activate)(struct tcpcb *, 300 uint32_t, u_int); 301 int (*tfb_tcp_timer_active)(struct tcpcb *, uint32_t); 302 void (*tfb_tcp_timer_stop)(struct tcpcb *, uint32_t); 303 void (*tfb_tcp_rexmit_tmr)(struct tcpcb *); 304 int (*tfb_tcp_handoff_ok)(struct tcpcb *); 305 void (*tfb_tcp_mtu_chg)(struct tcpcb *); 306 volatile uint32_t tfb_refcnt; 307 uint32_t tfb_flags; 308 uint8_t tfb_id; 309 }; 310 311 struct tcp_function { 312 TAILQ_ENTRY(tcp_function) tf_next; 313 char tf_name[TCP_FUNCTION_NAME_LEN_MAX]; 314 struct tcp_function_block *tf_fb; 315 }; 316 317 TAILQ_HEAD(tcp_funchead, tcp_function); 318 #endif /* _KERNEL */ 319 320 /* 321 * Flags and utility macros for the t_flags field. 322 */ 323 #define TF_ACKNOW 0x00000001 /* ack peer immediately */ 324 #define TF_DELACK 0x00000002 /* ack, but try to delay it */ 325 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */ 326 #define TF_NOOPT 0x00000008 /* don't use tcp options */ 327 #define TF_SENTFIN 0x00000010 /* have sent FIN */ 328 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */ 329 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */ 330 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */ 331 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */ 332 #define TF_SACK_PERMIT 0x00000200 /* other side said I could SACK */ 333 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */ 334 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */ 335 #define TF_NOPUSH 0x00001000 /* don't push */ 336 #define TF_PREVVALID 0x00002000 /* saved values for bad rxmit valid */ 337 #define TF_UNUSED1 0x00004000 /* unused */ 338 #define TF_GPUTINPROG 0x00008000 /* Goodput measurement in progress */ 339 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */ 340 #define TF_LQ_OVERFLOW 0x00020000 /* listen queue overflow */ 341 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */ 342 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */ 343 #define TF_FASTRECOVERY 0x00100000 /* in NewReno Fast Recovery */ 344 #define TF_WASFRECOVERY 0x00200000 /* was in NewReno Fast Recovery */ 345 #define TF_SIGNATURE 0x00400000 /* require MD5 digests (RFC2385) */ 346 #define TF_FORCEDATA 0x00800000 /* force out a byte */ 347 #define TF_TSO 0x01000000 /* TSO enabled on this connection */ 348 #define TF_TOE 0x02000000 /* this connection is offloaded */ 349 #define TF_UNUSED3 0x04000000 /* unused */ 350 #define TF_UNUSED4 0x08000000 /* unused */ 351 #define TF_UNUSED5 0x10000000 /* unused */ 352 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 353 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 354 #define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */ 355 356 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 357 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 358 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 359 360 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 361 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 362 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 363 364 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 365 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 366 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 367 368 #if defined(_KERNEL) && !defined(TCP_RFC7413) 369 #define IS_FASTOPEN(t_flags) (false) 370 #else 371 #define IS_FASTOPEN(t_flags) (t_flags & TF_FASTOPEN) 372 #endif 373 374 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 375 376 /* 377 * Flags for the t_oobflags field. 