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 #define TCP_END_BYTE_INFO 8 /* Bytes that makeup the "end information array" */ 47 /* Types of ending byte info */ 48 #define TCP_EI_EMPTY_SLOT 0 49 #define TCP_EI_STATUS_CLIENT_FIN 0x1 50 #define TCP_EI_STATUS_CLIENT_RST 0x2 51 #define TCP_EI_STATUS_SERVER_FIN 0x3 52 #define TCP_EI_STATUS_SERVER_RST 0x4 53 #define TCP_EI_STATUS_RETRAN 0x5 54 #define TCP_EI_STATUS_PROGRESS 0x6 55 #define TCP_EI_STATUS_PERSIST_MAX 0x7 56 #define TCP_EI_STATUS_KEEP_MAX 0x8 57 #define TCP_EI_STATUS_DATA_A_CLOSE 0x9 58 #define TCP_EI_STATUS_RST_IN_FRONT 0xa 59 #define TCP_EI_STATUS_2MSL 0xb 60 #define TCP_EI_STATUS_MAX_VALUE 0xb 61 62 /************************************************/ 63 /* Status bits we track to assure no duplicates, 64 * the bits here are not used by the code but 65 * for human representation. To check a bit we 66 * take and shift over by 1 minus the value (1-8). 67 */ 68 /************************************************/ 69 #define TCP_EI_BITS_CLIENT_FIN 0x001 70 #define TCP_EI_BITS_CLIENT_RST 0x002 71 #define TCP_EI_BITS_SERVER_FIN 0x004 72 #define TCP_EI_BITS_SERVER_RST 0x008 73 #define TCP_EI_BITS_RETRAN 0x010 74 #define TCP_EI_BITS_PROGRESS 0x020 75 #define TCP_EI_BITS_PRESIST_MAX 0x040 76 #define TCP_EI_BITS_KEEP_MAX 0x080 77 #define TCP_EI_BITS_DATA_A_CLO 0x100 78 #define TCP_EI_BITS_RST_IN_FR 0x200 /* a front state reset */ 79 #define TCP_EI_BITS_2MS_TIMER 0x400 /* 2 MSL timer expired */ 80 81 #if defined(_KERNEL) || defined(_WANT_TCPCB) 82 /* TCP segment queue entry */ 83 struct tseg_qent { 84 TAILQ_ENTRY(tseg_qent) tqe_q; 85 struct mbuf *tqe_m; /* mbuf contains packet */ 86 struct mbuf *tqe_last; /* last mbuf in chain */ 87 tcp_seq tqe_start; /* TCP Sequence number start */ 88 int tqe_len; /* TCP segment data length */ 89 uint32_t tqe_flags; /* The flags from the th->th_flags */ 90 uint32_t tqe_mbuf_cnt; /* Count of mbuf overhead */ 91 }; 92 TAILQ_HEAD(tsegqe_head, tseg_qent); 93 94 struct sackblk { 95 tcp_seq start; /* start seq no. of sack block */ 96 tcp_seq end; /* end seq no. */ 97 }; 98 99 struct sackhole { 100 tcp_seq start; /* start seq no. of hole */ 101 tcp_seq end; /* end seq no. */ 102 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */ 103 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */ 104 }; 105 106 struct sackhint { 107 struct sackhole *nexthole; 108 int32_t sack_bytes_rexmit; 109 tcp_seq last_sack_ack; /* Most recent/largest sacked ack */ 110 111 int32_t delivered_data; /* Newly acked data from last SACK */ 112 113 int32_t sacked_bytes; /* Total sacked bytes reported by the 114 * receiver via sack option 115 */ 116 uint32_t _pad1[1]; /* TBD */ 117 uint64_t _pad[1]; /* TBD */ 118 }; 119 120 #define SEGQ_EMPTY(tp) TAILQ_EMPTY(&(tp)->t_segq) 121 122 STAILQ_HEAD(tcp_log_stailq, tcp_log_mem); 123 124 /* 125 * Tcp control block, one per tcp; fields: 126 * Organized for 64 byte cacheline efficiency based 127 * on common tcp_input/tcp_output processing. 128 */ 129 struct tcpcb { 130 /* Cache line 1 */ 131 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 132 struct tcp_function_block *t_fb;/* TCP function call block */ 133 void *t_fb_ptr; /* Pointer to t_fb specific data */ 134 uint32_t t_maxseg:24, /* maximum segment size */ 135 t_logstate:8; /* State of "black box" logging */ 136 uint32_t t_port:16, /* Tunneling (over udp) port */ 137 t_state:4, /* state of this connection */ 138 t_idle_reduce : 1, 139 t_delayed_ack: 7, /* Delayed ack variable */ 140 t_fin_is_rst: 1, /* Are fin's treated as resets */ 141 t_log_state_set: 1, 142 bits_spare : 2; 143 u_int t_flags; 144 tcp_seq snd_una; /* sent but unacknowledged */ 145 tcp_seq snd_max; /* highest sequence number sent; 146 * used to recognize retransmits 147 */ 148 tcp_seq snd_nxt; /* send next */ 149 tcp_seq snd_up; /* send urgent pointer */ 150 uint32_t snd_wnd; /* send window */ 151 uint32_t snd_cwnd; /* congestion-controlled window */ 152 uint32_t t_peakrate_thr; /* pre-calculated peak rate threshold */ 153 /* Cache line 2 */ 154 u_int32_t ts_offset; /* our timestamp offset */ 155 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */ 156 int rcv_numsacks; /* # distinct sack blks present */ 157 u_int t_tsomax; /* TSO total burst length limit in bytes */ 158 u_int t_tsomaxsegcount; /* TSO maximum segment count */ 159 u_int t_tsomaxsegsize; /* TSO maximum segment size in bytes */ 160 tcp_seq rcv_nxt; /* receive next */ 161 tcp_seq rcv_adv; /* advertised window */ 162 uint32_t rcv_wnd; /* receive window */ 163 u_int t_flags2; /* More tcpcb flags storage */ 164 int t_srtt; /* smoothed round-trip time */ 165 int t_rttvar; /* variance in round-trip time */ 166 u_int32_t ts_recent; /* timestamp echo data */ 167 u_char snd_scale; /* window scaling for send window */ 168 u_char rcv_scale; /* window scaling for recv window */ 169 u_char snd_limited; /* segments limited transmitted */ 170 u_char request_r_scale; /* pending window scaling */ 171 tcp_seq last_ack_sent; 172 u_int t_rcvtime; /* inactivity time */ 173 /* Cache line 3 */ 174 tcp_seq rcv_up; /* receive urgent pointer */ 175 int t_segqlen; /* segment reassembly queue length */ 176 uint32_t t_segqmbuflen; /* Count of bytes mbufs on all entries */ 177 struct tsegqe_head t_segq; /* segment reassembly queue */ 178 struct mbuf *t_in_pkt; 179 struct mbuf *t_tail_pkt; 180 struct tcp_timer *t_timers; /* All the TCP timers in one struct */ 181 struct vnet *t_vnet; /* back pointer to parent vnet */ 182 uint32_t snd_ssthresh; /* snd_cwnd size threshold for 183 * for slow start exponential to 184 * linear switch 185 */ 186 tcp_seq snd_wl1; /* window update