tcp_input.c

嵌入式操作系统ECOS的网络开发包

			 * we have enough buffer space to take it.
			 */
#ifdef TCP_SACK
			/* Clean receiver SACK report if present */
			if (!tp->sack_disable && tp->rcv_numsacks)
				tcp_clean_sackreport(tp);
#endif /* TCP_SACK */
			++tcpstat.tcps_preddat;
			tp->rcv_nxt += tlen;
			tcpstat.tcps_rcvpack++;
			tcpstat.tcps_rcvbyte += tlen;
			ND6_HINT(tp);
			/*
			 * Drop TCP, IP headers and TCP options then add data
			 * to socket buffer.
			 */
			m_adj(m, iphlen + off);
			sbappend(&so->so_rcv, m);
			sorwakeup(so);
			if (th->th_flags & TH_PUSH)
				tp->t_flags |= TF_ACKNOW;
			else
				tp->t_flags |= TF_DELACK;
			return;
		}
	}

	/*
	 * Compute mbuf offset to TCP data segment.
	 */
	hdroptlen = iphlen + off;

	/*
	 * 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 = imax(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.
	 * If it is from this socket, drop it, it must be forged.
	 * 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: {
		struct mbuf *am;
		register struct sockaddr_in *sin;
#ifdef INET6
		register struct sockaddr_in6 *sin6;
#endif /* INET6 */

		if (tiflags & TH_RST)
			goto drop;
		if (tiflags & TH_ACK)
			goto dropwithreset;
		if ((tiflags & TH_SYN) == 0)
			goto drop;
		if (th->th_dport == th->th_sport) {
#ifdef INET6
		  if (is_ipv6) {
		    if (IN6_ARE_ADDR_EQUAL(&ipv6->ip6_src, &ipv6->ip6_dst))
		      goto drop;
		  } else {
#endif /* INET6 */
		    if (ti->ti_dst.s_addr == ti->ti_src.s_addr)
		      goto drop;
#ifdef INET6
		  }
#endif /* INET6 */
		}

		/*
		 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
		 * in_broadcast() should never return true on a received
		 * packet with M_BCAST not set.
		 */
		if (m->m_flags & (M_BCAST|M_MCAST))
		  goto drop;
#ifdef INET6
		if (is_ipv6) {
			/* XXX What about IPv6 Anycasting ?? :-(  rja */
			if (IN6_IS_ADDR_MULTICAST(&ipv6->ip6_dst))
				goto drop;
		} else
#endif /* INET6 */
		if (IN_MULTICAST(ti->ti_dst.s_addr))
			goto drop;
		am = m_get(M_DONTWAIT, MT_SONAME);	/* XXX */
		if (am == NULL)
			goto drop;
#ifdef INET6
		if (is_ipv6) {
		  /*
		   * This is probably the place to set the tp->pf value.
		   * (Don't forget to do it in the v4 code as well!)
		   *
		   * Also, remember to blank out things like flowlabel, or
		   * set flowlabel for accepted sockets in v6.
		   *
		   * FURTHERMORE, this is PROBABLY the place where the whole
		   * business of key munging is set up for passive
		   * connections.
		   */
		  am->m_len = sizeof(struct sockaddr_in6);
		  sin6 = mtod(am, struct sockaddr_in6 *);
		  sin6->sin6_family = AF_INET6;
		  sin6->sin6_len = sizeof(struct sockaddr_in6);
		  sin6->sin6_addr = ipv6->ip6_src;
		  sin6->sin6_port = th->th_sport;
		  sin6->sin6_flowinfo = htonl(0x0fffffff) &
		    inp->inp_ipv6.ip6_flow;
		  laddr6 = inp->inp_laddr6;
		  if (IN6_IS_ADDR_UNSPECIFIED(&inp->inp_laddr6))
		    inp->inp_laddr6 = ipv6->ip6_dst;
		  /* This is a good optimization. */
		  if (in6_pcbconnect(inp, am)) {
		    inp->inp_laddr6 = laddr6;
		    (void) m_free(am);
		    goto drop;
		  } /* endif in6_pcbconnect() */
		  tp->pf = PF_INET6;
		} else {
		  /*
		   * Letting v4 incoming datagrams to reach valid 
		   * PF_INET6 sockets causes some overhead here.
		   */
		  if (inp->inp_flags & INP_IPV6) {
		    if (!(inp->inp_flags & (INP_IPV6_UNDEC|INP_IPV6_MAPPED))) {
		      (void) m_free(am);
		      goto drop;
		    }

		    am->m_len = sizeof(struct sockaddr_in6);
		    
		    sin6 = mtod(am, struct sockaddr_in6 *);
		    sin6->sin6_family = AF_INET6;
		    sin6->sin6_len = sizeof(*sin6);
		    CREATE_IPV6_MAPPED(sin6->sin6_addr, ti->ti_src.s_addr);
		    sin6->sin6_port = th->th_sport;
		    sin6->sin6_flowinfo = 0;

		    laddr6 = inp->inp_laddr6;
		    if (inp->inp_laddr.s_addr == INADDR_ANY)
		      CREATE_IPV6_MAPPED(inp->inp_laddr6, ti->ti_dst.s_addr);
		    
