sys/netinet/tcp_timer.h

   1 /*
   2  * Copyright (c) 1982, 1986, 1993
   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  * 3. 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_timer.h 8.1 (Berkeley) 6/10/93
  30  * $FreeBSD: src/sys/netinet/tcp_timer.h,v 1.18.2.1 2002/08/16 22:16:39 dillon Exp $
  31  * $DragonFly: src/sys/netinet/tcp_timer.h,v 1.3 2003/08/23 11:18:00 rob Exp $
  32  */
  33
  34 #ifndef _NETINET_TCP_TIMER_H_
  35 #define _NETINET_TCP_TIMER_H_
  36
  37 /*
  38  * The TCPT_REXMT timer is used to force retransmissions.
  39  * The TCP has the TCPT_REXMT timer set whenever segments
  40  * have been sent for which ACKs are expected but not yet
  41  * received.  If an ACK is received which advances tp->snd_una,
  42  * then the retransmit timer is cleared (if there are no more
  43  * outstanding segments) or reset to the base value (if there
  44  * are more ACKs expected).  Whenever the retransmit timer goes off,
  45  * we retransmit one unacknowledged segment, and do a backoff
  46  * on the retransmit timer.
  47  *
  48  * The TCPT_PERSIST timer is used to keep window size information
  49  * flowing even if the window goes shut.  If all previous transmissions
  50  * have been acknowledged (so that there are no retransmissions in progress),
  51  * and the window is too small to bother sending anything, then we start
  52  * the TCPT_PERSIST timer.  When it expires, if the window is nonzero,
  53  * we go to transmit state.  Otherwise, at intervals send a single byte
  54  * into the peer's window to force him to update our window information.
  55  * We do this at most as often as TCPT_PERSMIN time intervals,
  56  * but no more frequently than the current estimate of round-trip
  57  * packet time.  The TCPT_PERSIST timer is cleared whenever we receive
  58  * a window update from the peer.
  59  *
  60  * The TCPT_KEEP timer is used to keep connections alive.  If an
  61  * connection is idle (no segments received) for TCPTV_KEEP_INIT amount of time,
  62  * but not yet established, then we drop the connection.  Once the connection
  63  * is established, if the connection is idle for TCPTV_KEEP_IDLE time
  64  * (and keepalives have been enabled on the socket), we begin to probe
  65  * the connection.  We force the peer to send us a segment by sending:
  66  *      <SEQ=SND.UNA-1><ACK=RCV.NXT><CTL=ACK>
  67  * This segment is (deliberately) outside the window, and should elicit
  68  * an ack segment in response from the peer.  If, despite the TCPT_KEEP
  69  * initiated segments we cannot elicit a response from a peer in TCPT_MAXIDLE
  70  * amount of time probing, then we drop the connection.
  71  */
  72
  73 /*
  74  * Time constants.
  75  */
  76 #define TCPTV_MSL       ( 30*hz)                /* max seg lifetime (hah!) */
  77 #define TCPTV_SRTTBASE  0                       /* base roundtrip time;
  78                                                    if 0, no idea yet */
  79 #define TCPTV_RTOBASE3  (  3*hz)                /* assumed RTO if no info */
  80 #define TCPTV_RTOBASE1  (  1*hz)                /* assumed RTO (RFC 6298) */
  81 #define TCPTV_SRTTDFLT  (  3*hz)                /* assumed RTT if no info */
  82
  83 #define TCPTV_PERSMIN   (  5*hz)                /* retransmit persistence */
  84 #define TCPTV_PERSMAX   ( 60*hz)                /* maximum persist interval */
  85
  86 #define TCPTV_KEEP_INIT ( 75*hz)                /* initial connect keepalive */
  87 #define TCPTV_KEEP_IDLE (120*60*hz)             /* dflt time before probing */
  88 #define TCPTV_KEEPINTVL ( 75*hz)                /* default probe interval */
  89 #define TCPTV_KEEPCNT   8                       /* max probes before drop */
  90
  91 /*
  92  * Minimum retransmit timer is 3 ticks, for algorithmic stability.
