oops, 7.19 wasn't finished; back part out for now, but leave TOS/TTL

[unix-history] / usr / src / sys / netinet / tcp_var.h

diff --git a/usr/src/sys/netinet/tcp_var.h b/usr/src/sys/netinet/tcp_var.h
index 604140c..b06a058 100644 (file)
--- a/usr/src/sys/netinet/tcp_var.h
+++ b/usr/src/sys/netinet/tcp_var.h
@@ -1,27 +1,85 @@
-/*     tcp_var.h       4.9     81/11/26        */
+/*
+ * Copyright (c) 1982, 1986 Regents of the University of California.
+ * All rights reserved.
+ *
+ * %sccs.include.redist.c%
+ *
+ *     @(#)tcp_var.h   7.10 (Berkeley) %G%
+ */
+
+/*
+ * TCP configuration:  This is a half-assed attempt to make TCP
+ * self-configure for a few varieties of 4.2 and 4.3-based unixes.
+ * If you don't have a) a 4.3bsd vax or b) a 3.x Sun (x<6), check
+ * this carefully (it's probably not right).  Please send me mail
+ * if you run into configuration problems.
+ *  - Van Jacobson (van@lbl-csam.arpa)
+ */
+
+#ifndef BSD
+#define BSD 42 /* if we're not 4.3, pretend we're 4.2 */
+#define OLDSTAT        /* set if we have to use old netstat binaries */
+#endif
+
+/* #define OLDSTAT     /* set if we have to use old netstat binaries */
+
+#if sun || BSD < 43
+#define TCP_COMPAT_42  /* set if we have to interop w/4.2 systems */
+#endif
+
+#ifndef SB_MAX
+#ifdef SB_MAXCOUNT
+#define        SB_MAX  SB_MAXCOUNT     /* Sun has to be a little bit different... */
+#else
+#define SB_MAX 32767           /* XXX */
+#endif SB_MAXCOUNT
+#endif SB_MAX
+
+/*
+ * Bill Nowicki pointed out that the page size (CLBYTES) has
+ * nothing to do with the mbuf cluster size.  So, we followed
+ * Sun's lead and made the new define MCLBYTES stand for the mbuf
+ * cluster size.  The following define makes up backwards compatible
+ * with 4.3 and 4.2.  If CLBYTES is >1024 on your machine, check
+ * this against the mbuf cluster definitions in /usr/include/sys/mbuf.h.
+ */
+#ifndef MCLBYTES
+#define        MCLBYTES CLBYTES        /* XXX */
+#endif
+
+/*
+ * The routine in_localaddr is broken in Sun's 3.4.  We redefine ours
+ * (in tcp_input.c) so we use can it but won't have a name conflict.
+ */
+#ifdef sun
+#define in_localaddr tcp_in_localaddr
+#endif
+
+/* --------------- end of TCP config ---------------- */

 
 /*
  * Kernel variables for tcp.
  */
 
 /*
 
 /*
  * Kernel variables for tcp.
  */
 
 /*

- * Tcp control block.
+ * Tcp control block, one per tcp; fields:

  */
 struct tcpcb {
        struct  tcpiphdr *seg_next;     /* sequencing queue */
        struct  tcpiphdr *seg_prev;
  */
 struct tcpcb {
        struct  tcpiphdr *seg_next;     /* sequencing queue */
        struct  tcpiphdr *seg_prev;

-       int     t_state;                /* state of this connection */
-       int     t_seqcnt;               /* count of chars in seq queue */
+       short   t_state;                /* state of this connection */

        short   t_timer[TCPT_NTIMERS];  /* tcp timers */
        short   t_timer[TCPT_NTIMERS];  /* tcp timers */

-       struct  mbuf *t_tcpopt;         /* tcp options */
-       struct  mbuf *t_ipopt;          /* ip options */
-       short   t_maxseg;               /* maximum segment size */
+       short   t_rxtshift;             /* log(2) of rexmt exp. backoff */
+       short   t_rxtcur;               /* current retransmit value */
+       short   t_dupacks;              /* consecutive dup acks recd */
+       u_short t_maxseg;               /* maximum segment size */
+       char    t_force;                /* 1 if forcing out a byte */

        u_char  t_flags;
        u_char  t_flags;

