1 /* $NetBSD: tcp_var.h,v 1.33 1998/01/05 10:32:14 thorpej 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 /* 39 * Kernel variables for tcp. 40 */ 41 42 /* 43 * Tcp control block, one per tcp; fields: 44 */ 45 struct tcpcb { 46 struct ipqehead segq; /* sequencing queue */ 47 short t_state; /* state of this connection */ 48 short t_timer[TCPT_NTIMERS]; /* tcp timers */ 49 short t_rxtshift; /* log(2) of rexmt exp. backoff */ 50 short t_rxtcur; /* current retransmit value */ 51 short t_dupacks; /* consecutive dup acks recd */ 52 u_short t_peermss; /* peer's maximum segment size */ 53 u_short t_ourmss; /* our's maximum segment size */ 54 u_short t_segsz; /* current segment size in use */ 55 char t_force; /* 1 if forcing out a byte */ 56 u_short t_flags; 57 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 58 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 59 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 60 #define TF_NOOPT 0x0008 /* don't use tcp options */ 61 #define TF_SENTFIN 0x0010 /* have sent FIN */ 62 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 63 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 64 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 65 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 66 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 67 68 struct tcpiphdr *t_template; /* skeletal packet for transmit */ 69 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 70 LIST_ENTRY(tcpcb) t_delack; /* delayed ACK queue */ 71 /* 72 * The following fields are used as in the protocol specification. 73 * See RFC783, Dec. 1981, page 21. 74 */ 75 /* send sequence variables */ 76 tcp_seq snd_una; /* send unacknowledged */ 77 tcp_seq snd_nxt; /* send next */ 78 tcp_seq snd_up; /* send urgent pointer */ 79 tcp_seq snd_wl1; /* window update seg seq number */ 80 tcp_seq snd_wl2; /* window update seg ack number */ 81 tcp_seq iss; /* initial send sequence number */ 82 u_long snd_wnd; /* send window */ 83 /* receive sequence variables */ 84 u_long rcv_wnd; /* receive window */ 85 tcp_seq rcv_nxt; /* receive next */ 86 tcp_seq rcv_up; /* receive urgent pointer */ 87 tcp_seq irs; /* initial receive sequence number */ 88 /* 89 * Additional variables for this implementation. 90 */ 91 /* receive variables */ 92 tcp_seq rcv_adv; /* advertised window */ 93 /* retransmit variables */ 94 tcp_seq snd_max; /* highest sequence number sent; 95 * used to recognize retransmits 96 */ 97 /* congestion control (for slow start, source quench, retransmit after loss) */ 98 u_long snd_cwnd; /* congestion-controlled window */ 99 u_long snd_ssthresh; /* snd_cwnd size threshhold for 100 * for slow start exponential to 101 * linear switch 102 */ 103 /* 104 * transmit timing stuff. See below for scale of srtt and rttvar. 105 * "Variance" is actually smoothed difference. 106 */ 107 short t_idle; /* inactivity time */ 108 short t_rtt; /* round trip time */ 109 tcp_seq t_rtseq; /* sequence number being timed */ 110 short t_srtt; /* smoothed round-trip time */ 111 short t_rttvar; /* variance in round-trip time */ 112 short t_rttmin; /* minimum rtt allowed */ 113 u_long max_sndwnd; /* largest window peer has offered */ 114 115 /* out-of-band data */ 116 char t_oobflags; /* have some */ 117 char t_iobc; /* input character */ 118 #define TCPOOB_HAVEDATA 0x01 119 #define TCPOOB_HADDATA 0x02 120 short t_softerror; /* possible error not yet reported */ 121 122 /* RFC 1323 variables */ 123 u_char snd_scale; /* window scaling for send window */ 124 u_char rcv_scale; /* window scaling for recv window */ 125 u_char request_r_scale; /* pending window scaling */ 126 u_char requested_s_scale; 127 u_int32_t ts_recent; /* timestamp echo data */ 128 u_int32_t ts_recent_age; /* when last updated */ 129 tcp_seq last_ack_sent; 130 131 /* TUBA stuff */ 132 caddr_t t_tuba_pcb; /* next level down pcb for TCP over z */ 133 }; 134 135 /* 136 * Queue for delayed ACK processing. 