X-Git-Url: https://git.saurik.com/apple/xnu.git/blobdiff_plain/5d5c5d0d5b79ade9a973d55186ffda2638ba2b6e..7ee9d059c4eecf68ae4f8b0fb99ae2471eda79af:/bsd/netinet/tcp_var.h diff --git a/bsd/netinet/tcp_var.h b/bsd/netinet/tcp_var.h index bc920c2bc..4066829cb 100644 --- a/bsd/netinet/tcp_var.h +++ b/bsd/netinet/tcp_var.h @@ -1,31 +1,29 @@ /* - * Copyright (c) 2000 Apple Computer, Inc. All rights reserved. + * Copyright (c) 2000-2011 Apple Inc. All rights reserved. * - * @APPLE_LICENSE_OSREFERENCE_HEADER_START@ + * @APPLE_OSREFERENCE_LICENSE_HEADER_START@ * - * This file contains Original Code and/or Modifications of Original Code - * as defined in and that are subject to the Apple Public Source License - * Version 2.0 (the 'License'). You may not use this file except in - * compliance with the License. The rights granted to you under the - * License may not be used to create, or enable the creation or - * redistribution of, unlawful or unlicensed copies of an Apple operating - * system, or to circumvent, violate, or enable the circumvention or - * violation of, any terms of an Apple operating system software license - * agreement. - * - * Please obtain a copy of the License at - * http://www.opensource.apple.com/apsl/ and read it before using this - * file. - * - * The Original Code and all software distributed under the License are - * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER - * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, - * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. - * Please see the License for the specific language governing rights and + * This file contains Original Code and/or Modifications of Original Code + * as defined in and that are subject to the Apple Public Source License + * Version 2.0 (the 'License'). You may not use this file except in + * compliance with the License. The rights granted to you under the License + * may not be used to create, or enable the creation or redistribution of, + * unlawful or unlicensed copies of an Apple operating system, or to + * circumvent, violate, or enable the circumvention or violation of, any + * terms of an Apple operating system software license agreement. + * + * Please obtain a copy of the License at + * http://www.opensource.apple.com/apsl/ and read it before using this file. + * + * The Original Code and all software distributed under the License are + * distributed on an 'AS IS' basis, WITHOUT WARRANTY OF ANY KIND, EITHER + * EXPRESS OR IMPLIED, AND APPLE HEREBY DISCLAIMS ALL SUCH WARRANTIES, + * INCLUDING WITHOUT LIMITATION, ANY WARRANTIES OF MERCHANTABILITY, + * FITNESS FOR A PARTICULAR PURPOSE, QUIET ENJOYMENT OR NON-INFRINGEMENT. + * Please see the License for the specific language governing rights and * limitations under the License. - * - * @APPLE_LICENSE_OSREFERENCE_HEADER_END@ + * + * @APPLE_OSREFERENCE_LICENSE_HEADER_END@ */ /* * Copyright (c) 1982, 1986, 1993, 1994, 1995 @@ -70,9 +68,67 @@ #include #include +#if defined(__LP64__) +#define _TCPCB_PTR(x) u_int32_t +#define _TCPCB_LIST_HEAD(name, type) \ +struct name { \ + u_int32_t lh_first; \ +}; +#else +#define _TCPCB_PTR(x) x +#define _TCPCB_LIST_HEAD(name, type) LIST_HEAD(name, type) +#endif + +#define TCP_RETRANSHZ 1000 /* granularity of TCP timestamps, 1ms */ +#define TCP_TIMERHZ 100 /* frequency of TCP fast timer, 100 ms */ + +/* Minimum time quantum within which the timers are coalesced */ +#define TCP_FASTTIMER_QUANTUM TCP_TIMERHZ /* fast mode, once every 100ms */ +#define TCP_SLOWTIMER_QUANTUM TCP_RETRANSHZ / PR_SLOWHZ /* slow mode, once every 500ms */ + +#define TCP_RETRANSHZ_TO_USEC 1000 + #ifdef KERNEL_PRIVATE +#define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */ + +/* Base RTT is stored for N_MIN_RTT_HISTORY slots. This is used to + * estimate expected minimum RTT for delay based congestion control + * algorithms. + */ +#define N_RTT_BASE 5 -#define N_TIME_WAIT_SLOTS 128 /* must be power of 2 */ +/* Always allow at least 4 packets worth of recv window when adjusting + * recv window using inter-packet arrival jitter. + */ +#define MIN_IAJ_WIN 4 + +/* A variation in delay of this many milliseconds is tolerable. This limit has to + * be low but greater than zero. We also use standard deviation on jitter to adjust + * this limit for different link and connection types. + */ +#define ALLOWED_IAJ 5 + +/* Ignore the first few packets on a connection until the ACK clock gets going + */ +#define IAJ_IGNORE_PKTCNT 40 + +/* Let the accumulated IAJ value increase by this threshold at most. This limit + * will control how many ALLOWED_IAJ measurements a receiver will have to see + * before opening the receive window + */ +#define ACC_IAJ_HIGH_THRESH 100 + +/* When accumulated IAJ reaches this value, the receiver starts to react by + * closing the window + */ +#define ACC_IAJ_REACT_LIMIT 200 + +/* If the number of small packets (smaller than IAJ packet size) seen on a + * connection