1 /* $NetBSD: tcp_var.h,v 1.25 1997/10/17 22:12:46 kml Exp $ */ 2 3 /* 4 * Copyright (c) 1982, 1986, 1993, 1994 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 8.3 (Berkeley) 4/10/94 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_maxseg; /* peer's maximum segment size */ 53 u_short t_ourmss; /* our's maximum segment size */ 54 char t_force; /* 1 if forcing out a byte */ 55 u_short t_flags; 56 #define TF_ACKNOW 0x0001 /* ack peer immediately */ 57 #define TF_DELACK 0x0002 /* ack, but try to delay it */ 58 #define TF_NODELAY 0x0004 /* don't delay packets to coalesce */ 59 #define TF_NOOPT 0x0008 /* don't use tcp options */ 60 #define TF_SENTFIN 0x0010 /* have sent FIN */ 61 #define TF_REQ_SCALE 0x0020 /* have/will request window scaling */ 62 #define TF_RCVD_SCALE 0x0040 /* other side has requested scaling */ 63 #define TF_REQ_TSTMP 0x0080 /* have/will request timestamps */ 64 #define TF_RCVD_TSTMP 0x0100 /* a timestamp was received in SYN */ 65 #define TF_SACK_PERMIT 0x0200 /* other side said I could SACK */ 66 67 struct tcpiphdr *t_template; /* skeletal packet for transmit */ 68 struct inpcb *t_inpcb; /* back pointer to internet pcb */ 69 /* 70 * The following fields are used as in the protocol specification. 71 * See RFC783, Dec. 1981, page 21. 72 */ 73 /* send sequence variables */ 74 tcp_seq snd_una; /* send unacknowledged */ 75 tcp_seq snd_nxt; /* send next */ 76 tcp_seq snd_up; /* send urgent pointer */ 77 tcp_seq snd_wl1; /* window update seg seq number */ 78 tcp_seq snd_wl2; /* window update seg ack number */ 79 tcp_seq iss; /* initial send sequence number */ 80 u_long snd_wnd; /* send window */ 81 /* receive sequence variables */ 82 u_long rcv_wnd; /* receive window */ 83 tcp_seq rcv_nxt; /* receive next */ 84 tcp_seq rcv_up; /* receive urgent pointer */ 85 tcp_seq irs; /* initial receive sequence number */ 86 /* 87 * Additional variables for this implementation. 88 */ 89 /* receive variables */ 90 tcp_seq rcv_adv; /* advertised window */ 91 /* retransmit variables */ 92 tcp_seq snd_max; /* highest sequence number sent; 93 * used to recognize retransmits 94 */ 95 /* congestion control (for slow start, source quench, retransmit after loss) */ 96 u_long snd_cwnd; /* congestion-controlled window */ 97 u_long snd_ssthresh; /* snd_cwnd size threshhold for 98 * for slow start exponential to 99 * linear switch 100 */ 101 /* 102 * transmit timing stuff. See below for scale of srtt and rttvar. 103 * "Variance" is actually smoothed difference. 104 */ 105 short t_idle; /* inactivity time */ 106 short t_rtt; /* round trip time */ 107 tcp_seq t_rtseq; /* sequence number being timed */ 108 short t_srtt; /* smoothed round-trip time */ 109 short t_rttvar; /* variance in round-trip time */ 110 short t_rttmin; /* minimum rtt allowed */ 111 u_long max_sndwnd; /* largest window peer has offered */ 112 113 /* out-of-band data */ 114 char t_oobflags; /* have some */ 115 char t_iobc; /* input character */ 116 #define TCPOOB_HAVEDATA 0x01 117 #define TCPOOB_HADDATA 0x02 118 short t_softerror; /* possible error not yet reported */ 119 120 /* RFC 1323 variables */ 121 u_char snd_scale; /* window scaling for send window */ 122 u_char rcv_scale; /* window scaling for recv window */ 123 u_char request_r_scale; /* pending window scaling */ 124 u_char requested_s_scale; 125 u_int32_t ts_recent; /* timestamp echo data */ 126 u_int32_t ts_recent_age; /* when last updated */ 127 tcp_seq last_ack_sent; 128 129 /* TUBA stuff */ 130 caddr_t t_tuba_pcb; /* next level down pcb for TCP over z */ 131 }; 132 133 /* 134 * Handy way of passing around TCP option info. 135 */ 136 struct tcp_opt_info { 137 int ts_present; 138 u_int32_t ts_val; 139 u_int32_t ts_ecr; 140 u_int16_t maxseg; 141 }; 142 143 /* 144 * This structure should not exceed 32 bytes. 