rtoss - Rev 73

Subversion Repositories:
Rev:
/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994
 *      The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *      @(#)tcp_input.c 8.5 (Berkeley) 4/10/94
 * tcp_input.c,v 1.10 1994/10/13 18:36:32 wollman Exp
 */


/*
 * Changes and additions relating to SLiRP
 * Copyright (c) 1995 Danny Gasparovski.
 *
 * Please read the file COPYRIGHT for the
 * terms and conditions of the copyright.
 */


#include <slirp.h>
#include "ip_icmp.h"

#define TCPREXMTTHRESH 3

#define TCP_PAWS_IDLE   (24 * 24 * 60 * 60 * PR_SLOWHZ)

/* for modulo comparisons of timestamps */
#define TSTMP_LT(a,b)   ((int)((a)-(b)) < 0)
#define TSTMP_GEQ(a,b)  ((int)((a)-(b)) >= 0)

/*
 * Insert segment ti into reassembly queue of tcp with
 * control block tp.  Return TH_FIN if reassembly now includes
 * a segment with FIN.  The macro form does the common case inline
 * (segment is the next to be received on an established connection,
 * and the queue is empty), avoiding linkage into and removal
 * from the queue and repetition of various conversions.
 * Set DELACK for segments received in order, but ack immediately
 * when segments are out of order (so fast retransmit can work).
 */

#ifdef TCP_ACK_HACK
#define TCP_REASS(tp, ti, m, so, flags) {\
       if ((ti)->ti_seq == (tp)->rcv_nxt && \
           tcpfrag_list_empty(tp) && \
           (tp)->t_state == TCPS_ESTABLISHED) {\
               if (ti->ti_flags & TH_PUSH) \
                       tp->t_flags |= TF_ACKNOW; \
               else \
                       tp->t_flags |= TF_DELACK; \
               (tp)->rcv_nxt += (ti)->ti_len; \
               flags = (ti)->ti_flags & TH_FIN; \
               if (so->so_emu) { \
                       if (tcp_emu((so),(m))) sbappend((so), (m)); \
               } else \
                       sbappend((so), (m)); \
        } else {\
               (flags) = tcp_reass((tp), (ti), (m)); \
               tp->t_flags |= TF_ACKNOW; \
       } \
}

#else
#define TCP_REASS(tp, ti, m, so, flags) { \
        if ((ti)->ti_seq == (tp)->rcv_nxt && \
        tcpfrag_list_empty(tp) && \
            (tp)->t_state == TCPS_ESTABLISHED) { \
                tp->t_flags |= TF_DELACK; \
                (tp)->rcv_nxt += (ti)->ti_len; \
                flags = (ti)->ti_flags & TH_FIN; \
                if (so->so_emu) { \
                        if (tcp_emu((so),(m))) sbappend(so, (m)); \
                } else \
                        sbappend((so), (m)); \
        } else { \
                (flags) = tcp_reass((tp), (ti), (m)); \
                tp->t_flags |= TF_ACKNOW; \
        } \
}

#endif
static void tcp_dooptions(struct tcpcb *tp, u_char *cp, int cnt,
                          struct tcpiphdr *ti);
static void tcp_xmit_timer(register struct tcpcb *tp, int rtt);

static int
tcp_reass(register struct tcpcb *tp, register struct tcpiphdr *ti,
          struct mbuf *m)
{
        register struct tcpiphdr *q;
        struct socket *so = tp->t_socket;
        int flags;

        /*
         * Call with ti==NULL after become established to
         * force pre-ESTABLISHED data up to user socket.
         */

        if (ti == NULL)
                goto present;

        /*
         * Find a segment which begins after this one does.
         */

        for (q = tcpfrag_list_first(tp); !tcpfrag_list_end(q, tp);
            q = tcpiphdr_next(q))
                if (SEQ_GT(q->ti_seq, ti->ti_seq))
                        break;

        /*
         * If there is a preceding segment, it may provide some of
         * our data already.  If so, drop the data from the incoming
         * segment.  If it provides all of our data, drop us.
         */

        if (!tcpfrag_list_end(tcpiphdr_prev(q), tp)) {
                register int i;
                q = tcpiphdr_prev(q);
                /* conversion to int (in i) handles seq wraparound */
                i = q->ti_seq + q->ti_len - ti->ti_seq;
                if (i > 0) {
                        if (i >= ti->ti_len) {
                                m_freem(m);
                                /*
                                 * Try to present any queued data
                                 * at the left window edge to the user.
                                 * This is needed after the 3-WHS
                                 * completes.
                                 */

                                goto present;   /* ??? */
                        }
                        m_adj(m, i);
                        ti->ti_len -= i;
                        ti->ti_seq += i;
                }
                q = tcpiphdr_next(q);
        }
        ti->ti_mbuf = m;

        /*
         * While we overlap succeeding segments trim them or,
         * if they are completely covered, dequeue them.
         */

        while (!tcpfrag_list_end(q, tp)) {
                register int i = (ti->ti_seq + ti->ti_len) - q->ti_seq;
                if (i <= 0)
                        break;
                if (i < q->ti_len) {
                        q->ti_seq += i;
                        q->ti_len -= i;
                        m_adj(q->ti_mbuf, i);
                        break;
                }
                q = tcpiphdr_next(q);
                m = tcpiphdr_prev(q)->ti_mbuf;
                remque(tcpiphdr2qlink(tcpiphdr_prev(q)));
                m_freem(m);
        }

        /*
         * Stick new segment in its place.
         */

        insque(tcpiphdr2qlink(ti), tcpiphdr2qlink(tcpiphdr_prev(q)));

present:
        /*
         * Present data to user, advancing rcv_nxt through
         * completed sequence space.
         */

        if (!TCPS_HAVEESTABLISHED(tp->t_state))
                return (0);
        ti = tcpfrag_list_first(tp);
        if (tcpfrag_list_end(ti, tp) || ti->ti_seq != tp->rcv_nxt)
                return (0);
        if (tp->t_state == TCPS_SYN_RECEIVED && ti->ti_len)
                return (0);
        do {
                tp->rcv_nxt += ti->ti_len;
                flags = ti->ti_flags & TH_FIN;
                remque(tcpiphdr2qlink(ti));
                m = ti->ti_mbuf;
                ti = tcpiphdr_next(ti);
                if (so->so_state & SS_FCANTSENDMORE)
                        m_freem(m);
                else {
                        if (so->so_emu) {
                                if (tcp_emu(so,m)) sbappend(so, m);
                        } else
                                sbappend(so, m);
                }
        } while (ti != (struct tcpiphdr *)tp && ti->ti_seq == tp->rcv_nxt);
        return (flags);
}