is more than this threshold, reset the size and learn it again. + * This is needed because the sender might send smaller segments after PMTU + * discovery and the receiver has to learn the new size. + */ +#define RESET_IAJ_SIZE_THRESH 20 /* * Kernel variables for tcp. @@ -123,9 +179,8 @@ struct tcptemp { struct tcpcb { struct tsegqe_head t_segq; int t_dupacks; /* consecutive dup acks recd */ - struct tcptemp *unused; /* unused now: was t_template */ - - int t_timer[TCPT_NTIMERS]; /* tcp timers */ + uint32_t t_timer[TCPT_NTIMERS]; /* tcp timers */ + struct tcptimerentry tentry; /* entry in timer list */ struct inpcb *t_inpcb; /* back pointer to internet pcb */ int t_state; /* state of this connection */ @@ -147,16 +202,21 @@ struct tcpcb { #define TF_RCVD_CC 0x04000 /* a CC was received in SYN */ #define TF_SENDCCNEW 0x08000 /* send CCnew instead of CC in SYN */ #define TF_MORETOCOME 0x10000 /* More data to be appended to sock */ -#define TF_LQ_OVERFLOW 0x20000 /* listen queue overflow */ +#define TF_LOCAL 0x20000 /* connection to a host on local link */ #define TF_RXWIN0SENT 0x40000 /* sent a receiver win 0 in response */ #define TF_SLOWLINK 0x80000 /* route is a on a modem speed link */ - - -#define TF_LASTIDLE 0x100000 /* connection was previously idle */ +#define TF_LASTIDLE 0x100000 /* connection was previously idle */ #define TF_FASTRECOVERY 0x200000 /* in NewReno Fast Recovery */ #define TF_WASFRECOVERY 0x400000 /* was in NewReno Fast Recovery */ #define TF_SIGNATURE 0x800000 /* require MD5 digests (RFC2385) */ #define TF_MAXSEGSNT 0x1000000 /* last segment sent was a full segment */ +#define TF_SENDINPROG 0x2000000 /* send is in progress */ +#define TF_PMTUD 0x4000000 /* Perform Path MTU Discovery for this connection */ +#define TF_CLOSING 0x8000000 /* pending tcp close */ +#define TF_TSO 0x10000000 /* TCP Segment Offloading is enable on this connection */ +#define TF_BLACKHOLE 0x20000000 /* Path MTU Discovery Black Hole detection */ +#define TF_TIMER_ONLIST 0x40000000 /* pcb is on tcp_timer_list */ +#define TF_STRETCHACK 0x80000000 /* receiver is going to delay acks */ int t_force; /* 1 if forcing out a byte */ @@ -174,22 +234,29 @@ struct tcpcb { tcp_seq rcv_nxt; /* receive next */ tcp_seq rcv_adv; /* advertised window */ - u_long rcv_wnd; /* receive window */ + u_int32_t rcv_wnd; /* receive window */ tcp_seq rcv_up; /* receive urgent pointer */ - u_long snd_wnd; /* send window */ - u_long snd_cwnd; /* congestion-controlled window */ - u_long snd_ssthresh; /* snd_cwnd size threshold for + u_int32_t snd_wnd; /* send window */ + u_int32_t snd_cwnd; /* congestion-controlled window */ + u_int32_t snd_bwnd; /* bandwidth-controlled window */ + u_int32_t snd_ssthresh; /* snd_cwnd size threshold for * for slow start exponential to * linear switch */ + u_int32_t snd_bandwidth; /* calculated bandwidth or 0 */ + tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ + u_int t_maxopd; /* mss plus options */ - u_long t_rcvtime; /* inactivity time */ - u_long t_starttime; /* time connection was established */ - int t_rtttime; /* round trip time */ + u_int32_t t_rcvtime; /* time at which a packet was received */ + u_int32_t t_starttime; /* time connection was established */ + int t_rtttime; /* tcp clock when rtt calculation was started */ tcp_seq t_rtseq; /* sequence number being timed */ + int t_bw_rtttime; /* used for bandwidth calculation */ + tcp_seq t_bw_rtseq; /* used for bandwidth calculation */ + int t_rxtcur; /* current retransmit value (ticks) */ u_int t_maxseg; /* maximum segment size */ int t_srtt; /* smoothed round-trip time */ @@ -197,8 +264,12 @@ struct tcpcb { int t_rxtshift; /* log(2) of rexmt exp. backoff */ u_int t_rttmin; /* minimum rtt allowed */ - u_long t_rttupdated; /* number of times rtt sampled */ - u_long max_sndwnd; /* largest window peer has offered */ + u_int t_rttbest; /* best rtt we've seen */ + u_int t_rttcur; /* most recent value of rtt */ + u_int32_t t_rttupdated; /* number of times rtt sampled */ + u_int32_t rxt_conndroptime; /* retxmt conn gets dropped after this time, when set */ + u_int32_t rxt_start; /* time at a connection starts retransmitting */ + u_int32_t max_sndwnd; /* largest window peer has offered */ int t_softerror; /* possible error not yet reported */ /* out-of-band data */ @@ -211,37 +282,54 @@ struct tcpcb { u_char rcv_scale; /* window scaling for recv window */ u_char request_r_scale; /* pending window scaling */ u_char requested_s_scale; - u_long ts_recent; /* timestamp echo data */ + u_int16_t tcp_cc_index; /* index of congestion control algorithm */ + u_int32_t ts_recent; /* timestamp echo data */ - u_long ts_recent_age; /* when last updated */ + u_int32_t ts_recent_age; /* when last updated */ tcp_seq last_ack_sent; /* RFC 1644 variables */ tcp_cc cc_send; /* send connection count */ tcp_cc cc_recv; /* receive connection count */ - tcp_seq snd_recover; /* for use in NewReno Fast Recovery */ +/* RFC 3465 variables */ + u_int32_t t_bytes_acked; /* ABC "bytes_acked" parameter */ /* experimental */ - u_long snd_cwnd_prev; /* cwnd prior to retransmit */ - u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ - u_long t_badrxtwin; /* window for retransmit recovery */ + u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ + u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ + u_int32_t t_badrxtwin; /* window for retransmit recovery */ int t_keepidle; /* keepalive idle timer (override global if > 0) */ int t_lastchain; /* amount of packets chained last time around */ + int t_unacksegs; /* received but unacked segments: used for delaying acks */ + u_int32_t t_persist_timeout; /* ZWP persistence limit as set by PERSIST_TIMEOUT */ + u_int32_t t_persist_stop; /* persistence limit deadline if triggered by ZWP */ + /* 3529618 MSS overload prevention */ - u_long rcv_reset; - u_long rcv_pps; - u_long rcv_byps; + u_int32_t rcv_reset; + u_int32_t rcv_pps; + u_int32_t rcv_byps; + u_int32_t rcv_maxbyps; + +/* Receiver state for stretch-ack algorithm */ + u_int32_t rcv_unackwin; /* to measure win for stretching acks */ + u_int32_t rcv_by_unackwin; /* bytes seen during the last ack-stretching win */ + u_int16_t rcv_waitforss; /* wait for packets during slow-start */ + u_int16_t ecn_flags; +#define TE_SETUPSENT 0x01 /* Indicate we have sent ECN-SETUP SYN or SYN-ACK */ +#define TE_SETUPRECEIVED 0x02 /* Indicate we have received ECN-SETUP SYN or SYN-ACK */ +#define TE_SENDIPECT 0x04 /* Indicate we haven't sent or received non-ECN-setup SYN or SYN-ACK */ +#define TE_SENDCWR 0x08 /* Indicate that the next non-retransmit should have the TCP CWR flag set */ +#define TE_SENDECE 0x10 /* Indicate that the next packet should have the TCP ECE flag set */ tcp_seq snd_high; /* for use in NewReno Fast Recovery */ tcp_seq snd_high_prev; /* snd_high prior to retransmit */ - tcp_seq snd_recover_prev; /* snd_recover prior to retransmit */ u_char snd_limited; /* segments limited transmitted */ /* anti DoS counters */ - u_long rcv_second; /* start of interval second */ + u_int32_t rcv_second; /* start of interval second */ + /* SACK related state */ int sack_enable; /* enable SACK for this connection */ int snd_numholes; /* number of holes seen by sender */ - TAILQ_HEAD(sackhole_head, sackhole) snd_holes; /* SACK scoreboard (sorted) */ tcp_seq snd_fack; /* last seq number(+1) sack'd by rcv'r*/ @@ -250,13 +338,65 @@ struct tcpcb { tcp_seq sack_newdata; /* New data xmitted in this recovery episode starts at this seq number */ struct sackhint sackhint; /* SACK scoreboard hint */ - int t_rttlow; /* smallest observerved RTT */ + + u_int32_t t_pktlist_sentlen; /* total bytes in transmit chain */ + struct mbuf *t_pktlist_head; /* First packet in transmit chain */ + struct mbuf *t_pktlist_tail; /* Last packet in transmit chain */ + + int t_keepinit; /* connection timeout, i.e. idle time in SYN_SENT or SYN_RECV state */ + u_int32_t tso_max_segment_size; /* TCP Segment Offloading maximum segment unit for NIC */ + u_int t_pmtud_saved_maxopd; /* MSS saved before performing PMTU-D BlackHole detection */ + + struct + { + u_int32_t rxduplicatebytes; + u_int32_t rxoutoforderbytes; + u_int32_t txretransmitbytes; + u_int32_t unused_pad_to_8; + } t_stat; + + /* Background congestion related state */ + uint32_t rtt_hist[N_RTT_BASE]; /* history of minimum RTT */ + uint32_t rtt_count; /* Number of RTT samples in recent base history */ + uint32_t bg_ssthresh; /* Slow start threshold until delay increases */ + uint32_t t_flagsext; /* Another field to accommodate more flags */ +#define TF_RXTFINDROP 0x1 /* Drop conn after retransmitting FIN 3 times */ +#define TF_RCVUNACK_WAITSS 0x2 /* set when the receiver should not stretch acks */ + +#if TRAFFIC_MGT + /* Inter-arrival jitter related state */ + uint32_t iaj_rcv_ts; /* tcp clock when the first packet was received */ + uint16_t iaj_size; /* Size of packet for iaj measurement */ + uint16_t iaj_small_pkt; /* Count of packets smaller than iaj_size */ + uint16_t iaj_pktcnt; /* packet count, to avoid throttling initially */ + uint16_t acc_iaj; /* Accumulated iaj */ + tcp_seq iaj_rwintop; /* recent max advertised window */ + uint32_t avg_iaj; /* Mean */ + uint32_t std_dev_iaj; /* Standard deviation */ +#endif /* TRAFFIC_MGT */ }; #define IN_FASTRECOVERY(tp) (tp->t_flags & TF_FASTRECOVERY) #define ENTER_FASTRECOVERY(tp) tp->t_flags |= TF_FASTRECOVERY #define EXIT_FASTRECOVERY(tp) tp->t_flags &= ~TF_FASTRECOVERY +#if CONFIG_DTRACE +enum tcp_cc_event { + TCP_CC_CWND_INIT, + TCP_CC_INSEQ_ACK_RCVD, + TCP_CC_ACK_RCVD, + TCP_CC_ENTER_FASTRECOVERY, + TCP_CC_IN_FASTRECOVERY, + TCP_CC_EXIT_FASTRECOVERY, + TCP_CC_PARTIAL_ACK, + TCP_CC_IDLE_TIMEOUT, + TCP_CC_REXMT_TIMEOUT, + TCP_CC_ECN_RCVD, + TCP_CC_BAD_REXMT_RECOVERY, + TCP_CC_OUTPUT_ERROR, + TCP_CC_CHANGE_ALGO +}; +#endif /* CONFIG_DTRACE */ /* * Structure to hold TCP options that are only used