378 */ 379 #define TCPOOB_HAVEDATA 0x01 380 #define TCPOOB_HADDATA 0x02 381 382 /* 383 * Flags for the extended TCP flags field, t_flags2 384 */ 385 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 386 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 387 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 388 #define TF2_LOG_AUTO 0x00000008 /* Session is auto-logging. */ 389 #define TF2_DROP_AF_DATA 0x00000010 /* Drop after all data ack'd */ 390 #define TF2_ECN_PERMIT 0x00000020 /* connection ECN-ready */ 391 #define TF2_ECN_SND_CWR 0x00000040 /* ECN CWR in queue */ 392 #define TF2_ECN_SND_ECE 0x00000080 /* ECN ECE in queue */ 393 #define TF2_ACE_PERMIT 0x00000100 /* Accurate ECN mode */ 394 395 /* 396 * Structure to hold TCP options that are only used during segment 397 * processing (in tcp_input), but not held in the tcpcb. 398 * It's basically used to reduce the number of parameters 399 * to tcp_dooptions and tcp_addoptions. 400 * The binary order of the to_flags is relevant for packing of the 401 * options in tcp_addoptions. 402 */ 403 struct tcpopt { 404 u_int32_t to_flags; /* which options are present */ 405 #define TOF_MSS 0x0001 /* maximum segment size */ 406 #define TOF_SCALE 0x0002 /* window scaling */ 407 #define TOF_SACKPERM 0x0004 /* SACK permitted */ 408 #define TOF_TS 0x0010 /* timestamp */ 409 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 410 #define TOF_SACK 0x0080 /* Peer sent SACK option */ 411 #define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */ 412 #define TOF_MAXOPT 0x0200 413 u_int32_t to_tsval; /* new timestamp */ 414 u_int32_t to_tsecr; /* reflected timestamp */ 415 u_char *to_sacks; /* pointer to the first SACK blocks */ 416 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 417 u_int8_t *to_tfo_cookie; /* pointer to the TFO cookie */ 418 u_int16_t to_mss; /* maximum segment size */ 419 u_int8_t to_wscale; /* window scaling */ 420 u_int8_t to_nsacks; /* number of SACK blocks */ 421 u_int8_t to_tfo_len; /* TFO cookie length */ 422 u_int32_t to_spare; /* UTO */ 423 }; 424 425 /* 426 * Flags for tcp_dooptions. 427 */ 428 #define TO_SYN 0x01 /* parse SYN-only options */ 429 430 struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 431 uint32_t rmx_mtu; /* MTU for this path */ 432 uint32_t rmx_ssthresh; /* outbound gateway buffer limit */ 433 uint32_t rmx_rtt; /* estimated round trip time */ 434 uint32_t rmx_rttvar; /* estimated rtt variance */ 435 uint32_t rmx_cwnd; /* congestion window */ 436 uint32_t rmx_sendpipe; /* outbound delay-bandwidth product */ 437 uint32_t rmx_recvpipe; /* inbound delay-bandwidth product */ 438 }; 439 440 /* 441 * Used by tcp_maxmtu() to communicate interface specific features 442 * and limits at the time of connection setup. 443 */ 444 struct tcp_ifcap { 445 int ifcap; 446 u_int tsomax; 447 u_int tsomaxsegcount; 448 u_int tsomaxsegsize; 449 }; 450 451 #ifndef _NETINET_IN_PCB_H_ 452 struct in_conninfo; 453 #endif /* _NETINET_IN_PCB_H_ */ 454 455 struct tcptw { 456 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 457 tcp_seq snd_nxt; 458 tcp_seq rcv_nxt; 459 tcp_seq iss; 460 tcp_seq irs; 461 u_short last_win; /* cached window value */ 462 short tw_so_options; /* copy of so_options */ 463 struct ucred *tw_cred; /* user credentials */ 464 u_int32_t t_recent; 465 u_int32_t ts_offset; /* our timestamp offset */ 466 u_int t_starttime; 467 int tw_time; 468 TAILQ_ENTRY(tcptw) tw_2msl; 469 void *tw_pspare; /* TCP_SIGNATURE */ 470 u_int *tw_spare; /* TCP_SIGNATURE */ 471 }; 472 473 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 474 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 475 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 476 477 /* 478 * The smoothed round-trip time and estimated variance 479 * are stored as fixed point numbers scaled by the values below. 480 * For convenience, these scales are also used in smoothing the average 481 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 482 * With these scales, srtt has 3 bits to the right of the binary point, 483 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 484 * binary point, and is smoothed with an ALPHA of 0.75. 