seg seq number */ 187 /* Cache line 4 */ 188 tcp_seq snd_wl2; /* window update seg ack number */ 189 190 tcp_seq irs; /* initial receive sequence number */ 191 tcp_seq iss; /* initial send sequence number */ 192 u_int t_acktime; /* RACK and BBR incoming new data was acked */ 193 u_int t_sndtime; /* time last data was sent */ 194 u_int ts_recent_age; /* when last updated */ 195 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ 196 uint16_t cl4_spare; /* Spare to adjust CL 4 */ 197 char t_oobflags; /* have some */ 198 char t_iobc; /* input character */ 199 int t_rxtcur; /* current retransmit value (ticks) */ 200 201 int t_rxtshift; /* log(2) of rexmt exp. backoff */ 202 u_int t_rtttime; /* RTT measurement start time */ 203 204 tcp_seq t_rtseq; /* sequence number being timed */ 205 u_int t_starttime; /* time connection was established */ 206 u_int t_fbyte_in; /* ticks time when first byte queued in */ 207 u_int t_fbyte_out; /* ticks time when first byte queued out */ 208 209 u_int t_pmtud_saved_maxseg; /* pre-blackhole MSS */ 210 int t_blackhole_enter; /* when to enter blackhole detection */ 211 int t_blackhole_exit; /* when to exit blackhole detection */ 212 u_int t_rttmin; /* minimum rtt allowed */ 213 214 u_int t_rttbest; /* best rtt we've seen */ 215 216 int t_softerror; /* possible error not yet reported */ 217 uint32_t max_sndwnd; /* largest window peer has offered */ 218 /* Cache line 5 */ 219 uint32_t snd_cwnd_prev; /* cwnd prior to retransmit */ 220 uint32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ 221 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ 222 int t_sndzerowin; /* zero-window updates sent */ 223 u_long t_rttupdated; /* number of times rtt sampled */ 224 int snd_numholes; /* number of holes seen by sender */ 225 u_int t_badrxtwin; /* window for retransmit recovery */ 226 TAILQ_HEAD(sackhole_head, sackhole) snd_holes; 227 /* SACK scoreboard (sorted) */ 228 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ 229 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */ 230 struct sackhint sackhint; /* SACK scoreboard hint */ 231 int t_rttlow; /* smallest observerved RTT */ 232 int rfbuf_cnt; /* recv buffer autoscaling byte count */ 233 struct toedev *tod; /* toedev handling this connection */ 234 int t_sndrexmitpack; /* retransmit packets sent */ 235 int t_rcvoopack; /* out-of-order packets received */ 236 void *t_toe; /* TOE pcb pointer */ 237 struct cc_algo *cc_algo; /* congestion control algorithm */ 238 struct cc_var *ccv; /* congestion control specific vars */ 239 struct osd *osd; /* storage for Khelp module data */ 240 int t_bytes_acked; /* # bytes acked during current RTT */ 241 u_int t_maxunacktime; 242 u_int t_keepinit; /* time to establish connection */ 243 u_int t_keepidle; /* time before keepalive probes begin */ 244 u_int t_keepintvl; /* interval between keepalives */ 245 u_int t_keepcnt; /* number of keepalives before close */ 246 int t_dupacks; /* consecutive dup acks recd */ 247 int t_lognum; /* Number of log entries */ 248 int t_loglimit; /* Maximum number of log entries */ 249 int64_t t_pacing_rate; /* bytes / sec, -1 => unlimited */ 250 struct tcp_log_stailq t_logs; /* Log buffer */ 251 struct tcp_log_id_node *t_lin; 252 struct tcp_log_id_bucket *t_lib; 253 const char *t_output_caller; /* Function that called tcp_output */ 254 struct statsblob *t_stats; /* Per-connection stats */ 255 uint32_t t_logsn; /* Log "serial number" */ 256 uint32_t gput_ts; /* Time goodput measurement started */ 257 tcp_seq gput_seq; /* Outbound measurement seq */ 258 tcp_seq gput_ack; /* Inbound measurement ack */ 259 int32_t t_stats_gput_prev; /* XXXLAS: Prev gput measurement */ 260 uint8_t t_tfo_client_cookie_len; /* TCP Fast Open client cookie length */ 261 uint32_t t_end_info_status; /* Status flag of end info */ 262 unsigned int *t_tfo_pending; /* TCP Fast Open server pending counter */ 263 union { 264 uint8_t client[TCP_FASTOPEN_MAX_COOKIE_LEN]; 265 uint64_t server; 266 } t_tfo_cookie; /* TCP Fast Open cookie to send */ 267 union { 268 uint8_t t_end_info_bytes[TCP_END_BYTE_INFO]; 269 uint64_t t_end_info; 270 }; 271 #ifdef TCPPCAP 272 struct mbufq t_inpkts; /* List of saved input packets. */ 273 struct mbufq t_outpkts; /* List of saved output packets. */ 274 #endif 275 }; 276 #endif /* _KERNEL || _WANT_TCPCB */ 277 278 #ifdef _KERNEL 279 struct tcptemp { 280 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */ 281 struct tcphdr tt_t; 282 }; 283 284 /* Minimum map entries limit value, if set */ 285 #define TCP_MIN_MAP_ENTRIES_LIMIT 128 286 287 /* 288 * TODO: We yet need to brave plowing in 289 * to tcp_input() and the pru_usrreq() block. 290 * Right now these go to the old standards which 291 * are somewhat ok, but in the long term may 292 * need to be changed. If we do tackle tcp_input() 293 * then we need to get rid of the tcp_do_segment() 294 * function below. 295 */ 296 /* Flags for tcp functions */ 297 #define TCP_FUNC_BEING_REMOVED 0x01 /* Can no longer be referenced */ 298 299 /* 300 * If defining the optional tcp_timers, in the 301 * tfb_tcp_timer_stop call you must use the 302 * callout_async_drain() function with the 303 * tcp_timer_discard callback. You should check 304 * the return of callout_async_drain() and if 0 305 * increment tt_draincnt. Since the timer sub-system 306 * does not know your callbacks you must provide a 307 * stop_all function that loops through and calls 308 * tcp_timer_stop() with each of your defined timers. 