/*
 * TCP input routine, follows pages 65-76 of the
 * protocol specification dated September, 1981 very closely.
 */

void
tcp_input(struct mbuf *m, int iphlen, struct socket *inso)
{
        struct ip save_ip, *ip;
        register struct tcpiphdr *ti;
        caddr_t optp = NULL;
        int optlen = 0;
        int len, tlen, off;
        register struct tcpcb *tp = NULL;
        register int tiflags;
        struct socket *so = NULL;
        int todrop, acked, ourfinisacked, needoutput = 0;
        int iss = 0;
        u_long tiwin;
        int ret;
    struct ex_list *ex_ptr;
    Slirp *slirp;

        DEBUG_CALL("tcp_input");
        DEBUG_ARGS((dfd," m = %8lx  iphlen = %2d  inso = %lx\n",
                    (long )m, iphlen, (long )inso ));

        /*
         * If called with m == 0, then we're continuing the connect
         */

        if (m == NULL) {
                so = inso;
                slirp = so->slirp;

                /* Re-set a few variables */
                tp = sototcpcb(so);
                m = so->so_m;
                so->so_m = NULL;
                ti = so->so_ti;
                tiwin = ti->ti_win;
                tiflags = ti->ti_flags;

                goto cont_conn;
        }
        slirp = m->slirp;

        /*
         * Get IP and TCP header together in first mbuf.
         * Note: IP leaves IP header in first mbuf.
         */

        ti = mtod(m, struct tcpiphdr *);
        if (iphlen > sizeof(struct ip )) {
          ip_stripoptions(m, (struct mbuf *)0);
          iphlen=sizeof(struct ip );
        }
        /* XXX Check if too short */


        /*
         * Save a copy of the IP header in case we want restore it
         * for sending an ICMP error message in response.
         */

        ip=mtod(m, struct ip *);
        save_ip = *ip;
        save_ip.ip_len+= iphlen;

        /*
         * Checksum extended TCP header and data.
         */

        tlen = ((struct ip *)ti)->ip_len;
        tcpiphdr2qlink(ti)->next = tcpiphdr2qlink(ti)->prev = NULL;
        memset(&ti->ti_i.ih_mbuf, 0 , sizeof(struct mbuf_ptr));
        ti->ti_x1 = 0;
        ti->ti_len = htons((u_int16_t)tlen);
        len = sizeof(struct ip ) + tlen;
        if(cksum(m, len)) {
          goto drop;
        }

        /*
         * Check that TCP offset makes sense,
         * pull out TCP options and adjust length.              XXX
         */

        off = ti->ti_off << 2;
        if (off < sizeof (struct tcphdr) || off > tlen) {
          goto drop;
        }
        tlen -= off;
        ti->ti_len = tlen;
        if (off > sizeof (struct tcphdr)) {
          optlen = off - sizeof (struct tcphdr);
          optp = mtod(m, caddr_t) + sizeof (struct tcpiphdr);
        }
        tiflags = ti->ti_flags;

        /*
         * Convert TCP protocol specific fields to host format.
         */

        NTOHL(ti->ti_seq);
        NTOHL(ti->ti_ack);
        NTOHS(ti->ti_win);
        NTOHS(ti->ti_urp);

        /*
         * Drop TCP, IP headers and TCP options.
         */

        m->m_data += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
        m->m_len  -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);

    if (slirp->restricted) {
        for (ex_ptr = slirp->exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
            if (ex_ptr->ex_fport == ti->ti_dport &&
                ti->ti_dst.s_addr == ex_ptr->ex_addr.s_addr) {
                break;
            }
        }
        if (!ex_ptr)
            goto drop;
    }
        /*
         * Locate pcb for segment.
         */

findso:
        so = slirp->tcp_last_so;
        if (so->so_fport != ti->ti_dport ||
            so->so_lport != ti->ti_sport ||
            so->so_laddr.s_addr != ti->ti_src.s_addr ||
            so->so_faddr.s_addr != ti->ti_dst.s_addr) {
                so = solookup(&slirp->tcb, ti->ti_src, ti->ti_sport,
                               ti->ti_dst, ti->ti_dport);
                if (so)
                        slirp->tcp_last_so = so;
        }

        /*
         * If the state is CLOSED (i.e., TCB does not exist) then
         * all data in the incoming segment is discarded.
         * If the TCB exists but is in CLOSED state, it is embryonic,
         * but should either do a listen or a connect soon.
         *
         * state == CLOSED means we've done socreate() but haven't
         * attached it to a protocol yet...
         *
         * XXX If a TCB does not exist, and the TH_SYN flag is
         * the only flag set, then create a session, mark it
         * as if it was LISTENING, and continue...
         */

        if (so == NULL) {
          if ((tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) != TH_SYN)
            goto dropwithreset;

          if ((so = socreate(slirp)) == NULL)
            goto dropwithreset;
          if (tcp_attach(so) < 0) {
            free(so); /* Not sofree (if it failed, it's not insqued) */
            goto dropwithreset;
          }

          sbreserve(&so->so_snd, TCP_SNDSPACE);
          sbreserve(&so->so_rcv, TCP_RCVSPACE);

          so->so_laddr = ti->ti_src;
          so->so_lport = ti->ti_sport;
          so->so_faddr = ti->ti_dst;
          so->so_fport = ti->ti_dport;

          if ((so->so_iptos = tcp_tos(so)) == 0)
            so->so_iptos = ((struct ip *)ti)->ip_tos;

          tp = sototcpcb(so);
          tp->t_state = TCPS_LISTEN;
        }

        /*
         * If this is a still-connecting socket, this probably
         * a retransmit of the SYN.  Whether it's a retransmit SYN
         * or something else, we nuke it.
         */

        if (so->so_state & SS_ISFCONNECTING)
                goto drop;

        tp = sototcpcb(so);

        /* XXX Should never fail */
        if (tp == NULL)
                goto dropwithreset;
        if (tp->t_state == TCPS_CLOSED)
                goto drop;

        tiwin = ti->ti_win;