485 */ 486 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 487 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 488 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 489 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 490 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 491 492 /* 493 * The initial retransmission should happen at rtt + 4 * rttvar. 494 * Because of the way we do the smoothing, srtt and rttvar 495 * will each average +1/2 tick of bias. When we compute 496 * the retransmit timer, we want 1/2 tick of rounding and 497 * 1 extra tick because of +-1/2 tick uncertainty in the 498 * firing of the timer. The bias will give us exactly the 499 * 1.5 tick we need. But, because the bias is 500 * statistical, we have to test that we don't drop below 501 * the minimum feasible timer (which is 2 ticks). 502 * This version of the macro adapted from a paper by Lawrence 503 * Brakmo and Larry Peterson which outlines a problem caused 504 * by insufficient precision in the original implementation, 505 * which results in inappropriately large RTO values for very 506 * fast networks. 507 */ 508 #define TCP_REXMTVAL(tp) \ 509 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 510 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 511 512 /* 513 * TCP statistics. 514 * Many of these should be kept per connection, 515 * but that's inconvenient at the moment. 516 */ 517 struct tcpstat { 518 uint64_t tcps_connattempt; /* connections initiated */ 519 uint64_t tcps_accepts; /* connections accepted */ 520 uint64_t tcps_connects; /* connections established */ 521 uint64_t tcps_drops; /* connections dropped */ 522 uint64_t tcps_conndrops; /* embryonic connections dropped */ 523 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 524 uint64_t tcps_closed; /* conn. closed (includes drops) */ 525 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 526 uint64_t tcps_rttupdated; /* times we succeeded */ 527 uint64_t tcps_delack; /* delayed acks sent */ 528 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 529 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 530 uint64_t tcps_persisttimeo; /* persist timeouts */ 531 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 532 uint64_t tcps_keepprobe; /* keepalive probes sent */ 533 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 534 535 uint64_t tcps_sndtotal; /* total packets sent */ 536 uint64_t tcps_sndpack; /* data packets sent */ 537 uint64_t tcps_sndbyte; /* data bytes sent */ 538 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 539 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 540 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 541 uint64_t tcps_sndacks; /* ack-only packets sent */ 542 uint64_t tcps_sndprobe; /* window probes sent */ 543 uint64_t tcps_sndurg; /* packets sent with URG only */ 544 uint64_t tcps_sndwinup; /* window update-only packets sent */ 545 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 546 547 uint64_t tcps_rcvtotal; /* total packets received */ 548 uint64_t tcps_rcvpack; /* packets received in sequence */ 549 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 550 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 551 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 552 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ 553 uint64_t tcps_rcvshort; /* packets received too short */ 554 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 555 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 556 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 557 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 558 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 559 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 560 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 561 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 562 