tcp_input.c

qemu虚拟机代码

	/* XXX Should never fail */
	if (tp == 0)
		goto dropwithreset;
	if (tp->t_state == TCPS_CLOSED)
		goto drop;
	
	/* Unscale the window into a 32-bit value. */
/*	if ((tiflags & TH_SYN) == 0)
 *		tiwin = ti->ti_win << tp->snd_scale;
 *	else
 */
		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] = tcp_keepintvl;
	else
	   tp->t_timer[TCPT_KEEP] = tcp_keepidle;

	/*
	 * 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); 
/* , */
/*			&ts_present, &ts_val, &ts_ecr); */

	/* 
	 * 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 &&
/*	    (!ts_present || TSTMP_GEQ(ts_val, tp->ts_recent)) && */
	    ti->ti_seq == tp->rcv_nxt &&
	    tiwin && tiwin == tp->snd_wnd &&
	    tp->snd_nxt == tp->snd_max) {
		/* 
		 * If last ACK falls within this segment's sequence numbers,
		 *  record the timestamp.
		 */
/*		if (ts_present && SEQ_LEQ(ti->ti_seq, tp->last_ack_sent) &&
 *		   SEQ_LT(tp->last_ack_sent, ti->ti_seq + ti->ti_len)) {
 *			tp->ts_recent_age = tcp_now;
 *			tp->ts_recent = ts_val;
 *		}
 */
		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.
				 */
				++tcpstat.tcps_predack;
/*				if (ts_present)
 *					tcp_xmit_timer(tp, tcp_now-ts_ecr+1);
 *				else 
 */				     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;
				tcpstat.tcps_rcvackpack++;
				tcpstat.tcps_rcvackbyte += acked;
				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;

				/* 
				 * There's room in so_snd, sowwakup will read()
				 * from the socket if we can
				 */
/*				if (so->so_snd.sb_flags & SB_NOTIFY)
 *					sowwakeup(so);
 */
				/* 
				 * 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 &&
		    tp->seg_next == (tcpiphdrp_32)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.
			 */
			++tcpstat.tcps_preddat;
			tp->rcv_nxt += ti->ti_len;
			tcpstat.tcps_rcvpack++;
			tcpstat.tcps_rcvbyte += ti->ti_len;
			/*
			 * Add data to socket buffer.
			 */
			if (so->so_emu) {
				if (tcp_emu(so,m)) sbappend(so, m);
			} else
				sbappend(so, m);
			
			/* 
			 * XXX This is called when data arrives.  Later, check
			 * if we can actually write() to the socket
			 * XXX Need to check? It's be NON_BLOCKING
			 */
/*			sorwakeup(so); */
			
			/*
			 * 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&htonl(0xffffff00)) == special_addr.s_addr) {
	    int lastbyte=ntohl(so->so_faddr.s_addr) & 0xff;
	    if (lastbyte!=CTL_ALIAS && lastbyte!=CTL_DNS) {
#if 0
	      if(lastbyte==CTL_CMD || lastbyte==CTL_EXEC) {
		/* Command or exec adress */
		so->so_state |= SS_CTL;
	      } else 
#endif
              {
		/* May be an add exec */
		struct ex_list *ex_ptr;
		for(ex_ptr = exec_list; ex_ptr; ex_ptr = ex_ptr->ex_next) {
		  if(ex_ptr->ex_fport == so->so_fport && 
		     lastbyte == ex_ptr->ex_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);
	  /* , */
	  /*				&ts_present, &ts_val, &ts_ecr); */
	  
	  if (iss)
	    tp->iss = iss;
	  else 
	    tp->iss = tcp_iss;
	  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;
	  tcpstat.tcps_accepts++;
	  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)) {
			tcpstat.tcps_connects++;
			soisfconnected(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 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;
			tcpstat.tcps_rcvpackafterwin++;
			tcpstat.tcps_rcvbyteafterwin += todrop;
		}
		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.
	 * 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 += ti->ti_len;
 *			tcpstat.tcps_pawsdrop++;
 *			goto dropafterack;
 *		}
 *	}
 */

	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.