137 */ 138 LIST_HEAD(tcp_delack_head, tcpcb); 139 #ifdef _KERNEL 140 extern struct tcp_delack_head tcp_delacks; 141 142 #define TCP_SET_DELACK(tp) \ 143 do { \ 144 if (((tp)->t_flags & TF_DELACK) == 0) { \ 145 (tp)->t_flags |= TF_DELACK; \ 146 LIST_INSERT_HEAD(&tcp_delacks, (tp), t_delack); \ 147 } \ 148 } while (0) 149 150 #define TCP_CLEAR_DELACK(tp) \ 151 do { \ 152 if ((tp)->t_flags & TF_DELACK) { \ 153 (tp)->t_flags &= ~TF_DELACK; \ 154 LIST_REMOVE((tp), t_delack); \ 155 } \ 156 } while (0) 157 #endif /* _KERNEL */ 158 159 /* 160 * Handy way of passing around TCP option info. 161 */ 162 struct tcp_opt_info { 163 int ts_present; 164 u_int32_t ts_val; 165 u_int32_t ts_ecr; 166 u_int16_t maxseg; 167 }; 168 169 /* 170 * This structure should not exceed 32 bytes. 171 * XXX On the Alpha, it's already 36-bytes, which rounds to 40. 172 * XXX Need to eliminate the pointer. 173 * 174 * XXX We've blown 32 bytes on non-Alpha systems, too, since we're 175 * XXX storing the maxseg we advertised to the peer. Should we 176 * XXX create another malloc bucket? Should we care? 177 */ 178 struct syn_cache { 179 struct syn_cache *sc_next; 180 u_int32_t sc_tstmp : 1, 181 sc_hash : 31; 182 struct in_addr sc_src; 183 struct in_addr sc_dst; 184 tcp_seq sc_irs; 185 tcp_seq sc_iss; 186 u_int16_t sc_sport; 187 u_int16_t sc_dport; 188 u_int16_t sc_peermaxseg; 189 u_int16_t sc_ourmaxseg; 190 u_int8_t sc_timer; 191 u_int8_t sc_request_r_scale : 4, 192 sc_requested_s_scale : 4; 193 }; 194 195 struct syn_cache_head { 196 struct syn_cache *sch_first; /* First entry in the bucket */ 197 struct syn_cache *sch_last; /* Last entry in the bucket */ 198 struct syn_cache_head *sch_headq; /* The next non-empty bucket */ 199 int16_t sch_timer_sum; /* Total time in this bucket */ 200 u_int16_t sch_length; /* @ # elements in bucket */ 201 }; 202 203 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 204 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 205 206 /* 207 * The smoothed round-trip time and estimated variance 208 * are stored as fixed point numbers scaled by the values below. 209 * For convenience, these scales are also used in smoothing the average 210 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 211 * With these scales, srtt has 3 bits to the right of the binary point, 212 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 213 * binary point, and is smoothed with an ALPHA of 0.75. 214 */ 215 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 216 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 217 218 /* 219 * The initial retransmission should happen at rtt + 4 * rttvar. 220 * Because of the way we do the smoothing, srtt and rttvar 221 * will each average +1/2 tick of bias. When we compute 222 * the retransmit timer, we want 1/2 tick of rounding and 223 * 1 extra tick because of +-1/2 tick uncertainty in the 224 * firing of the timer. The bias will give us exactly the 225 * 1.5 tick we need. But, because the bias is 226 * statistical, we have to test that we don't drop below 227 * the minimum feasible timer (which is 2 ticks). 228 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 229 * is the same as the multiplier for rttvar. 230 */ 231 #define TCP_REXMTVAL(tp) \ 232 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 233 234 #ifdef _KERNEL 235 /* 236 * Compute the initial window for slow start. 237 */ 238 extern int tcp_init_win; 239 #define TCP_INITIAL_WINDOW(segsz) \ 240 ((tcp_init_win == 0) ? (min(4 * (segsz), max(2 * (segsz), 4380))) : \ 241 ((segsz) * tcp_init_win)) 242 #endif /* _KERNEL */ 243 244 /* 245 * TCP statistics. 246 * Many of these should be kept per connection, 247 * but that's inconvenient at the moment. 