        /*
         * Segment received on connection.
         * Reset idle time and keep-alive timer.
         */

        tp->t_idle = 0;
        if (SO_OPTIONS)
           tp->t_timer[TCPT_KEEP] = TCPTV_KEEPINTVL;
        else
           tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_IDLE;

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

        if (optp && tp->t_state != TCPS_LISTEN)
                tcp_dooptions(tp, (u_char *)optp, optlen, ti);

        /*
         * Header prediction: check for the two common cases
         * of a uni-directional data xfer.  If the packet has
         * no control flags, is in-sequence, the window didn't
         * change and we're not retransmitting, it's a
         * candidate.  If the length is zero and the ack moved
         * forward, we're the sender side of the xfer.  Just
         * free the data acked & wake any higher level process
         * that was blocked waiting for space.  If the length
         * is non-zero and the ack didn't move, we're the
         * receiver side.  If we're getting packets in-order
         * (the reassembly queue is empty), add the data to
         * the socket buffer and note that we need a delayed ack.
         *
         * XXX Some of these tests are not needed
         * eg: the tiwin == tp->snd_wnd prevents many more
         * predictions.. with no *real* advantage..
         */

        if (tp->t_state == TCPS_ESTABLISHED &&
            (tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ACK)) == TH_ACK &&
            ti->ti_seq == tp->rcv_nxt &&
            tiwin && tiwin == tp->snd_wnd &&
            tp->snd_nxt == tp->snd_max) {
                if (ti->ti_len == 0) {
                        if (SEQ_GT(ti->ti_ack, tp->snd_una) &&
                            SEQ_LEQ(ti->ti_ack, tp->snd_max) &&
                            tp->snd_cwnd >= tp->snd_wnd) {
                                /*
                                 * this is a pure ack for outstanding data.
                                 */

                                if (tp->t_rtt &&
                                    SEQ_GT(ti->ti_ack, tp->t_rtseq))
                                        tcp_xmit_timer(tp, tp->t_rtt);
                                acked = ti->ti_ack - tp->snd_una;
                                sbdrop(&so->so_snd, acked);
                                tp->snd_una = ti->ti_ack;
                                m_freem(m);

                                /*
                                 * If all outstanding data are acked, stop
                                 * retransmit timer, otherwise restart timer
                                 * using current (possibly backed-off) value.
                                 * If process is waiting for space,
                                 * wakeup/selwakeup/signal.  If data
                                 * are ready to send, let tcp_output
                                 * decide between more output or persist.
                                 */

                                if (tp->snd_una == tp->snd_max)
                                        tp->t_timer[TCPT_REXMT] = 0;
                                else if (tp->t_timer[TCPT_PERSIST] == 0)
                                        tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;

                                /*
                                 * This is called because sowwakeup might have
                                 * put data into so_snd.  Since we don't so sowwakeup,
                                 * we don't need this.. XXX???
                                 */

                                if (so->so_snd.sb_cc)
                                        (void) tcp_output(tp);

                                return;
                        }
                } else if (ti->ti_ack == tp->snd_una &&
                    tcpfrag_list_empty(tp) &&
                    ti->ti_len <= sbspace(&so->so_rcv)) {
                        /*
                         * this is a pure, in-sequence data packet
                         * with nothing on the reassembly queue and
                         * we have enough buffer space to take it.
                         */

                        tp->rcv_nxt += ti->ti_len;
                        /*
                         * Add data to socket buffer.
                         */

                        if (so->so_emu) {
                                if (tcp_emu(so,m)) sbappend(so, m);
                        } else
                                sbappend(so, m);

                        /*
                         * If this is a short packet, then ACK now - with Nagel
                         *      congestion avoidance sender won't send more until
                         *      he gets an ACK.
                         *
                         * It is better to not delay acks at all to maximize
                         * TCP throughput.  See RFC 2581.
                         */

                        tp->t_flags |= TF_ACKNOW;
                        tcp_output(tp);
                        return;
                }
        } /* header prediction */
        /*
         * Calculate amount of space in receive window,
         * and then do TCP input processing.
         * Receive window is amount of space in rcv queue,
         * but not less than advertised window.
         */

        { int win;
          win = sbspace(&so->so_rcv);
          if (win < 0)
            win = 0;
          tp->rcv_wnd = max(win, (int)(tp->rcv_adv - tp->rcv_nxt));
        }

        switch (tp->t_state) {

        /*
         * If the state is LISTEN then ignore segment if it contains an RST.
         * If the segment contains an ACK then it is bad and send a RST.
         * If it does not contain a SYN then it is not interesting; drop it.
         * Don't bother responding if the destination was a broadcast.
         * Otherwise initialize tp->rcv_nxt, and tp->irs, select an initial
         * tp->iss, and send a segment:
         *     <SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK>
         * Also initialize tp->snd_nxt to tp->iss+1 and tp->snd_una to tp->iss.
         * Fill in remote peer address fields if not previously specified.
         * Enter SYN_RECEIVED state, and process any other fields of this
         * segment in this state.
         */

        case TCPS_LISTEN: {

          if (tiflags & TH_RST)
            goto drop;
          if (tiflags & TH_ACK)
            goto dropwithreset;
          if ((tiflags & TH_SYN) == 0)
            goto drop;

          /*
           * This has way too many gotos...
           * But a bit of spaghetti code never hurt anybody :)
           */


          /*
           * If this is destined for the control address, then flag to
           * tcp_ctl once connected, otherwise connect
           */

          if ((so->so_faddr.s_addr & slirp->vnetwork_mask.s_addr) ==
              slirp->vnetwork_addr.s_addr) {
            if (so->so_faddr.s_addr != slirp->vhost_addr.s_addr &&
                so->so_faddr.s_addr != slirp->vnameserver_addr.s_addr) {
                /* May be an add exec */
                for (ex_ptr = slirp->exec_list; ex_ptr;
                     ex_ptr = ex_ptr->ex_next) {
                  if(ex_ptr->ex_fport == so->so_fport &&
                     so->so_faddr.s_addr == ex_ptr->ex_addr.s_addr) {
                    so->so_state |= SS_CTL;
                    break;
                  }
                }
                if (so->so_state & SS_CTL) {
                    goto cont_input;
                }
            }
            /* CTL_ALIAS: Do nothing, tcp_fconnect will be called on it */
          }

          if (so->so_emu & EMU_NOCONNECT) {
            so->so_emu &= ~EMU_NOCONNECT;
            goto cont_input;
          }

