1/*-
2 * Copyright (c) 1982, 1986, 1993, 1994, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
| 1/*-
2 * Copyright (c) 1982, 1986, 1993, 1994, 1995
3 * The Regents of the University of California. All rights reserved.
4 *
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
7 * are met:
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
16 *
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE.
28 *
29 * @(#)tcp_var.h 8.4 (Berkeley) 5/24/95
|
30 * $FreeBSD: head/sys/netinet/tcp_var.h 204838 2010-03-07 15:58:44Z bz $
| 30 * $FreeBSD: head/sys/netinet/tcp_var.h 207369 2010-04-29 11:52:42Z bz $
|
31 */
32
33#ifndef _NETINET_TCP_VAR_H_
34#define _NETINET_TCP_VAR_H_
35
36#include <netinet/tcp.h>
37
38#ifdef _KERNEL
39#include <net/vnet.h>
40
41/*
42 * Kernel variables for tcp.
43 */
44VNET_DECLARE(int, tcp_do_rfc1323);
| 31 */
32
33#ifndef _NETINET_TCP_VAR_H_
34#define _NETINET_TCP_VAR_H_
35
36#include <netinet/tcp.h>
37
38#ifdef _KERNEL
39#include <net/vnet.h>
40
41/*
42 * Kernel variables for tcp.
43 */
44VNET_DECLARE(int, tcp_do_rfc1323);
|
| 45#define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323)
46
|
45VNET_DECLARE(int, tcp_reass_qsize);
46VNET_DECLARE(struct uma_zone *, tcp_reass_zone);
| 47VNET_DECLARE(int, tcp_reass_qsize);
48VNET_DECLARE(struct uma_zone *, tcp_reass_zone);
|
47#define V_tcp_do_rfc1323 VNET(tcp_do_rfc1323)
| |
48#define V_tcp_reass_qsize VNET(tcp_reass_qsize)
49#define V_tcp_reass_zone VNET(tcp_reass_zone)
| 49#define V_tcp_reass_qsize VNET(tcp_reass_qsize)
50#define V_tcp_reass_zone VNET(tcp_reass_zone)
|
50
| |
51#endif /* _KERNEL */
52
53/* TCP segment queue entry */
54struct tseg_qent {
55 LIST_ENTRY(tseg_qent) tqe_q;
56 int tqe_len; /* TCP segment data length */
57 struct tcphdr *tqe_th; /* a pointer to tcp header */
58 struct mbuf *tqe_m; /* mbuf contains packet */
59};
60LIST_HEAD(tsegqe_head, tseg_qent);
61
62struct sackblk {
63 tcp_seq start; /* start seq no. of sack block */
64 tcp_seq end; /* end seq no. */
65};
66
67struct sackhole {
68 tcp_seq start; /* start seq no. of hole */
69 tcp_seq end; /* end seq no. */
70 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
71 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
72};
73
74struct sackhint {
75 struct sackhole *nexthole;
76 int sack_bytes_rexmit;
77
78 int ispare; /* explicit pad for 64bit alignment */
79 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */
80};
81
82struct tcptemp {
83 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
84 struct tcphdr tt_t;
85};
86
87#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
88
89/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
90#ifdef INET6
91#define ND6_HINT(tp) \
92do { \
93 if ((tp) && (tp)->t_inpcb && \
94 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \
95 nd6_nud_hint(NULL, NULL, 0); \
96} while (0)
97#else
98#define ND6_HINT(tp)
99#endif
100
101/*
102 * Tcp control block, one per tcp; fields:
103 * Organized for 16 byte cacheline efficiency.
104 */
105struct tcpcb {
106 struct tsegqe_head t_segq; /* segment reassembly queue */
107 void *t_pspare[2]; /* new reassembly queue */
108 int t_segqlen; /* segment reassembly queue length */
109 int t_dupacks; /* consecutive dup acks recd */
110
111 struct tcp_timer *t_timers; /* All the TCP timers in one struct */
112
113 struct inpcb *t_inpcb; /* back pointer to internet pcb */
114 int t_state; /* state of this connection */
115 u_int t_flags;
116
117 struct vnet *t_vnet; /* back pointer to parent vnet */
118
119 tcp_seq snd_una; /* send unacknowledged */
120 tcp_seq snd_max; /* highest sequence number sent;
121 * used to recognize retransmits
122 */
123 tcp_seq snd_nxt; /* send next */
124 tcp_seq snd_up; /* send urgent pointer */
125
126 tcp_seq snd_wl1; /* window update seg seq number */
127 tcp_seq snd_wl2; /* window update seg ack number */
128 tcp_seq iss; /* initial send sequence number */
129 tcp_seq irs; /* initial receive sequence number */
130
131 tcp_seq rcv_nxt; /* receive next */
132 tcp_seq rcv_adv; /* advertised window */
133 u_long rcv_wnd; /* receive window */
134 tcp_seq rcv_up; /* receive urgent pointer */
135
136 u_long snd_wnd; /* send window */
137 u_long snd_cwnd; /* congestion-controlled window */
138 u_long snd_bwnd; /* bandwidth-controlled window */
139 u_long snd_ssthresh; /* snd_cwnd size threshold for
140 * for slow start exponential to
141 * linear switch
142 */
143 u_long snd_bandwidth; /* calculated bandwidth or 0 */
144 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
145
146 u_int t_maxopd; /* mss plus options */
147
148 u_int t_rcvtime; /* inactivity time */
149 u_int t_starttime; /* time connection was established */
150 u_int t_rtttime; /* RTT measurement start time */
151 tcp_seq t_rtseq; /* sequence number being timed */
152
153 u_int t_bw_rtttime; /* used for bandwidth calculation */
154 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */
155
156 int t_rxtcur; /* current retransmit value (ticks) */
157 u_int t_maxseg; /* maximum segment size */
158 int t_srtt; /* smoothed round-trip time */
159 int t_rttvar; /* variance in round-trip time */
160
161 int t_rxtshift; /* log(2) of rexmt exp. backoff */
162 u_int t_rttmin; /* minimum rtt allowed */
163 u_int t_rttbest; /* best rtt we've seen */
164 u_long t_rttupdated; /* number of times rtt sampled */
165 u_long max_sndwnd; /* largest window peer has offered */
166
167 int t_softerror; /* possible error not yet reported */
168/* out-of-band data */
169 char t_oobflags; /* have some */
170 char t_iobc; /* input character */
171/* RFC 1323 variables */
172 u_char snd_scale; /* window scaling for send window */
173 u_char rcv_scale; /* window scaling for recv window */
174 u_char request_r_scale; /* pending window scaling */
175 u_int32_t ts_recent; /* timestamp echo data */
176 u_int ts_recent_age; /* when last updated */
177 u_int32_t ts_offset; /* our timestamp offset */
178
179 tcp_seq last_ack_sent;
180/* experimental */
181 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
182 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
183 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
184 u_int t_badrxtwin; /* window for retransmit recovery */
185 u_char snd_limited; /* segments limited transmitted */
186/* SACK related state */
187 int snd_numholes; /* number of holes seen by sender */
188 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
189 /* SACK scoreboard (sorted) */
190 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
191 int rcv_numsacks; /* # distinct sack blks present */
192 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
193 tcp_seq sack_newdata; /* New data xmitted in this recovery
194 episode starts at this seq number */
195 struct sackhint sackhint; /* SACK scoreboard hint */
196 int t_rttlow; /* smallest observerved RTT */
197 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
198 int rfbuf_cnt; /* recv buffer autoscaling byte count */
199 struct toe_usrreqs *t_tu; /* offload operations vector */
200 void *t_toe; /* TOE pcb pointer */
201 int t_bytes_acked; /* # bytes acked during current RTT */
202
203 int t_ispare; /* explicit pad for 64bit alignment */
204 void *t_pspare2[6]; /* 2 CC / 4 TBD */
205 uint64_t _pad[12]; /* 7 UTO, 5 TBD (1-2 CC/RTT?) */
206};
207
208/*
209 * Flags and utility macros for the t_flags field.