309 * Adding a tfb_tcp_handoff_ok function allows the socket 310 * option to change stacks to query you even if the 311 * connection is in a later stage. You return 0 to 312 * say you can take over and run your stack, you return 313 * non-zero (an error number) to say no you can't. 314 * If the function is undefined you can only change 315 * in the early states (before connect or listen). 316 * tfb_tcp_fb_fini is changed to add a flag to tell 317 * the old stack if the tcb is being destroyed or 318 * not. A one in the flag means the TCB is being 319 * destroyed, a zero indicates its transitioning to 320 * another stack (via socket option). 321 */ 322 struct tcp_function_block { 323 char tfb_tcp_block_name[TCP_FUNCTION_NAME_LEN_MAX]; 324 int (*tfb_tcp_output)(struct tcpcb *); 325 int (*tfb_tcp_output_wtime)(struct tcpcb *, const struct timeval *); 326 void (*tfb_tcp_do_segment)(struct mbuf *, struct tcphdr *, 327 struct socket *, struct tcpcb *, 328 int, int, uint8_t); 329 int (*tfb_do_queued_segments)(struct socket *, struct tcpcb *, int); 330 int (*tfb_do_segment_nounlock)(struct mbuf *, struct tcphdr *, 331 struct socket *, struct tcpcb *, 332 int, int, uint8_t, 333 int, struct timeval *); 334 void (*tfb_tcp_hpts_do_segment)(struct mbuf *, struct tcphdr *, 335 struct socket *, struct tcpcb *, 336 int, int, uint8_t, 337 int, struct timeval *); 338 int (*tfb_tcp_ctloutput)(struct socket *so, struct sockopt *sopt, 339 struct inpcb *inp, struct tcpcb *tp); 340 /* Optional memory allocation/free routine */ 341 int (*tfb_tcp_fb_init)(struct tcpcb *); 342 void (*tfb_tcp_fb_fini)(struct tcpcb *, int); 343 /* Optional timers, must define all if you define one */ 344 int (*tfb_tcp_timer_stop_all)(struct tcpcb *); 345 void (*tfb_tcp_timer_activate)(struct tcpcb *, 346 uint32_t, u_int); 347 int (*tfb_tcp_timer_active)(struct tcpcb *, uint32_t); 348 void (*tfb_tcp_timer_stop)(struct tcpcb *, uint32_t); 349 void (*tfb_tcp_rexmit_tmr)(struct tcpcb *); 350 int (*tfb_tcp_handoff_ok)(struct tcpcb *); 351 void (*tfb_tcp_mtu_chg)(struct tcpcb *); 352 int (*tfb_pru_options)(struct tcpcb *, int); 353 volatile uint32_t tfb_refcnt; 354 uint32_t tfb_flags; 355 uint8_t tfb_id; 356 }; 357 358 struct tcp_function { 359 TAILQ_ENTRY(tcp_function) tf_next; 360 char tf_name[TCP_FUNCTION_NAME_LEN_MAX]; 361 struct tcp_function_block *tf_fb; 362 }; 363 364 TAILQ_HEAD(tcp_funchead, tcp_function); 365 #endif /* _KERNEL */ 366 367 /* 368 * Flags and utility macros for the t_flags field. 369 */ 370 #define TF_ACKNOW 0x00000001 /* ack peer immediately */ 371 #define TF_DELACK 0x00000002 /* ack, but try to delay it */ 372 #define TF_NODELAY 0x00000004 /* don't delay packets to coalesce */ 373 #define TF_NOOPT 0x00000008 /* don't use tcp options */ 374 #define TF_SENTFIN 0x00000010 /* have sent FIN */ 375 #define TF_REQ_SCALE 0x00000020 /* have/will request window scaling */ 376 #define TF_RCVD_SCALE 0x00000040 /* other side has requested scaling */ 377 #define TF_REQ_TSTMP 0x00000080 /* have/will request timestamps */ 378 #define TF_RCVD_TSTMP 0x00000100 /* a timestamp was received in SYN */ 379 #define TF_SACK_PERMIT 0x00000200 /* other side said I could SACK */ 380 #define TF_NEEDSYN 0x00000400 /* send SYN (implicit state) */ 381 #define TF_NEEDFIN 0x00000800 /* send FIN (implicit state) */ 382 #define TF_NOPUSH 0x00001000 /* don't push */ 383 #define TF_PREVVALID 0x00002000 /* saved values for bad rxmit valid */ 384 #define TF_UNUSED1 0x00004000 /* unused */ 385 #define TF_GPUTINPROG 0x00008000 /* Goodput measurement in progress */ 386 #define TF_MORETOCOME 0x00010000 /* More data to be appended to sock */ 387 #define TF_LQ_OVERFLOW 0x00020000 /* listen queue overflow */ 388 #define TF_LASTIDLE 0x00040000 /* connection was previously idle */ 389 #define TF_RXWIN0SENT 0x00080000 /* sent a receiver win 0 in response */ 390 #define TF_FASTRECOVERY 0x00100000 /* in NewReno Fast Recovery */ 391 #define TF_WASFRECOVERY 0x00200000 /* was in NewReno Fast Recovery */ 392 #define TF_SIGNATURE 0x00400000 /* require MD5 digests (RFC2385) */ 393 #define TF_FORCEDATA 0x00800000 /* force out a byte */ 394 #define TF_TSO 0x01000000 /* TSO enabled on this connection */ 395 #define TF_TOE 0x02000000 /* this connection is offloaded */ 396 #define TF_UNUSED3 0x04000000 /* unused */ 397 #define TF_UNUSED4 0x08000000 /* unused */ 398 #define TF_UNUSED5 0x10000000 /* unused */ 399 #define TF_CONGRECOVERY 0x20000000 /* congestion recovery mode */ 400 #define TF_WASCRECOVERY 0x40000000 /* was in congestion recovery */ 401 #define TF_FASTOPEN 0x80000000 /* TCP Fast Open indication */ 402 403 #define IN_FASTRECOVERY(t_flags) (t_flags & TF_FASTRECOVERY) 404 #define ENTER_FASTRECOVERY(t_flags) t_flags |= TF_FASTRECOVERY 405 #define EXIT_FASTRECOVERY(t_flags) t_flags &= ~TF_FASTRECOVERY 406 407 #define IN_CONGRECOVERY(t_flags) (t_flags & TF_CONGRECOVERY) 408 #define ENTER_CONGRECOVERY(t_flags) t_flags |= TF_CONGRECOVERY 409 #define EXIT_CONGRECOVERY(t_flags) t_flags &= ~TF_CONGRECOVERY 410 411 #define IN_RECOVERY(t_flags) (t_flags & (TF_CONGRECOVERY | TF_FASTRECOVERY)) 412 #define ENTER_RECOVERY(t_flags) t_flags |= (TF_CONGRECOVERY | TF_FASTRECOVERY) 413 #define EXIT_RECOVERY(t_flags) t_flags &= ~(TF_CONGRECOVERY | TF_FASTRECOVERY) 414 415 #if defined(_KERNEL) && !defined(TCP_RFC7413) 416 #define IS_FASTOPEN(t_flags) (false) 417 #else 418 #define IS_FASTOPEN(t_flags) (t_flags & TF_FASTOPEN) 419 #endif 420 421 #define BYTES_THIS_ACK(tp, th) (th->th_ack - tp->snd_una) 422 423 /* 424 * Flags for the t_oobflags field. 