248 */ 249 struct tcpstat { 250 u_long tcps_connattempt; /* connections initiated */ 251 u_long tcps_accepts; /* connections accepted */ 252 u_long tcps_connects; /* connections established */ 253 u_long tcps_drops; /* connections dropped */ 254 u_long tcps_conndrops; /* embryonic connections dropped */ 255 u_long tcps_closed; /* conn. closed (includes drops) */ 256 u_long tcps_segstimed; /* segs where we tried to get rtt */ 257 u_long tcps_rttupdated; /* times we succeeded */ 258 u_long tcps_delack; /* delayed acks sent */ 259 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 260 u_long tcps_rexmttimeo; /* retransmit timeouts */ 261 u_long tcps_persisttimeo; /* persist timeouts */ 262 u_long tcps_keeptimeo; /* keepalive timeouts */ 263 u_long tcps_keepprobe; /* keepalive probes sent */ 264 u_long tcps_keepdrops; /* connections dropped in keepalive */ 265 u_long tcps_persistdrops; /* connections dropped in persist */ 266 u_long tcps_connsdrained; /* connections drained due to memory 267 shortage */ 268 269 u_long tcps_sndtotal; /* total packets sent */ 270 u_long tcps_sndpack; /* data packets sent */ 271 u_long tcps_sndbyte; /* data bytes sent */ 272 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 273 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 274 u_long tcps_sndacks; /* ack-only packets sent */ 275 u_long tcps_sndprobe; /* window probes sent */ 276 u_long tcps_sndurg; /* packets sent with URG only */ 277 u_long tcps_sndwinup; /* window update-only packets sent */ 278 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 279 280 u_long tcps_rcvtotal; /* total packets received */ 281 u_long tcps_rcvpack; /* packets received in sequence */ 282 u_long tcps_rcvbyte; /* bytes received in sequence */ 283 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 284 u_long tcps_rcvbadoff; /* packets received with bad offset */ 285 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 286 u_long tcps_rcvshort; /* packets received too short */ 287 u_long tcps_rcvduppack; /* duplicate-only packets received */ 288 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 289 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 290 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 291 u_long tcps_rcvoopack; /* out-of-order packets received */ 292 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 293 u_long tcps_rcvpackafterwin; /* packets with data after window */ 294 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 295 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 296 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 297 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 298 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 299 u_long tcps_rcvackpack; /* rcvd ack packets */ 300 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 301 u_long tcps_rcvwinupd; /* rcvd window update packets */ 302 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 303 u_long tcps_predack; /* times hdr predict ok for acks */ 304 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 305 306 u_long tcps_pcbhashmiss; /* input packets missing pcb hash */ 307 u_long tcps_noport; /* no socket on port */ 308 u_long tcps_badsyn; /* received ack for which we have 309 no SYN in compressed state */ 310 311 /* These statistics deal with the SYN cache. */ 312 u_long tcps_sc_added; /* # of entries added */ 313 u_long tcps_sc_completed; /* # of connections completed */ 314 u_long tcps_sc_timed_out; /* # of entries timed out */ 315 u_long tcps_sc_overflowed; /* # dropped due to overflow */ 316 u_long tcps_sc_reset; /* # dropped due to RST */ 317 u_long tcps_sc_unreach; /* # dropped due to ICMP unreach */ 318 u_long tcps_sc_bucketoverflow; /* # dropped due to bucket overflow */ 319 u_long tcps_sc_aborted; /* # of entries aborted (no mem) */ 320 u_long tcps_sc_dupesyn; /* # of duplicate SYNs received */ 321 u_long tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 322 u_long tcps_sc_collisions; /* # of hash collisions */ 323 }; 324 325 /* 326 * Names for TCP sysctl objects. 