tcp_input.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

			m -= (tp->mdev >> 2);   /* similar update on mdev */
			/* This is similar to one of Eifel findings.
			 * Eifel blocks mdev updates when rtt decreases.
			 * This solution is a bit different: we use finer gain
			 * for mdev in this case (alpha*beta).
			 * Like Eifel it also prevents growth of rto,
			 * but also it limits too fast rto decreases,
			 * happening in pure Eifel.
			 */
			if (m > 0)
				m >>= 3;
		} else {
			m -= (tp->mdev >> 2);   /* similar update on mdev */
		}
		tp->mdev += m;	    	/* mdev = 3/4 mdev + 1/4 new */
		if (tp->mdev > tp->mdev_max) {
			tp->mdev_max = tp->mdev;
			if (tp->mdev_max > tp->rttvar)
				tp->rttvar = tp->mdev_max;
		}
		if (after(tp->snd_una, tp->rtt_seq)) {
			if (tp->mdev_max < tp->rttvar)
				tp->rttvar -= (tp->rttvar-tp->mdev_max)>>2;
			tp->rtt_seq = tp->snd_una;
			tp->mdev_max = TCP_RTO_MIN;
		}
	} else {
		/* no previous measure. */
		tp->srtt = m<<3;	/* take the measured time to be rtt */
		tp->mdev = m<<2;	/* make sure rto = 3*rtt */
		tp->mdev_max = tp->rttvar = max(tp->mdev, TCP_RTO_MIN);
		tp->rtt_seq = tp->snd_nxt;
	}
}

/* Calculate rto without backoff.  This is the second half of Van Jacobson's
 * routine referred to above.
 */
static __inline__ void tcp_set_rto(struct tcp_opt *tp)
{
	/* Old crap is replaced with new one. 8)
	 *
	 * More seriously:
	 * 1. If rtt variance happened to be less 50msec, it is hallucination.
	 *    It cannot be less due to utterly erratic ACK generation made
	 *    at least by solaris and freebsd. "Erratic ACKs" has _nothing_
	 *    to do with delayed acks, because at cwnd>2 true delack timeout
	 *    is invisible. Actually, Linux-2.4 also generates erratic
	 *    ACKs in some curcumstances.
	 */
	tp->rto = (tp->srtt >> 3) + tp->rttvar;

	/* 2. Fixups made earlier cannot be right.
	 *    If we do not estimate RTO correctly without them,
	 *    all the algo is pure shit and should be replaced
	 *    with correct one. It is exaclty, which we pretend to do.
	 */
}

/* NOTE: clamping at TCP_RTO_MIN is not required, current algo
 * guarantees that rto is higher.
 */
static __inline__ void tcp_bound_rto(struct tcp_opt *tp)
{
	if (tp->rto > TCP_RTO_MAX)
		tp->rto = TCP_RTO_MAX;
}

/* Save metrics learned by this TCP session.
   This function is called only, when TCP finishes sucessfully
   i.e. when it enters TIME-WAIT or goes from LAST-ACK to CLOSE.
 */
void tcp_update_metrics(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct dst_entry *dst = __sk_dst_get(sk);

	dst_confirm(dst);

	if (dst && (dst->flags&DST_HOST)) {
		int m;

		if (tp->backoff || !tp->srtt) {
			/* This session failed to estimate rtt. Why?
			 * Probably, no packets returned in time.
			 * Reset our results.
			 */
			if (!(dst->mxlock&(1<<RTAX_RTT)))
				dst->rtt = 0;
			return;
		}

		m = dst->rtt - tp->srtt;

		/* If newly calculated rtt larger than stored one,
		 * store new one. Otherwise, use EWMA. Remember,
		 * rtt overestimation is always better than underestimation.
		 */
		if (!(dst->mxlock&(1<<RTAX_RTT))) {
			if (m <= 0)
				dst->rtt = tp->srtt;
			else
				dst->rtt -= (m>>3);
		}

		if (!(dst->mxlock&(1<<RTAX_RTTVAR))) {
			if (m < 0)
				m = -m;