210 */
211#define TF_ACKNOW 0x000001 /* ack peer immediately */
212#define TF_DELACK 0x000002 /* ack, but try to delay it */
213#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */
214#define TF_NOOPT 0x000008 /* don't use tcp options */
215#define TF_SENTFIN 0x000010 /* have sent FIN */
216#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */
217#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */
218#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */
219#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */
220#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */
221#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
222#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
223#define TF_NOPUSH 0x001000 /* don't push */
224#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */
225#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
226#define TF_LASTIDLE 0x040000 /* connection was previously idle */
227#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */
228#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */
229#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */
230#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
231#define TF_FORCEDATA 0x800000 /* force out a byte */
232#define TF_TSO 0x1000000 /* TSO enabled on this connection */
233#define TF_TOE 0x2000000 /* this connection is offloaded */
234#define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */
235#define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */
236#define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */
237
238#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY)
239#define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY
240#define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY
241
242/*
243 * Flags for the t_oobflags field.
244 */
245#define TCPOOB_HAVEDATA 0x01
246#define TCPOOB_HADDATA 0x02
247
248#ifdef TCP_SIGNATURE
249/*
250 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
251 * for SADB verification and lookup.
252 */
253#define TCP_SIGLEN 16 /* length of computed digest in bytes */
254#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
255#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
256/*
257 * Only a single SA per host may be specified at this time. An SPI is
258 * needed in order for the KEY_ALLOCSA() lookup to work.
259 */
260#define TCP_SIG_SPI 0x1000
261#endif /* TCP_SIGNATURE */
262
263/*
264 * Structure to hold TCP options that are only used during segment
265 * processing (in tcp_input), but not held in the tcpcb.
266 * It's basically used to reduce the number of parameters
267 * to tcp_dooptions and tcp_addoptions.
268 * The binary order of the to_flags is relevant for packing of the
269 * options in tcp_addoptions.
270 */
271struct tcpopt {
272 u_int64_t to_flags; /* which options are present */
273#define TOF_MSS 0x0001 /* maximum segment size */
274#define TOF_SCALE 0x0002 /* window scaling */
275#define TOF_SACKPERM 0x0004 /* SACK permitted */
276#define TOF_TS 0x0010 /* timestamp */
277#define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */
278#define TOF_SACK 0x0080 /* Peer sent SACK option */
279#define TOF_MAXOPT 0x0100
280 u_int32_t to_tsval; /* new timestamp */
281 u_int32_t to_tsecr; /* reflected timestamp */
282 u_char *to_sacks; /* pointer to the first SACK blocks */
283 u_char *to_signature; /* pointer to the TCP-MD5 signature */
284 u_int16_t to_mss; /* maximum segment size */
285 u_int8_t to_wscale; /* window scaling */
286 u_int8_t to_nsacks; /* number of SACK blocks */
287};
288
289/*
290 * Flags for tcp_dooptions.
291 */
292#define TO_SYN 0x01 /* parse SYN-only options */
293
294struct hc_metrics_lite { /* must stay in sync with hc_metrics */
295 u_long rmx_mtu; /* MTU for this path */
296 u_long rmx_ssthresh; /* outbound gateway buffer limit */
297 u_long rmx_rtt; /* estimated round trip time */
298 u_long rmx_rttvar; /* estimated rtt variance */
299 u_long rmx_bandwidth; /* estimated bandwidth */
300 u_long rmx_cwnd; /* congestion window */
301 u_long rmx_sendpipe; /* outbound delay-bandwidth product */
302 u_long rmx_recvpipe; /* inbound delay-bandwidth product */
303};
304
305#ifndef _NETINET_IN_PCB_H_
306struct in_conninfo;
307#endif /* _NETINET_IN_PCB_H_ */
308
309struct tcptw {
310 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
311 tcp_seq snd_nxt;
312 tcp_seq rcv_nxt;
313 tcp_seq iss;
314 tcp_seq irs;
315 u_short last_win; /* cached window value */
316 u_short tw_so_options; /* copy of so_options */
317 struct ucred *tw_cred; /* user credentials */
318 u_int32_t t_recent;
319 u_int32_t ts_offset; /* our timestamp offset */
320 u_int t_starttime;
321 int tw_time;
322 TAILQ_ENTRY(tcptw) tw_2msl;
323};
324
325#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
326#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
327#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
328
329/*
330 * The smoothed round-trip time and estimated variance
331 * are stored as fixed point numbers scaled by the values below.
332 * For convenience, these scales are also used in smoothing the average
333 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
334 * With these scales, srtt has 3 bits to the right of the binary point,
335 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
336 * binary point, and is smoothed with an ALPHA of 0.75.
337 */
338#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
339#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
340#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
341#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
342#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
343
344/*
345 * The initial retransmission should happen at rtt + 4 * rttvar.
346 * Because of the way we do the smoothing, srtt and rttvar
347 * will each average +1/2 tick of bias. When we compute
348 * the retransmit timer, we want 1/2 tick of rounding and
349 * 1 extra tick because of +-1/2 tick uncertainty in the
350 * firing of the timer. The bias will give us exactly the
351 * 1.5 tick we need. But, because the bias is
352 * statistical, we have to test that we don't drop below
353 * the minimum feasible timer (which is 2 ticks).
354 * This version of the macro adapted from a paper by Lawrence
355 * Brakmo and Larry Peterson which outlines a problem caused
356 * by insufficient precision in the original implementation,
357 * which results in inappropriately large RTO values for very
358 * fast networks.
359 */
360#define TCP_REXMTVAL(tp) \
361 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
362 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
363
364/*
365 * TCP statistics.
366 * Many of these should be kept per connection,
367 * but that's inconvenient at the moment.
