remove fix for acking window probes: don't ack acks!

[unix-history] / usr / src / sys / netinet / tcp_input.c

diff --git a/usr/src/sys/netinet/tcp_input.c b/usr/src/sys/netinet/tcp_input.c
index 5f12bae..c092479 100644 (file)
--- a/usr/src/sys/netinet/tcp_input.c
+++ b/usr/src/sys/netinet/tcp_input.c
@@ -1,28 +1,28 @@
-/*     tcp_input.c     6.2     83/11/04        */
+/*     tcp_input.c     6.9     85/02/12        */

 
 

-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/mbuf.h"
-#include "../h/protosw.h"
-#include "../h/socket.h"
-#include "../h/socketvar.h"
-#include "../h/errno.h"
+#include "param.h"
+#include "systm.h"
+#include "mbuf.h"
+#include "protosw.h"
+#include "socket.h"
+#include "socketvar.h"
+#include "errno.h"

 
 #include "../net/if.h"
 #include "../net/route.h"
 
 
 #include "../net/if.h"
 #include "../net/route.h"
 

-#include "../netinet/in.h"
-#include "../netinet/in_pcb.h"
-#include "../netinet/in_systm.h"
-#include "../netinet/ip.h"
-#include "../netinet/ip_var.h"
-#include "../netinet/tcp.h"
-#include "../netinet/tcp_fsm.h"
-#include "../netinet/tcp_seq.h"
-#include "../netinet/tcp_timer.h"
-#include "../netinet/tcp_var.h"
-#include "../netinet/tcpip.h"
-#include "../netinet/tcp_debug.h"
+#include "in.h"
+#include "in_pcb.h"
+#include "in_systm.h"
+#include "ip.h"
+#include "ip_var.h"
+#include "tcp.h"
+#include "tcp_fsm.h"
+#include "tcp_seq.h"
+#include "tcp_timer.h"
+#include "tcp_var.h"
+#include "tcpip.h"
+#include "tcp_debug.h"

 
 int    tcpprintfs = 0;
 int    tcpcksum = 1;
 
 int    tcpprintfs = 0;
 int    tcpcksum = 1;


@@ -183,10 +183,11 @@ tcp_input(m0)
        tp->t_timer[TCPT_KEEP] = TCPTV_KEEP;
 
        /*
        tp->t_timer[TCPT_KEEP] = TCPTV_KEEP;
 
        /*

-        * Process options.
+        * Process options if not in LISTEN state,
+        * else do it below (after getting remote address).

         */
         */

-       if (om) {
-               tcp_dooptions(tp, om);
+       if (om && tp->t_state != TCPS_LISTEN) {
+               tcp_dooptions(tp, om, ti);

                om = 0;
        }
 
                om = 0;
        }
 


@@ -242,10 +243,15 @@ tcp_input(m0)
                tp->t_template = tcp_template(tp);
                if (tp->t_template == 0) {
                        in_pcbdisconnect(inp);
                tp->t_template = tcp_template(tp);
                if (tp->t_template == 0) {
                        in_pcbdisconnect(inp);

+                       dropsocket = 0;         /* socket is already gone */

                        inp->inp_laddr = laddr;
                        tp = 0;
                        goto drop;
                }
                        inp->inp_laddr = laddr;
                        tp = 0;
                        goto drop;
                }

+               if (om) {
+                       tcp_dooptions(tp, om, ti);
+                       om = 0;
+               }

                tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
                tp->irs = ti->ti_seq;
                tcp_sendseqinit(tp);
                tp->iss = tcp_iss; tcp_iss += TCP_ISSINCR/2;
                tp->irs = ti->ti_seq;
                tcp_sendseqinit(tp);


@@ -291,6 +297,7 @@ tcp_input(m0)
                if (SEQ_GT(tp->snd_una, tp->iss)) {
                        soisconnected(so);
                        tp->t_state = TCPS_ESTABLISHED;
                if (SEQ_GT(tp->snd_una, tp->iss)) {
                        soisconnected(so);
                        tp->t_state = TCPS_ESTABLISHED;

+                       tp->t_maxseg = MIN(tp->t_maxseg, tcp_mss(tp));