          if((tcp_fconnect(so) == -1) && (errno != EINPROGRESS) && (errno != EWOULDBLOCK)) {
            u_char code=ICMP_UNREACH_NET;
            DEBUG_MISC((dfd," tcp fconnect errno = %d-%s\n",
                        errno,strerror(errno)));
            if(errno == ECONNREFUSED) {
              /* ACK the SYN, send RST to refuse the connection */
              tcp_respond(tp, ti, m, ti->ti_seq+1, (tcp_seq)0,
                          TH_RST|TH_ACK);
            } else {
              if(errno == EHOSTUNREACH) code=ICMP_UNREACH_HOST;
              HTONL(ti->ti_seq);             /* restore tcp header */
              HTONL(ti->ti_ack);
              HTONS(ti->ti_win);
              HTONS(ti->ti_urp);
              m->m_data -= sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
              m->m_len  += sizeof(struct tcpiphdr)+off-sizeof(struct tcphdr);
              *ip=save_ip;
              icmp_error(m, ICMP_UNREACH,code, 0,strerror(errno));
            }
            tp = tcp_close(tp);
            m_free(m);
          } else {
            /*
             * Haven't connected yet, save the current mbuf
             * and ti, and return
             * XXX Some OS's don't tell us whether the connect()
             * succeeded or not.  So we must time it out.
             */

            so->so_m = m;
            so->so_ti = ti;
            tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
            tp->t_state = TCPS_SYN_RECEIVED;
          }
          return;

        cont_conn:
          /* m==NULL
           * Check if the connect succeeded
           */

          if (so->so_state & SS_NOFDREF) {
            tp = tcp_close(tp);
            goto dropwithreset;
          }
        cont_input:
          tcp_template(tp);

          if (optp)
            tcp_dooptions(tp, (u_char *)optp, optlen, ti);

          if (iss)
            tp->iss = iss;
          else
            tp->iss = slirp->tcp_iss;
          slirp->tcp_iss += TCP_ISSINCR/2;
          tp->irs = ti->ti_seq;
          tcp_sendseqinit(tp);
          tcp_rcvseqinit(tp);
          tp->t_flags |= TF_ACKNOW;
          tp->t_state = TCPS_SYN_RECEIVED;
          tp->t_timer[TCPT_KEEP] = TCPTV_KEEP_INIT;
          goto trimthenstep6;
        } /* case TCPS_LISTEN */

        /*
         * If the state is SYN_SENT:
         *      if seg contains an ACK, but not for our SYN, drop the input.
         *      if seg contains a RST, then drop the connection.
         *      if seg does not contain SYN, then drop it.
         * Otherwise this is an acceptable SYN segment
         *      initialize tp->rcv_nxt and tp->irs
         *      if seg contains ack then advance tp->snd_una
         *      if SYN has been acked change to ESTABLISHED else SYN_RCVD state
         *      arrange for segment to be acked (eventually)
         *      continue processing rest of data/controls, beginning with URG
         */

        case TCPS_SYN_SENT:
                if ((tiflags & TH_ACK) &&
                    (SEQ_LEQ(ti->ti_ack, tp->iss) ||
                     SEQ_GT(ti->ti_ack, tp->snd_max)))
                        goto dropwithreset;

                if (tiflags & TH_RST) {
                        if (tiflags & TH_ACK)
                                tp = tcp_drop(tp,0); /* XXX Check t_softerror! */
                        goto drop;
                }

                if ((tiflags & TH_SYN) == 0)
                        goto drop;
                if (tiflags & TH_ACK) {
                        tp->snd_una = ti->ti_ack;
                        if (SEQ_LT(tp->snd_nxt, tp->snd_una))
                                tp->snd_nxt = tp->snd_una;
                }

                tp->t_timer[TCPT_REXMT] = 0;
                tp->irs = ti->ti_seq;
                tcp_rcvseqinit(tp);
                tp->t_flags |= TF_ACKNOW;
                if (tiflags & TH_ACK && SEQ_GT(tp->snd_una, tp->iss)) {
                        soisfconnected(so);
                        tp->t_state = TCPS_ESTABLISHED;

                        (void) tcp_reass(tp, (struct tcpiphdr *)0,
                                (struct mbuf *)0);
                        /*
                         * if we didn't have to retransmit the SYN,
                         * use its rtt as our initial srtt & rtt var.
                         */

                        if (tp->t_rtt)
                                tcp_xmit_timer(tp, tp->t_rtt);
                } else
                        tp->t_state = TCPS_SYN_RECEIVED;

trimthenstep6:
                /*
                 * Advance ti->ti_seq to correspond to first data byte.
                 * If data, trim to stay within window,
                 * dropping FIN if necessary.
                 */

                ti->ti_seq++;
                if (ti->ti_len > tp->rcv_wnd) {
                        todrop = ti->ti_len - tp->rcv_wnd;
                        m_adj(m, -todrop);
                        ti->ti_len = tp->rcv_wnd;
                        tiflags &= ~TH_FIN;
                }
                tp->snd_wl1 = ti->ti_seq - 1;
                tp->rcv_up = ti->ti_seq;
                goto step6;
        } /* switch tp->t_state */
        /*
         * States other than LISTEN or SYN_SENT.
         * Check that at least some bytes of segment are within
         * receive window.  If segment begins before rcv_nxt,
         * drop leading data (and SYN); if nothing left, just ack.
         */

        todrop = tp->rcv_nxt - ti->ti_seq;
        if (todrop > 0) {
                if (tiflags & TH_SYN) {
                        tiflags &= ~TH_SYN;
                        ti->ti_seq++;
                        if (ti->ti_urp > 1)
                                ti->ti_urp--;
                        else
                                tiflags &= ~TH_URG;
                        todrop--;
                }
                /*
                 * Following if statement from Stevens, vol. 2, p. 960.
                 */

                if (todrop > ti->ti_len
                    || (todrop == ti->ti_len && (tiflags & TH_FIN) == 0)) {
                        /*
                         * Any valid FIN must be to the left of the window.
                         * At this point the FIN must be a duplicate or out
                         * of sequence; drop it.
                         */

                        tiflags &= ~TH_FIN;

                        /*
                         * Send an ACK to resynchronize and drop any data.
                         * But keep on processing for RST or ACK.
                         */