368 */
369struct tcpstat {
370 u_long tcps_connattempt; /* connections initiated */
371 u_long tcps_accepts; /* connections accepted */
372 u_long tcps_connects; /* connections established */
373 u_long tcps_drops; /* connections dropped */
374 u_long tcps_conndrops; /* embryonic connections dropped */
375 u_long tcps_minmssdrops; /* average minmss too low drops */
376 u_long tcps_closed; /* conn. closed (includes drops) */
377 u_long tcps_segstimed; /* segs where we tried to get rtt */
378 u_long tcps_rttupdated; /* times we succeeded */
379 u_long tcps_delack; /* delayed acks sent */
380 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
381 u_long tcps_rexmttimeo; /* retransmit timeouts */
382 u_long tcps_persisttimeo; /* persist timeouts */
383 u_long tcps_keeptimeo; /* keepalive timeouts */
384 u_long tcps_keepprobe; /* keepalive probes sent */
385 u_long tcps_keepdrops; /* connections dropped in keepalive */
386
387 u_long tcps_sndtotal; /* total packets sent */
388 u_long tcps_sndpack; /* data packets sent */
389 u_long tcps_sndbyte; /* data bytes sent */
390 u_long tcps_sndrexmitpack; /* data packets retransmitted */
391 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
392 u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */
393 u_long tcps_sndacks; /* ack-only packets sent */
394 u_long tcps_sndprobe; /* window probes sent */
395 u_long tcps_sndurg; /* packets sent with URG only */
396 u_long tcps_sndwinup; /* window update-only packets sent */
397 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
398
399 u_long tcps_rcvtotal; /* total packets received */
400 u_long tcps_rcvpack; /* packets received in sequence */
401 u_long tcps_rcvbyte; /* bytes received in sequence */
402 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
403 u_long tcps_rcvbadoff; /* packets received with bad offset */
404 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
405 u_long tcps_rcvshort; /* packets received too short */
406 u_long tcps_rcvduppack; /* duplicate-only packets received */
407 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
408 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
409 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
410 u_long tcps_rcvoopack; /* out-of-order packets received */
411 u_long tcps_rcvoobyte; /* out-of-order bytes received */
412 u_long tcps_rcvpackafterwin; /* packets with data after window */
413 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
414 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
415 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
416 u_long tcps_rcvdupack; /* rcvd duplicate acks */
417 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
418 u_long tcps_rcvackpack; /* rcvd ack packets */
419 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
420 u_long tcps_rcvwinupd; /* rcvd window update packets */
421 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
422 u_long tcps_predack; /* times hdr predict ok for acks */
423 u_long tcps_preddat; /* times hdr predict ok for data pkts */
424 u_long tcps_pcbcachemiss;
425 u_long tcps_cachedrtt; /* times cached RTT in route updated */
426 u_long tcps_cachedrttvar; /* times cached rttvar updated */
427 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
428 u_long tcps_usedrtt; /* times RTT initialized from route */
429 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
430 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
431 u_long tcps_persistdrop; /* timeout in persist state */
432 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
433 u_long tcps_mturesent; /* resends due to MTU discovery */
434 u_long tcps_listendrop; /* listen queue overflows */
435 u_long tcps_badrst; /* ignored RSTs in the window */
436
437 u_long tcps_sc_added; /* entry added to syncache */
438 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */
439 u_long tcps_sc_dupsyn; /* duplicate SYN packet */
440 u_long tcps_sc_dropped; /* could not reply to packet */
441 u_long tcps_sc_completed; /* successful extraction of entry */
442 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
443 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */
444 u_long tcps_sc_reset; /* RST removed entry from syncache */
445 u_long tcps_sc_stale; /* timed out or listen socket gone */
446 u_long tcps_sc_aborted; /* syncache entry aborted */
447 u_long tcps_sc_badack; /* removed due to bad ACK */
448 u_long tcps_sc_unreach; /* ICMP unreachable received */
449 u_long tcps_sc_zonefail; /* zalloc() failed */
450 u_long tcps_sc_sendcookie; /* SYN cookie sent */
451 u_long tcps_sc_recvcookie; /* SYN cookie received */
452
453 u_long tcps_hc_added; /* entry added to hostcache */
454 u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
455
456 u_long tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */
457
458 /* SACK related stats */
459 u_long tcps_sack_recovery_episode; /* SACK recovery episodes */
460 u_long tcps_sack_rexmits; /* SACK rexmit segments */
461 u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
462 u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */
463 u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */
464 u_long tcps_sack_sboverflow; /* times scoreboard overflowed */
465
466 /* ECN related stats */
467 u_long tcps_ecn_ce; /* ECN Congestion Experienced */
468 u_long tcps_ecn_ect0; /* ECN Capable Transport */
469 u_long tcps_ecn_ect1; /* ECN Capable Transport */
470 u_long tcps_ecn_shs; /* ECN successful handshakes */
471 u_long tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */
472
473 u_long _pad[12]; /* 6 UTO, 6 TBD */
474};
475
476#ifdef _KERNEL
477/*
478 * In-kernel consumers can use these accessor macros directly to update
479 * stats.
480 */
481#define TCPSTAT_ADD(name, val) V_tcpstat.name += (val)
482#define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1)
483
484/*
485 * Kernel module consumers must use this accessor macro.
486 */
487void kmod_tcpstat_inc(int statnum);
488#define KMOD_TCPSTAT_INC(name) \
489 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(u_long))
490#endif
491
492/*
493 * TCB structure exported to user-land via sysctl(3).
494 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
495 * included. Not all of our clients do.
496 */
497#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
498struct xtcp_timer {
499 int tt_rexmt; /* retransmit timer */
500 int tt_persist; /* retransmit persistence */
501 int tt_keep; /* keepalive */
502 int tt_2msl; /* 2*msl TIME_WAIT timer */
503 int tt_delack; /* delayed ACK timer */
504 int t_rcvtime; /* Time since last packet received */
505};
506struct xtcpcb {
507 size_t xt_len;
508 struct inpcb xt_inp;
509 struct tcpcb xt_tp;
510 struct xsocket xt_socket;
511 struct xtcp_timer xt_timer;
512 u_quad_t xt_alignment_hack;
513};
514#endif
515
516/*
517 * Names for TCP sysctl objects
518 */
519#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
520#define TCPCTL_MSSDFLT 3 /* MSS default */
521#define TCPCTL_STATS 4 /* statistics (read-only) */
522#define TCPCTL_RTTDFLT 5 /* default RTT estimate */
523#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
524#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
525#define TCPCTL_SENDSPACE 8 /* send buffer space */
526#define TCPCTL_RECVSPACE 9 /* receive buffer space */
527#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
528#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
529#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
530#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
531#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
532#define TCPCTL_DROP 15 /* drop tcp connection */
533#define TCPCTL_MAXID 16
534#define TCPCTL_FINWAIT2_TIMEOUT 17
535
536#define TCPCTL_NAMES { \
537 { 0, 0 }, \
538 { "rfc1323", CTLTYPE_INT }, \
539 { "mssdflt", CTLTYPE_INT }, \
540 { "stats", CTLTYPE_STRUCT }, \
541 { "rttdflt", CTLTYPE_INT }, \
542 { "keepidle", CTLTYPE_INT }, \
543 { "keepintvl", CTLTYPE_INT }, \
544 { "sendspace", CTLTYPE_INT }, \
545 { "recvspace", CTLTYPE_INT }, \
546 { "keepinit", CTLTYPE_INT }, \
547 { "pcblist", CTLTYPE_STRUCT }, \
548 { "delacktime", CTLTYPE_INT }, \
549 { "v6mssdflt", CTLTYPE_INT }, \
550 { "maxid", CTLTYPE_INT }, \
551}
552
553
554#ifdef _KERNEL
555#ifdef SYSCTL_DECL
556SYSCTL_DECL(_net_inet_tcp);
557SYSCTL_DECL(_net_inet_tcp_sack);
558MALLOC_DECLARE(M_TCPLOG);
559#endif
560
| 51#endif /* _KERNEL */
52
53/* TCP segment queue entry */
54struct tseg_qent {
55 LIST_ENTRY(tseg_qent) tqe_q;
56 int tqe_len; /* TCP segment data length */
57 struct tcphdr *tqe_th; /* a pointer to tcp header */
58 struct mbuf *tqe_m; /* mbuf contains packet */
59};
60LIST_HEAD(tsegqe_head, tseg_qent);
61
62struct sackblk {
63 tcp_seq start; /* start seq no. of sack block */
64 tcp_seq end; /* end seq no. */
65};
66
67struct sackhole {
68 tcp_seq start; /* start seq no. of hole */
69 tcp_seq end; /* end seq no. */
70 tcp_seq rxmit; /* next seq. no in hole to be retransmitted */
71 TAILQ_ENTRY(sackhole) scblink; /* scoreboard linkage */
72};
73
74struct sackhint {
75 struct sackhole *nexthole;
76 int sack_bytes_rexmit;
77
78 int ispare; /* explicit pad for 64bit alignment */
79 uint64_t _pad[2]; /* 1 sacked_bytes, 1 TBD */
80};
81
82struct tcptemp {
83 u_char tt_ipgen[40]; /* the size must be of max ip header, now IPv6 */
84 struct tcphdr tt_t;
85};
86
87#define tcp6cb tcpcb /* for KAME src sync over BSD*'s */
88
89/* Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint. */
90#ifdef INET6
91#define ND6_HINT(tp) \
92do { \
93 if ((tp) && (tp)->t_inpcb && \
94 ((tp)->t_inpcb->inp_vflag & INP_IPV6) != 0) \
95 nd6_nud_hint(NULL, NULL, 0); \
96} while (0)
97#else
98#define ND6_HINT(tp)
99#endif
100
101/*
102 * Tcp control block, one per tcp; fields:
103 * Organized for 16 byte cacheline efficiency.