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 563 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 564 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 565 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 566 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 567 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 568 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 569 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 570 uint64_t tcps_predack; /* times hdr predict ok for acks */ 571 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 572 uint64_t tcps_pcbcachemiss; 573 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 574 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 575 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 576 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 577 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 578 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 579 uint64_t tcps_persistdrop; /* timeout in persist state */ 580 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 581 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 582 uint64_t tcps_listendrop; /* listen queue overflows */ 583 uint64_t tcps_badrst; /* ignored RSTs in the window */ 584 585 uint64_t tcps_sc_added; /* entry added to syncache */ 586 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 587 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 588 uint64_t tcps_sc_dropped; /* could not reply to packet */ 589 uint64_t tcps_sc_completed; /* successful extraction of entry */ 590 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 591 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 592 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 593 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 594 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 595 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 596 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 597 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 598 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 599 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 600 601 uint64_t tcps_hc_added; /* entry added to hostcache */ 602 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 603 604 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 605 606 /* SACK related stats */ 607 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 608 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 609 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 610 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 611 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 612 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 613 614 /* ECN related stats */ 615 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 616 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 617 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 618 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 619 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 620 621 /* TCP_SIGNATURE related stats */ 622 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 623 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 624 uint64_t tcps_sig_err_buildsig; /* Failed to make signature */ 625 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 626 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 627 628 /* Path MTU Discovery Black Hole Detection related stats */ 629 uint64_t tcps_pmtud_blackhole_activated; /* Black Hole Count */ 630 uint64_t tcps_pmtud_blackhole_activated_min_mss; /* BH at min MSS Count */ 631 uint64_t tcps_pmtud_blackhole_failed; /* Black Hole Failure Count */ 632 633 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 634 }; 635 636 #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ 637 638 #ifdef _KERNEL 639 #define TI_UNLOCKED 1 640 #define TI_RLOCKED 2 641 #include <sys/counter.h> 642 643 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 644 /* 645 * In-kernel consumers can use these accessor macros directly to update 646 * stats. 