425 */ 426 #define TCPOOB_HAVEDATA 0x01 427 #define TCPOOB_HADDATA 0x02 428 429 /* 430 * Flags for the extended TCP flags field, t_flags2 431 */ 432 #define TF2_PLPMTU_BLACKHOLE 0x00000001 /* Possible PLPMTUD Black Hole. */ 433 #define TF2_PLPMTU_PMTUD 0x00000002 /* Allowed to attempt PLPMTUD. */ 434 #define TF2_PLPMTU_MAXSEGSNT 0x00000004 /* Last seg sent was full seg. */ 435 #define TF2_LOG_AUTO 0x00000008 /* Session is auto-logging. */ 436 #define TF2_DROP_AF_DATA 0x00000010 /* Drop after all data ack'd */ 437 #define TF2_ECN_PERMIT 0x00000020 /* connection ECN-ready */ 438 #define TF2_ECN_SND_CWR 0x00000040 /* ECN CWR in queue */ 439 #define TF2_ECN_SND_ECE 0x00000080 /* ECN ECE in queue */ 440 #define TF2_ACE_PERMIT 0x00000100 /* Accurate ECN mode */ 441 #define TF2_FBYTES_COMPLETE 0x00000400 /* We have first bytes in and out */ 442 /* 443 * Structure to hold TCP options that are only used during segment 444 * processing (in tcp_input), but not held in the tcpcb. 445 * It's basically used to reduce the number of parameters 446 * to tcp_dooptions and tcp_addoptions. 447 * The binary order of the to_flags is relevant for packing of the 448 * options in tcp_addoptions. 449 */ 450 struct tcpopt { 451 u_int32_t to_flags; /* which options are present */ 452 #define TOF_MSS 0x0001 /* maximum segment size */ 453 #define TOF_SCALE 0x0002 /* window scaling */ 454 #define TOF_SACKPERM 0x0004 /* SACK permitted */ 455 #define TOF_TS 0x0010 /* timestamp */ 456 #define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */ 457 #define TOF_SACK 0x0080 /* Peer sent SACK option */ 458 #define TOF_FASTOPEN 0x0100 /* TCP Fast Open (TFO) cookie */ 459 #define TOF_MAXOPT 0x0200 460 u_int32_t to_tsval; /* new timestamp */ 461 u_int32_t to_tsecr; /* reflected timestamp */ 462 u_char *to_sacks; /* pointer to the first SACK blocks */ 463 u_char *to_signature; /* pointer to the TCP-MD5 signature */ 464 u_int8_t *to_tfo_cookie; /* pointer to the TFO cookie */ 465 u_int16_t to_mss; /* maximum segment size */ 466 u_int8_t to_wscale; /* window scaling */ 467 u_int8_t to_nsacks; /* number of SACK blocks */ 468 u_int8_t to_tfo_len; /* TFO cookie length */ 469 u_int32_t to_spare; /* UTO */ 470 }; 471 472 /* 473 * Flags for tcp_dooptions. 474 */ 475 #define TO_SYN 0x01 /* parse SYN-only options */ 476 477 struct hc_metrics_lite { /* must stay in sync with hc_metrics */ 478 uint32_t rmx_mtu; /* MTU for this path */ 479 uint32_t rmx_ssthresh; /* outbound gateway buffer limit */ 480 uint32_t rmx_rtt; /* estimated round trip time */ 481 uint32_t rmx_rttvar; /* estimated rtt variance */ 482 uint32_t rmx_cwnd; /* congestion window */ 483 uint32_t rmx_sendpipe; /* outbound delay-bandwidth product */ 484 uint32_t rmx_recvpipe; /* inbound delay-bandwidth product */ 485 }; 486 487 /* 488 * Used by tcp_maxmtu() to communicate interface specific features 489 * and limits at the time of connection setup. 490 */ 491 struct tcp_ifcap { 492 int ifcap; 493 u_int tsomax; 494 u_int tsomaxsegcount; 495 u_int tsomaxsegsize; 496 }; 497 498 #ifndef _NETINET_IN_PCB_H_ 499 struct in_conninfo; 500 #endif /* _NETINET_IN_PCB_H_ */ 501 502 struct tcptw { 503 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */ 504 tcp_seq snd_nxt; 505 tcp_seq rcv_nxt; 506 tcp_seq iss; 507 tcp_seq irs; 508 u_short last_win; /* cached window value */ 509 short tw_so_options; /* copy of so_options */ 510 struct ucred *tw_cred; /* user credentials */ 511 u_int32_t t_recent; 512 u_int32_t ts_offset; /* our timestamp offset */ 513 u_int t_starttime; 514 int tw_time; 515 TAILQ_ENTRY(tcptw) tw_2msl; 516 void *tw_pspare; /* TCP_SIGNATURE */ 517 u_int *tw_spare; /* TCP_SIGNATURE */ 518 }; 519 520 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 521 #define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb) 522 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 523 524 /* 525 * The smoothed round-trip time and estimated variance 526 * are stored as fixed point numbers scaled by the values below. 527 * For convenience, these scales are also used in smoothing the average 528 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 529 * With these scales, srtt has 3 bits to the right of the binary point, 530 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 531 * binary point, and is smoothed with an ALPHA of 0.75. 532 */ 533 #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ 534 #define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ 535 #define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ 536 #define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ 537 #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ 538 539 /* 540 * The initial retransmission should happen at rtt + 4 * rttvar. 541 * Because of the way we do the smoothing, srtt and rttvar 542 * will each average +1/2 tick of bias. When we compute 543 * the retransmit timer, we want 1/2 tick of rounding and 544 * 1 extra tick because of +-1/2 tick uncertainty in the 545 * firing of the timer. The bias will give us exactly the 546 * 1.5 tick we need. But, because the bias is 547 * statistical, we have to test that we don't drop below 548 * the minimum feasible timer (which is 2 ticks). 549 * This version of the macro adapted from a paper by Lawrence 550 * Brakmo and Larry Peterson which outlines a problem caused 551 * by insufficient precision in the original implementation, 552 * which results in inappropriately large RTO values for very 553 * fast networks. 554 */ 555 #define TCP_REXMTVAL(tp) \ 556 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \ 557 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT) 558 559 /* 560 * TCP statistics. 561 * Many of these should be kept per connection, 562 * but that's inconvenient at the moment. 