104 */
105struct tcpcb {
106 struct tsegqe_head t_segq; /* segment reassembly queue */
107 void *t_pspare[2]; /* new reassembly queue */
108 int t_segqlen; /* segment reassembly queue length */
109 int t_dupacks; /* consecutive dup acks recd */
110
111 struct tcp_timer *t_timers; /* All the TCP timers in one struct */
112
113 struct inpcb *t_inpcb; /* back pointer to internet pcb */
114 int t_state; /* state of this connection */
115 u_int t_flags;
116
117 struct vnet *t_vnet; /* back pointer to parent vnet */
118
119 tcp_seq snd_una; /* send unacknowledged */
120 tcp_seq snd_max; /* highest sequence number sent;
121 * used to recognize retransmits
122 */
123 tcp_seq snd_nxt; /* send next */
124 tcp_seq snd_up; /* send urgent pointer */
125
126 tcp_seq snd_wl1; /* window update seg seq number */
127 tcp_seq snd_wl2; /* window update seg ack number */
128 tcp_seq iss; /* initial send sequence number */
129 tcp_seq irs; /* initial receive sequence number */
130
131 tcp_seq rcv_nxt; /* receive next */
132 tcp_seq rcv_adv; /* advertised window */
133 u_long rcv_wnd; /* receive window */
134 tcp_seq rcv_up; /* receive urgent pointer */
135
136 u_long snd_wnd; /* send window */
137 u_long snd_cwnd; /* congestion-controlled window */
138 u_long snd_bwnd; /* bandwidth-controlled window */
139 u_long snd_ssthresh; /* snd_cwnd size threshold for
140 * for slow start exponential to
141 * linear switch
142 */
143 u_long snd_bandwidth; /* calculated bandwidth or 0 */
144 tcp_seq snd_recover; /* for use in NewReno Fast Recovery */
145
146 u_int t_maxopd; /* mss plus options */
147
148 u_int t_rcvtime; /* inactivity time */
149 u_int t_starttime; /* time connection was established */
150 u_int t_rtttime; /* RTT measurement start time */
151 tcp_seq t_rtseq; /* sequence number being timed */
152
153 u_int t_bw_rtttime; /* used for bandwidth calculation */
154 tcp_seq t_bw_rtseq; /* used for bandwidth calculation */
155
156 int t_rxtcur; /* current retransmit value (ticks) */
157 u_int t_maxseg; /* maximum segment size */
158 int t_srtt; /* smoothed round-trip time */
159 int t_rttvar; /* variance in round-trip time */
160
161 int t_rxtshift; /* log(2) of rexmt exp. backoff */
162 u_int t_rttmin; /* minimum rtt allowed */
163 u_int t_rttbest; /* best rtt we've seen */
164 u_long t_rttupdated; /* number of times rtt sampled */
165 u_long max_sndwnd; /* largest window peer has offered */
166
167 int t_softerror; /* possible error not yet reported */
168/* out-of-band data */
169 char t_oobflags; /* have some */
170 char t_iobc; /* input character */
171/* RFC 1323 variables */
172 u_char snd_scale; /* window scaling for send window */
173 u_char rcv_scale; /* window scaling for recv window */
174 u_char request_r_scale; /* pending window scaling */
175 u_int32_t ts_recent; /* timestamp echo data */
176 u_int ts_recent_age; /* when last updated */
177 u_int32_t ts_offset; /* our timestamp offset */
178
179 tcp_seq last_ack_sent;
180/* experimental */
181 u_long snd_cwnd_prev; /* cwnd prior to retransmit */
182 u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */
183 tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */
184 u_int t_badrxtwin; /* window for retransmit recovery */
185 u_char snd_limited; /* segments limited transmitted */
186/* SACK related state */
187 int snd_numholes; /* number of holes seen by sender */
188 TAILQ_HEAD(sackhole_head, sackhole) snd_holes;
189 /* SACK scoreboard (sorted) */
190 tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/
191 int rcv_numsacks; /* # distinct sack blks present */
192 struct sackblk sackblks[MAX_SACK_BLKS]; /* seq nos. of sack blocks */
193 tcp_seq sack_newdata; /* New data xmitted in this recovery
194 episode starts at this seq number */
195 struct sackhint sackhint; /* SACK scoreboard hint */
196 int t_rttlow; /* smallest observerved RTT */
197 u_int32_t rfbuf_ts; /* recv buffer autoscaling timestamp */
198 int rfbuf_cnt; /* recv buffer autoscaling byte count */
199 struct toe_usrreqs *t_tu; /* offload operations vector */
200 void *t_toe; /* TOE pcb pointer */
201 int t_bytes_acked; /* # bytes acked during current RTT */
202
203 int t_ispare; /* explicit pad for 64bit alignment */
204 void *t_pspare2[6]; /* 2 CC / 4 TBD */
205 uint64_t _pad[12]; /* 7 UTO, 5 TBD (1-2 CC/RTT?) */
206};
207
208/*
209 * Flags and utility macros for the t_flags field.
210 */
211#define TF_ACKNOW 0x000001 /* ack peer immediately */
212#define TF_DELACK 0x000002 /* ack, but try to delay it */
213#define TF_NODELAY 0x000004 /* don't delay packets to coalesce */
214#define TF_NOOPT 0x000008 /* don't use tcp options */
215#define TF_SENTFIN 0x000010 /* have sent FIN */
216#define TF_REQ_SCALE 0x000020 /* have/will request window scaling */
217#define TF_RCVD_SCALE 0x000040 /* other side has requested scaling */
218#define TF_REQ_TSTMP 0x000080 /* have/will request timestamps */
219#define TF_RCVD_TSTMP 0x000100 /* a timestamp was received in SYN */
220#define TF_SACK_PERMIT 0x000200 /* other side said I could SACK */
221#define TF_NEEDSYN 0x000400 /* send SYN (implicit state) */
222#define TF_NEEDFIN 0x000800 /* send FIN (implicit state) */
223#define TF_NOPUSH 0x001000 /* don't push */
224#define TF_MORETOCOME 0x010000 /* More data to be appended to sock */
225#define TF_LQ_OVERFLOW 0x020000 /* listen queue overflow */
226#define TF_LASTIDLE 0x040000 /* connection was previously idle */
227#define TF_RXWIN0SENT 0x080000 /* sent a receiver win 0 in response */
228#define TF_FASTRECOVERY 0x100000 /* in NewReno Fast Recovery */
229#define TF_WASFRECOVERY 0x200000 /* was in NewReno Fast Recovery */
230#define TF_SIGNATURE 0x400000 /* require MD5 digests (RFC2385) */
231#define TF_FORCEDATA 0x800000 /* force out a byte */
232#define TF_TSO 0x1000000 /* TSO enabled on this connection */
233#define TF_TOE 0x2000000 /* this connection is offloaded */
234#define TF_ECN_PERMIT 0x4000000 /* connection ECN-ready */
235#define TF_ECN_SND_CWR 0x8000000 /* ECN CWR in queue */
236#define TF_ECN_SND_ECE 0x10000000 /* ECN ECE in queue */
237
238#define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY)
239#define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY
240#define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY
241
242/*
243 * Flags for the t_oobflags field.