during segment @@ -265,15 +405,15 @@ struct tcpcb { * to tcp_dooptions. */ struct tcpopt { - u_long to_flags; /* which options are present */ + u_int32_t to_flags; /* which options are present */ #define TOF_TS 0x0001 /* timestamp */ #define TOF_MSS 0x0010 #define TOF_SCALE 0x0020 #define TOF_SIGNATURE 0x0040 /* signature option present */ #define TOF_SIGLEN 0x0080 /* signature length valid (RFC2385) */ #define TOF_SACK 0x0100 /* Peer sent SACK option */ - u_long to_tsval; - u_long to_tsecr; + u_int32_t to_tsval; + u_int32_t to_tsecr; u_int16_t to_mss; u_int8_t to_requested_s_scale; u_int8_t to_nsacks; /* number of SACK blocks */ @@ -301,18 +441,19 @@ struct rmxp_tao { #define sototcpcb(so) (intotcpcb(sotoinpcb(so))) /* - * The smoothed round-trip time and estimated variance + * The rtt measured is in milliseconds as the timestamp granularity is + * a millisecond. The smoothed round-trip time and estimated variance * are stored as fixed point numbers scaled by the values below. * For convenience, these scales are also used in smoothing the average * (smoothed = (1/scale)sample + ((scale-1)/scale)smoothed). - * With these scales, srtt has 3 bits to the right of the binary point, - * and thus an "ALPHA" of 0.875. rttvar has 2 bits to the right of the + * With these scales, srtt has 5 bits to the right of the binary point, + * and thus an "ALPHA" of 0.875. rttvar has 4 bits to the right of the * binary point, and is smoothed with an ALPHA of 0.75. */ #define TCP_RTT_SCALE 32 /* multiplier for srtt; 3 bits frac. */ -#define TCP_RTT_SHIFT 5 /* shift for srtt; 3 bits frac. */ -#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 2 bits */ -#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 2 bits */ +#define TCP_RTT_SHIFT 5 /* shift for srtt; 5 bits frac. */ +#define TCP_RTTVAR_SCALE 16 /* multiplier for rttvar; 4 bits */ +#define TCP_RTTVAR_SHIFT 4 /* shift for rttvar; 4 bits */ #define TCP_DELTA_SHIFT 2 /* see tcp_input.c */ /* @@ -342,17 +483,21 @@ struct rmxp_tao { struct otcpcb { #else struct tseg_qent; -LIST_HEAD(tsegqe_head, tseg_qent); +_TCPCB_LIST_HEAD(tsegqe_head, tseg_qent); struct tcpcb { #endif /* KERNEL_PRIVATE */ +#if defined(KERNEL_PRIVATE) + u_int32_t t_segq; +#else struct tsegqe_head t_segq; +#endif /* KERNEL_PRIVATE */ int t_dupacks; /* consecutive dup acks recd */ - struct tcptemp *unused; /* unused now: was t_template */ + u_int32_t unused; /* unused now: was t_template */ - int t_timer[TCPT_NTIMERS]; /* tcp timers */ + int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ - struct inpcb *t_inpcb; /* back pointer to internet pcb */ + _TCPCB_PTR(struct inpcb *) t_inpcb; /* back pointer to internet pcb */ int t_state; /* state of this connection */ u_int t_flags; #define TF_ACKNOW 0x00001 /* ack peer immediately */ @@ -392,19 +537,19 @@ struct tcpcb { tcp_seq rcv_nxt; /* receive next */ tcp_seq rcv_adv; /* advertised window */ - u_long rcv_wnd; /* receive window */ + u_int32_t rcv_wnd; /* receive window */ tcp_seq rcv_up; /* receive urgent pointer */ - u_long snd_wnd; /* send window */ - u_long snd_cwnd; /* congestion-controlled window */ - u_long snd_ssthresh; /* snd_cwnd size threshold for + u_int32_t snd_wnd; /* send window */ + u_int32_t snd_cwnd; /* congestion-controlled window */ + u_int32_t snd_ssthresh; /* snd_cwnd size threshold for * for slow start exponential to * linear switch */ u_int t_maxopd; /* mss plus options */ - u_long t_rcvtime; /* inactivity time */ - u_long t_starttime; /* time connection was established */ + u_int32_t t_rcvtime; /* time at which a packet was received */ + u_int32_t t_starttime; /* time connection was established */ int t_rtttime; /* round trip time */ tcp_seq t_rtseq; /* sequence number being timed */ @@ -415,8 +560,8 @@ struct tcpcb { int t_rxtshift; /* log(2) of rexmt exp. backoff */ u_int t_rttmin; /* minimum rtt allowed */ - u_long t_rttupdated; /* number of times rtt sampled */ - u_long max_sndwnd; /* largest window peer has offered */ + u_int32_t t_rttupdated; /* number of times rtt sampled */ + u_int32_t max_sndwnd; /* largest window peer has offered */ int t_softerror; /* possible error not yet reported */ /* out-of-band data */ @@ -429,121 +574,317 @@ struct tcpcb { u_char rcv_scale; /* window scaling for recv window */ u_char request_r_scale; /* pending window scaling */ u_char requested_s_scale; - u_long ts_recent; /* timestamp echo data */ + u_int32_t ts_recent; /* timestamp echo data */ - u_long ts_recent_age; /* when last updated */ + u_int32_t ts_recent_age; /* when last updated */ tcp_seq last_ack_sent; /* RFC 1644 variables */ tcp_cc cc_send; /* send connection count */ tcp_cc cc_recv; /* receive connection count */ - tcp_seq snd_recover; /* for use in fast recovery */ + tcp_seq snd_recover; /* for use in fast recovery */ /* experimental */ - u_long snd_cwnd_prev; /* cwnd prior to retransmit */ - u_long snd_ssthresh_prev; /* ssthresh prior to retransmit */ - u_long