-#define        TF_ACKNOW       0x01                    /* ack peer immediately */
-#define        TF_DELACK       0x02                    /* ack, but try to delay it */
-#define        TF_PUSH         0x04                    /* push mode */
-#define        TF_URG          0x08                    /* urgent mode */
-#define        TF_KEEP         0x10                    /* use keep-alives */
+#define        TF_ACKNOW       0x01            /* ack peer immediately */
+#define        TF_DELACK       0x02            /* ack, but try to delay it */
+#define        TF_NODELAY      0x04            /* don't delay packets to coalesce */
+#define        TF_NOOPT        0x08            /* don't use tcp options */
+#define        TF_SENTFIN      0x10            /* have sent FIN */

        struct  tcpiphdr *t_template;   /* skeletal packet for transmit */
        struct  inpcb *t_inpcb;         /* back pointer to internet pcb */
 /*
        struct  tcpiphdr *t_template;   /* skeletal packet for transmit */
        struct  inpcb *t_inpcb;         /* back pointer to internet pcb */
 /*


@@ -31,14 +89,14 @@ struct tcpcb {
 /* send sequence variables */
        tcp_seq snd_una;                /* send unacknowledged */
        tcp_seq snd_nxt;                /* send next */
 /* send sequence variables */
        tcp_seq snd_una;                /* send unacknowledged */
        tcp_seq snd_nxt;                /* send next */

-       u_short snd_wnd;                /* send window */

        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 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 */

+       u_short snd_wnd;                /* send window */

 /* receive sequence variables */
 /* receive sequence variables */

-       tcp_seq rcv_nxt;                /* receive next */

        u_short rcv_wnd;                /* receive window */
        u_short rcv_wnd;                /* receive window */

+       tcp_seq rcv_nxt;                /* receive next */

        tcp_seq rcv_up;                 /* receive urgent pointer */
        tcp_seq irs;                    /* initial receive sequence number */
 /*
        tcp_seq rcv_up;                 /* receive urgent pointer */
        tcp_seq irs;                    /* initial receive sequence number */
 /*


@@ -47,30 +105,155 @@ struct tcpcb {
 /* receive variables */
        tcp_seq rcv_adv;                /* advertised window */
 /* retransmit variables */
 /* receive variables */
        tcp_seq rcv_adv;                /* advertised window */
 /* retransmit variables */

-       tcp_seq snd_max;                /* highest sequence number sent
-                                          used to recognize retransmits */
-       tcp_seq rxt_seq;
-       short   rxt_time;
-       short   rxt_cnt;
-};
+       tcp_seq snd_max;                /* highest sequence number sent;
+                                        * used to recognize retransmits
+                                        */
+/* congestion control (for slow start, source quench, retransmit after loss) */
+       u_short snd_cwnd;               /* congestion-controlled window */
+       u_short snd_ssthresh;           /* snd_cwnd size threshhold for
+                                        * for slow start exponential to
+                                        * linear switch
+                                        */
+/*
+ * transmit timing stuff.  See below for scale of srtt and rttvar.
+ * "Variance" is actually smoothed difference.
+ */
+       short   t_idle;                 /* inactivity time */
+       short   t_rtt;                  /* round trip time */
+       tcp_seq t_rtseq;                /* sequence number being timed */
+       short   t_srtt;                 /* smoothed round-trip time */
+       short   t_rttvar;               /* variance in round-trip time */
+       u_short t_rttmin;               /* minimum rtt allowed */
+       u_short max_sndwnd;             /* largest window peer has offered */

 
 

-#define        tcp_finisacked(tp)      0               /* XXX */
+/* out-of-band data */
+       char    t_oobflags;             /* have some */
+       char    t_iobc;                 /* input character */
+#define        TCPOOB_HAVEDATA 0x01
+#define        TCPOOB_HADDATA  0x02
+       short   t_softerror;            /* possible error not yet reported */
+};