                        tp->t_flags |= TF_ACKNOW;
                        todrop = ti->ti_len;
                }
                m_adj(m, todrop);
                ti->ti_seq += todrop;
                ti->ti_len -= todrop;
                if (ti->ti_urp > todrop)
                        ti->ti_urp -= todrop;
                else {
                        tiflags &= ~TH_URG;
                        ti->ti_urp = 0;
                }
        }
        /*
         * If new data are 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 segment ends after window, drop trailing data
         * (and PUSH and FIN); if nothing left, just ACK.
         */

        todrop = (ti->ti_seq+ti->ti_len) - (tp->rcv_nxt+tp->rcv_wnd);
        if (todrop > 0) {
                if (todrop >= ti->ti_len) {
                        /*
                         * If a new connection request is received
                         * while in TIME_WAIT, drop the old connection
                         * and start over if the sequence numbers
                         * are above the previous ones.
                         */

                        if (tiflags & TH_SYN &&
                            tp->t_state == TCPS_TIME_WAIT &&
                            SEQ_GT(ti->ti_seq, tp->rcv_nxt)) {
                                iss = tp->rcv_nxt + TCP_ISSINCR;
                                tp = tcp_close(tp);
                                goto findso;
                        }
                        /*
                         * If window is closed can only take segments at
                         * window edge, and have to drop data and PUSH from
                         * incoming segments.  Continue processing, but
                         * remember to ack.  Otherwise, drop segment
                         * and ack.
                         */

                        if (tp->rcv_wnd == 0 && ti->ti_seq == tp->rcv_nxt) {
                                tp->t_flags |= TF_ACKNOW;
                        } else {
                                goto dropafterack;
                        }
                }
                m_adj(m, -todrop);
                ti->ti_len -= todrop;
                tiflags &= ~(TH_PUSH|TH_FIN);
        }

        /*
         * If the RST bit is set examine the state:
         *    SYN_RECEIVED STATE:
         *      If passive open, return to LISTEN state.
         *      If active open, inform user that connection was refused.
         *    ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES:
         *      Inform user that connection was reset, and close tcb.
         *    CLOSING, LAST_ACK, TIME_WAIT STATES
         *      Close the tcb.
         */

        if (tiflags&TH_RST) switch (tp->t_state) {

        case TCPS_SYN_RECEIVED:
        case TCPS_ESTABLISHED:
        case TCPS_FIN_WAIT_1:
        case TCPS_FIN_WAIT_2:
        case TCPS_CLOSE_WAIT:
                tp->t_state = TCPS_CLOSED;
                tp = tcp_close(tp);
                goto drop;

        case TCPS_CLOSING:
        case TCPS_LAST_ACK:
        case TCPS_TIME_WAIT:
                tp = tcp_close(tp);
                goto drop;
        }

        /*
         * If a SYN is in the window, then this is an
         * error and we send an RST and drop the connection.
         */

        if (tiflags & TH_SYN) {
                tp = tcp_drop(tp,0);
                goto dropwithreset;
        }

        /*
         * If the ACK bit is off we drop the segment and return.
         */

        if ((tiflags & TH_ACK) == 0) goto drop;

        /*
         * Ack processing.
         */

        switch (tp->t_state) {
        /*
         * In SYN_RECEIVED state if the ack ACKs our SYN then enter
         * ESTABLISHED state and continue processing, otherwise
         * send an RST.  una<=ack<=max
         */

        case TCPS_SYN_RECEIVED:

                if (SEQ_GT(tp->snd_una, ti->ti_ack) ||
                    SEQ_GT(ti->ti_ack, tp->snd_max))
                        goto dropwithreset;
                tp->t_state = TCPS_ESTABLISHED;
                /*
                 * The sent SYN is ack'ed with our sequence number +1
                 * The first data byte already in the buffer will get
                 * lost if no correction is made.  This is only needed for
                 * SS_CTL since the buffer is empty otherwise.
                 * tp->snd_una++; or:
                 */

                tp->snd_una=ti->ti_ack;
                if (so->so_state & SS_CTL) {
                  /* So tcp_ctl reports the right state */
                  ret = tcp_ctl(so);
                  if (ret == 1) {
                    soisfconnected(so);
                    so->so_state &= ~SS_CTL;   /* success XXX */
                  } else if (ret == 2) {
                    so->so_state &= SS_PERSISTENT_MASK;
                    so->so_state |= SS_NOFDREF; /* CTL_CMD */
                  } else {
                    needoutput = 1;
                    tp->t_state = TCPS_FIN_WAIT_1;
                  }
                } else {
                  soisfconnected(so);
                }

                (void) tcp_reass(tp, (struct tcpiphdr *)0, (struct mbuf *)0);
                tp->snd_wl1 = ti->ti_seq - 1;
                /* Avoid ack processing; snd_una==ti_ack  =>  dup ack */
                goto synrx_to_est;
                /* fall into ... */

        /*
         * In ESTABLISHED state: drop duplicate ACKs; ACK out of range
         * ACKs.  If the ack is in the range
         *      tp->snd_una < ti->ti_ack <= tp->snd_max
         * then advance tp->snd_una to ti->ti_ack and drop
         * data from the retransmission queue.  If this ACK reflects
         * more up to date window information we update our window information.
         */

        case TCPS_ESTABLISHED:
        case TCPS_FIN_WAIT_1:
        case TCPS_FIN_WAIT_2:
        case TCPS_CLOSE_WAIT:
        case TCPS_CLOSING:
        case TCPS_LAST_ACK:
        case TCPS_TIME_WAIT:

                if (SEQ_LEQ(ti->ti_ack, tp->snd_una)) {
                        if (ti->ti_len == 0 && tiwin == tp->snd_wnd) {
                          DEBUG_MISC((dfd," dup ack  m = %lx  so = %lx \n",
                                      (long )m, (long )so));
                                /*
                                 * If we have outstanding data (other than
                                 * a window probe), this is a completely
                                 * duplicate ack (ie, window info didn't
                                 * change), the ack is the biggest we've
                                 * seen and we've seen exactly our rexmt
                                 * threshold of them, assume a packet
                                 * has been dropped and retransmit it.
                                 * Kludge snd_nxt & the congestion
                                 * window so we send only this one
                                 * packet.
                                 *
                                 * We know we're losing at the current
                                 * window size so do congestion avoidance
                                 * (set ssthresh to half the current window
                                 * and pull our congestion window back to
                                 * the new ssthresh).
                                 *
                                 * Dup acks mean that packets have left the
                                 * network (they're now cached at the receiver)
                                 * so bump cwnd by the amount in the receiver
                                 * to keep a constant cwnd packets in the
                                 * network.
                                 */