t_badrxtwin; /* window for retransmit recovery */ + u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ + u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ + u_int32_t t_badrxtwin; /* window for retransmit recovery */ }; + /* * TCP statistics. * Many of these should be kept per connection, * but that's inconvenient at the moment. */ struct tcpstat { - u_long tcps_connattempt; /* connections initiated */ - u_long tcps_accepts; /* connections accepted */ - u_long tcps_connects; /* connections established */ - u_long tcps_drops; /* connections dropped */ - u_long tcps_conndrops; /* embryonic connections dropped */ - u_long tcps_closed; /* conn. closed (includes drops) */ - u_long tcps_segstimed; /* segs where we tried to get rtt */ - u_long tcps_rttupdated; /* times we succeeded */ - u_long tcps_delack; /* delayed acks sent */ - u_long tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ - u_long tcps_rexmttimeo; /* retransmit timeouts */ - u_long tcps_persisttimeo; /* persist timeouts */ - u_long tcps_keeptimeo; /* keepalive timeouts */ - u_long tcps_keepprobe; /* keepalive probes sent */ - u_long tcps_keepdrops; /* connections dropped in keepalive */ - - u_long tcps_sndtotal; /* total packets sent */ - u_long tcps_sndpack; /* data packets sent */ - u_long tcps_sndbyte; /* data bytes sent */ - u_long tcps_sndrexmitpack; /* data packets retransmitted */ - u_long tcps_sndrexmitbyte; /* data bytes retransmitted */ - u_long tcps_sndacks; /* ack-only packets sent */ - u_long tcps_sndprobe; /* window probes sent */ - u_long tcps_sndurg; /* packets sent with URG only */ - u_long tcps_sndwinup; /* window update-only packets sent */ - u_long tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ - - u_long tcps_rcvtotal; /* total packets received */ - u_long tcps_rcvpack; /* packets received in sequence */ - u_long tcps_rcvbyte; /* bytes received in sequence */ - u_long tcps_rcvbadsum; /* packets received with ccksum errs */ - u_long tcps_rcvbadoff; /* packets received with bad offset */ - u_long tcps_rcvmemdrop; /* packets dropped for lack of memory */ - u_long tcps_rcvshort; /* packets received too short */ - u_long tcps_rcvduppack; /* duplicate-only packets received */ - u_long tcps_rcvdupbyte; /* duplicate-only bytes received */ - u_long tcps_rcvpartduppack; /* packets with some duplicate data */ - u_long tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ - u_long tcps_rcvoopack; /* out-of-order packets received */ - u_long tcps_rcvoobyte; /* out-of-order bytes received */ - u_long tcps_rcvpackafterwin; /* packets with data after window */ - u_long tcps_rcvbyteafterwin; /* bytes rcvd after window */ - u_long tcps_rcvafterclose; /* packets rcvd after "close" */ - u_long tcps_rcvwinprobe; /* rcvd window probe packets */ - u_long tcps_rcvdupack; /* rcvd duplicate acks */ - u_long tcps_rcvacktoomuch; /* rcvd acks for unsent data */ - u_long tcps_rcvackpack; /* rcvd ack packets */ - u_long tcps_rcvackbyte; /* bytes acked by rcvd acks */ - u_long tcps_rcvwinupd; /* rcvd window update packets */ - u_long tcps_pawsdrop; /* segments dropped due to PAWS */ - u_long tcps_predack; /* times hdr predict ok for acks */ - u_long tcps_preddat; /* times hdr predict ok for data pkts */ - u_long tcps_pcbcachemiss; - u_long tcps_cachedrtt; /* times cached RTT in route updated */ - u_long tcps_cachedrttvar; /* times cached rttvar updated */ - u_long tcps_cachedssthresh; /* times cached ssthresh updated */ - u_long tcps_usedrtt; /* times RTT initialized from route */ - u_long tcps_usedrttvar; /* times RTTVAR initialized from rt */ - u_long tcps_usedssthresh; /* times ssthresh initialized from rt*/ - u_long tcps_persistdrop; /* timeout in persist state */ - u_long tcps_badsyn; /* bogus SYN, e.g. premature ACK */ - u_long tcps_mturesent; /* resends due to MTU discovery */ - u_long tcps_listendrop; /* listen queue overflows */ + u_int32_t tcps_connattempt; /* connections initiated */ + u_int32_t tcps_accepts; /* connections accepted */ + u_int32_t tcps_connects; /* connections established */ + u_int32_t tcps_drops; /* connections dropped */ + u_int32_t tcps_conndrops; /* embryonic connections dropped */ + u_int32_t tcps_closed; /* conn. closed (includes drops) */ + u_int32_t tcps_segstimed; /* segs where we tried to get rtt */ + u_int32_t tcps_rttupdated; /* times we succeeded */ + u_int32_t tcps_delack; /* delayed acks sent */ + u_int32_t tcps_timeoutdrop; /* conn. dropped in rxmt timeout */ + u_int32_t tcps_rexmttimeo; /* retransmit timeouts */ + u_int32_t tcps_persisttimeo; /* persist timeouts */ + u_int32_t tcps_keeptimeo; /* keepalive timeouts */ + u_int32_t tcps_keepprobe; /* keepalive probes sent */ + u_int32_t tcps_keepdrops; /* connections dropped in keepalive */ + + u_int32_t tcps_sndtotal; /* total packets sent */ + u_int32_t tcps_sndpack; /* data packets sent */ + u_int32_t tcps_sndbyte; /* data bytes sent */ + u_int32_t tcps_sndrexmitpack; /* data