tcp_var.h revision 1.34
1 /* $NetBSD: tcp_var.h,v 1.34 1998/02/10 01:27:19 perry Exp $ */ 2 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. All advertising materials mentioning features or use of this software 16 * must display the following acknowledgement: 17 * This product includes software developed by the University of 18 * California, Berkeley and its contributors. 19 * 4. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95 36 */ 37 38 #ifndef _NETINET_TCP_VAR_H_ 39 #define _NETINET_TCP_VAR_H_ 40 41 /* 42 * Kernel variables for tcp. 43 */ 44 45 /* 46 * Tcp control block, one per tcp; fields: 47 */ 48 struct tcpcb { 49 struct ipqehead segq; /* sequencing queue */ 50 short t_state; /* state of this connection */ 51 short t_timer[TCPT_NTIMERS]; /* tcp timers */ 52 short t_rxtshift; /* log(2) of rexmt exp. backoff */ 53 short t_rxtcur; /* current retransmit value */ 54 short t_dupacks; /* consecutive dup acks recd */ 55 u_short t_peermss; /* peer's maximum segment size */ 56 u_short t_ourmss; /* our's maximum segment size */ 57 u_short t_segsz; /* current segment size in use */ 58 char t_force; /* 1 if forcing out a byte */ 59 u_short t_flags; 60 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 61 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 62 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 63 #define TF_NOOPT 0x0008 /* don't use tcp options */ 64 #define TF_SENTFIN 0x0010 /* have sent FIN */ 65 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 66 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 67 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 68 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 69 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 70 71 struct tcpiphdr *t_template; /* skeletal packet for transmit */ 72 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 73 LIST_ENTRY(tcpcb) t_delack; /* delayed ACK queue */ 74 /* 75 * The following fields are used as in the protocol specification. 76 * See RFC783, Dec. 1981, page 21. 77 */ 78 /* send sequence variables */ 79 tcp_seq snd_una; /* send unacknowledged */ 80 tcp_seq snd_nxt; /* send next */ 81 tcp_seq snd_up; /* send urgent pointer */ 82 tcp_seq snd_wl1; /* window update seg seq number */ 83 tcp_seq snd_wl2; /* window update seg ack number */ 84 tcp_seq iss; /* initial send sequence number */ 85 u_long snd_wnd; /* send window */ 86 /* receive sequence variables */ 87 u_long rcv_wnd; /* receive window */ 88 tcp_seq rcv_nxt; /* receive next */ 89 tcp_seq rcv_up; /* receive urgent pointer */ 90 tcp_seq irs; /* initial receive sequence number */ 91 /* 92 * Additional variables for this implementation. 93 */ 94 /* receive variables */ 95 tcp_seq rcv_adv; /* advertised window */ 96 /* retransmit variables */ 97 tcp_seq snd_max; /* highest sequence number sent; 98 * used to recognize retransmits 99 */ 100 /* congestion control (for slow start, source quench, retransmit after loss) */ 101 u_long snd_cwnd; /* congestion-controlled window */ 102 u_long snd_ssthresh; /* snd_cwnd size threshhold for 103 * for slow start exponential to 104 * linear switch 105 */ 106 /* 107 * transmit timing stuff. See below for scale of srtt and rttvar. 108 * "Variance" is actually smoothed difference. 109 */ 110 short t_idle; /* inactivity time */ 111 short t_rtt; /* round trip time */ 112 tcp_seq t_rtseq; /* sequence number being timed */ 113 short t_srtt; /* smoothed round-trip time */ 114 short t_rttvar; /* variance in round-trip time */ 115 short t_rttmin; /* minimum rtt allowed */ 116 u_long max_sndwnd; /* largest window peer has offered */ 117 118 /* out-of-band data */ 119 char t_oobflags; /* have some */ 120 char t_iobc; /* input character */ 121 #define TCPOOB_HAVEDATA 0x01 122 #define TCPOOB_HADDATA 0x02 123 short t_softerror; /* possible error not yet reported */ 124 125 /* RFC 1323 variables */ 126 u_char snd_scale; /* window scaling for send window */ 127 u_char rcv_scale; /* window scaling for recv window */ 128 u_char request_r_scale; /* pending window scaling */ 129 u_char requested_s_scale; 130 u_int32_t ts_recent; /* timestamp echo data */ 131 u_int32_t ts_recent_age; /* when last updated */ 132 tcp_seq last_ack_sent; 133 134 /* TUBA stuff */ 135 caddr_t t_tuba_pcb; /* next level down pcb for TCP over z */ 136 }; 137 138 /* 139 * Queue for delayed ACK processing. 