                        (void) tcp_reass(tp, (struct tcpiphdr *)0);
                } else
                        tp->t_state = TCPS_SYN_RECEIVED;
                        (void) tcp_reass(tp, (struct tcpiphdr *)0);
                } else
                        tp->t_state = TCPS_SYN_RECEIVED;


@@ -313,6 +320,16 @@ trimthenstep6:
                goto step6;
        }
 
                goto step6;
        }
 

+       /*
+        * If data is received on a connection after the
+        * user processes are gone, then RST the other end.
+        */
+       if ((so->so_state & SS_NOFDREF) && tp->t_state > TCPS_CLOSE_WAIT &&
+           ti->ti_len) {
+               tp = tcp_close(tp);
+               goto dropwithreset;
+       }
+

        /*
         * States other than LISTEN or SYN_SENT.
         * First check that at least some bytes of segment are within 
        /*
         * States other than LISTEN or SYN_SENT.
         * First check that at least some bytes of segment are within 


@@ -376,16 +393,6 @@ trimthenstep6:
                }
        }
 
                }
        }
 

-       /*
-        * If data is received on a connection after the
-        * user processes are gone, then RST the other end.
-        */
-       if ((so->so_state & SS_NOFDREF) && tp->t_state > TCPS_CLOSE_WAIT &&
-           ti->ti_len) {
-               tp = tcp_close(tp);
-               goto dropwithreset;
-       }
-

        /*
         * If the RST bit is set examine the state:
         *    SYN_RECEIVED STATE:
        /*
         * If the RST bit is set examine the state:
         *    SYN_RECEIVED STATE:


@@ -451,6 +458,7 @@ trimthenstep6:
                tp->t_timer[TCPT_REXMT] = 0;
                soisconnected(so);
                tp->t_state = TCPS_ESTABLISHED;
                tp->t_timer[TCPT_REXMT] = 0;
                soisconnected(so);
                tp->t_state = TCPS_ESTABLISHED;

+               tp->t_maxseg = MIN(tp->t_maxseg, tcp_mss(tp));

                (void) tcp_reass(tp, (struct tcpiphdr *)0);
                tp->snd_wl1 = ti->ti_seq - 1;
                /* fall into ... */
                (void) tcp_reass(tp, (struct tcpiphdr *)0);
                tp->snd_wl1 = ti->ti_seq - 1;
                /* fall into ... */


@@ -497,9 +505,13 @@ trimthenstep6:
                else {
                        TCPT_RANGESET(tp->t_timer[TCPT_REXMT],
                            tcp_beta * tp->t_srtt, TCPTV_MIN, TCPTV_MAX);
                else {
                        TCPT_RANGESET(tp->t_timer[TCPT_REXMT],
                            tcp_beta * tp->t_srtt, TCPTV_MIN, TCPTV_MAX);

-                       tp->t_rtt = 1;

                        tp->t_rxtshift = 0;
                }
                        tp->t_rxtshift = 0;
                }

+               /*
+                * When new data is acked, open the congestion window a bit.
+                */
+               if (acked > 0)
+                       tp->snd_cwnd = MIN(11 * tp->snd_cwnd / 10, 65535);

                if (acked > so->so_snd.sb_cc) {
                        tp->snd_wnd -= so->so_snd.sb_cc;
                        sbdrop(&so->so_snd, so->so_snd.sb_cc);
                if (acked > so->so_snd.sb_cc) {
                        tp->snd_wnd -= so->so_snd.sb_cc;
                        sbdrop(&so->so_snd, so->so_snd.sb_cc);


@@ -530,6 +542,12 @@ trimthenstep6:
                                if (so->so_state & SS_CANTRCVMORE)
                                        soisdisconnected(so);
                                tp->t_state = TCPS_FIN_WAIT_2;
                                if (so->so_state & SS_CANTRCVMORE)
                                        soisdisconnected(so);
                                tp->t_state = TCPS_FIN_WAIT_2;

+                               /*
+                                * This is contrary to the specification,
+                                * but if we haven't gotten our FIN in 
+                                * 5 minutes, it's not forthcoming.
+                                */
+                               tp->t_timer[TCPT_2MSL] = 5 * 60 * PR_SLOWHZ;

                        }
                        break;
 
                        }
                        break;
 