563 */ 564 struct tcpstat { 565 uint64_t tcps_connattempt; /* connections initiated */ 566 uint64_t tcps_accepts; /* connections accepted */ 567 uint64_t tcps_connects; /* connections established */ 568 uint64_t tcps_drops; /* connections dropped */ 569 uint64_t tcps_conndrops; /* embryonic connections dropped */ 570 uint64_t tcps_minmssdrops; /* average minmss too low drops */ 571 uint64_t tcps_closed; /* conn. closed (includes drops) */ 572 uint64_t tcps_segstimed; /* segs where we tried to get rtt */ 573 uint64_t tcps_rttupdated; /* times we succeeded */ 574 uint64_t tcps_delack; /* delayed acks sent */ 575 uint64_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 576 uint64_t tcps_rexmttimeo; /* retransmit timeouts */ 577 uint64_t tcps_persisttimeo; /* persist timeouts */ 578 uint64_t tcps_keeptimeo; /* keepalive timeouts */ 579 uint64_t tcps_keepprobe; /* keepalive probes sent */ 580 uint64_t tcps_keepdrops; /* connections dropped in keepalive */ 581 582 uint64_t tcps_sndtotal; /* total packets sent */ 583 uint64_t tcps_sndpack; /* data packets sent */ 584 uint64_t tcps_sndbyte; /* data bytes sent */ 585 uint64_t tcps_sndrexmitpack; /* data packets retransmitted */ 586 uint64_t tcps_sndrexmitbyte; /* data bytes retransmitted */ 587 uint64_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ 588 uint64_t tcps_sndacks; /* ack-only packets sent */ 589 uint64_t tcps_sndprobe; /* window probes sent */ 590 uint64_t tcps_sndurg; /* packets sent with URG only */ 591 uint64_t tcps_sndwinup; /* window update-only packets sent */ 592 uint64_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 593 594 uint64_t tcps_rcvtotal; /* total packets received */ 595 uint64_t tcps_rcvpack; /* packets received in sequence */ 596 uint64_t tcps_rcvbyte; /* bytes received in sequence */ 597 uint64_t tcps_rcvbadsum; /* packets received with ccksum errs */ 598 uint64_t tcps_rcvbadoff; /* packets received with bad offset */ 599 uint64_t tcps_rcvreassfull; /* packets dropped for no reass space */ 600 uint64_t tcps_rcvshort; /* packets received too short */ 601 uint64_t tcps_rcvduppack; /* duplicate-only packets received */ 602 uint64_t tcps_rcvdupbyte; /* duplicate-only bytes received */ 603 uint64_t tcps_rcvpartduppack; /* packets with some duplicate data */ 604 uint64_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 605 uint64_t tcps_rcvoopack; /* out-of-order packets received */ 606 uint64_t tcps_rcvoobyte; /* out-of-order bytes received */ 607 uint64_t tcps_rcvpackafterwin; /* packets with data after window */ 608 uint64_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ 609 uint64_t tcps_rcvafterclose; /* packets rcvd after "close" */ 610 uint64_t tcps_rcvwinprobe; /* rcvd window probe packets */ 611 uint64_t tcps_rcvdupack; /* rcvd duplicate acks */ 612 uint64_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 613 uint64_t tcps_rcvackpack; /* rcvd ack packets */ 614 uint64_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ 615 uint64_t tcps_rcvwinupd; /* rcvd window update packets */ 616 uint64_t tcps_pawsdrop; /* segments dropped due to PAWS */ 617 uint64_t tcps_predack; /* times hdr predict ok for acks */ 618 uint64_t tcps_preddat; /* times hdr predict ok for data pkts */ 619 uint64_t tcps_pcbcachemiss; 620 uint64_t tcps_cachedrtt; /* times cached RTT in route updated */ 621 uint64_t tcps_cachedrttvar; /* times cached rttvar updated */ 622 uint64_t tcps_cachedssthresh; /* times cached ssthresh updated */ 623 uint64_t tcps_usedrtt; /* times RTT initialized from route */ 624 uint64_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ 625 uint64_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ 626 uint64_t tcps_persistdrop; /* timeout in persist state */ 627 uint64_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ 628 uint64_t tcps_mturesent; /* resends due to MTU discovery */ 629 uint64_t tcps_listendrop; /* listen queue overflows */ 630 uint64_t tcps_badrst; /* ignored RSTs in the window */ 631 632 uint64_t tcps_sc_added; /* entry added to syncache */ 633 uint64_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ 634 uint64_t tcps_sc_dupsyn; /* duplicate SYN packet */ 635 uint64_t tcps_sc_dropped; /* could not reply to packet */ 636 uint64_t tcps_sc_completed; /* successful extraction of entry */ 637 uint64_t tcps_sc_bucketoverflow;/* syncache per-bucket limit hit */ 638 uint64_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ 639 uint64_t tcps_sc_reset; /* RST removed entry from syncache */ 640 uint64_t tcps_sc_stale; /* timed out or listen socket gone */ 641 uint64_t tcps_sc_aborted; /* syncache entry aborted */ 642 uint64_t tcps_sc_badack; /* removed due to bad ACK */ 643 uint64_t tcps_sc_unreach; /* ICMP unreachable received */ 644 uint64_t tcps_sc_zonefail; /* zalloc() failed */ 645 uint64_t tcps_sc_sendcookie; /* SYN cookie sent */ 646 uint64_t tcps_sc_recvcookie; /* SYN cookie received */ 647 648 uint64_t tcps_hc_added; /* entry added to hostcache */ 649 uint64_t tcps_hc_bucketoverflow;/* hostcache per bucket limit hit */ 650 651 uint64_t tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */ 652 653 /* SACK related stats */ 654 uint64_t tcps_sack_recovery_episode; /* SACK recovery episodes */ 655 uint64_t tcps_sack_rexmits; /* SACK rexmit segments */ 656 uint64_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ 657 uint64_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ 658 uint64_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ 659 uint64_t tcps_sack_sboverflow; /* times scoreboard overflowed */ 660 661 /* ECN related stats */ 662 uint64_t tcps_ecn_ce; /* ECN Congestion Experienced */ 663 uint64_t tcps_ecn_ect0; /* ECN Capable Transport */ 664 uint64_t tcps_ecn_ect1; /* ECN Capable Transport */ 665 uint64_t