 
 #define        intotcpcb(ip)   ((struct tcpcb *)(ip)->inp_ppcb)
 #define        sototcpcb(so)   (intotcpcb(sotoinpcb(so)))
 
 
 #define        intotcpcb(ip)   ((struct tcpcb *)(ip)->inp_ppcb)
 #define        sototcpcb(so)   (intotcpcb(sotoinpcb(so)))
 

-#define        TCP_ISSINCR     128             /* increment for iss each second */
+/*
+ * 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
+ * binary point, and is smoothed with an ALPHA of 0.75.
+ */
+#define        TCP_RTT_SCALE           8       /* multiplier for srtt; 3 bits frac. */
+#define        TCP_RTT_SHIFT           3       /* shift for srtt; 3 bits frac. */
+#define        TCP_RTTVAR_SCALE        4       /* multiplier for rttvar; 2 bits */
+#define        TCP_RTTVAR_SHIFT        2       /* multiplier for rttvar; 2 bits */
+
+/*
+ * The initial retransmission should happen at rtt + 4 * rttvar.
+ * Because of the way we do the smoothing, srtt and rttvar
+ * will each average +1/2 tick of bias.  When we compute
+ * the retransmit timer, we want 1/2 tick of rounding and
+ * 1 extra tick because of +-1/2 tick uncertainty in the
+ * firing of the timer.  The bias will give us exactly the
+ * 1.5 tick we need.  But, because the bias is
+ * statistical, we have to test that we don't drop below
+ * the minimum feasible timer (which is 2 ticks).
+ * This macro assumes that the value of TCP_RTTVAR_SCALE
+ * is the same as the multiplier for rttvar.
+ */
+#define        TCP_REXMTVAL(tp) \
+       (((tp)->t_srtt >> TCP_RTT_SHIFT) + (tp)->t_rttvar)

 
 

+/* XXX
+ * We want to avoid doing m_pullup on incoming packets but that
+ * means avoiding dtom on the tcp reassembly code.  That in turn means
+ * keeping an mbuf pointer in the reassembly queue (since we might
+ * have a cluster).  As a quick hack, the source & destination
+ * port numbers (which are no longer needed once we've located the
+ * tcpcb) are overlayed with an mbuf pointer.
+ */
+#define REASS_MBUF(ti) (*(struct mbuf **)&((ti)->ti_t))
+
+/*
+ * TCP statistics.
+ * Many of these should be kept per connection,
+ * but that's inconvenient at the moment.
+ */

 struct tcpstat {
 struct tcpstat {

-       int     tcps_badsum;
-       int     tcps_badoff;
-       int     tcps_hdrops;
+#ifdef OLDSTAT
+       /*
+        * Declare statistics the same as in 4.3
+        * at the start of tcpstat (same size and
+        * position) for netstat.
+        */
+       int     tcps_rcvbadsum;
+       int     tcps_rcvbadoff;
+       int     tcps_rcvshort;

        int     tcps_badsegs;
        int     tcps_unack;
        int     tcps_badsegs;
        int     tcps_unack;

+#define        tcps_badsum     tcps_rcvbadsum
+#define        tcps_badoff     tcps_rcvbadoff
+#define        tcps_hdrops     tcps_rcvshort
+
+#endif OLDSTAT
+       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_rcvshort;          /* packets received too short */
+#endif
+       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 */

 };
 
 #ifdef KERNEL
 struct inpcb tcb;              /* head of queue of active tcpcb's */
 struct tcpstat tcpstat;        /* tcp statistics */
 struct tcpiphdr *tcp_template();
 };
 
 #ifdef KERNEL
 struct inpcb tcb;              /* head of queue of active tcpcb's */
 struct tcpstat tcpstat;        /* tcp statistics */
 struct tcpiphdr *tcp_template();

+struct tcpcb *tcp_close(), *tcp_drop();
+struct tcpcb *tcp_timers(), *tcp_disconnect(), *tcp_usrclosed();

 #endif
 #endif