327 */ 328 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 329 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 330 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 331 #define TCPCTL_MSSDFLT 4 /* default seg size */ 332 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 333 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 334 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 335 #define TCPCTL_INIT_WIN 8 /* initial window */ 336 #define TCPCTL_MAXID 9 337 338 #define TCPCTL_NAMES { \ 339 { 0, 0 }, \ 340 { "rfc1323", CTLTYPE_INT }, \ 341 { "sendspace", CTLTYPE_INT }, \ 342 { "recvspace", CTLTYPE_INT }, \ 343 { "mssdflt", CTLTYPE_INT }, \ 344 { "syn_cache_limit", CTLTYPE_INT }, \ 345 { "syn_bucket_limit", CTLTYPE_INT }, \ 346 { "syn_cache_interval", CTLTYPE_INT },\ 347 { "init_win", CTLTYPE_INT }, \ 348 } 349 350 #ifdef _KERNEL 351 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 352 struct tcpstat tcpstat; /* tcp statistics */ 353 u_int32_t tcp_now; /* for RFC 1323 timestamps */ 354 extern int tcp_do_rfc1323; /* enabled/disabled? */ 355 extern int tcp_mssdflt; /* default seg size */ 356 extern int tcp_init_win; /* initial window */ 357 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 358 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 359 extern int tcp_syn_cache_interval; /* compressed state timer */ 360 361 extern int tcp_syn_cache_size; 362 extern int tcp_syn_cache_timeo; 363 extern struct syn_cache_head tcp_syn_cache[], *tcp_syn_cache_first; 364 extern u_long syn_cache_count; 365 366 int tcp_attach __P((struct socket *)); 367 void tcp_canceltimers __P((struct tcpcb *)); 368 struct tcpcb * 369 tcp_close __P((struct tcpcb *)); 370 void *tcp_ctlinput __P((int, struct sockaddr *, void *)); 371 int tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **)); 372 struct tcpcb * 373 tcp_disconnect __P((struct tcpcb *)); 374 struct tcpcb * 375 tcp_drop __P((struct tcpcb *, int)); 376 void tcp_dooptions __P((struct tcpcb *, 377 u_char *, int, struct tcpiphdr *, struct tcp_opt_info *)); 378 void tcp_drain __P((void)); 379 void tcp_established __P((struct tcpcb *)); 380 void tcp_fasttimo __P((void)); 381 void tcp_init __P((void)); 382 void tcp_input __P((struct mbuf *, ...)); 383 int tcp_mss_to_advertise __P((const struct tcpcb *)); 384 void tcp_mss_from_peer __P((struct tcpcb *, int)); 385 void tcp_mtudisc __P((struct inpcb *, int)); 386 struct tcpcb * 387 tcp_newtcpcb __P((struct inpcb *)); 388 void tcp_notify __P((struct inpcb *, int)); 389 int tcp_output __P((struct tcpcb *)); 390 void tcp_pulloutofband __P((struct socket *, 391 struct tcpiphdr *, struct mbuf *)); 392 void tcp_quench __P((struct inpcb *, int)); 393 int tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *)); 394 int tcp_respond __P((struct tcpcb *, 395 struct tcpiphdr *, struct mbuf *, tcp_seq, tcp_seq, int)); 396 void tcp_rmx_rtt __P((struct tcpcb *)); 397 void tcp_setpersist __P((struct tcpcb *)); 398 void tcp_slowtimo __P((void)); 399 struct tcpiphdr * 400 tcp_template __P((struct tcpcb *)); 401 struct tcpcb * 402 tcp_timers __P((struct tcpcb *, int)); 403 void tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int)); 404 struct tcpcb * 405 tcp_usrclosed __P((struct tcpcb *)); 406 int tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t)); 407 int tcp_usrreq __P((struct socket *, 408 int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *)); 409 void tcp_xmit_timer __P((struct tcpcb *, int)); 410 tcp_seq tcp_new_iss __P((void *, u_long, tcp_seq)); 411 412 int syn_cache_add __P((struct socket *, struct mbuf *, u_char *, 413 int, struct tcp_opt_info *)); 414 void syn_cache_unreach __P((struct ip *, struct tcphdr *)); 415 struct socket * 416 syn_cache_get __P((struct socket *so, struct mbuf *)); 417 void syn_cache_insert __P((struct syn_cache *, struct syn_cache ***, 418 struct syn_cache_head **)); 419 struct syn_cache * 420 syn_cache_lookup __P((struct tcpiphdr *, struct syn_cache ***, 421 struct syn_cache_head **)); 422 void syn_cache_reset __P((struct tcpiphdr *)); 423 int syn_cache_respond __P((struct syn_cache *, struct mbuf *, 424 register struct tcpiphdr *, long, u_long)); 425 void syn_cache_timer __P((int)); 426 427 #endif 428