		    /*
		     * The pcb initially has the v6 default hoplimit
		     * set. We're sending v4 packets so we need to set
		     * the v4 ttl and tos.
		     */
		    inp->inp_ip.ip_ttl = ip_defttl;
		    inp->inp_ip.ip_tos = 0;
		    
		    if (in6_pcbconnect(inp, am)) {
		      inp->inp_laddr6 = laddr6;
		      (void) m_freem(am);
		      goto drop;
		    }
		    tp->pf = PF_INET;
		  } else { 
#endif /* INET6 */
		am->m_len = sizeof (struct sockaddr_in);
		sin = mtod(am, struct sockaddr_in *);
		sin->sin_family = AF_INET;
		sin->sin_len = sizeof(*sin);
		sin->sin_addr = ti->ti_src;
		sin->sin_port = ti->ti_sport;
		bzero((caddr_t)sin->sin_zero, sizeof(sin->sin_zero));
		laddr = inp->inp_laddr;
		if (inp->inp_laddr.s_addr == INADDR_ANY)
			inp->inp_laddr = ti->ti_dst;
		if (in_pcbconnect(inp, am)) {
			inp->inp_laddr = laddr;
			(void) m_free(am);
			goto drop;
		}
		(void) m_free(am);
		tp->pf = PF_INET;
#ifdef INET6
		  }  /* if (inp->inp_flags & INP_IPV6) */
		} /* if (is_ipv6) */
#endif /* INET6 */
		tp->t_template = tcp_template(tp);
		if (tp->t_template == 0) {
			tp = tcp_drop(tp, ENOBUFS);
			dropsocket = 0;		/* socket is already gone */
			goto drop;
		}
		if (optp)
			tcp_dooptions(tp, optp, optlen, th,
				&ts_present, &ts_val, &ts_ecr);
#ifdef TCP_SACK
		/*
		 * If peer did not send a SACK_PERMITTED option (i.e., if
		 * tcp_dooptions() did not set TF_SACK_PERMIT), set 
                 * sack_disable to 1 if it is currently 0.
                 */
                if (!tp->sack_disable)
                        if ((tp->t_flags & TF_SACK_PERMIT) == 0) 
                                tp->sack_disable = 1;
#endif

		if (iss)
			tp->iss = iss;
		else
			tp->iss = tcp_iss;
#ifdef TCP_COMPAT_42
		tcp_iss += TCP_ISSINCR/2;
#else /* TCP_COMPAT_42 */
		tcp_iss += arc4random() % TCP_ISSINCR + 1;
#endif /* !TCP_COMPAT_42 */
		tp->irs = th->th_seq;
		tcp_sendseqinit(tp);
#if defined (TCP_SACK) || defined (TCP_NEWRENO)
		tp->snd_last = tp->snd_una;
#endif /* TCP_SACK || TCP_NEWRENO */
#if defined(TCP_SACK) && defined(TCP_FACK)
		tp->snd_fack = tp->snd_una;
		tp->retran_data = 0;
		tp->snd_awnd = 0;
#endif /* TCP_FACK */
		tcp_rcvseqinit(tp);
		tp->t_flags |= TF_ACKNOW;
		tp->t_state = TCPS_SYN_RECEIVED;
		tp->t_timer[TCPT_KEEP] = tcptv_keep_init;
		dropsocket = 0;		/* committed to socket */
		tcpstat.tcps_accepts++;
		goto trimthenstep6;
		}

	/*
	 * If the state is SYN_RECEIVED:
	 * 	if seg contains SYN/ACK, send an RST.
	 *	if seg contains an ACK, but not for our SYN/ACK, send an RST
  	 */

	case TCPS_SYN_RECEIVED:
		if (tiflags & TH_ACK) {
			if (tiflags & TH_SYN) {
				tcpstat.tcps_badsyn++;
				goto dropwithreset;
			}
			if (SEQ_LEQ(th->th_ack, tp->snd_una) ||
			    SEQ_GT(th->th_ack, tp->snd_max))
				goto dropwithreset;
		}
		break;