packets retransmitted */ + u_int32_t tcps_sndrexmitbyte; /* data bytes retransmitted */ + u_int32_t tcps_sndacks; /* ack-only packets sent */ + u_int32_t tcps_sndprobe; /* window probes sent */ + u_int32_t tcps_sndurg; /* packets sent with URG only */ + u_int32_t tcps_sndwinup; /* window update-only packets sent */ + u_int32_t tcps_sndctrl; /* control (SYN|FIN|RST) packets sent */ + + u_int32_t tcps_rcvtotal; /* total packets received */ + u_int32_t tcps_rcvpack; /* packets received in sequence */ + u_int32_t tcps_rcvbyte; /* bytes received in sequence */ + u_int32_t tcps_rcvbadsum; /* packets received with ccksum errs */ + u_int32_t tcps_rcvbadoff; /* packets received with bad offset */ + u_int32_t tcps_rcvmemdrop; /* packets dropped for lack of memory */ + u_int32_t tcps_rcvshort; /* packets received too short */ + u_int32_t tcps_rcvduppack; /* duplicate-only packets received */ + u_int32_t tcps_rcvdupbyte; /* duplicate-only bytes received */ + u_int32_t tcps_rcvpartduppack; /* packets with some duplicate data */ + u_int32_t tcps_rcvpartdupbyte; /* dup. bytes in part-dup. packets */ + u_int32_t tcps_rcvoopack; /* out-of-order packets received */ + u_int32_t tcps_rcvoobyte; /* out-of-order bytes received */ + u_int32_t tcps_rcvpackafterwin; /* packets with data after window */ + u_int32_t tcps_rcvbyteafterwin; /* bytes rcvd after window */ + u_int32_t tcps_rcvafterclose; /* packets rcvd after "close" */ + u_int32_t tcps_rcvwinprobe; /* rcvd window probe packets */ + u_int32_t tcps_rcvdupack; /* rcvd duplicate acks */ + u_int32_t tcps_rcvacktoomuch; /* rcvd acks for unsent data */ + u_int32_t tcps_rcvackpack; /* rcvd ack packets */ + u_int32_t tcps_rcvackbyte; /* bytes acked by rcvd acks */ + u_int32_t tcps_rcvwinupd; /* rcvd window update packets */ + u_int32_t tcps_pawsdrop; /* segments dropped due to PAWS */ + u_int32_t tcps_predack; /* times hdr predict ok for acks */ + u_int32_t tcps_preddat; /* times hdr predict ok for data pkts */ + u_int32_t tcps_pcbcachemiss; + u_int32_t tcps_cachedrtt; /* times cached RTT in route updated */ + u_int32_t tcps_cachedrttvar; /* times cached rttvar updated */ + u_int32_t tcps_cachedssthresh; /* times cached ssthresh updated */ + u_int32_t tcps_usedrtt; /* times RTT initialized from route */ + u_int32_t tcps_usedrttvar; /* times RTTVAR initialized from rt */ + u_int32_t tcps_usedssthresh; /* times ssthresh initialized from rt*/ + u_int32_t tcps_persistdrop; /* timeout in persist state */ + u_int32_t tcps_badsyn; /* bogus SYN, e.g. premature ACK */ + u_int32_t tcps_mturesent; /* resends due to MTU discovery */ + u_int32_t tcps_listendrop; /* listen queue overflows */ + + /* new stats from FreeBSD 5.4 sync up */ + u_int32_t tcps_minmssdrops; /* average minmss too low drops */ + u_int32_t tcps_sndrexmitbad; /* unnecessary packet retransmissions */ + u_int32_t tcps_badrst; /* ignored RSTs in the window */ + + u_int32_t tcps_sc_added; /* entry added to syncache */ + u_int32_t tcps_sc_retransmitted; /* syncache entry was retransmitted */ + u_int32_t tcps_sc_dupsyn; /* duplicate SYN packet */ + u_int32_t tcps_sc_dropped; /* could not reply to packet */ + u_int32_t tcps_sc_completed; /* successful extraction of entry */ + u_int32_t tcps_sc_bucketoverflow; /* syncache per-bucket limit hit */ + u_int32_t tcps_sc_cacheoverflow; /* syncache cache limit hit */ + u_int32_t tcps_sc_reset; /* RST removed entry from syncache */ + u_int32_t tcps_sc_stale; /* timed out or listen socket gone */ + u_int32_t tcps_sc_aborted; /* syncache entry aborted */ + u_int32_t tcps_sc_badack; /* removed due to bad ACK */ + u_int32_t tcps_sc_unreach; /* ICMP unreachable received */ + u_int32_t tcps_sc_zonefail; /* zalloc() failed */ + u_int32_t tcps_sc_sendcookie; /* SYN cookie sent */ + u_int32_t tcps_sc_recvcookie; /* SYN cookie received */ + + u_int32_t tcps_hc_added; /* entry added to hostcache */ + u_int32_t tcps_hc_bucketoverflow; /* hostcache per bucket limit hit */ /* SACK related stats */ - u_long tcps_sack_recovery_episode; /* SACK recovery episodes */ - u_long tcps_sack_rexmits; /* SACK rexmit segments */ - u_long tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ - u_long tcps_sack_rcv_blocks; /* SACK blocks (options) received */ - u_long tcps_sack_send_blocks; /* SACK blocks (options) sent */ - u_long tcps_sack_sboverflow; /* SACK sendblock overflow */ - + u_int32_t tcps_sack_recovery_episode; /* SACK recovery episodes */ + u_int32_t tcps_sack_rexmits; /* SACK rexmit segments */ + u_int32_t tcps_sack_rexmit_bytes; /* SACK rexmit bytes */ + u_int32_t tcps_sack_rcv_blocks; /* SACK blocks (options) received */ + u_int32_t tcps_sack_send_blocks; /* SACK blocks (options) sent */ + u_int32_t tcps_sack_sboverflow; /* SACK sendblock overflow */ + + u_int32_t tcps_bg_rcvtotal; /* total background packets received */ + u_int32_t tcps_rxtfindrop; /* drop conn after retransmitting FIN */ }; +#pragma pack(4) + /* * TCB structure