244 */
245#define TCPOOB_HAVEDATA 0x01
246#define TCPOOB_HADDATA 0x02
247
248#ifdef TCP_SIGNATURE
249/*
250 * Defines which are needed by the xform_tcp module and tcp_[in|out]put
251 * for SADB verification and lookup.
252 */
253#define TCP_SIGLEN 16 /* length of computed digest in bytes */
254#define TCP_KEYLEN_MIN 1 /* minimum length of TCP-MD5 key */
255#define TCP_KEYLEN_MAX 80 /* maximum length of TCP-MD5 key */
256/*
257 * Only a single SA per host may be specified at this time. An SPI is
258 * needed in order for the KEY_ALLOCSA() lookup to work.
259 */
260#define TCP_SIG_SPI 0x1000
261#endif /* TCP_SIGNATURE */
262
263/*
264 * Structure to hold TCP options that are only used during segment
265 * processing (in tcp_input), but not held in the tcpcb.
266 * It's basically used to reduce the number of parameters
267 * to tcp_dooptions and tcp_addoptions.
268 * The binary order of the to_flags is relevant for packing of the
269 * options in tcp_addoptions.
270 */
271struct tcpopt {
272 u_int64_t to_flags; /* which options are present */
273#define TOF_MSS 0x0001 /* maximum segment size */
274#define TOF_SCALE 0x0002 /* window scaling */
275#define TOF_SACKPERM 0x0004 /* SACK permitted */
276#define TOF_TS 0x0010 /* timestamp */
277#define TOF_SIGNATURE 0x0040 /* TCP-MD5 signature option (RFC2385) */
278#define TOF_SACK 0x0080 /* Peer sent SACK option */
279#define TOF_MAXOPT 0x0100
280 u_int32_t to_tsval; /* new timestamp */
281 u_int32_t to_tsecr; /* reflected timestamp */
282 u_char *to_sacks; /* pointer to the first SACK blocks */
283 u_char *to_signature; /* pointer to the TCP-MD5 signature */
284 u_int16_t to_mss; /* maximum segment size */
285 u_int8_t to_wscale; /* window scaling */
286 u_int8_t to_nsacks; /* number of SACK blocks */
287};
288
289/*
290 * Flags for tcp_dooptions.
291 */
292#define TO_SYN 0x01 /* parse SYN-only options */
293
294struct hc_metrics_lite { /* must stay in sync with hc_metrics */
295 u_long rmx_mtu; /* MTU for this path */
296 u_long rmx_ssthresh; /* outbound gateway buffer limit */
297 u_long rmx_rtt; /* estimated round trip time */
298 u_long rmx_rttvar; /* estimated rtt variance */
299 u_long rmx_bandwidth; /* estimated bandwidth */
300 u_long rmx_cwnd; /* congestion window */
301 u_long rmx_sendpipe; /* outbound delay-bandwidth product */
302 u_long rmx_recvpipe; /* inbound delay-bandwidth product */
303};
304
305#ifndef _NETINET_IN_PCB_H_
306struct in_conninfo;
307#endif /* _NETINET_IN_PCB_H_ */
308
309struct tcptw {
310 struct inpcb *tw_inpcb; /* XXX back pointer to internet pcb */
311 tcp_seq snd_nxt;
312 tcp_seq rcv_nxt;
313 tcp_seq iss;
314 tcp_seq irs;
315 u_short last_win; /* cached window value */
316 u_short tw_so_options; /* copy of so_options */
317 struct ucred *tw_cred; /* user credentials */
318 u_int32_t t_recent;
319 u_int32_t ts_offset; /* our timestamp offset */
320 u_int t_starttime;
321 int tw_time;
322 TAILQ_ENTRY(tcptw) tw_2msl;
323};
324
325#define intotcpcb(ip) ((struct tcpcb *)(ip)->inp_ppcb)
326#define intotw(ip) ((struct tcptw *)(ip)->inp_ppcb)
327#define sototcpcb(so) (intotcpcb(sotoinpcb(so)))
328
329/*
330 * The smoothed round-trip time and estimated variance
331 * are stored as fixed point numbers scaled by the values below.
332 * For convenience, these scales are also used in smoothing the average
333 * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed).
334 * With these scales, srtt has 3 bits to the right of the binary point,
335 * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the
336 * binary point, and is smoothed with an ALPHA of 0.75.
337 */
338#define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */
339#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */
340#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */
341#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */
342#define TCP_DELTA_SHIFT 2 /* see tcp_input.c */
343
344/*
345 * The initial retransmission should happen at rtt + 4 * rttvar.
346 * Because of the way we do the smoothing, srtt and rttvar
347 * will each average +1/2 tick of bias. When we compute
348 * the retransmit timer, we want 1/2 tick of rounding and
349 * 1 extra tick because of +-1/2 tick uncertainty in the
350 * firing of the timer. The bias will give us exactly the
351 * 1.5 tick we need. But, because the bias is
352 * statistical, we have to test that we don't drop below
353 * the minimum feasible timer (which is 2 ticks).
354 * This version of the macro adapted from a paper by Lawrence
355 * Brakmo and Larry Peterson which outlines a problem caused
356 * by insufficient precision in the original implementation,
357 * which results in inappropriately large RTO values for very
358 * fast networks.
359 */
360#define TCP_REXMTVAL(tp) \
361 max((tp)->t_rttmin, (((tp)->t_srtt >> (TCP_RTT_SHIFT - TCP_DELTA_SHIFT)) \
362 + (tp)->t_rttvar) >> TCP_DELTA_SHIFT)
363
364/*
365 * TCP statistics.
366 * Many of these should be kept per connection,
367 * but that's inconvenient at the moment.