647 */ 648 #define TCPSTAT_ADD(name, val) \ 649 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 650 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 651 652 /* 653 * Kernel module consumers must use this accessor macro. 654 */ 655 void kmod_tcpstat_add(int statnum, int val); 656 #define KMOD_TCPSTAT_ADD(name, val) \ 657 kmod_tcpstat_add(offsetof(struct tcpstat, name) / sizeof(uint64_t), val) 658 #define KMOD_TCPSTAT_INC(name) KMOD_TCPSTAT_ADD(name, 1) 659 660 /* 661 * Running TCP connection count by state. 662 */ 663 VNET_DECLARE(counter_u64_t, tcps_states[TCP_NSTATES]); 664 #define V_tcps_states VNET(tcps_states) 665 #define TCPSTATES_INC(state) counter_u64_add(V_tcps_states[state], 1) 666 #define TCPSTATES_DEC(state) counter_u64_add(V_tcps_states[state], -1) 667 668 /* 669 * TCP specific helper hook point identifiers. 670 */ 671 #define HHOOK_TCP_EST_IN 0 672 #define HHOOK_TCP_EST_OUT 1 673 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 674 675 struct tcp_hhook_data { 676 struct tcpcb *tp; 677 struct tcphdr *th; 678 struct tcpopt *to; 679 uint32_t len; 680 int tso; 681 tcp_seq curack; 682 }; 683 #ifdef TCP_HHOOK 684 void hhook_run_tcp_est_out(struct tcpcb *tp, 685 struct tcphdr *th, struct tcpopt *to, 686 uint32_t len, int tso); 687 #endif 688 #endif 689 690 /* 691 * TCB structure exported to user-land via sysctl(3). 692 * 693 * Fields prefixed with "xt_" are unique to the export structure, and fields 694 * with "t_" or other prefixes match corresponding fields of 'struct tcpcb'. 695 * 696 * Legend: 697 * (s) - used by userland utilities in src 698 * (p) - used by utilities in ports 699 * (3) - is known to be used by third party software not in ports 700 * (n) - no known usage 701 * 702 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 703 * included. Not all of our clients do. 704 */ 705 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 706 struct xtcpcb { 707 ksize_t xt_len; /* length of this structure */ 708 struct xinpcb xt_inp; 709 char xt_stack[TCP_FUNCTION_NAME_LEN_MAX]; /* (s) */ 710 char xt_logid[TCP_LOG_ID_LEN]; /* (s) */ 711 int64_t spare64[8]; 712 int32_t t_state; /* (s,p) */ 713 uint32_t t_flags; /* (s,p) */ 714 int32_t t_sndzerowin; /* (s) */ 715 int32_t t_sndrexmitpack; /* (s) */ 716 int32_t t_rcvoopack; /* (s) */ 717 int32_t t_rcvtime; /* (s) */ 718 int32_t tt_rexmt; /* (s) */ 719 int32_t tt_persist; /* (s) */ 720 int32_t tt_keep; /* (s) */ 721 int32_t tt_2msl; /* (s) */ 722 int32_t tt_delack; /* (s) */ 723 int32_t t_logstate; /* (3) */ 724 int32_t spare32[32]; 725 } __aligned(8); 726 727 #ifdef _KERNEL 728 void tcp_inptoxtp(const struct inpcb *, struct xtcpcb *); 729 #endif 730 #endif 731 732 /* 733 * TCP function information (name-to-id mapping, aliases, and refcnt) 734 * exported to user-land via sysctl(3). 735 */ 736 struct tcp_function_info { 737 uint32_t tfi_refcnt; 738 uint8_t tfi_id; 739 char tfi_name[TCP_FUNCTION_NAME_LEN_MAX]; 740 char tfi_alias[TCP_FUNCTION_NAME_LEN_MAX]; 741 }; 742 743 /* 744 * Identifiers for TCP sysctl nodes 745 */ 746 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 747 #define TCPCTL_MSSDFLT 3 /* MSS default */ 748 #define TCPCTL_STATS 4 /* statistics */ 749 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 750 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 751 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 752 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 753 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 754 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 755 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 756 