exported to user-land via sysctl(3). * Evil hack: declare only if in_pcb.h and sys/socketvar.h have been * included. Not all of our clients do. */ -struct xtcpcb { - size_t xt_len; + +struct xtcpcb { + u_int32_t xt_len; #ifdef KERNEL_PRIVATE - struct inpcb_compat xt_inp; + struct inpcb_compat xt_inp; #else - struct inpcb xt_inp; + struct inpcb xt_inp; #endif #ifdef KERNEL_PRIVATE - struct otcpcb xt_tp; + struct otcpcb xt_tp; #else - struct tcpcb xt_tp; + struct tcpcb xt_tp; #endif - struct xsocket xt_socket; - u_quad_t xt_alignment_hack; + struct xsocket xt_socket; + u_quad_t xt_alignment_hack; +}; + +#if !CONFIG_EMBEDDED + +struct xtcpcb64 { + u_int32_t xt_len; + struct xinpcb64 xt_inpcb; + + u_int64_t t_segq; + int t_dupacks; /* consecutive dup acks recd */ + + int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ + + int t_state; /* state of this connection */ + u_int t_flags; + + int t_force; /* 1 if forcing out a byte */ + + tcp_seq snd_una; /* send unacknowledged */ + tcp_seq snd_max; /* highest sequence number sent; + * used to recognize retransmits + */ + tcp_seq snd_nxt; /* send next */ + tcp_seq snd_up; /* send urgent pointer */ + + tcp_seq snd_wl1; /* window update seg seq number */ + tcp_seq snd_wl2; /* window update seg ack number */ + tcp_seq iss; /* initial send sequence number */ + tcp_seq irs; /* initial receive sequence number */ + + tcp_seq rcv_nxt; /* receive next */ + tcp_seq rcv_adv; /* advertised window */ + u_int32_t rcv_wnd; /* receive window */ + tcp_seq rcv_up; /* receive urgent pointer */ + + u_int32_t snd_wnd; /* send window */ + u_int32_t snd_cwnd; /* congestion-controlled window */ + u_int32_t snd_ssthresh; /* snd_cwnd size threshold for + * for slow start exponential to + * linear switch + */ + u_int t_maxopd; /* mss plus options */ + + u_int32_t t_rcvtime; /* time at which a packet was received */ + u_int32_t t_starttime; /* time connection was established */ + int t_rtttime; /* round trip time */ + tcp_seq t_rtseq; /* sequence number being timed */ + + int t_rxtcur; /* current retransmit value (ticks) */ + u_int t_maxseg; /* maximum segment size */ + int t_srtt; /* smoothed round-trip time */ + int t_rttvar; /* variance in round-trip time */ + + int t_rxtshift; /* log(2) of rexmt exp. backoff */ + u_int t_rttmin; /* minimum rtt allowed */ + u_int32_t t_rttupdated; /* number of times rtt sampled */ + u_int32_t max_sndwnd; /* largest window peer has offered */ + + int t_softerror; /* possible error not yet reported */ +/* out-of-band data */ + char t_oobflags; /* have some */ + char t_iobc; /* input character */ +/* RFC 1323 variables */ + u_char snd_scale; /* window scaling for send window */ + u_char rcv_scale; /* window scaling for recv window */ + u_char request_r_scale; /* pending window scaling */ + u_char requested_s_scale; + u_int32_t ts_recent; /* timestamp echo data */ + + u_int32_t ts_recent_age; /* when last updated */ + tcp_seq last_ack_sent; +/* RFC 1644 variables */ + tcp_cc cc_send; /* send connection count */ + tcp_cc cc_recv; /* receive connection count */ + tcp_seq snd_recover; /* for use in fast recovery */ +/* experimental */ + u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ + u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ + u_int32_t t_badrxtwin; /* window for retransmit recovery */ + + u_quad_t xt_alignment_hack; +}; + +#endif /* !CONFIG_EMBEDDED */ + +#ifdef PRIVATE + +struct xtcpcb_n { + u_int32_t xt_len; + u_int32_t xt_kind; /* XSO_TCPCB */ + + u_int64_t t_segq; + int t_dupacks; /* consecutive dup acks recd */ + + int t_timer[TCPT_NTIMERS_EXT]; /* tcp timers */ + + int t_state; /* state of this connection */ + u_int t_flags; + + int t_force; /* 1 if forcing out a byte */ + + tcp_seq snd_una; /* send unacknowledged */ + tcp_seq snd_max; /* highest sequence number sent; + * used to recognize retransmits + */ + tcp_seq snd_nxt; /* send next */ + tcp_seq snd_up; /* send urgent pointer */ + + tcp_seq snd_wl1; /* window update seg seq number */ + tcp_seq snd_wl2; /* window update seg ack number */ + tcp_seq iss; /* initial send sequence number */ + tcp_seq irs; /* initial receive sequence number */ + + tcp_seq rcv_nxt; /* receive next */ + tcp_seq rcv_adv; /* advertised window */ + u_int32_t rcv_wnd; /* receive window */ + tcp_seq rcv_up; /* receive urgent pointer */ + + u_int32_t snd_wnd; /* send window */ + u_int32_t snd_cwnd; /* congestion-controlled window */ + u_int32_t snd_ssthresh; /* snd_cwnd size threshold for + * for slow start exponential to + * linear switch + */ + u_int t_maxopd; /* mss plus options */ + + u_int32_t t_rcvtime; /* time at which a packet was received */ + u_int32_t t_starttime; /* time connection was established */ + int t_rtttime; /* round trip time */ + tcp_seq t_rtseq; /* sequence number being timed */ + + int t_rxtcur; /* current retransmit value (ticks) */ + u_int t_maxseg; /* maximum segment size */ + int t_srtt; /* smoothed round-trip time */ + int t_rttvar; /* variance in round-trip time */ + + int t_rxtshift; /* log(2) of rexmt exp. backoff */ + u_int t_rttmin; /* minimum rtt allowed */ + u_int32_t t_rttupdated; /* number of times rtt sampled */ + u_int32_t max_sndwnd; /* largest window peer has offered */ + + int t_softerror; /* possible error not yet reported */ + /* out-of-band data */ + char t_oobflags; /* have some */ + char t_iobc; /* input character */ + /* RFC 1323 variables */ + u_char snd_scale; /* window scaling for send window */ + u_char rcv_scale; /* window scaling for recv window */ + u_char request_r_scale; /* pending window scaling */ + u_char requested_s_scale; + u_int32_t ts_recent; /* timestamp echo data */ + + u_int32_t ts_recent_age; /* when last updated */ + tcp_seq last_ack_sent; + /* RFC 1644 variables */ + tcp_cc cc_send; /* send connection count */ + tcp_cc cc_recv; /* receive connection count */ + tcp_seq snd_recover; /* for use in fast recovery */ + /* experimental */ + u_int32_t snd_cwnd_prev; /* cwnd prior to retransmit */ + u_int32_t snd_ssthresh_prev; /* ssthresh prior to retransmit */ + u_int32_t t_badrxtwin; /* window for retransmit recovery */ }; +#endif /* PRIVATE */ + +#pragma pack() + /* * Names for TCP sysctl objects */ @@ -563,6 +904,11 @@ struct xtcpcb { #define TCPCTL_MAXID 14 #ifdef KERNEL_PRIVATE +#define TCP_PKTLIST_CLEAR(tp) { \ + (tp)->t_pktlist_head = (tp)->t_pktlist_tail = NULL; \ + (tp)->t_lastchain = (tp)->t_pktlist_sentlen = 0; \ +} + #define TCPCTL_NAMES { \ { 0, 0 }, \ { "rfc1323", CTLTYPE_INT }, \ @@ -590,16 +936,23 @@ extern struct tcpstat tcpstat; /* tcp statistics */ extern int tcp_mssdflt; /* XXX */ extern int tcp_minmss; extern int tcp_minmssoverload; -extern int tcp_do_newreno; extern int ss_fltsz; extern int ss_fltsz_local; +extern int tcp_do_rfc3390; /* Calculate ss_fltsz according to RFC 3390 */ #ifdef __APPLE__ -extern u_long tcp_now; /* for RFC 1323 timestamps */ +extern u_int32_t tcp_now; /* for RFC 1323 timestamps */ +extern struct timeval tcp_uptime; +extern lck_spin_t *tcp_uptime_lock; + extern int tcp_delack_enabled; #endif /* __APPLE__ */ extern int tcp_do_sack; /* SACK enabled/disabled */ +#if CONFIG_IFEF_NOWINDOWSCALE +extern int tcp_obey_ifef_nowindowscale; +#endif + void tcp_canceltimers(struct tcpcb *); struct tcpcb * tcp_close(struct tcpcb *); @@ -608,25 +961,26 @@ int tcp_ctloutput(struct socket *, struct sockopt *); struct tcpcb * tcp_drop(struct tcpcb *, int); void tcp_drain(void); -void tcp_fasttimo(void); struct rmxp_tao * tcp_gettaocache(struct inpcb *); -void tcp_init(void); +void tcp_init(void) __attribute__((section("__TEXT, initcode"))); void tcp_input(struct mbuf *, int); -void tcp_mss(struct tcpcb *, int); +void tcp_mss(struct tcpcb *, int, unsigned int); int tcp_mssopt(struct tcpcb *); void tcp_drop_syn_sent(struct inpcb *, int); void tcp_mtudisc(struct inpcb *, int); struct tcpcb * tcp_newtcpcb(struct inpcb *); int tcp_output(struct tcpcb *); -void tcp_quench(struct inpcb *, int); void tcp_respond(struct tcpcb *, void *, - struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int); -struct rtentry * - tcp_rtlookup(struct inpcb *); + struct tcphdr *, struct mbuf *, tcp_seq, tcp_seq, int, + unsigned int, unsigned int); +struct rtentry *tcp_rtlookup(struct inpcb *, unsigned int); void tcp_setpersist(struct tcpcb *); void tcp_slowtimo(void); +void tcp_check_timer_state(struct tcpcb *tp); +void tcp_run_timerlist(void *arg1, void *arg2); + struct tcptemp * tcp_maketemplate(struct tcpcb *); void tcp_fillheaders(struct tcpcb *, void *, void *); @@ -634,16 +988,25 @@ struct tcpcb * tcp_timers(struct tcpcb *, int); void tcp_trace(int, int, struct tcpcb *, void *, struct tcphdr *, int); +void tcp_sack_doack(struct tcpcb *, struct tcpopt *, tcp_seq); void tcp_update_sack_list(struct tcpcb *tp, tcp_seq rcv_laststart, tcp_seq rcv_lastend); void tcp_clean_sackreport(struct tcpcb *tp); void tcp_sack_adjust(struct tcpcb *tp); struct sackhole *tcp_sack_output(struct tcpcb *tp, int *sack_bytes_rexmt); void tcp_sack_partialack(struct tcpcb *, struct tcphdr *); void tcp_free_sackholes(struct tcpcb *tp); +int32_t tcp_sbspace(struct tcpcb *tp); +void tcp_set_tso(struct tcpcb *tp, struct ifnet *ifp); +void tcp_reset_stretch_ack(struct tcpcb *tp); + +#if TRAFFIC_MGT +void reset_acc_iaj(struct tcpcb *tp); +#endif /* TRAFFIC_MGT */ +int tcp_lock (struct socket *, int, void *); +int tcp_unlock (struct socket *, int, void *); +void calculate_tcp_clock(void); -int tcp_lock (struct socket *, int, int); -int tcp_unlock (struct socket *, int, int); #ifdef _KERN_LOCKS_H_ lck_mtx_t * tcp_getlock (struct socket *, int); #else @@ -652,8 +1015,8 @@ void * tcp_getlock (struct socket *, int); extern struct pr_usrreqs tcp_usrreqs; -extern u_long tcp_sendspace; -extern u_long tcp_recvspace; +extern u_int32_t tcp_sendspace; +extern u_int32_t tcp_recvspace; tcp_seq tcp_new_isn(struct tcpcb *); #endif /* KERNEL_RPIVATE */