  93  * TCPT_RANGESET() will add another TCPTV_CPU_VAR to deal with
  94  * the expected worst-case processing variances by the kernels
  95  * representing the end points.  Such variances do not always show
  96  * up in the srtt because the timestamp is often calculated at
  97  * the interface rather then at the TCP layer.  This value is
  98  * typically 50ms.  However, it is also possible that delayed
  99  * acks (typically 100ms) could create issues so we set the slop
 100  * to ~160ms to try to cover it.  Note that, properly speaking,
 101  * delayed-acks should not create a major issue for interactive
 102  * environments which 'P'ush the last segment, at least as
 103  * long as implementations do the required 'at least one ack
 104  * for every two packets' for the non-interactive streaming case.
 105  *
 106  * The prior minimum of 1*hz (1 second) badly breaks throughput
 107  * on any networks faster then a modem that has minor (e.g. 1%)
 108  * packet loss.
 109  */
 110 #define TCPTV_MIN       ( hz/33 )               /* minimum allowable value */
 111 #define TCPTV_CPU_VAR   ( hz/6 )                /* cpu variance (~160ms) */
 112 #define TCPTV_REXMTMAX  ( 64*hz)                /* max allowable REXMT value */
 113
 114 #define TCPTV_TWTRUNC   8                       /* RTO factor to truncate TW */
 115
 116 #define TCP_LINGERTIME  120                     /* linger at most 2 minutes */
 117
 118 #define TCP_MAXRXTSHIFT 12                      /* maximum retransmits */
 119
 120 #define TCPTV_DELACK    ( hz/10)                /* 100ms timeout */
 121
 122 #ifdef  TCPTIMERS
 123 static const char *tcptimers[] =
 124     { "REXMT", "PERSIST", "KEEP", "2MSL" };
 125 #endif
 126
 127 /*
 128  * Force a time value to be in a certain range.
 129  */
 130 #define TCPT_RANGESET(tv, value, tvmin, tvmax) do { \
 131         (tv) = (value) + tcp_rexmit_slop; \
 132         if ((u_long)(tv) < (u_long)(tvmin)) \
 133                 (tv) = (tvmin); \
 134         else if ((u_long)(tv) > (u_long)(tvmax)) \
 135                 (tv) = (tvmax); \
 136 } while(0)
 137
 138 #ifdef _KERNEL
 139
 140 #define TCPTV_RTOBASE \
 141         (tcp_low_rtobase ? TCPTV_RTOBASE1 : TCPTV_RTOBASE3)
 142
 143 struct tcpcb;
 144 struct netmsg_tcp_timer {
 145         struct netmsg_base tt_msg;
 146         struct tcpcb    *tt_tcb;
 147         int             tt_cpuid;               /* owner cpuid */
 148         lwkt_port_t     tt_msgport;             /* owner msgport */
 149         uint32_t        tt_tasks;               /* pending tasks */
 150         uint32_t        tt_running_tasks;       /* running tasks */
 151         uint32_t        tt_prev_tasks;          /* prev pending tasks (debug) */
 152 };
 153
 154 struct tcp_callout {
 155         struct callout  tc_callout;
 156         uint32_t        tc_task;                /* callout's task id */
 157 };
 158
 159 extern int tcp_keepinit;                /* time to establish connection */
 160 extern int tcp_keepidle;                /* time before keepalive probes begin */
 161 extern int tcp_keepintvl;               /* time between keepalive probes */
 162 extern int tcp_keepcnt;                 /* maximum number of keepalive probes */
 163 extern int tcp_delacktime;              /* time before sending a delayed ACK */
 164 extern int tcp_maxpersistidle;
 165 extern int tcp_rexmit_min;
 166 extern int tcp_rexmit_slop;
 167 extern int tcp_msl;
 168 extern int tcp_ttl;                     /* time to live for TCP segs */
 169 extern int tcp_backoff[];               /* rxt backoff for data */
 170 extern int tcp_syn_backoff[];           /* rxt backoff for SYN */
 171 extern int tcp_syn_backoff_low[];       /* rxt backoff for SYN w/ RFC6298 */
 172
 173 void    tcp_timer_2msl(void *xtp);
 174 void    tcp_timer_keep(void *xtp);
 175 void    tcp_timer_persist(void *xtp);
 176 void    tcp_timer_rexmt(void *xtp);
 177 void    tcp_timer_delack(void *xtp);
 178 void    tcp_inittimers(struct tcpcb *);
 179
 180 struct lwkt_port;
 181 void    tcp_create_timermsg(struct tcpcb *, struct lwkt_port *);
 182 void    tcp_destroy_timermsg(struct tcpcb *);
 183
 184 #endif /* _KERNEL */
 185
 186 #endif /* !_NETINET_TCP_TIMER_H_ */