                                if (tp->t_timer[TCPT_REXMT] == 0 ||
                                    ti->ti_ack != tp->snd_una)
                                        tp->t_dupacks = 0;
                                else if (++tp->t_dupacks == TCPREXMTTHRESH) {
                                        tcp_seq onxt = tp->snd_nxt;
                                        u_int win =
                                            min(tp->snd_wnd, tp->snd_cwnd) / 2 /
                                                tp->t_maxseg;

                                        if (win < 2)
                                                win = 2;
                                        tp->snd_ssthresh = win * tp->t_maxseg;
                                        tp->t_timer[TCPT_REXMT] = 0;
                                        tp->t_rtt = 0;
                                        tp->snd_nxt = ti->ti_ack;
                                        tp->snd_cwnd = tp->t_maxseg;
                                        (void) tcp_output(tp);
                                        tp->snd_cwnd = tp->snd_ssthresh +
                                               tp->t_maxseg * tp->t_dupacks;
                                        if (SEQ_GT(onxt, tp->snd_nxt))
                                                tp->snd_nxt = onxt;
                                        goto drop;
                                } else if (tp->t_dupacks > TCPREXMTTHRESH) {
                                        tp->snd_cwnd += tp->t_maxseg;
                                        (void) tcp_output(tp);
                                        goto drop;
                                }
                        } else
                                tp->t_dupacks = 0;
                        break;
                }
        synrx_to_est:
                /*
                 * If the congestion window was inflated to account
                 * for the other side's cached packets, retract it.
                 */

                if (tp->t_dupacks > TCPREXMTTHRESH &&
                    tp->snd_cwnd > tp->snd_ssthresh)
                        tp->snd_cwnd = tp->snd_ssthresh;
                tp->t_dupacks = 0;
                if (SEQ_GT(ti->ti_ack, tp->snd_max)) {
                        goto dropafterack;
                }
                acked = ti->ti_ack - tp->snd_una;

                /*
                 * If transmit timer is running and timed sequence
                 * number was acked, update smoothed round trip time.
                 * Since we now have an rtt measurement, cancel the
                 * timer backoff (cf., Phil Karn's retransmit alg.).
                 * Recompute the initial retransmit timer.
                 */

                if (tp->t_rtt && SEQ_GT(ti->ti_ack, tp->t_rtseq))
                        tcp_xmit_timer(tp,tp->t_rtt);

                /*
                 * If all outstanding data is acked, stop retransmit
                 * timer and remember to restart (more output or persist).
                 * If there is more data to be acked, restart retransmit
                 * timer, using current (possibly backed-off) value.
                 */

                if (ti->ti_ack == tp->snd_max) {
                        tp->t_timer[TCPT_REXMT] = 0;
                        needoutput = 1;
                } else if (tp->t_timer[TCPT_PERSIST] == 0)
                        tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
                /*
                 * When new data is acked, open the congestion window.
                 * If the window gives us less than ssthresh packets
                 * in flight, open exponentially (maxseg per packet).
                 * Otherwise open linearly: maxseg per window
                 * (maxseg^2 / cwnd per packet).
                 */

                {
                  register u_int cw = tp->snd_cwnd;
                  register u_int incr = tp->t_maxseg;

                  if (cw > tp->snd_ssthresh)
                    incr = incr * incr / cw;
                  tp->snd_cwnd = min(cw + incr, TCP_MAXWIN<<tp->snd_scale);
                }
                if (acked > so->so_snd.sb_cc) {
                        tp->snd_wnd -= so->so_snd.sb_cc;
                        sbdrop(&so->so_snd, (int )so->so_snd.sb_cc);
                        ourfinisacked = 1;
                } else {
                        sbdrop(&so->so_snd, acked);
                        tp->snd_wnd -= acked;
                        ourfinisacked = 0;
                }
                tp->snd_una = ti->ti_ack;
                if (SEQ_LT(tp->snd_nxt, tp->snd_una))
                        tp->snd_nxt = tp->snd_una;

                switch (tp->t_state) {

                /*
                 * In FIN_WAIT_1 STATE in addition to the processing
                 * for the ESTABLISHED state if our FIN is now acknowledged
                 * then enter FIN_WAIT_2.
                 */

                case TCPS_FIN_WAIT_1:
                        if (ourfinisacked) {
                                /*
                                 * If we can't receive any more
                                 * data, then closing user can proceed.
                                 * Starting the timer is contrary to the
                                 * specification, but if we don't get a FIN
                                 * we'll hang forever.
                                 */

                                if (so->so_state & SS_FCANTRCVMORE) {
                                        tp->t_timer[TCPT_2MSL] = TCP_MAXIDLE;
                                }
                                tp->t_state = TCPS_FIN_WAIT_2;
                        }
                        break;

                /*
                 * In CLOSING STATE in addition to the processing for
                 * the ESTABLISHED state if the ACK acknowledges our FIN
                 * then enter the TIME-WAIT state, otherwise ignore
                 * the segment.
                 */

                case TCPS_CLOSING:
                        if (ourfinisacked) {
                                tp->t_state = TCPS_TIME_WAIT;
                                tcp_canceltimers(tp);
                                tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
                        }
                        break;

                /*
                 * In LAST_ACK, we may still be waiting for data to drain
                 * and/or to be acked, as well as for the ack of our FIN.
                 * If our FIN is now acknowledged, delete the TCB,
                 * enter the closed state and return.
                 */

                case TCPS_LAST_ACK:
                        if (ourfinisacked) {
                                tp = tcp_close(tp);
                                goto drop;
                        }
                        break;

                /*
                 * In TIME_WAIT state the only thing that should arrive
                 * is a retransmission of the remote FIN.  Acknowledge
                 * it and restart the finack timer.
                 */

                case TCPS_TIME_WAIT:
                        tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
                        goto dropafterack;
                }
        } /* switch(tp->t_state) */

step6:
        /*
         * Update window information.
         * Don't look at window if no ACK: TAC's send garbage on first SYN.
         */

        if ((tiflags & TH_ACK) &&
            (SEQ_LT(tp->snd_wl1, ti->ti_seq) ||
            (tp->snd_wl1 == ti->ti_seq && (SEQ_LT(tp->snd_wl2, ti->ti_ack) ||
            (tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd))))) {
                tp->snd_wnd = tiwin;
                tp->snd_wl1 = ti->ti_seq;
                tp->snd_wl2 = ti->ti_ack;
                if (tp->snd_wnd > tp->max_sndwnd)
                        tp->max_sndwnd = tp->snd_wnd;
                needoutput = 1;
        }