145 * XXX On the Alpha, it's already 36-bytes, which rounds to 40. 146 * XXX Need to eliminate the pointer. 147 * 148 * XXX We've blown 32 bytes on non-Alpha systems, too, since we're 149 * XXX storing the maxseg we advertised to the peer. Should we 150 * XXX create another malloc bucket? Should we care? 151 */ 152 struct syn_cache { 153 struct syn_cache *sc_next; 154 u_int32_t sc_tstmp : 1, 155 sc_hash : 31; 156 struct in_addr sc_src; 157 struct in_addr sc_dst; 158 tcp_seq sc_irs; 159 tcp_seq sc_iss; 160 u_int16_t sc_sport; 161 u_int16_t sc_dport; 162 u_int16_t sc_peermaxseg; 163 u_int16_t sc_ourmaxseg; 164 u_int8_t sc_timer; 165 u_int8_t sc_request_r_scale : 4, 166 sc_requested_s_scale : 4; 167 }; 168 169 struct syn_cache_head { 170 struct syn_cache *sch_first; /* First entry in the bucket */ 171 struct syn_cache *sch_last; /* Last entry in the bucket */ 172 struct syn_cache_head *sch_headq; /* The next non-empty bucket */ 173 int16_t sch_timer_sum; /* Total time in this bucket */ 174 u_int16_t sch_length; /* @ # elements in bucket */ 175 }; 176 177 #define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb) 178 #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) 179 180 /* 181 * The smoothed round-trip time and estimated variance 182 * are stored as fixed point numbers scaled by the values below. 183 * For convenience, these scales are also used in smoothing the average 184 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). 185 * With these scales, srtt has 3 bits to the right of the binary point, 186 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the 187 * binary point, and is smoothed with an ALPHA of 0.75. 188 */ 189 #define TCP_RTT_SHIFT 3 /* shift for srtt; 3 bits frac. */ 190 #define TCP_RTTVAR_SHIFT 2 /* multiplier for rttvar; 2 bits */ 191 192 /* 193 * The initial retransmission should happen at rtt + 4 * rttvar. 194 * Because of the way we do the smoothing, srtt and rttvar 195 * will each average +1/2 tick of bias. When we compute 196 * the retransmit timer, we want 1/2 tick of rounding and 197 * 1 extra tick because of +-1/2 tick uncertainty in the 198 * firing of the timer. The bias will give us exactly the 199 * 1.5 tick we need. But, because the bias is 200 * statistical, we have to test that we don't drop below 201 * the minimum feasible timer (which is 2 ticks). 202 * This macro assumes that the value of 1<<TCP_RTTVAR_SHIFT 203 * is the same as the multiplier for rttvar. 204 */ 205 #define TCP_REXMTVAL(tp) \ 206 ((((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar) >> 2) 207 208 /* 209 * TCP statistics. 210 * Many of these should be kept per connection, 211 * but that's inconvenient at the moment. 212 */ 213 struct tcpstat { 214 u_long tcps_connattempt; /* connections initiated */ 215 u_long tcps_accepts; /* connections accepted */ 216 u_long tcps_connects; /* connections established */ 217 u_long tcps_drops; /* connections dropped */ 218 u_long tcps_conndrops; /* embryonic connections dropped */ 219 u_long tcps_closed; /* conn. closed (includes drops) */ 220 u_long tcps_segstimed; /* segs where we tried to get rtt */ 221 u_long tcps_rttupdated; /* times we succeeded */ 222 u_long tcps_delack; /* delayed acks sent */ 223 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ 224 u_long tcps_rexmttimeo; /* retransmit timeouts */ 225 u_long tcps_persisttimeo; /* persist timeouts */ 226 u_long tcps_keeptimeo; /* keepalive timeouts */ 227 u_long tcps_keepprobe; /* keepalive probes sent */ 228 u_long tcps_keepdrops; /* connections