368 */
369struct tcpstat {
370 u_long tcps_connattempt; /* connections initiated */
371 u_long tcps_accepts; /* connections accepted */
372 u_long tcps_connects; /* connections established */
373 u_long tcps_drops; /* connections dropped */
374 u_long tcps_conndrops; /* embryonic connections dropped */
375 u_long tcps_minmssdrops; /* average minmss too low drops */
376 u_long tcps_closed; /* conn. closed (includes drops) */
377 u_long tcps_segstimed; /* segs where we tried to get rtt */
378 u_long tcps_rttupdated; /* times we succeeded */
379 u_long tcps_delack; /* delayed acks sent */
380 u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */
381 u_long tcps_rexmttimeo; /* retransmit timeouts */
382 u_long tcps_persisttimeo; /* persist timeouts */
383 u_long tcps_keeptimeo; /* keepalive timeouts */
384 u_long tcps_keepprobe; /* keepalive probes sent */
385 u_long tcps_keepdrops; /* connections dropped in keepalive */
386
387 u_long tcps_sndtotal; /* total packets sent */
388 u_long tcps_sndpack; /* data packets sent */
389 u_long tcps_sndbyte; /* data bytes sent */
390 u_long tcps_sndrexmitpack; /* data packets retransmitted */
391 u_long tcps_sndrexmitbyte; /* data bytes retransmitted */
392 u_long tcps_sndrexmitbad; /* unnecessary packet retransmissions */
393 u_long tcps_sndacks; /* ack-only packets sent */
394 u_long tcps_sndprobe; /* window probes sent */
395 u_long tcps_sndurg; /* packets sent with URG only */
396 u_long tcps_sndwinup; /* window update-only packets sent */
397 u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */
398
399 u_long tcps_rcvtotal; /* total packets received */
400 u_long tcps_rcvpack; /* packets received in sequence */
401 u_long tcps_rcvbyte; /* bytes received in sequence */
402 u_long tcps_rcvbadsum; /* packets received with ccksum errs */
403 u_long tcps_rcvbadoff; /* packets received with bad offset */
404 u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */
405 u_long tcps_rcvshort; /* packets received too short */
406 u_long tcps_rcvduppack; /* duplicate-only packets received */
407 u_long tcps_rcvdupbyte; /* duplicate-only bytes received */
408 u_long tcps_rcvpartduppack; /* packets with some duplicate data */
409 u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */
410 u_long tcps_rcvoopack; /* out-of-order packets received */
411 u_long tcps_rcvoobyte; /* out-of-order bytes received */
412 u_long tcps_rcvpackafterwin; /* packets with data after window */
413 u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */
414 u_long tcps_rcvafterclose; /* packets rcvd after "close" */
415 u_long tcps_rcvwinprobe; /* rcvd window probe packets */
416 u_long tcps_rcvdupack; /* rcvd duplicate acks */
417 u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */
418 u_long tcps_rcvackpack; /* rcvd ack packets */
419 u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */
420 u_long tcps_rcvwinupd; /* rcvd window update packets */
421 u_long tcps_pawsdrop; /* segments dropped due to PAWS */
422 u_long tcps_predack; /* times hdr predict ok for acks */
423 u_long tcps_preddat; /* times hdr predict ok for data pkts */
424 u_long tcps_pcbcachemiss;
425 u_long tcps_cachedrtt; /* times cached RTT in route updated */
426 u_long tcps_cachedrttvar; /* times cached rttvar updated */
427 u_long tcps_cachedssthresh; /* times cached ssthresh updated */
428 u_long tcps_usedrtt; /* times RTT initialized from route */
429 u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */
430 u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/
431 u_long tcps_persistdrop; /* timeout in persist state */
432 u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */
433 u_long tcps_mturesent; /* resends due to MTU discovery */
434 u_long tcps_listendrop; /* listen queue overflows */
435 u_long tcps_badrst; /* ignored RSTs in the window */
436
437 u_long tcps_sc_added; /* entry added to syncache */
438 u_long tcps_sc_retransmitted; /* syncache entry was retransmitted */
439 u_long tcps_sc_dupsyn; /* duplicate SYN packet */
440 u_long tcps_sc_dropped; /* could not reply to packet */
441 u_long tcps_sc_completed; /* successful extraction of entry */
442 u_long tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */
443 u_long tcps_sc_cacheoverflow; /* syncache cache limit hit */
444 u_long tcps_sc_reset; /* RST removed entry from syncache */
445 u_long tcps_sc_stale; /* timed out or listen socket gone */
446 u_long tcps_sc_aborted; /* syncache entry aborted */
447 u_long tcps_sc_badack; /* removed due to bad ACK */
448 u_long tcps_sc_unreach; /* ICMP unreachable received */
449 u_long tcps_sc_zonefail; /* zalloc() failed */
450 u_long tcps_sc_sendcookie; /* SYN cookie sent */
451 u_long tcps_sc_recvcookie; /* SYN cookie received */
452
453 u_long tcps_hc_added; /* entry added to hostcache */
454 u_long tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */
455
456 u_long tcps_finwait2_drops; /* Drop FIN_WAIT_2 connection after time limit */
457
458 /* SACK related stats */
459 u_long tcps_sack_recovery_episode; /* SACK recovery episodes */
460 u_long tcps_sack_rexmits; /* SACK rexmit segments */
461 u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */
462 u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */
463 u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */
464 u_long tcps_sack_sboverflow; /* times scoreboard overflowed */
465
466 /* ECN related stats */
467 u_long tcps_ecn_ce; /* ECN Congestion Experienced */
468 u_long tcps_ecn_ect0; /* ECN Capable Transport */
469 u_long tcps_ecn_ect1; /* ECN Capable Transport */
470 u_long tcps_ecn_shs; /* ECN successful handshakes */
471 u_long tcps_ecn_rcwnd; /* # times ECN reduced the cwnd */
472
473 u_long _pad[12]; /* 6 UTO, 6 TBD */
474};
475
476#ifdef _KERNEL
477/*
478 * In-kernel consumers can use these accessor macros directly to update
479 * stats.
480 */
481#define TCPSTAT_ADD(name, val) V_tcpstat.name += (val)
482#define TCPSTAT_INC(name) TCPSTAT_ADD(name, 1)
483
484/*
485 * Kernel module consumers must use this accessor macro.
486 */
487void kmod_tcpstat_inc(int statnum);
488#define KMOD_TCPSTAT_INC(name) \
489 kmod_tcpstat_inc(offsetof(struct tcpstat, name) / sizeof(u_long))
490#endif
491
492/*
493 * TCB structure exported to user-land via sysctl(3).
494 * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been
495 * included. Not all of our clients do.