        /*
         * Process segments with URG.
         */

        if ((tiflags & TH_URG) && ti->ti_urp &&
            TCPS_HAVERCVDFIN(tp->t_state) == 0) {
                /*
                 * This is a kludge, but if we receive and accept
                 * random urgent pointers, we'll crash in
                 * soreceive.  It's hard to imagine someone
                 * actually wanting to send this much urgent data.
                 */

                if (ti->ti_urp + so->so_rcv.sb_cc > so->so_rcv.sb_datalen) {
                        ti->ti_urp = 0;
                        tiflags &= ~TH_URG;
                        goto dodata;
                }
                /*
                 * If this segment advances the known urgent pointer,
                 * then mark the data stream.  This should not happen
                 * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since
                 * a FIN has been received from the remote side.
                 * In these states we ignore the URG.
                 *
                 * According to RFC961 (Assigned Protocols),
                 * the urgent pointer points to the last octet
                 * of urgent data.  We continue, however,
                 * to consider it to indicate the first octet
                 * of data past the urgent section as the original
                 * spec states (in one of two places).
                 */

                if (SEQ_GT(ti->ti_seq+ti->ti_urp, tp->rcv_up)) {
                        tp->rcv_up = ti->ti_seq + ti->ti_urp;
                        so->so_urgc =  so->so_rcv.sb_cc +
                                (tp->rcv_up - tp->rcv_nxt); /* -1; */
                        tp->rcv_up = ti->ti_seq + ti->ti_urp;

                }
        } else
                /*
                 * If no out of band data is expected,
                 * pull receive urgent pointer along
                 * with the receive window.
                 */

                if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
                        tp->rcv_up = tp->rcv_nxt;
dodata:

        /*
         * Process the segment text, merging it into the TCP sequencing queue,
         * and arranging for acknowledgment of receipt if necessary.
         * This process logically involves adjusting tp->rcv_wnd as data
         * is presented to the user (this happens in tcp_usrreq.c,
         * case PRU_RCVD).  If a FIN has already been received on this
         * connection then we just ignore the text.
         */

        if ((ti->ti_len || (tiflags&TH_FIN)) &&
            TCPS_HAVERCVDFIN(tp->t_state) == 0) {
                TCP_REASS(tp, ti, m, so, tiflags);
                /*
                 * Note the amount of data that peer has sent into
                 * our window, in order to estimate the sender's
                 * buffer size.
                 */

                len = so->so_rcv.sb_datalen - (tp->rcv_adv - tp->rcv_nxt);
        } else {
                m_free(m);
                tiflags &= ~TH_FIN;
        }

        /*
         * If FIN is received ACK the FIN and let the user know
         * that the connection is closing.
         */

        if (tiflags & TH_FIN) {
                if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
                        /*
                         * If we receive a FIN we can't send more data,
                         * set it SS_FDRAIN
                         * Shutdown the socket if there is no rx data in the
                         * buffer.
                         * soread() is called on completion of shutdown() and
                         * will got to TCPS_LAST_ACK, and use tcp_output()
                         * to send the FIN.
                         */

                        sofwdrain(so);

                        tp->t_flags |= TF_ACKNOW;
                        tp->rcv_nxt++;
                }
                switch (tp->t_state) {

                /*
                 * In SYN_RECEIVED and ESTABLISHED STATES
                 * enter the CLOSE_WAIT state.
                 */

                case TCPS_SYN_RECEIVED:
                case TCPS_ESTABLISHED:
                  if(so->so_emu == EMU_CTL)        /* no shutdown on socket */
                    tp->t_state = TCPS_LAST_ACK;
                  else
                    tp->t_state = TCPS_CLOSE_WAIT;
                  break;

                /*
                 * If still in FIN_WAIT_1 STATE FIN has not been acked so
                 * enter the CLOSING state.
                 */

                case TCPS_FIN_WAIT_1:
                        tp->t_state = TCPS_CLOSING;
                        break;

                /*
                 * In FIN_WAIT_2 state enter the TIME_WAIT state,
                 * starting the time-wait timer, turning off the other
                 * standard timers.
                 */

                case TCPS_FIN_WAIT_2:
                        tp->t_state = TCPS_TIME_WAIT;
                        tcp_canceltimers(tp);
                        tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
                        break;

                /*
                 * In TIME_WAIT state restart the 2 MSL time_wait timer.
                 */

                case TCPS_TIME_WAIT:
                        tp->t_timer[TCPT_2MSL] = 2 * TCPTV_MSL;
                        break;
                }
        }

        /*
         * If this is a small packet, then ACK now - with Nagel
         *      congestion avoidance sender won't send more until
         *      he gets an ACK.
         *
         * See above.
         */

        if (ti->ti_len && (unsigned)ti->ti_len <= 5 &&
            ((struct tcpiphdr_2 *)ti)->first_char == (char)27) {
                tp->t_flags |= TF_ACKNOW;
        }

        /*
         * Return any desired output.
         */

        if (needoutput || (tp->t_flags & TF_ACKNOW)) {
                (void) tcp_output(tp);
        }
        return;

dropafterack:
        /*
         * Generate an ACK dropping incoming segment if it occupies
         * sequence space, where the ACK reflects our state.
         */

        if (tiflags & TH_RST)
                goto drop;
        m_freem(m);
        tp->t_flags |= TF_ACKNOW;
        (void) tcp_output(tp);
        return;

dropwithreset:
        /* reuses m if m!=NULL, m_free() unnecessary */
        if (tiflags & TH_ACK)
                tcp_respond(tp, ti, m, (tcp_seq)0, ti->ti_ack, TH_RST);
        else {
                if (tiflags & TH_SYN) ti->ti_len++;
                tcp_respond(tp, ti, m, ti->ti_seq+ti->ti_len, (tcp_seq)0,
                    TH_RST|TH_ACK);
        }

        return;

drop:
        /*
         * Drop space held by incoming segment and return.
         */

        m_free(m);

        return;
}

static void
tcp_dooptions(struct tcpcb *tp, u_char *cp, int cnt, struct tcpiphdr *ti)
{
        u_int16_t mss;
        int opt, optlen;