140 */ 141 LIST_HEAD(tcp_delack_head, tcpcb); 142 #ifdef _KERNEL 143 extern struct tcp_delack_head tcp_delacks; 144 145 #define TCP_SET_DELACK(tp) \ 146 do { \ 147 if (((tp)->t_flags & TF_DELACK) == 0) { \ 148 (tp)->t_flags |= TF_DELACK; \ 149 LIST_INSERT_HEAD(&tcp_delacks, (tp), t_delack); \ 150 } \ 151 } while (0) 152 153 #define TCP_CLEAR_DELACK(tp) \ 154 do { \ 155 if ((tp)->t_flags & TF_DELACK) { \ 156 (tp)->t_flags &= ~TF_DELACK; \ 157 LIST_REMOVE((tp), t_delack); \ 158 } \ 159 } while (0) 160 #endif /* _KERNEL */ 161 162 /* 163 * Handy way of passing around TCP option info. 164 */ 165 struct tcp_opt_info { 166 int ts_present; 167 u_int32_t ts_val; 168 u_int32_t ts_ecr; 169 u_int16_t maxseg; 170 }; 171 172 /* 173 * This structure should not exceed 32 bytes. 174 * XXX On the Alpha, it's already 36-bytes, which rounds to 40. 175 * XXX Need to eliminate the pointer. 176 * 177 * XXX We've blown 32 bytes on non-Alpha systems, too, since we're 178 * XXX storing the maxseg we advertised to the peer. Should we 179 * XXX create another malloc bucket? Should we care? 180 */ 181 struct syn_cache { 182 struct syn_cache *sc_next; 183 u_int32_t sc_tstmp : 1, 184 sc_hash : 31; 185 struct in_addr sc_src; 186 struct in_addr sc_dst; 187 tcp_seq sc_irs; 188 tcp_seq sc_iss; 189 u_int16_t sc_sport; 190 u_int16_t sc_dport; 191 u_int16_t sc_peermaxseg; 192 u_int16_t sc_ourmaxseg; 193 u_int8_t sc_timer; 194 u_int8_t sc_request_r_scale : 4, 195 sc_requested_s_scale : 4; 196 }; 197 198 struct syn_cache_head { 199 struct syn_cache *sch_first; /* First entry in the bucket */ 200 struct syn_cache *sch_last; /* Last entry in the bucket */ 201 struct syn_cache_head *sch_headq; /* The next non-empty bucket */ 202 int16_t sch_timer_sum; /* Total time in this bucket */ 203 u_int16_t sch_length; /* @ # elements in bucket */ 204 }; 205 206 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 207 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 208 209 /* 210 * The smoothed round-trip time and estimated variance 211 * are stored as fixed point numbers scaled by the values below. 212 * For convenience, these scales are also used in smoothing the average 213 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 214 * With these scales, srtt has 3 bits to the right of the binary point, 215 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 216 * binary point, and is smoothed with an ALPHA of 0.75. 217 */ 218 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 219 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 220 221 /* 222 * The initial retransmission should happen at rtt + 4 * rttvar. 223 * Because of the way we do the smoothing, srtt and rttvar 224 * will each average +1/2 tick of bias. When we compute 225 * the retransmit timer, we want 1/2 tick of rounding and 226 * 1 extra tick because of +-1/2 tick uncertainty in the 227 * firing of the timer. The bias will give us exactly the 228 * 1.5 tick we need. But, because the bias is 229 * statistical, we have to test that we don't drop below 230 * the minimum feasible timer (which is 2 ticks). 231 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 232 * is the same as the multiplier for rttvar. 233 */ 234 #define TCP_REXMTVAL(tp) \ 235 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 236 237 #ifdef _KERNEL 238 /* 239 * Compute the initial window for slow start. 240 */ 241 extern int tcp_init_win; 242 #define TCP_INITIAL_WINDOW(segsz) \ 243 ((tcp_init_win == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 244 ((segsz) * tcp_init_win)) 245 #endif /* _KERNEL */ 246 247 /* 248 * TCP statistics. 249 * Many of these should be kept per connection, 250 * but that's inconvenient at the moment. 