dropped in keepalive */ 229 230 u_long tcps_sndtotal; /* total packets sent */ 231 u_long tcps_sndpack; /* data packets sent */ 232 u_long tcps_sndbyte; /* data bytes sent */ 233 u_long tcps_sndrexmitpack; /* data packets retransmitted */ 234 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ 235 u_long tcps_sndacks; /* ack-only packets sent */ 236 u_long tcps_sndprobe; /* window probes sent */ 237 u_long tcps_sndurg; /* packets sent with URG only */ 238 u_long tcps_sndwinup; /* window update-only packets sent */ 239 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ 240 241 u_long tcps_rcvtotal; /* total packets received */ 242 u_long tcps_rcvpack; /* packets received in sequence */ 243 u_long tcps_rcvbyte; /* bytes received in sequence */ 244 u_long tcps_rcvbadsum; /* packets received with ccksum errs */ 245 u_long tcps_rcvbadoff; /* packets received with bad offset */ 246 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ 247 u_long tcps_rcvshort; /* packets received too short */ 248 u_long tcps_rcvduppack; /* duplicate-only packets received */ 249 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ 250 u_long tcps_rcvpartduppack; /* packets with some duplicate data */ 251 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ 252 u_long tcps_rcvoopack; /* out-of-order packets received */ 253 u_long tcps_rcvoobyte; /* out-of-order bytes received */ 254 u_long tcps_rcvpackafterwin; /* packets with data after window */ 255 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ 256 u_long tcps_rcvafterclose; /* packets rcvd after "close" */ 257 u_long tcps_rcvwinprobe; /* rcvd window probe packets */ 258 u_long tcps_rcvdupack; /* rcvd duplicate acks */ 259 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ 260 u_long tcps_rcvackpack; /* rcvd ack packets */ 261 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ 262 u_long tcps_rcvwinupd; /* rcvd window update packets */ 263 u_long tcps_pawsdrop; /* segments dropped due to PAWS */ 264 u_long tcps_predack; /* times hdr predict ok for acks */ 265 u_long tcps_preddat; /* times hdr predict ok for data pkts */ 266 267 u_long tcps_pcbhashmiss; /* input packets missing pcb hash */ 268 u_long tcps_noport; /* no socket on port */ 269 u_long tcps_badsyn; /* received ack for which we have 270 no SYN in compressed state */ 271 272 /* These statistics deal with the SYN cache. */ 273 u_long tcps_sc_added; /* # of entries added */ 274 u_long tcps_sc_completed; /* # of connections completed */ 275 u_long tcps_sc_timed_out; /* # of entries timed out */ 276 u_long tcps_sc_overflowed; /* # dropped due to overflow */ 277 u_long tcps_sc_reset; /* # dropped due to RST */ 278 u_long tcps_sc_unreach; /* # dropped due to ICMP unreach */ 279 u_long tcps_sc_bucketoverflow; /* # dropped due to bucket overflow */ 280 u_long tcps_sc_aborted; /* # of entries aborted (no mem) */ 281 u_long tcps_sc_dupesyn; /* # of duplicate SYNs received */ 282 u_long tcps_sc_dropped; /* # of SYNs dropped (no route/mem) */ 283 u_long tcps_sc_collisions; /* # of hash collisions */ 284 }; 285 286 /* 287 * Names for TCP sysctl objects. 288 */ 289 #define TCPCTL_RFC1323 1 /* RFC1323 timestamps/scaling */ 290 #define TCPCTL_SENDSPACE 2 /* default send buffer */ 291 #define TCPCTL_RECVSPACE 3 /* default recv buffer */ 292 #define TCPCTL_MSSDFLT 4 /* default seg size */ 293 #define TCPCTL_SYN_CACHE_LIMIT 5 /* max size of comp. state engine */ 294 #define TCPCTL_SYN_BUCKET_LIMIT 6 /* max size of hash bucket */ 295 #define TCPCTL_SYN_CACHE_INTER 7 /* interval of comp. state timer */ 296 #define TCPCTL_MAXID 8 297 298 #define TCPCTL_NAMES { \ 299 { 0, 0 }, \ 300 { "rfc1323", CTLTYPE_INT }, \ 301 { "sendspace", CTLTYPE_INT }, \ 302 { "recvspace", CTLTYPE_INT }, \ 303 { "mssdflt", CTLTYPE_INT }, \ 304 { "syn_cache_limit", CTLTYPE_INT }, \ 305 { "syn_bucket_limit", CTLTYPE_INT }, \ 306 { "syn_cache_interval", CTLTYPE_INT },\ 307 } 308 309 #ifdef _KERNEL 310 struct inpcbtable tcbtable; /* head of queue of active tcpcb's */ 311 struct tcpstat tcpstat; /* tcp statistics */ 312 u_int32_t tcp_now; /* for RFC 1323 timestamps */ 313 extern int tcp_do_rfc1323; /* enabled/disabled? */ 314 extern int tcp_mssdflt; /* default seg size */ 315 extern int tcp_syn_cache_limit; /* max entries for compressed state engine */ 316 extern int tcp_syn_bucket_limit;/* max entries per hash bucket */ 317 extern int tcp_syn_cache_interval; /* compressed state timer */ 318 319 extern int tcp_syn_cache_size; 320 extern int tcp_syn_cache_timeo; 321 extern struct syn_cache_head tcp_syn_cache[], *tcp_syn_cache_first; 322 extern u_long syn_cache_count; 323 324 int tcp_attach __P((struct socket *)); 325 void tcp_canceltimers __P((struct tcpcb *)); 326 struct tcpcb * 327 tcp_close __P((struct tcpcb *)); 328 void *tcp_ctlinput __P((int, struct sockaddr *, void *)); 329 int tcp_ctloutput __P((int, struct socket *, int, int, struct mbuf **)); 330 struct tcpcb * 331 tcp_disconnect __P((struct tcpcb *)); 332 struct tcpcb * 333 tcp_drop __P((struct tcpcb *, int)); 334 void tcp_dooptions __P((struct tcpcb *, 335 u_char *, int, struct tcpiphdr *, struct tcp_opt_info *)); 336 void tcp_drain __P((void)); 337 void tcp_established __P((struct tcpcb *)); 338 void tcp_fasttimo __P((void)); 339 void tcp_init __P((void)); 340 void tcp_input __P((struct mbuf *, ...)); 341 int tcp_mss_to_advertise __P((const struct tcpcb *)); 342 void tcp_mss_from_peer __P((struct tcpcb *, int)); 343 void tcp_mtudisc __P((struct inpcb *, int)); 344 struct tcpcb * 345 tcp_newtcpcb __P((struct inpcb *)); 346 void tcp_notify __P((struct inpcb *, int)); 347 int tcp_output __P((struct tcpcb *)); 348 void tcp_pulloutofband __P((struct socket *, 349 struct tcpiphdr *, struct mbuf *)); 350 void tcp_quench __P((struct inpcb *, int)); 351 int tcp_reass __P((struct tcpcb *, struct tcpiphdr *, struct mbuf *)); 352 int tcp_respond __P((struct tcpcb *, 353 struct tcpiphdr *, struct mbuf *, tcp_seq, tcp_seq, int)); 354 void tcp_rmx_rtt __P((struct tcpcb *)); 355 void tcp_setpersist __P((struct tcpcb *)); 356 void tcp_slowtimo __P((void)); 357 struct tcpiphdr * 358 tcp_template __P((struct tcpcb *)); 359 struct tcpcb * 360 tcp_timers __P((struct tcpcb *, int)); 361 void tcp_trace __P((int, int, struct tcpcb *, struct tcpiphdr *, int)); 362 struct tcpcb * 363 tcp_usrclosed __P((struct tcpcb *)); 364 int tcp_sysctl __P((int *, u_int, void *, size_t *, void *, size_t)); 365 int tcp_usrreq __P((struct socket *, 366 int, struct mbuf *, struct mbuf *, struct mbuf *, struct proc *)); 367 void tcp_xmit_timer __P((struct tcpcb *, int)); 368 tcp_seq tcp_new_iss __P((void *, u_long, tcp_seq)); 369 370 int syn_cache_add __P((struct socket *, struct mbuf *, u_char *, 371 int, struct tcp_opt_info *)); 372 void syn_cache_unreach __P((struct ip *, struct tcphdr *)); 373 struct socket * 374 syn_cache_get __P((struct socket *so, struct mbuf *)); 375 void syn_cache_insert __P((struct syn_cache *, struct syn_cache ***, 376 struct syn_cache_head **)); 377 struct syn_cache * 378 syn_cache_lookup __P((struct tcpiphdr *, struct syn_cache ***, 379 struct syn_cache_head **)); 380 void syn_cache_reset __P((struct tcpiphdr *)); 381 int syn_cache_respond __P((struct syn_cache *, struct mbuf *, 382 register struct tcpiphdr *, long, u_long)); 383 void syn_cache_timer __P((int)); 384 385 #endif 386