tcps_ecn_shs; /* ECN successful handshakes */ 666 uint64_t tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */ 667 668 /* TCP_SIGNATURE related stats */ 669 uint64_t tcps_sig_rcvgoodsig; /* Total matching signature received */ 670 uint64_t tcps_sig_rcvbadsig; /* Total bad signature received */ 671 uint64_t tcps_sig_err_buildsig; /* Failed to make signature */ 672 uint64_t tcps_sig_err_sigopt; /* No signature expected by socket */ 673 uint64_t tcps_sig_err_nosigopt; /* No signature provided by segment */ 674 675 /* Path MTU Discovery Black Hole Detection related stats */ 676 uint64_t tcps_pmtud_blackhole_activated; /* Black Hole Count */ 677 uint64_t tcps_pmtud_blackhole_activated_min_mss; /* BH at min MSS Count */ 678 uint64_t tcps_pmtud_blackhole_failed; /* Black Hole Failure Count */ 679 680 uint64_t _pad[12]; /* 6 UTO, 6 TBD */ 681 }; 682 683 #define tcps_rcvmemdrop tcps_rcvreassfull /* compat */ 684 685 #ifdef _KERNEL 686 #define TI_UNLOCKED 1 687 #define TI_RLOCKED 2 688 #include <sys/counter.h> 689 690 VNET_PCPUSTAT_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */ 691 /* 692 * In-kernel consumers can use these accessor macros directly to update 693 * stats. 694 */ 695 #define TCPSTAT_ADD(name, val) \ 696 VNET_PCPUSTAT_ADD(struct tcpstat, tcpstat, name, (val)) 697 #define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1) 698 699 /* 700 * Kernel module consumers must use this accessor macro. 701 */ 702 void kmod_tcpstat_add(int statnum, int val); 703 #define KMOD_TCPSTAT_ADD(name, val) \ 704 kmod_tcpstat_add(offsetof(struct tcpstat, name) / sizeof(uint64_t), val) 705 #define KMOD_TCPSTAT_INC(name) KMOD_TCPSTAT_ADD(name, 1) 706 707 /* 708 * Running TCP connection count by state. 709 */ 710 VNET_DECLARE(counter_u64_t, tcps_states[TCP_NSTATES]); 711 #define V_tcps_states VNET(tcps_states) 712 #define TCPSTATES_INC(state) counter_u64_add(V_tcps_states[state], 1) 713 #define TCPSTATES_DEC(state) counter_u64_add(V_tcps_states[state], -1) 714 715 /* 716 * TCP specific helper hook point identifiers. 717 */ 718 #define HHOOK_TCP_EST_IN 0 719 #define HHOOK_TCP_EST_OUT 1 720 #define HHOOK_TCP_LAST HHOOK_TCP_EST_OUT 721 722 struct tcp_hhook_data { 723 struct tcpcb *tp; 724 struct tcphdr *th; 725 struct tcpopt *to; 726 uint32_t len; 727 int tso; 728 tcp_seq curack; 729 }; 730 #ifdef TCP_HHOOK 731 void hhook_run_tcp_est_out(struct tcpcb *tp, 732 struct tcphdr *th, struct tcpopt *to, 733 uint32_t len, int tso); 734 #endif 735 #endif 736 737 /* 738 * TCB structure exported to user-land via sysctl(3). 739 * 740 * Fields prefixed with "xt_" are unique to the export structure, and fields 741 * with "t_" or other prefixes match corresponding fields of 'struct tcpcb'. 742 * 743 * Legend: 744 * (s) - used by userland utilities in src 745 * (p) - used by utilities in ports 746 * (3) - is known to be used by third party software not in ports 747 * (n) - no known usage 748 * 749 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been 750 * included. Not all of our clients do. 751 */ 752 #if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_) 753 struct xtcpcb { 754 ksize_t xt_len; /* length of this structure */ 755 struct xinpcb xt_inp; 756 char xt_stack[TCP_FUNCTION_NAME_LEN_MAX]; /* (s) */ 757 char xt_logid[TCP_LOG_ID_LEN]; /* (s) */ 758 char xt_cc[TCP_CA_NAME_MAX]; /* (s) */ 759 int64_t spare64[6]; 760 int32_t t_state; /* (s,p) */ 761 uint32_t t_flags; /* (s,p) */ 762 int32_t t_sndzerowin; /* (s) */ 763 int32_t t_sndrexmitpack; /* (s) */ 764 int32_t t_rcvoopack; /* (s) */ 765 int32_t t_rcvtime; /* (s) */ 766 int32_t tt_rexmt; /* (s) */ 767 int32_t tt_persist; /* (s) */ 768 int32_t tt_keep; /* (s) */ 769 int32_t tt_2msl; /* (s) */ 770 int32_t tt_delack; /* (s) */ 771 int32_t t_logstate; /* (3) */ 772 uint32_t t_snd_cwnd; /* (s) */ 773 uint32_t t_snd_ssthresh; /* (s) */ 774 uint32_t t_maxseg; /* (s) */ 775 uint32_t t_rcv_wnd; /* (s) */ 776 uint32_t t_snd_wnd; /* (s) */ 777 uint32_t xt_ecn; /* (s) */ 778 int32_t spare32[26]; 779 } __aligned(8); 780 781 #ifdef _KERNEL 782 void tcp_inptoxtp(const struct inpcb *, struct xtcpcb *); 783 #endif 784 #endif 785 786 /* 787 * TCP function information (name-to-id mapping, aliases, and refcnt) 788 * exported to user-land via sysctl(3). 789 */ 790 struct tcp_function_info { 791 uint32_t tfi_refcnt; 792 uint8_t tfi_id; 793 char tfi_name[TCP_FUNCTION_NAME_LEN_MAX]; 794 char tfi_alias[TCP_FUNCTION_NAME_LEN_MAX]; 795 }; 796 797 /* 798 * Identifiers for TCP sysctl nodes 799 */ 800 #define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */ 801 #define TCPCTL_MSSDFLT 3 /* MSS default */ 802 #define TCPCTL_STATS 4 /* statistics */ 803 #define TCPCTL_RTTDFLT 5 /* default RTT estimate */ 804 #define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */ 805 #define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */ 806 #define TCPCTL_SENDSPACE 8 /* send buffer space */ 807 #define TCPCTL_RECVSPACE 9 /* receive buffer space */ 808 #define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */ 809 #define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */ 810 #define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */ 811 #define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */ 812 #define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */ 813 #define TCPCTL_DROP 15 /* drop tcp connection */ 814 #define TCPCTL_STATES 16 /* connection counts by TCP state */ 815 816 #ifdef _KERNEL 817 #ifdef SYSCTL_DECL 818 SYSCTL_DECL(_net_inet_tcp); 819 SYSCTL_DECL(_net_inet_tcp_sack); 820 MALLOC_DECLARE(M_TCPLOG); 821 #endif 822 823 VNET_DECLARE(int, tcp_log_in_vain); 824 #define V_tcp_log_in_vain VNET(tcp_log_in_vain) 825 826 /* 827 * Global TCP tunables shared between different stacks. 