496 */
497#if defined(_NETINET_IN_PCB_H_) && defined(_SYS_SOCKETVAR_H_)
498struct xtcp_timer {
499 int tt_rexmt; /* retransmit timer */
500 int tt_persist; /* retransmit persistence */
501 int tt_keep; /* keepalive */
502 int tt_2msl; /* 2*msl TIME_WAIT timer */
503 int tt_delack; /* delayed ACK timer */
504 int t_rcvtime; /* Time since last packet received */
505};
506struct xtcpcb {
507 size_t xt_len;
508 struct inpcb xt_inp;
509 struct tcpcb xt_tp;
510 struct xsocket xt_socket;
511 struct xtcp_timer xt_timer;
512 u_quad_t xt_alignment_hack;
513};
514#endif
515
516/*
517 * Names for TCP sysctl objects
518 */
519#define TCPCTL_DO_RFC1323 1 /* use RFC-1323 extensions */
520#define TCPCTL_MSSDFLT 3 /* MSS default */
521#define TCPCTL_STATS 4 /* statistics (read-only) */
522#define TCPCTL_RTTDFLT 5 /* default RTT estimate */
523#define TCPCTL_KEEPIDLE 6 /* keepalive idle timer */
524#define TCPCTL_KEEPINTVL 7 /* interval to send keepalives */
525#define TCPCTL_SENDSPACE 8 /* send buffer space */
526#define TCPCTL_RECVSPACE 9 /* receive buffer space */
527#define TCPCTL_KEEPINIT 10 /* timeout for establishing syn */
528#define TCPCTL_PCBLIST 11 /* list of all outstanding PCBs */
529#define TCPCTL_DELACKTIME 12 /* time before sending delayed ACK */
530#define TCPCTL_V6MSSDFLT 13 /* MSS default for IPv6 */
531#define TCPCTL_SACK 14 /* Selective Acknowledgement,rfc 2018 */
532#define TCPCTL_DROP 15 /* drop tcp connection */
533#define TCPCTL_MAXID 16
534#define TCPCTL_FINWAIT2_TIMEOUT 17
535
536#define TCPCTL_NAMES { \
537 { 0, 0 }, \
538 { "rfc1323", CTLTYPE_INT }, \
539 { "mssdflt", CTLTYPE_INT }, \
540 { "stats", CTLTYPE_STRUCT }, \
541 { "rttdflt", CTLTYPE_INT }, \
542 { "keepidle", CTLTYPE_INT }, \
543 { "keepintvl", CTLTYPE_INT }, \
544 { "sendspace", CTLTYPE_INT }, \
545 { "recvspace", CTLTYPE_INT }, \
546 { "keepinit", CTLTYPE_INT }, \
547 { "pcblist", CTLTYPE_STRUCT }, \
548 { "delacktime", CTLTYPE_INT }, \
549 { "v6mssdflt", CTLTYPE_INT }, \
550 { "maxid", CTLTYPE_INT }, \
551}
552
553
554#ifdef _KERNEL
555#ifdef SYSCTL_DECL
556SYSCTL_DECL(_net_inet_tcp);
557SYSCTL_DECL(_net_inet_tcp_sack);
558MALLOC_DECLARE(M_TCPLOG);
559#endif
560
|
561extern int tcp_log_in_vain;
562
| |
563VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */
564VNET_DECLARE(struct inpcbinfo, tcbinfo);
565VNET_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */
| 561VNET_DECLARE(struct inpcbhead, tcb); /* queue of active tcpcb's */
562VNET_DECLARE(struct inpcbinfo, tcbinfo);
563VNET_DECLARE(struct tcpstat, tcpstat); /* tcp statistics */
|
| 564extern int tcp_log_in_vain;
|
566VNET_DECLARE(int, tcp_mssdflt); /* XXX */
567VNET_DECLARE(int, tcp_minmss);
568VNET_DECLARE(int, tcp_delack_enabled);
569VNET_DECLARE(int, tcp_do_newreno);
570VNET_DECLARE(int, path_mtu_discovery);
571VNET_DECLARE(int, ss_fltsz);
572VNET_DECLARE(int, ss_fltsz_local);
| 565VNET_DECLARE(int, tcp_mssdflt); /* XXX */
566VNET_DECLARE(int, tcp_minmss);
567VNET_DECLARE(int, tcp_delack_enabled);
568VNET_DECLARE(int, tcp_do_newreno);
569VNET_DECLARE(int, path_mtu_discovery);
570VNET_DECLARE(int, ss_fltsz);
571VNET_DECLARE(int, ss_fltsz_local);
|
573
| |
574#define V_tcb VNET(tcb)
575#define V_tcbinfo VNET(tcbinfo)
576#define V_tcpstat VNET(tcpstat)
577#define V_tcp_mssdflt VNET(tcp_mssdflt)
578#define V_tcp_minmss VNET(tcp_minmss)
579#define V_tcp_delack_enabled VNET(tcp_delack_enabled)
580#define V_tcp_do_newreno VNET(tcp_do_newreno)
581#define V_path_mtu_discovery VNET(path_mtu_discovery)
582#define V_ss_fltsz VNET(ss_fltsz)
583#define V_ss_fltsz_local VNET(ss_fltsz_local)
584
| 572#define V_tcb VNET(tcb)
573#define V_tcbinfo VNET(tcbinfo)
574#define V_tcpstat VNET(tcpstat)
575#define V_tcp_mssdflt VNET(tcp_mssdflt)
576#define V_tcp_minmss VNET(tcp_minmss)
577#define V_tcp_delack_enabled VNET(tcp_delack_enabled)
578#define V_tcp_do_newreno VNET(tcp_do_newreno)
579#define V_path_mtu_discovery VNET(path_mtu_discovery)
580#define V_ss_fltsz VNET(ss_fltsz)
581#define V_ss_fltsz_local VNET(ss_fltsz_local)
582
|
585VNET_DECLARE(int, blackhole);
586VNET_DECLARE(int, drop_synfin);
587VNET_DECLARE(int, tcp_do_rfc3042);
588VNET_DECLARE(int, tcp_do_rfc3390);
589VNET_DECLARE(int, tcp_insecure_rst);
590VNET_DECLARE(int, tcp_do_autorcvbuf);
591VNET_DECLARE(int, tcp_autorcvbuf_inc);
592VNET_DECLARE(int, tcp_autorcvbuf_max);
593VNET_DECLARE(int, tcp_do_rfc3465);
594VNET_DECLARE(int, tcp_abc_l_var);
595
596#define V_blackhole VNET(blackhole)
597#define V_drop_synfin VNET(drop_synfin)
598#define V_tcp_do_rfc3042 VNET(tcp_do_rfc3042)
599#define V_tcp_do_rfc3390 VNET(tcp_do_rfc3390)
600#define V_tcp_insecure_rst VNET(tcp_insecure_rst)
601#define V_tcp_do_autorcvbuf VNET(tcp_do_autorcvbuf)
602#define V_tcp_autorcvbuf_inc VNET(tcp_autorcvbuf_inc)
603#define V_tcp_autorcvbuf_max VNET(tcp_autorcvbuf_max)
604#define V_tcp_do_rfc3465 VNET(tcp_do_rfc3465)
605#define V_tcp_abc_l_var VNET(tcp_abc_l_var)
606
607VNET_DECLARE(int, tcp_do_tso);
608VNET_DECLARE(int, tcp_do_autosndbuf);
609VNET_DECLARE(int, tcp_autosndbuf_inc);
610VNET_DECLARE(int, tcp_autosndbuf_max);
611
612#define V_tcp_do_tso VNET(tcp_do_tso)
613#define V_tcp_do_autosndbuf VNET(tcp_do_autosndbuf)
614#define V_tcp_autosndbuf_inc VNET(tcp_autosndbuf_inc)
615#define V_tcp_autosndbuf_max VNET(tcp_autosndbuf_max)
616
617VNET_DECLARE(int, nolocaltimewait);
618
619#define V_nolocaltimewait VNET(nolocaltimewait)
620
| |
621VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */
| 583VNET_DECLARE(int, tcp_do_sack); /* SACK enabled/disabled */
|
622VNET_DECLARE(int, tcp_sack_maxholes);
623VNET_DECLARE(int, tcp_sack_globalmaxholes);
624VNET_DECLARE(int, tcp_sack_globalholes);
| |
625VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */
| 584VNET_DECLARE(int, tcp_sc_rst_sock_fail); /* RST on sock alloc failure */
|
626VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */
627VNET_DECLARE(int, tcp_ecn_maxretries);
628
| |
629#define V_tcp_do_sack VNET(tcp_do_sack)
| 585#define V_tcp_do_sack VNET(tcp_do_sack)
|
630#define V_tcp_sack_maxholes VNET(tcp_sack_maxholes)
631#define V_tcp_sack_globalmaxholes VNET(tcp_sack_globalmaxholes)
632#define V_tcp_sack_globalholes VNET(tcp_sack_globalholes)
| |
633#define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail)
| 586#define V_tcp_sc_rst_sock_fail VNET(tcp_sc_rst_sock_fail)
|
| 587
588VNET_DECLARE(int, tcp_do_ecn); /* TCP ECN enabled/disabled */
589VNET_DECLARE(int, tcp_ecn_maxretries);
|
634#define V_tcp_do_ecn VNET(tcp_do_ecn)
635#define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries)
636
637int tcp_addoptions(struct tcpopt *, u_char *);
638struct tcpcb *
639 tcp_close(struct tcpcb *);