        DEBUG_CALL("tcp_dooptions");
        DEBUG_ARGS((dfd," tp = %lx  cnt=%i \n", (long )tp, cnt));

        for (; cnt > 0; cnt -= optlen, cp += optlen) {
                opt = cp[0];
                if (opt == TCPOPT_EOL)
                        break;
                if (opt == TCPOPT_NOP)
                        optlen = 1;
                else {
                        optlen = cp[1];
                        if (optlen <= 0)
                                break;
                }
                switch (opt) {

                default:
                        continue;

                case TCPOPT_MAXSEG:
                        if (optlen != TCPOLEN_MAXSEG)
                                continue;
                        if (!(ti->ti_flags & TH_SYN))
                                continue;
                        memcpy((char *) &mss, (char *) cp + 2, sizeof(mss));
                        NTOHS(mss);
                        (void) tcp_mss(tp, mss);        /* sets t_maxseg */
                        break;
                }
        }
}


/*
 * Pull out of band byte out of a segment so
 * it doesn't appear in the user's data queue.
 * It is still reflected in the segment length for
 * sequencing purposes.
 */


#ifdef notdef

void
tcp_pulloutofband(so, ti, m)
        struct socket *so;
        struct tcpiphdr *ti;
        register struct mbuf *m;
{
        int cnt = ti->ti_urp - 1;

        while (cnt >= 0) {
                if (m->m_len > cnt) {
                        char *cp = mtod(m, caddr_t) + cnt;
                        struct tcpcb *tp = sototcpcb(so);

                        tp->t_iobc = *cp;
                        tp->t_oobflags |= TCPOOB_HAVEDATA;
                        memcpy(sp, cp+1, (unsigned)(m->m_len - cnt - 1));
                        m->m_len--;
                        return;
                }
                cnt -= m->m_len;
                m = m->m_next; /* XXX WRONG! Fix it! */
                if (m == 0)
                        break;
        }
        panic("tcp_pulloutofband");
}

#endif /* notdef */

/*
 * Collect new round-trip time estimate
 * and update averages and current timeout.
 */


static void
tcp_xmit_timer(register struct tcpcb *tp, int rtt)
{
        register short delta;

        DEBUG_CALL("tcp_xmit_timer");
        DEBUG_ARG("tp = %lx", (long)tp);
        DEBUG_ARG("rtt = %d", rtt);

        if (tp->t_srtt != 0) {
                /*
                 * srtt is stored as fixed point with 3 bits after the
                 * binary point (i.e., scaled by 8).  The following magic
                 * is equivalent to the smoothing algorithm in rfc793 with
                 * an alpha of .875 (srtt = rtt/8 + srtt*7/8 in fixed
                 * point).  Adjust rtt to origin 0.
                 */

                delta = rtt - 1 - (tp->t_srtt >> TCP_RTT_SHIFT);
                if ((tp->t_srtt += delta) <= 0)
                        tp->t_srtt = 1;
                /*
                 * We accumulate a smoothed rtt variance (actually, a
                 * smoothed mean difference), then set the retransmit
                 * timer to smoothed rtt + 4 times the smoothed variance.
                 * rttvar is stored as fixed point with 2 bits after the
                 * binary point (scaled by 4).  The following is
                 * equivalent to rfc793 smoothing with an alpha of .75
                 * (rttvar = rttvar*3/4 + |delta| / 4).  This replaces
                 * rfc793's wired-in beta.
                 */

                if (delta < 0)
                        delta = -delta;
                delta -= (tp->t_rttvar >> TCP_RTTVAR_SHIFT);
                if ((tp->t_rttvar += delta) <= 0)
                        tp->t_rttvar = 1;
        } else {
                /*
                 * No rtt measurement yet - use the unsmoothed rtt.
                 * Set the variance to half the rtt (so our first
                 * retransmit happens at 3*rtt).
                 */

                tp->t_srtt = rtt << TCP_RTT_SHIFT;
                tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT - 1);
        }
        tp->t_rtt = 0;
        tp->t_rxtshift = 0;

        /*
         * the retransmit 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).
         */

        TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp),
            (short)tp->t_rttmin, TCPTV_REXMTMAX); /* XXX */

        /*
         * We received an ack for a packet that wasn't retransmitted;
         * it is probably safe to discard any error indications we've
         * received recently.  This isn't quite right, but close enough
         * for now (a route might have failed after we sent a segment,
         * and the return path might not be symmetrical).
         */

        tp->t_softerror = 0;
}

/*
 * Determine a reasonable value for maxseg size.
 * If the route is known, check route for mtu.
 * If none, use an mss that can be handled on the outgoing
 * interface without forcing IP to fragment; if bigger than
 * an mbuf cluster (MCLBYTES), round down to nearest multiple of MCLBYTES
 * to utilize large mbufs.  If no route is found, route has no mtu,
 * or the destination isn't local, use a default, hopefully conservative
 * size (usually 512 or the default IP max size, but no more than the mtu
 * of the interface), as we can't discover anything about intervening
 * gateways or networks.  We also initialize the congestion/slow start
 * window to be a single segment if the destination isn't local.
 * While looking at the routing entry, we also initialize other path-dependent
 * parameters from pre-set or cached values in the routing entry.
 */


int
tcp_mss(struct tcpcb *tp, u_int offer)
{
        struct socket *so = tp->t_socket;
        int mss;

        DEBUG_CALL("tcp_mss");
        DEBUG_ARG("tp = %lx", (long)tp);
        DEBUG_ARG("offer = %d", offer);

        mss = min(IF_MTU, IF_MRU) - sizeof(struct tcpiphdr);
        if (offer)
                mss = min(mss, offer);
        mss = max(mss, 32);
        if (mss < tp->t_maxseg || offer != 0)
           tp->t_maxseg = mss;

        tp->snd_cwnd = mss;

        sbreserve(&so->so_snd, TCP_SNDSPACE + ((TCP_SNDSPACE % mss) ?
                                               (mss - (TCP_SNDSPACE % mss)) :
                                               0));
        sbreserve(&so->so_rcv, TCP_RCVSPACE + ((TCP_RCVSPACE % mss) ?
                                               (mss - (TCP_RCVSPACE % mss)) :
                                               0));

        DEBUG_MISC((dfd, " returning mss = %d\n", mss));

        return mss;
}