828 * Please keep the list sorted. 829 */ 830 VNET_DECLARE(int, drop_synfin); 831 VNET_DECLARE(int, path_mtu_discovery); 832 VNET_DECLARE(int, tcp_abc_l_var); 833 VNET_DECLARE(int, tcp_autorcvbuf_max); 834 VNET_DECLARE(int, tcp_autosndbuf_inc); 835 VNET_DECLARE(int, tcp_autosndbuf_max); 836 VNET_DECLARE(int, tcp_delack_enabled); 837 VNET_DECLARE(int, tcp_do_autorcvbuf); 838 VNET_DECLARE(int, tcp_do_autosndbuf); 839 VNET_DECLARE(int, tcp_do_ecn); 840 VNET_DECLARE(int, tcp_do_newcwv); 841 VNET_DECLARE(int, tcp_do_rfc1323); 842 VNET_DECLARE(int, tcp_do_rfc3042); 843 VNET_DECLARE(int, tcp_do_rfc3390); 844 VNET_DECLARE(int, tcp_do_rfc3465); 845 VNET_DECLARE(int, tcp_do_rfc6675_pipe); 846 VNET_DECLARE(int, tcp_do_sack); 847 VNET_DECLARE(int, tcp_do_tso); 848 VNET_DECLARE(int, tcp_ecn_maxretries); 849 VNET_DECLARE(int, tcp_initcwnd_segments); 850 VNET_DECLARE(int, tcp_insecure_rst); 851 VNET_DECLARE(int, tcp_insecure_syn); 852 VNET_DECLARE(uint32_t, tcp_map_entries_limit); 853 VNET_DECLARE(uint32_t, tcp_map_split_limit); 854 VNET_DECLARE(int, tcp_minmss); 855 VNET_DECLARE(int, tcp_mssdflt); 856 #ifdef STATS 857 VNET_DECLARE(int, tcp_perconn_stats_dflt_tpl); 858 VNET_DECLARE(int, tcp_perconn_stats_enable); 859 #endif /* STATS */ 860 VNET_DECLARE(int, tcp_recvspace); 861 VNET_DECLARE(int, tcp_sack_globalholes); 862 VNET_DECLARE(int, tcp_sack_globalmaxholes); 863 VNET_DECLARE(int, tcp_sack_maxholes); 864 VNET_DECLARE(int, tcp_sc_rst_sock_fail); 865 VNET_DECLARE(int, tcp_sendspace); 866 VNET_DECLARE(struct inpcbhead, tcb); 867 VNET_DECLARE(struct inpcbinfo, tcbinfo); 868 869 #define V_tcp_do_newcwv VNET(tcp_do_newcwv) 870 #define V_drop_synfin VNET(drop_synfin) 871 #define V_path_mtu_discovery VNET(path_mtu_discovery) 872 #define V_tcb VNET(tcb) 873 #define V_tcbinfo VNET(tcbinfo) 874 #define V_tcp_abc_l_var VNET(tcp_abc_l_var) 875 #define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max) 876 #define V_tcp_autosndbuf_inc VNET(tcp_autosndbuf_inc) 877 #define V_tcp_autosndbuf_max VNET(tcp_autosndbuf_max) 878 #define V_tcp_delack_enabled VNET(tcp_delack_enabled) 879 #define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf) 880 #define V_tcp_do_autosndbuf VNET(tcp_do_autosndbuf) 881 #define V_tcp_do_ecn VNET(tcp_do_ecn) 882 #define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323) 883 #define V_tcp_ts_offset_per_conn VNET(tcp_ts_offset_per_conn) 884 #define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042) 885 #define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390) 886 #define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465) 887 #define V_tcp_do_rfc6675_pipe VNET(tcp_do_rfc6675_pipe) 888 #define V_tcp_do_sack VNET(tcp_do_sack) 889 #define V_tcp_do_tso VNET(tcp_do_tso) 890 #define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries) 891 #define V_tcp_initcwnd_segments VNET(tcp_initcwnd_segments) 892 #define V_tcp_insecure_rst VNET(tcp_insecure_rst) 893 #define V_tcp_insecure_syn VNET(tcp_insecure_syn) 894 #define V_tcp_map_entries_limit VNET(tcp_map_entries_limit) 895 #define V_tcp_map_split_limit VNET(tcp_map_split_limit) 896 #define V_tcp_minmss VNET(tcp_minmss) 897 #define V_tcp_mssdflt VNET(tcp_mssdflt) 898 #ifdef STATS 899 #define V_tcp_perconn_stats_dflt_tpl VNET(tcp_perconn_stats_dflt_tpl) 900 #define V_tcp_perconn_stats_enable VNET(tcp_perconn_stats_enable) 901 #endif /* STATS */ 902 #define V_tcp_recvspace VNET(tcp_recvspace) 903 #define V_tcp_sack_globalholes VNET(tcp_sack_globalholes) 904 #define V_tcp_sack_globalmaxholes VNET(tcp_sack_globalmaxholes) 905 #define V_tcp_sack_maxholes VNET(tcp_sack_maxholes) 906 #define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail) 907 #define V_tcp_sendspace VNET(tcp_sendspace) 908 #define V_tcp_udp_tunneling_overhead VNET(tcp_udp_tunneling_overhead) 909 #define V_tcp_udp_tunneling_port VNET(tcp_udp_tunneling_port) 910 911 #ifdef TCP_HHOOK 912 VNET_DECLARE(struct hhook_head *, tcp_hhh[HHOOK_TCP_LAST + 1]); 913 #define V_tcp_hhh VNET(tcp_hhh) 914 #endif 915 916 int tcp_addoptions(struct tcpopt *, u_char *); 917 int tcp_ccalgounload(struct cc_algo *unload_algo); 918 struct tcpcb * 919 tcp_close(struct tcpcb *); 920 void tcp_discardcb(struct tcpcb *); 921 void tcp_twstart(struct tcpcb *); 922 void tcp_twclose(struct tcptw *, int); 923 void tcp_ctlinput(int, struct sockaddr *, void *); 924 int tcp_ctloutput(struct socket *, struct sockopt *); 925 struct tcpcb * 926 tcp_drop(struct tcpcb *, int); 927 void tcp_drain(void); 928 void tcp_init(void); 929 void tcp_fini(void *); 930 char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *, 931 const void *); 932 char *tcp_log_vain(struct in_conninfo *, struct tcphdr *, void *, 933 const void *); 934 int tcp_reass(struct tcpcb *, struct tcphdr *, tcp_seq *, int *, struct mbuf *); 935 void tcp_reass_global_init(void); 936 void tcp_reass_flush(struct tcpcb *); 937 void tcp_dooptions(struct tcpopt *, u_char *, int, int); 938 void tcp_dropwithreset(struct mbuf *, struct tcphdr *, 939 struct tcpcb *, int, int); 940 void tcp_pulloutofband(struct socket *, 941 struct tcphdr *, struct mbuf *, int); 942 void tcp_xmit_timer(struct tcpcb *, int); 943 void tcp_newreno_partial_ack(struct tcpcb *, struct tcphdr *); 944 void cc_ack_received(struct tcpcb *tp, struct tcphdr *th, 945 uint16_t nsegs, uint16_t type); 946 void cc_conn_init(struct tcpcb *tp); 947 void cc_post_recovery(struct tcpcb *tp, struct tcphdr *th); 948 void cc_ecnpkt_handler(struct tcpcb *tp, struct tcphdr *th, uint8_t iptos); 949 void cc_cong_signal(struct