251 */ 252 struct tcpstat { 253 u_long tcps_connattempt; /* connections initiated */ 254 u_long tcps_accepts; /* connections accepted */ 255 u_long tcps_connects; /* connections established */ 256 u_long tcps_drops; /* connections dropped */ 257 u_long tcps_conndrops; /* embryonic connections dropped */ 258 u_long tcps_closed; /* conn. closed (includes drops) */ 259 u_long tcps_segstimed; /* segs where we tried to get rtt */ 260 u_long tcps_rttupdated; /* times we succeeded */ 261 u_long tcps_delack; /* delayed acks sent */ 262 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 263 u_long tcps_rexmttimeo; /* retransmit timeouts */ 264 u_long tcps_persisttimeo; /* persist timeouts */ 265 u_long tcps_keeptimeo; /* keepalive timeouts */ 266 u_long tcps_keepprobe; /* keepalive probes sent */ 267 u_long tcps_keepdrops; /* connections dropped in keepalive */ 268 u_long tcps_persistdrops; /* connections dropped in persist */ 269 u_long tcps_connsdrained; /* connections drained due to memory 270 shortage */ 271 272 u_long tcps_sndtotal; /* total packets sent */ 273 u_long tcps_sndpack; /* data packets sent */ 274 u_long tcps_sndbyte; /* data bytes sent */ 275 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 276 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 277 u_long tcps_sndacks; /* ack-only packets sent */ 278 u_long tcps_sndprobe; /* window probes sent */ 279 u_long tcps_sndurg; /* packets sent with URG only */ 280 u_long tcps_sndwinup; /* window update-only packets sent */ 281 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 282 283 u_long tcps_rcvtotal; /* total packets received */ 284 u_long tcps_rcvpack; /* packets received in sequence */ 285 u_long tcps_rcvbyte; /* bytes received in sequence */ 286 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 287 u_long tcps_rcvbadoff; /* packets received with bad offset */ 288 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 289 u_long tcps_rcvshort; /* packets received too short */ 290 u_long tcps_rcvduppack; /* duplicate-only packets received */ 291 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 292 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 293 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 294 u_long tcps_rcvoopack; /* out-of-order packets received */ 295 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 296 u_long tcps_rcvpackafterwin; /* packets with data after window */ 297 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 298 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 299 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 300 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 301 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 302 u_long tcps_rcvackpack; /* rcvd ack packets */ 303 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 304 u_long tcps_rcvwinupd; /* rcvd window update packets */ 305 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 306 u_long tcps_predack; /* times hdr predict ok for acks */ 307 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 308 309 u_long tcps_pcbhashmiss; /* input packets missing pcb hash */ 310 u_long tcps_noport; /* no socket on port */ 311 u_long tcps_badsyn; /* received ack for which we have 312 no SYN in compressed state */ 313 314 /* These statistics deal with the SYN cache. */ 315 u_long tcps_sc_added; /* # of entries added */ 316 u_long tcps_sc_completed; /* # of connections completed */ 317 u_long tcps_sc_timed_out; /* # of entries timed out */ 318 u_long tcps_sc_overflowed; /* # dropped due to overflow */ 319 u_long tcps_sc_reset; /* # dropped due to RST */ 320 u_long tcps_sc_unreach; /* # dropped due to ICMP unreach */ 321 u_long tcps_sc_bucketoverflow; /* # dropped due to bucket overflow */ 322 u_long tcps_sc_aborted; /* # of entries aborted (no mem) */ 323 u_long tcps_sc_dupesyn; /* # of duplicate SYNs received */ 324 u_long tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 325 u_long tcps_sc_collisions; /* # of hash collisions */ 326 }; 327 328 /* 329 * Names for TCP sysctl objects. 