			/* Scale deviation to rttvar fixed point */
			m >>= 1;
			if (m < tp->mdev)
				m = tp->mdev;

			if (m >= dst->rttvar)
				dst->rttvar = m;
			else
				dst->rttvar -= (dst->rttvar - m)>>2;
		}

		if (tp->snd_ssthresh >= 0xFFFF) {
			/* Slow start still did not finish. */
			if (dst->ssthresh &&
			    !(dst->mxlock&(1<<RTAX_SSTHRESH)) &&
			    (tp->snd_cwnd>>1) > dst->ssthresh)
				dst->ssthresh = (tp->snd_cwnd>>1);
			if (!(dst->mxlock&(1<<RTAX_CWND)) &&
			    tp->snd_cwnd > dst->cwnd)
				dst->cwnd = tp->snd_cwnd;
		} else if (tp->snd_cwnd > tp->snd_ssthresh &&
			   tp->ca_state == TCP_CA_Open) {
			/* Cong. avoidance phase, cwnd is reliable. */
			if (!(dst->mxlock&(1<<RTAX_SSTHRESH)))
				dst->ssthresh = max(tp->snd_cwnd>>1, tp->snd_ssthresh);
			if (!(dst->mxlock&(1<<RTAX_CWND)))
				dst->cwnd = (dst->cwnd + tp->snd_cwnd)>>1;
		} else {
			/* Else slow start did not finish, cwnd is non-sense,
			   ssthresh may be also invalid.
			 */
			if (!(dst->mxlock&(1<<RTAX_CWND)))
				dst->cwnd = (dst->cwnd + tp->snd_ssthresh)>>1;
			if (dst->ssthresh &&
			    !(dst->mxlock&(1<<RTAX_SSTHRESH)) &&
			    tp->snd_ssthresh > dst->ssthresh)
				dst->ssthresh = tp->snd_ssthresh;
		}

		if (!(dst->mxlock&(1<<RTAX_REORDERING))) {
			if (dst->reordering < tp->reordering &&
			    tp->reordering != sysctl_tcp_reordering)
				dst->reordering = tp->reordering;
		}
	}
}

/* Increase initial CWND conservatively: if estimated
 * RTT is low enough (<20msec) or if we have some preset ssthresh.
 *
 * Numbers are taken from RFC1414.
 */
__u32 tcp_init_cwnd(struct tcp_opt *tp)
{
	__u32 cwnd;

	if (tp->mss_cache > 1460)
		return 2;

	cwnd = (tp->mss_cache > 1095) ? 3 : 4;

	if (!tp->srtt || (tp->snd_ssthresh >= 0xFFFF && tp->srtt > ((HZ/50)<<3)))
		cwnd = 2;
	else if (cwnd > tp->snd_ssthresh)
		cwnd = tp->snd_ssthresh;

	return min(cwnd, tp->snd_cwnd_clamp);
}

/* Initialize metrics on socket. */

static void tcp_init_metrics(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct dst_entry *dst = __sk_dst_get(sk);

	if (dst == NULL)
		goto reset;

	dst_confirm(dst);

	if (dst->mxlock&(1<<RTAX_CWND))
		tp->snd_cwnd_clamp = dst->cwnd;
	if (dst->ssthresh) {
		tp->snd_ssthresh = dst->ssthresh;
		if (tp->snd_ssthresh > tp->snd_cwnd_clamp)
			tp->snd_ssthresh = tp->snd_cwnd_clamp;
	}
	if (dst->reordering && tp->reordering != dst->reordering) {
		tp->sack_ok &= ~2;
		tp->reordering = dst->reordering;
	}

	if (dst->rtt == 0)
		goto reset;

	if (!tp->srtt && dst->rtt < (TCP_TIMEOUT_INIT<<3))
		goto reset;