tcpcb *tp, struct tcphdr *th, uint32_t type); 950 #ifdef TCP_HHOOK 951 void hhook_run_tcp_est_in(struct tcpcb *tp, 952 struct tcphdr *th, struct tcpopt *to); 953 #endif 954 955 int tcp_input(struct mbuf **, int *, int); 956 int tcp_autorcvbuf(struct mbuf *, struct tcphdr *, struct socket *, 957 struct tcpcb *, int); 958 void tcp_do_segment(struct mbuf *, struct tcphdr *, 959 struct socket *, struct tcpcb *, int, int, uint8_t); 960 961 int register_tcp_functions(struct tcp_function_block *blk, int wait); 962 int register_tcp_functions_as_names(struct tcp_function_block *blk, 963 int wait, const char *names[], int *num_names); 964 int register_tcp_functions_as_name(struct tcp_function_block *blk, 965 const char *name, int wait); 966 int deregister_tcp_functions(struct tcp_function_block *blk, bool quiesce, 967 bool force); 968 struct tcp_function_block *find_and_ref_tcp_functions(struct tcp_function_set *fs); 969 void tcp_switch_back_to_default(struct tcpcb *tp); 970 struct tcp_function_block * 971 find_and_ref_tcp_fb(struct tcp_function_block *fs); 972 int tcp_default_ctloutput(struct socket *so, struct sockopt *sopt, struct inpcb *inp, struct tcpcb *tp); 973 974 extern counter_u64_t tcp_inp_lro_direct_queue; 975 extern counter_u64_t tcp_inp_lro_wokeup_queue; 976 extern counter_u64_t tcp_inp_lro_compressed; 977 extern counter_u64_t tcp_inp_lro_single_push; 978 extern counter_u64_t tcp_inp_lro_locks_taken; 979 extern counter_u64_t tcp_inp_lro_sack_wake; 980 981 #ifdef NETFLIX_EXP_DETECTION 982 /* Various SACK attack thresholds */ 983 extern int32_t tcp_force_detection; 984 extern int32_t tcp_sack_to_ack_thresh; 985 extern int32_t tcp_sack_to_move_thresh; 986 extern int32_t tcp_restoral_thresh; 987 extern int32_t tcp_sad_decay_val; 988 extern int32_t tcp_sad_pacing_interval; 989 extern int32_t tcp_sad_low_pps; 990 extern int32_t tcp_map_minimum; 991 extern int32_t tcp_attack_on_turns_on_logging; 992 #endif 993 994 uint32_t tcp_maxmtu(struct in_conninfo *, struct tcp_ifcap *); 995 uint32_t tcp_maxmtu6(struct in_conninfo *, struct tcp_ifcap *); 996 u_int tcp_maxseg(const struct tcpcb *); 997 void tcp_mss_update(struct tcpcb *, int, int, struct hc_metrics_lite *, 998 struct tcp_ifcap *); 999 void tcp_mss(struct tcpcb *, int); 1000 int tcp_mssopt(struct in_conninfo *); 1001 struct inpcb * 1002 tcp_drop_syn_sent(struct inpcb *, int); 1003 struct tcpcb * 1004 tcp_newtcpcb(struct inpcb *); 1005 int tcp_output(struct tcpcb *); 1006 void tcp_state_change(struct tcpcb *, int); 1007 void tcp_respond(struct tcpcb *, void *, 1008 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); 1009 void tcp_tw_init(void); 1010 #ifdef VIMAGE 1011 void tcp_tw_destroy(void); 1012 #endif 1013 void tcp_tw_zone_change(void); 1014 int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *, 1015 struct mbuf *, int); 1016 void tcp_setpersist(struct tcpcb *); 1017 void tcp_slowtimo(void); 1018 struct tcptemp * 1019 tcpip_maketemplate(struct inpcb *); 1020 void tcpip_fillheaders(struct inpcb *, void *, void *); 1021 void tcp_timer_activate(struct tcpcb *, uint32_t, u_int); 1022 int tcp_timer_suspend(struct tcpcb *, uint32_t); 1023 void tcp_timers_unsuspend(struct tcpcb *, uint32_t); 1024 int tcp_timer_active(struct tcpcb *, uint32_t); 1025 void tcp_timer_stop(struct tcpcb *, uint32_t); 1026 void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int); 1027 int inp_to_cpuid(struct inpcb *inp); 1028 /* 1029 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo) 1030 */ 1031 void tcp_hc_init(void); 1032 #ifdef VIMAGE 1033 void tcp_hc_destroy(void); 1034 #endif 1035 void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *); 1036 uint32_t tcp_hc_getmtu(struct in_conninfo *); 1037 void tcp_hc_updatemtu(struct in_conninfo *, uint32_t); 1038 void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *); 1039 1040 extern struct pr_usrreqs tcp_usrreqs; 1041 1042 uint32_t tcp_new_ts_offset(struct in_conninfo *); 1043 tcp_seq tcp_new_isn(struct in_conninfo *); 1044 1045 int tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); 1046 void tcp_update_dsack_list(struct tcpcb *, tcp_seq, tcp_seq); 1047 void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); 1048 void tcp_clean_dsack_blocks(struct tcpcb *tp); 1049 void tcp_clean_sackreport(struct tcpcb *tp); 1050 void tcp_sack_adjust(struct tcpcb *tp); 1051 struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); 1052 void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); 1053 void tcp_free_sackholes(struct tcpcb *tp); 1054 int tcp_newreno(struct tcpcb *, struct tcphdr *); 1055 int tcp_compute_pipe(struct tcpcb *); 1056 uint32_t tcp_compute_initwnd(uint32_t); 1057 void tcp_sndbuf_autoscale(struct tcpcb *, struct socket *, uint32_t); 1058 int tcp_stats_sample_rollthedice(struct tcpcb *tp, void *seed_bytes, 1059 size_t seed_len); 1060 struct mbuf * 1061 tcp_m_copym(struct mbuf *m, int32_t off0, int32_t *plen, 1062 int32_t seglimit, int32_t segsize, struct sockbuf *sb, bool hw_tls); 1063 1064 int tcp_stats_init(void); 1065 void tcp_log_end_status(struct tcpcb *tp, uint8_t status); 1066 1067 static inline void 1068 tcp_fields_to_host(struct tcphdr *th) 1069 { 1070 1071 th->th_seq = ntohl(th->th_seq); 1072 th->th_ack = ntohl(th->th_ack); 1073 th->th_win = ntohs(th->th_win); 1074 th->th_urp = ntohs(th->th_urp); 1075 } 1076 1077 static inline void 1078 tcp_fields_to_net(struct tcphdr *th) 1079 { 1080 1081 th->th_seq = htonl(th->th_seq); 1082 th->th_ack = htonl(th->th_ack); 1083 th->th_win = htons(th->th_win); 1084 th->th_urp = htons(th->th_urp); 1085 } 1086 #endif /* _KERNEL */ 1087 1088 #endif /* _NETINET_TCP_VAR_H_ */ 1089