330 */ 331 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 332 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 333 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 334 #define TCPCTL_MSSDFLT 4 /* default seg size */ 335 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 336 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 337 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 338 #define TCPCTL_INIT_WIN 8 /* initial window */ 339 #define TCPCTL_MAXID 9 340 341 #define TCPCTL_NAMES { \ 342 { 0, 0 }, \ 343 { "rfc1323", CTLTYPE_INT }, \ 344 { "sendspace", CTLTYPE_INT }, \ 345 { "recvspace", CTLTYPE_INT }, \ 346 { "mssdflt", CTLTYPE_INT }, \ 347 { "syn_cache_limit", CTLTYPE_INT }, \ 348 { "syn_bucket_limit", CTLTYPE_INT }, \ 349 { "syn_cache_interval", CTLTYPE_INT },\ 350 { "init_win", CTLTYPE_INT }, \ 351 } 352 353 #ifdef _KERNEL 354 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 355 struct tcpstat tcpstat; /* tcp statistics */ 356 u_int32_t tcp_now; /* for RFC 1323 timestamps */ 357 extern int tcp_do_rfc1323; /* enabled/disabled? */ 358 extern int tcp_mssdflt; /* default seg size */ 359 extern int tcp_init_win; /* initial window */ 360 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 361 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 362 extern int tcp_syn_cache_interval; /* compressed state timer */ 363 364 extern int tcp_syn_cache_size; 365 extern int tcp_syn_cache_timeo; 366 extern struct syn_cache_head tcp_syn_cache[], *tcp_syn_cache_first; 367 extern u_long syn_cache_count; 368 369 int tcp_attach __P((struct socket *)); 370 void tcp_canceltimers __P((struct tcpcb *)); 371 struct tcpcb * 372 tcp_close __P((struct tcpcb *)); 373 void *tcp_ctlinput __P((int, struct sockaddr *, void *)); 374 int tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **)); 375 struct tcpcb * 376 tcp_disconnect __P((struct tcpcb *)); 377 struct tcpcb * 378 tcp_drop __P((struct tcpcb *, int)); 379 void tcp_dooptions __P((struct tcpcb *, 380 u_char *, int, struct tcpiphdr *, struct tcp_opt_info *)); 381 void tcp_drain __P((void)); 382 void tcp_established __P((struct tcpcb *)); 383 void tcp_fasttimo __P((void)); 384 void tcp_init __P((void)); 385 void tcp_input __P((struct mbuf *, ...)); 386 int tcp_mss_to_advertise __P((const struct tcpcb *)); 387 void tcp_mss_from_peer __P((struct tcpcb *, int)); 388 void tcp_mtudisc __P((struct inpcb *, int)); 389 struct tcpcb * 390 tcp_newtcpcb __P((struct inpcb *)); 391 void tcp_notify __P((struct inpcb *, int)); 392 int tcp_output __P((struct tcpcb *)); 393 void tcp_pulloutofband __P((struct socket *, 394 struct tcpiphdr *, struct mbuf *)); 395 void tcp_quench __P((struct inpcb *, int)); 396 int tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *)); 397 int tcp_respond __P((struct tcpcb *, 398 struct tcpiphdr *, struct mbuf *, tcp_seq, tcp_seq, int)); 399 void tcp_rmx_rtt __P((struct tcpcb *)); 400 void tcp_setpersist __P((struct tcpcb *)); 401 void tcp_slowtimo __P((void)); 402 struct tcpiphdr * 403 tcp_template __P((struct tcpcb *)); 404 struct tcpcb * 405 tcp_timers __P((struct tcpcb *, int)); 406 void tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int)); 407 struct tcpcb * 408 tcp_usrclosed __P((struct tcpcb *)); 409 int tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t)); 410 int tcp_usrreq __P((struct socket *, 411 int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *)); 412 void tcp_xmit_timer __P((struct tcpcb *, int)); 413 tcp_seq tcp_new_iss __P((void *, u_long, tcp_seq)); 414 415 int syn_cache_add __P((struct socket *, struct mbuf *, u_char *, 416 int, struct tcp_opt_info *)); 417 void syn_cache_unreach __P((struct ip *, struct tcphdr *)); 418 struct socket * 419 syn_cache_get __P((struct socket *so, struct mbuf *)); 420 void syn_cache_insert __P((struct syn_cache *, struct syn_cache ***, 421 struct syn_cache_head **)); 422 struct syn_cache * 423 syn_cache_lookup __P((struct tcpiphdr *, struct syn_cache ***, 424 struct syn_cache_head **)); 425 void syn_cache_reset __P((struct tcpiphdr *)); 426 int syn_cache_respond __P((struct syn_cache *, struct mbuf *, 427 register struct tcpiphdr *, long, u_long)); 428 void syn_cache_timer __P((int)); 429 430 #endif 431 432 #endif /* _NETINET_TCP_VAR_H_ */ 433
Indexes created Wed Sep 24 22:09:59 GMT 2025