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 757 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 758 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 759 #define TCPCTL_DROP 15 /* drop tcp connection */ 760 #define TCPCTL_STATES 16 /* connection counts by TCP state */ 761 762 #ifdef _KERNEL 763 #ifdef SYSCTL_DECL 764 SYSCTL_DECL(_net_inet_tcp); 765 SYSCTL_DECL(_net_inet_tcp_sack); 766 MALLOC_DECLARE(M_TCPLOG); 767 #endif 768 769 VNET_DECLARE(int, tcp_log_in_vain); 770 #define V_tcp_log_in_vain VNET(tcp_log_in_vain) 771 772 /* 773 * Global TCP tunables shared between different stacks. 774 * Please keep the list sorted. 775 */ 776 VNET_DECLARE(int, drop_synfin); 777 VNET_DECLARE(int, path_mtu_discovery); 778 VNET_DECLARE(int, tcp_abc_l_var); 779 VNET_DECLARE(int, tcp_autorcvbuf_max); 780 VNET_DECLARE(int, tcp_autosndbuf_inc); 781 VNET_DECLARE(int, tcp_autosndbuf_max); 782 VNET_DECLARE(int, tcp_delack_enabled); 783 VNET_DECLARE(int, tcp_do_autorcvbuf); 784 VNET_DECLARE(int, tcp_do_autosndbuf); 785 VNET_DECLARE(int, tcp_do_ecn); 786 VNET_DECLARE(int, tcp_do_newcwv); 787 VNET_DECLARE(int, tcp_do_rfc1323); 788 VNET_DECLARE(int, tcp_do_rfc3042); 789 VNET_DECLARE(int, tcp_do_rfc3390); 790 VNET_DECLARE(int, tcp_do_rfc3465); 791 VNET_DECLARE(int, tcp_do_rfc6675_pipe); 792 VNET_DECLARE(int, tcp_do_sack); 793 VNET_DECLARE(int, tcp_do_tso); 794 VNET_DECLARE(int, tcp_ecn_maxretries); 795 VNET_DECLARE(int, tcp_initcwnd_segments); 796 VNET_DECLARE(int, tcp_insecure_rst); 797 VNET_DECLARE(int, tcp_insecure_syn); 798 VNET_DECLARE(uint32_t, tcp_map_entries_limit); 799 VNET_DECLARE(uint32_t, tcp_map_split_limit); 800 VNET_DECLARE(int, tcp_minmss); 801 VNET_DECLARE(int, tcp_mssdflt); 802 #ifdef STATS 803 VNET_DECLARE(int, tcp_perconn_stats_dflt_tpl); 804 VNET_DECLARE(int, tcp_perconn_stats_enable); 805 #endif /* STATS */ 806 VNET_DECLARE(int, tcp_recvspace); 807 VNET_DECLARE(int, tcp_sack_globalholes); 808 VNET_DECLARE(int, tcp_sack_globalmaxholes); 809 VNET_DECLARE(int, tcp_sack_maxholes); 810 VNET_DECLARE(int, tcp_sc_rst_sock_fail); 811 VNET_DECLARE(int, tcp_sendspace); 812 VNET_DECLARE(struct inpcbhead, tcb); 813 VNET_DECLARE(struct inpcbinfo, tcbinfo); 814 815 #define V_tcp_do_newcwv VNET(tcp_do_newcwv) 816 #define V_drop_synfin VNET(drop_synfin) 817 #define V_path_mtu_discovery VNET(path_mtu_discovery) 818 #define V_tcb VNET(tcb) 819 #define V_tcbinfo VNET(tcbinfo) 820 #define V_tcp_abc_l_var VNET(tcp_abc_l_var) 821 #define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max) 822 #define V_tcp_autosndbuf_inc VNET(tcp_autosndbuf_inc) 823 #define V_tcp_autosndbuf_max VNET(tcp_autosndbuf_max) 824 #define V_tcp_delack_enabled VNET(tcp_delack_enabled) 825 #define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf) 826 #define V_tcp_do_autosndbuf VNET(tcp_do_autosndbuf) 827 #define V_tcp_do_ecn VNET(tcp_do_ecn) 828 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 829 #define V_tcp_ts_offset_per_conn VNET(tcp_ts_offset_per_conn) 830 #define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042) 831 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 832 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 833 #define V_tcp_do_rfc6675_pipe VNET(tcp_do_rfc6675_pipe) 834 #define V_tcp_do_sack VNET(tcp_do_sack) 835 #define V_tcp_do_tso VNET(tcp_do_tso) 836 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 837 #define V_tcp_initcwnd_segments VNET(tcp_initcwnd_segments) 838 #define V_tcp_insecure_rst VNET(tcp_insecure_rst) 839 #define V_tcp_insecure_syn VNET(tcp_insecure_syn) 840 #define V_tcp_map_entries_limit VNET(tcp_map_entries_limit) 841 #define V_tcp_map_split_limit VNET(tcp_map_split_limit) 842 #define V_tcp_minmss VNET(tcp_minmss) 843 #define V_tcp_mssdflt VNET(tcp_mssdflt) 844 #ifdef STATS 845 #define V_tcp_perconn_stats_dflt_tpl