640void tcp_discardcb(struct tcpcb *);
641void tcp_twstart(struct tcpcb *);
642#if 0
643int tcp_twrecycleable(struct tcptw *tw);
644#endif
645void tcp_twclose(struct tcptw *_tw, int _reuse);
646void tcp_ctlinput(int, struct sockaddr *, void *);
647int tcp_ctloutput(struct socket *, struct sockopt *);
648struct tcpcb *
649 tcp_drop(struct tcpcb *, int);
650void tcp_drain(void);
651void tcp_init(void);
652#ifdef VIMAGE
653void tcp_destroy(void);
654#endif
655void tcp_fini(void *);
656char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
657 const void *);
658int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *);
659void tcp_reass_init(void);
660#ifdef VIMAGE
661void tcp_reass_destroy(void);
662#endif
663void tcp_input(struct mbuf *, int);
664u_long tcp_maxmtu(struct in_conninfo *, int *);
665u_long tcp_maxmtu6(struct in_conninfo *, int *);
666void tcp_mss_update(struct tcpcb *, int, struct hc_metrics_lite *, int *);
667void tcp_mss(struct tcpcb *, int);
668int tcp_mssopt(struct in_conninfo *);
669struct inpcb *
670 tcp_drop_syn_sent(struct inpcb *, int);
671struct inpcb *
672 tcp_mtudisc(struct inpcb *, int);
673struct tcpcb *
674 tcp_newtcpcb(struct inpcb *);
675int tcp_output(struct tcpcb *);
676void tcp_respond(struct tcpcb *, void *,
677 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
678void tcp_tw_init(void);
679#ifdef VIMAGE
680void tcp_tw_destroy(void);
681#endif
682void tcp_tw_zone_change(void);
683int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
684 struct mbuf *, int);
685int tcp_twrespond(struct tcptw *, int);
686void tcp_setpersist(struct tcpcb *);
687#ifdef TCP_SIGNATURE
688int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
689#endif
690void tcp_slowtimo(void);
691struct tcptemp *
692 tcpip_maketemplate(struct inpcb *);
693void tcpip_fillheaders(struct inpcb *, void *, void *);
694void tcp_timer_activate(struct tcpcb *, int, u_int);
695int tcp_timer_active(struct tcpcb *, int);
696void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
697void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
698/*
699 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
700 */
701void tcp_hc_init(void);
702#ifdef VIMAGE
703void tcp_hc_destroy(void);
704#endif
705void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
706u_long tcp_hc_getmtu(struct in_conninfo *);
707void tcp_hc_updatemtu(struct in_conninfo *, u_long);
708void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
709
710extern struct pr_usrreqs tcp_usrreqs;
711extern u_long tcp_sendspace;
712extern u_long tcp_recvspace;
713tcp_seq tcp_new_isn(struct tcpcb *);
714
715void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
716void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
717void tcp_clean_sackreport(struct tcpcb *tp);
718void tcp_sack_adjust(struct tcpcb *tp);
719struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
720void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
721void tcp_free_sackholes(struct tcpcb *tp);
722int tcp_newreno(struct tcpcb *, struct tcphdr *);
723u_long tcp_seq_subtract(u_long, u_long );
724
725#endif /* _KERNEL */
726
727#endif /* _NETINET_TCP_VAR_H_ */
| 590#define V_tcp_do_ecn VNET(tcp_do_ecn)
591#define V_tcp_ecn_maxretries VNET(tcp_ecn_maxretries)
592
593int tcp_addoptions(struct tcpopt *, u_char *);
594struct tcpcb *
595 tcp_close(struct tcpcb *);
596void tcp_discardcb(struct tcpcb *);
597void tcp_twstart(struct tcpcb *);
598#if 0
599int tcp_twrecycleable(struct tcptw *tw);
600#endif
601void tcp_twclose(struct tcptw *_tw, int _reuse);
602void tcp_ctlinput(int, struct sockaddr *, void *);
603int tcp_ctloutput(struct socket *, struct sockopt *);
604struct tcpcb *
605 tcp_drop(struct tcpcb *, int);
606void tcp_drain(void);
607void tcp_init(void);
608#ifdef VIMAGE
609void tcp_destroy(void);
610#endif
611void tcp_fini(void *);
612char *tcp_log_addrs(struct in_conninfo *, struct tcphdr *, void *,
613 const void *);
614int tcp_reass(struct tcpcb *, struct tcphdr *, int *, struct mbuf *);
615void tcp_reass_init(void);
616#ifdef VIMAGE
617void tcp_reass_destroy(void);
618#endif
619void tcp_input(struct mbuf *, int);
620u_long tcp_maxmtu(struct in_conninfo *, int *);
621u_long tcp_maxmtu6(struct in_conninfo *, int *);
622void tcp_mss_update(struct tcpcb *, int, struct hc_metrics_lite *, int *);
623void tcp_mss(struct tcpcb *, int);
624int tcp_mssopt(struct in_conninfo *);
625struct inpcb *
626 tcp_drop_syn_sent(struct inpcb *, int);
627struct inpcb *
628 tcp_mtudisc(struct inpcb *, int);
629struct tcpcb *
630 tcp_newtcpcb(struct inpcb *);
631int tcp_output(struct tcpcb *);
632void tcp_respond(struct tcpcb *, void *,
633 struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int);
634void tcp_tw_init(void);
635#ifdef VIMAGE
636void tcp_tw_destroy(void);
637#endif
638void tcp_tw_zone_change(void);
639int tcp_twcheck(struct inpcb *, struct tcpopt *, struct tcphdr *,
640 struct mbuf *, int);
641int tcp_twrespond(struct tcptw *, int);
642void tcp_setpersist(struct tcpcb *);
643#ifdef TCP_SIGNATURE
644int tcp_signature_compute(struct mbuf *, int, int, int, u_char *, u_int);
645#endif
646void tcp_slowtimo(void);
647struct tcptemp *
648 tcpip_maketemplate(struct inpcb *);
649void tcpip_fillheaders(struct inpcb *, void *, void *);
650void tcp_timer_activate(struct tcpcb *, int, u_int);
651int tcp_timer_active(struct tcpcb *, int);
652void tcp_trace(short, short, struct tcpcb *, void *, struct tcphdr *, int);
653void tcp_xmit_bandwidth_limit(struct tcpcb *tp, tcp_seq ack_seq);
654/*
655 * All tcp_hc_* functions are IPv4 and IPv6 (via in_conninfo)
656 */
657void tcp_hc_init(void);
658#ifdef VIMAGE
659void tcp_hc_destroy(void);
660#endif
661void tcp_hc_get(struct in_conninfo *, struct hc_metrics_lite *);
662u_long tcp_hc_getmtu(struct in_conninfo *);
663void tcp_hc_updatemtu(struct in_conninfo *, u_long);
664void tcp_hc_update(struct in_conninfo *, struct hc_metrics_lite *);
665
666extern struct pr_usrreqs tcp_usrreqs;
667extern u_long tcp_sendspace;
668extern u_long tcp_recvspace;
669tcp_seq tcp_new_isn(struct tcpcb *);
670
671void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq);
672void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend);
673void tcp_clean_sackreport(struct tcpcb *tp);
674void tcp_sack_adjust(struct tcpcb *tp);
675struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt);
676void tcp_sack_partialack(struct tcpcb *, struct tcphdr *);
677void tcp_free_sackholes(struct tcpcb *tp);
678int tcp_newreno(struct tcpcb *, struct tcphdr *);
679u_long tcp_seq_subtract(u_long, u_long );
680
681#endif /* _KERNEL */
682
683#endif /* _NETINET_TCP_VAR_H_ */
|