	/*
	 * 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(th->th_ack, tp->iss) ||
		     SEQ_GT(th->th_ack, tp->snd_max)))
			goto dropwithreset;
		if (tiflags & TH_RST) {
			if (tiflags & TH_ACK)
				tp = tcp_drop(tp, ECONNREFUSED);
			goto drop;
		}
		if ((tiflags & TH_SYN) == 0)
			goto drop;
		if (tiflags & TH_ACK) {
			tp->snd_una = th->th_ack;
			if (SEQ_LT(tp->snd_nxt, tp->snd_una))
				tp->snd_nxt = tp->snd_una;
		}
		tp->t_timer[TCPT_REXMT] = 0;
		tp->irs = th->th_seq;
		tcp_rcvseqinit(tp);
		tp->t_flags |= TF_ACKNOW;
#ifdef TCP_SACK
                /*
                 * If we've sent a SACK_PERMITTED option, and the peer
                 * also replied with one, then TF_SACK_PERMIT should have
                 * been set in tcp_dooptions().  If it was not, disable SACKs.
                 */
                if (!tp->sack_disable)
                        if ((tp->t_flags & TF_SACK_PERMIT) == 0) 
                                tp->sack_disable = 1;
#endif
		if (tiflags & TH_ACK && SEQ_GT(tp->snd_una, tp->iss)) {
			tcpstat.tcps_connects++;
			soisconnected(so);
			tp->t_state = TCPS_ESTABLISHED;
			/* Do window scaling on this connection? */
			if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
				(TF_RCVD_SCALE|TF_REQ_SCALE)) {
				tp->snd_scale = tp->requested_s_scale;
				tp->rcv_scale = tp->request_r_scale;
			}
			(void) tcp_reass(tp, (struct tcphdr *)0,
				(struct mbuf *)0, &tlen);
			/*
			 * 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);
			/*
			 * Since new data was acked (the SYN), open the
			 * congestion window by one MSS.  We do this
			 * here, because we won't go through the normal
			 * ACK processing below.  And since this is the
			 * start of the connection, we know we are in
			 * the exponential phase of slow-start.
			 */
			tp->snd_cwnd += tp->t_maxseg;
		} 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.
		 */
		th->th_seq++;
		if (tlen > tp->rcv_wnd) {
			todrop = tlen - tp->rcv_wnd;
			m_adj(m, -todrop);
			tlen = tp->rcv_wnd;
			tiflags &= ~TH_FIN;
			tcpstat.tcps_rcvpackafterwin++;
			tcpstat.tcps_rcvbyteafterwin += todrop;
		}
		tp->snd_wl1 = th->th_seq - 1;
		tp->rcv_up = th->th_seq;
		goto step6;
	}

	/*
	 * States other than LISTEN or SYN_SENT.
	 * First check timestamp, if present.
	 * Then 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.
	 * 
	 * RFC 1323 PAWS: If we have a timestamp reply on this segment
	 * and it's less than ts_recent, drop it.
	 */
	if (ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent &&
	    TSTMP_LT(ts_val, tp->ts_recent)) {

		/* Check to see if ts_recent is over 24 days old.  */
		if ((int)(tcp_now - tp->ts_recent_age) > TCP_PAWS_IDLE) {
			/*
			 * Invalidate ts_recent.  If this segment updates
			 * ts_recent, the age will be reset later and ts_recent
			 * will get a valid value.  If it does not, setting
			 * ts_recent to zero will at least satisfy the
			 * requirement that zero be placed in the timestamp
			 * echo reply when ts_recent isn't valid.  The
			 * age isn't reset until we get a valid ts_recent
			 * because we don't want out-of-order segments to be
			 * dropped when ts_recent is old.
			 */
			tp->ts_recent = 0;
		} else {
			tcpstat.tcps_rcvduppack++;
			tcpstat.tcps_rcvdupbyte += tlen;
			tcpstat.tcps_pawsdrop++;
			goto dropafterack;
		}
	}

	todrop = tp->rcv_nxt - th->th_seq;
	if (todrop > 0) {
		if (tiflags & TH_SYN) {
			tiflags &= ~TH_SYN;
			th->th_seq++;
			if (th->th_urp > 1) 
				th->th_urp--;
			else
				tiflags &= ~TH_URG;
			todrop--;
		}
		if (todrop >= tlen ||
		    (todrop == tlen && (tiflags & TH_FIN) == 0)) {
			/*
			 * Any valid FIN must be to the left of the
			 * window.  At this point, FIN must be a
			 * duplicate or out-of-sequence, so drop it.
			 */
			tiflags &= ~TH_FIN;
			/*
			 * Send ACK to resynchronize, and drop any data,
			 * but keep on processing for RST or ACK.
			 */
			tp->t_flags |= TF_ACKNOW;
			tcpstat.tcps_rcvdupbyte += todrop = tlen;
			tcpstat.tcps_rcvduppack++;
		} else {
			tcpstat.tcps_rcvpartduppack++;
			tcpstat.tcps_rcvpartdupbyte += todrop;
		}
		hdroptlen += todrop;	/* drop from head afterwards */
		th->th_seq += todrop;
		tlen -= todrop;
		if (th->th_urp > todrop)
			th->th_urp -= todrop;
		else {
			tiflags &= ~TH_URG;
			th->th_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 && tlen) {
		tp = tcp_close(tp);
		tcpstat.tcps_rcvafterclose++;
		goto dropwithreset;
	}

	/*
	 * If segment ends after window, drop trailing data
	 * (and PUSH and FIN); if nothing left, just ACK.
	 */
	todrop = (th->th_seq + tlen) - (tp->rcv_nxt+tp->rcv_wnd);
	if (todrop > 0) {
		tcpstat.tcps_rcvpackafterwin++;
		if (todrop >= tlen) {
			tcpstat.tcps_rcvbyteafterwin += tlen;
			/*
			 * If a new connection request is received
			 * while in TIME_WAIT, drop the old connection
			 * and start over if the sequence numbers