VNET(tcp_perconn_stats_dflt_tpl) 846 #define V_tcp_perconn_stats_enable VNET(tcp_perconn_stats_enable) 847 #endif /* STATS */ 848 #define V_tcp_recvspace VNET(tcp_recvspace) 849 #define V_tcp_sack_globalholes VNET(tcp_sack_globalholes) 850 #define V_tcp_sack_globalmaxholes VNET(tcp_sack_globalmaxholes) 851 #define V_tcp_sack_maxholes VNET(tcp_sack_maxholes) 852 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 853 #define V_tcp_sendspace VNET(tcp_sendspace) 854 #define V_tcp_udp_tunneling_overhead VNET(tcp_udp_tunneling_overhead) 855 #define V_tcp_udp_tunneling_port VNET(tcp_udp_tunneling_port) 856 857 #ifdef TCP_HHOOK 858 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 859 #define V_tcp_hhh VNET(tcp_hhh) 860 #endif 861 862 int tcp_addoptions(struct tcpopt *, u_char *); 863 int tcp_ccalgounload(struct cc_algo *unload_algo); 864 struct tcpcb * 865 tcp_close(struct tcpcb *); 866 void tcp_discardcb(struct tcpcb *); 867 void tcp_twstart(struct tcpcb *); 868 void tcp_twclose(struct tcptw *, int); 869 void tcp_ctlinput(int, struct sockaddr *, void *); 870 int tcp_ctloutput(struct socket *, struct sockopt *); 871 struct tcpcb * 872 tcp_drop(struct tcpcb *, int); 873 void tcp_drain(void); 874 void tcp_init(void); 875 void tcp_fini(void *); 876 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 877 const void *); 878 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 879 const void *); 880 int tcp_reass(struct tcpcb *, struct tcphdr *, tcp_seq *, int *, struct mbuf *); 881 void tcp_reass_global_init(void); 882 void tcp_reass_flush(struct tcpcb *); 883 void tcp_dooptions(struct tcpopt *, u_char *, int, int); 884 void tcp_dropwithreset(struct mbuf *, struct tcphdr *, 885 struct tcpcb *, int, int); 886 void tcp_pulloutofband(struct socket *, 887 struct tcphdr *, struct mbuf *, int); 888 void tcp_xmit_timer(struct tcpcb *, int); 889 void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *); 890 void cc_ack_received(struct tcpcb *tp, struct tcphdr *th, 891 uint16_t nsegs, uint16_t type); 892 void cc_conn_init(struct tcpcb *tp); 893 void cc_post_recovery(struct tcpcb *tp, struct tcphdr *th); 894 void cc_ecnpkt_handler(struct tcpcb *tp, struct tcphdr *th, uint8_t iptos); 895 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 896 #ifdef TCP_HHOOK 897 void hhook_run_tcp_est_in(struct tcpcb *tp, 898 struct tcphdr *th, struct tcpopt *to); 899 #endif 900 901 int tcp_input(struct mbuf **, int *, int); 902 int tcp_autorcvbuf(struct mbuf *, struct tcphdr *, struct socket *, 903 struct tcpcb *, int); 904 void tcp_do_segment(struct mbuf *, struct tcphdr *, 905 struct socket *, struct tcpcb *, int, int, uint8_t); 906 907 int register_tcp_functions(struct tcp_function_block *blk, int wait); 908 int register_tcp_functions_as_names(struct tcp_function_block *blk, 909 int wait, const char *names[], int *num_names); 910 int register_tcp_functions_as_name(struct tcp_function_block *blk, 911 const char *name, int wait); 912 int deregister_tcp_functions(struct tcp_function_block *blk, bool quiesce, 913 bool force); 914 struct tcp_function_block *find_and_ref_tcp_functions(struct tcp_function_set *fs); 915 void tcp_switch_back_to_default(struct tcpcb *tp); 916 struct tcp_function_block * 917 find_and_ref_tcp_fb(struct tcp_function_block *fs); 918 int tcp_default_ctloutput(struct socket *so, struct sockopt *sopt, struct inpcb *inp, struct tcpcb *tp); 919 920 extern counter_u64_t tcp_inp_lro_direct_queue; 921 extern counter_u64_t tcp_inp_lro_wokeup_queue; 922 extern counter_u64_t tcp_inp_lro_compressed; 923 extern counter_u64_t tcp_inp_lro_single_push; 924 extern counter_u64_t tcp_inp_lro_locks_taken; 925 extern counter_u64_t tcp_inp_lro_sack_wake; 926 927 #ifdef NETFLIX_EXP_DETECTION 928 /* Various SACK attack thresholds */ 929 extern int32_t tcp_force_detection; 930 extern int32_t