@@ -596,7 +614,7 @@ step6:
                 * soreceive.  It's hard to imagine someone
                 * actually wanting to send this much urgent data.
                 */
                 * soreceive.  It's hard to imagine someone
                 * actually wanting to send this much urgent data.
                 */

-               if (ti->ti_urp > tp->t_maxseg) {        /* XXX */
+               if (ti->ti_urp + (unsigned) so->so_rcv.sb_cc > 32767) {

                        ti->ti_urp = 0;                 /* XXX */
                        tiflags &= ~TH_URG;             /* XXX */
                        ti->ti_flags &= ~TH_URG;        /* XXX */
                        ti->ti_urp = 0;                 /* XXX */
                        tiflags &= ~TH_URG;             /* XXX */
                        ti->ti_flags &= ~TH_URG;        /* XXX */


@@ -759,9 +777,10 @@ drop:
        return;
 }
 
        return;
 }
 

-tcp_dooptions(tp, om)
+tcp_dooptions(tp, om, ti)

        struct tcpcb *tp;
        struct mbuf *om;
        struct tcpcb *tp;
        struct mbuf *om;

+       struct tcpiphdr *ti;

 {
        register u_char *cp;
        int opt, optlen, cnt;
 {
        register u_char *cp;
        int opt, optlen, cnt;


@@ -787,8 +806,11 @@ tcp_dooptions(tp, om)
                case TCPOPT_MAXSEG:
                        if (optlen != 4)
                                continue;
                case TCPOPT_MAXSEG:
                        if (optlen != 4)
                                continue;

+                       if (!(ti->ti_flags & TH_SYN))
+                               continue;

                        tp->t_maxseg = *(u_short *)(cp + 2);
                        tp->t_maxseg = ntohs((u_short)tp->t_maxseg);
                        tp->t_maxseg = *(u_short *)(cp + 2);
                        tp->t_maxseg = ntohs((u_short)tp->t_maxseg);

+                       tp->t_maxseg = MIN(tp->t_maxseg, tcp_mss(tp));

                        break;
                }
        }
                        break;
                }
        }


@@ -931,3 +953,51 @@ drop:
        m_freem(dtom(ti));
        return (0);
 }
        m_freem(dtom(ti));
        return (0);
 }

+
+/*
+ *  Determine a reasonable value for maxseg size.
+ *  If the route is known, use one that can be handled
+ *  on the given interface without forcing IP to fragment.
+ *  If bigger than a page (CLSIZE), round down to nearest pagesize
+ *  to utilize pagesize mbufs.
+ *  If interface pointer is unavailable, or the destination isn't local,
+ *  use a conservative size (512 or the default IP max size),
+ *  as we can't discover anything about intervening gateways or networks.
+ *
+ *  This is ugly, and doesn't belong at this level, but has to happen somehow.
+ */
+tcp_mss(tp)
+register struct tcpcb *tp;
+{
+       struct route *ro;
+       struct ifnet *ifp;
+       int mss;
+       struct inpcb *inp;
+
+       inp = tp->t_inpcb;
+       ro = &inp->inp_route;
+       if ((ro->ro_rt == (struct rtentry *)0) ||
+           (ifp = ro->ro_rt->rt_ifp) == (struct ifnet *)0) {
+               /* No route yet, so try to acquire one */
+               if (inp->inp_faddr.s_addr != INADDR_ANY) {
+                       ro->ro_dst.sa_family = AF_INET;
+                       ((struct sockaddr_in *) &ro->ro_dst)->sin_addr =
+                               inp->inp_faddr;
+                       rtalloc(ro);
+               }
+               if ((ro->ro_rt == 0) || (ifp = ro->ro_rt->rt_ifp) == 0)
+                       return (TCP_MSS);
+       }
+
+       mss = ifp->if_mtu - sizeof(struct tcpiphdr);
+#if    (CLBYTES & (CLBYTES - 1)) == 0
+       if (mss > CLBYTES)
+               mss &= ~(CLBYTES-1);
+#else
+       if (mss > CLBYTES)
+               mss = mss / CLBYTES * CLBYTES;
+#endif
+       if (in_localaddr(tp->t_inpcb->inp_faddr))
+               return(mss);
+       return (MIN(mss, TCP_MSS));
+}