tcp_sack_to_ack_thresh; 931 extern int32_t tcp_sack_to_move_thresh; 932 extern int32_t tcp_restoral_thresh; 933 extern int32_t tcp_sad_decay_val; 934 extern int32_t tcp_sad_pacing_interval; 935 extern int32_t tcp_sad_low_pps; 936 extern int32_t tcp_map_minimum; 937 extern int32_t tcp_attack_on_turns_on_logging; 938 #endif 939 940 uint32_t tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 941 uint32_t tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 942 u_int tcp_maxseg(const struct tcpcb *); 943 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 944 struct tcp_ifcap *); 945 void tcp_mss(struct tcpcb *, int); 946 int tcp_mssopt(struct in_conninfo *); 947 struct inpcb * 948 tcp_drop_syn_sent(struct inpcb *, int); 949 struct tcpcb * 950 tcp_newtcpcb(struct inpcb *); 951 int tcp_output(struct tcpcb *); 952 void tcp_state_change(struct tcpcb *, int); 953 void tcp_respond(struct tcpcb *, void *, 954 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 955 void tcp_tw_init(void); 956 #ifdef VIMAGE 957 void tcp_tw_destroy(void); 958 #endif 959 void tcp_tw_zone_change(void); 960 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 961 struct mbuf *, int); 962 void tcp_setpersist(struct tcpcb *); 963 void tcp_slowtimo(void); 964 struct tcptemp * 965 tcpip_maketemplate(struct inpcb *); 966 void tcpip_fillheaders(struct inpcb *, void *, void *); 967 void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 968 int tcp_timer_suspend(struct tcpcb *, uint32_t); 969 void tcp_timers_unsuspend(struct tcpcb *, uint32_t); 970 int tcp_timer_active(struct tcpcb *, uint32_t); 971 void tcp_timer_stop(struct tcpcb *, uint32_t); 972 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 973 int inp_to_cpuid(struct inpcb *inp); 974 /* 975 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 976 */ 977 void tcp_hc_init(void); 978 #ifdef VIMAGE 979 void tcp_hc_destroy(void); 980 #endif 981 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 982 uint32_t tcp_hc_getmtu(struct in_conninfo *); 983 void tcp_hc_updatemtu(struct in_conninfo *, uint32_t); 984 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 985 986 extern struct pr_usrreqs tcp_usrreqs; 987 988 uint32_t tcp_new_ts_offset(struct in_conninfo *); 989 tcp_seq tcp_new_isn(struct in_conninfo *); 990 991 int tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 992 void tcp_update_dsack_list(struct tcpcb *, tcp_seq, tcp_seq); 993 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 994 void tcp_clean_dsack_blocks(struct tcpcb *tp); 995 void tcp_clean_sackreport(struct tcpcb *tp); 996 void tcp_sack_adjust(struct tcpcb *tp); 997 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 998 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 999 void tcp_free_sackholes(struct tcpcb *tp); 1000 int tcp_newreno(struct tcpcb *, struct tcphdr *); 1001 int tcp_compute_pipe(struct tcpcb *); 1002 uint32_t tcp_compute_initwnd(uint32_t); 1003 void tcp_sndbuf_autoscale(struct tcpcb *, struct socket *, uint32_t); 1004 int tcp_stats_sample_rollthedice(struct tcpcb *tp, void *seed_bytes, 1005 size_t seed_len); 1006 struct mbuf * 1007 tcp_m_copym(struct mbuf *m, int32_t off0, int32_t *plen, 1008 int32_t seglimit, int32_t segsize, struct sockbuf *sb, bool hw_tls); 1009 1010 int tcp_stats_init(void); 1011 1012 static inline void 1013 tcp_fields_to_host(struct tcphdr *th) 1014 { 1015 1016 th->th_seq = ntohl(th->th_seq); 1017 th->th_ack = ntohl(th->th_ack); 1018 th->th_win = ntohs(th->th_win); 1019 th->th_urp = ntohs(th->th_urp); 1020 } 1021 1022 static inline void 1023 tcp_fields_to_net(struct tcphdr *th) 1024 { 1025 1026 th->th_seq = htonl(th->th_seq); 1027 th->th_ack = htonl(th->th_ack); 1028 th->th_win = htons(th->th_win); 1029 th->th_urp = htons(th->th_urp); 1030 } 1031 #endif /* _KERNEL */ 1032 1033 #endif /* _NETINET_TCP_VAR_H_ */ 1034