	/* Initial rtt is determined from SYN,SYN-ACK.
	 * The segment is small and rtt may appear much
	 * less than real one. Use per-dst memory
	 * to make it more realistic.
	 *
	 * A bit of theory. RTT is time passed after "normal" sized packet
	 * is sent until it is ACKed. In normal curcumstances sending small
	 * packets force peer to delay ACKs and calculation is correct too.
	 * The algorithm is adaptive and, provided we follow specs, it
	 * NEVER underestimate RTT. BUT! If peer tries to make some clever
	 * tricks sort of "quick acks" for time long enough to decrease RTT
	 * to low value, and then abruptly stops to do it and starts to delay
	 * ACKs, wait for troubles.
	 */
	if (dst->rtt > tp->srtt)
		tp->srtt = dst->rtt;
	if (dst->rttvar > tp->mdev) {
		tp->mdev = dst->rttvar;
		tp->mdev_max = tp->rttvar = max(tp->mdev, TCP_RTO_MIN);
	}
	tcp_set_rto(tp);
	tcp_bound_rto(tp);
	if (tp->rto < TCP_TIMEOUT_INIT && !tp->saw_tstamp)
		goto reset;
	tp->snd_cwnd = tcp_init_cwnd(tp);
	tp->snd_cwnd_stamp = tcp_time_stamp;
	return;

reset:
	/* Play conservative. If timestamps are not
	 * supported, TCP will fail to recalculate correct
	 * rtt, if initial rto is too small. FORGET ALL AND RESET!
	 */
	if (!tp->saw_tstamp && tp->srtt) {
		tp->srtt = 0;
		tp->mdev = tp->mdev_max = tp->rttvar = TCP_TIMEOUT_INIT;
		tp->rto = TCP_TIMEOUT_INIT;
	}
}

static void tcp_update_reordering(struct tcp_opt *tp, int metric, int ts)
{
	if (metric > tp->reordering) {
		tp->reordering = min(TCP_MAX_REORDERING, metric);

		/* This exciting event is worth to be remembered. 8) */
		if (ts)
			NET_INC_STATS_BH(TCPTSReorder);
		else if (IsReno(tp))
			NET_INC_STATS_BH(TCPRenoReorder);
		else if (IsFack(tp))
			NET_INC_STATS_BH(TCPFACKReorder);
		else
			NET_INC_STATS_BH(TCPSACKReorder);
#if FASTRETRANS_DEBUG > 1
		printk(KERN_DEBUG "Disorder%d %d %u f%u s%u rr%d\n",
		       tp->sack_ok, tp->ca_state,
		       tp->reordering, tp->fackets_out, tp->sacked_out,
		       tp->undo_marker ? tp->undo_retrans : 0);
#endif
		/* Disable FACK yet. */
		tp->sack_ok &= ~2;
	}
}

/* This procedure tags the retransmission queue when SACKs arrive.
 *
 * We have three tag bits: SACKED(S), RETRANS(R) and LOST(L).
 * Packets in queue with these bits set are counted in variables
 * sacked_out, retrans_out and lost_out, correspondingly.
 *
 * Valid combinations are:
 * Tag  InFlight	Description
 * 0	1		- orig segment is in flight.
 * S	0		- nothing flies, orig reached receiver.
 * L	0		- nothing flies, orig lost by net.
 * R	2		- both orig and retransmit are in flight.
 * L|R	1		- orig is lost, retransmit is in flight.
 * S|R  1		- orig reached receiver, retrans is still in flight.
 * (L|S|R is logically valid, it could occur when L|R is sacked,
 *  but it is equivalent to plain S and code short-curcuits it to S.
 *  L|S is logically invalid, it would mean -1 packet in flight 8))
 *
 * These 6 states form finite state machine, controlled by the following events:
 * 1. New ACK (+SACK) arrives. (tcp_sacktag_write_queue())
 * 2. Retransmission. (tcp_retransmit_skb(), tcp_xmit_retransmit_queue())
 * 3. Loss detection event of one of three flavors:
 *	A. Scoreboard estimator decided the packet is lost.
 *	   A'. Reno "three dupacks" marks head of queue lost.
 *	   A''. Its FACK modfication, head until snd.fack is lost.
 *	B. SACK arrives sacking data transmitted after never retransmitted
 *	   hole was sent out.
 *	C. SACK arrives sacking SND.NXT at the moment, when the
 *	   segment was retransmitted.
 * 4. D-SACK added new rule: D-SACK changes any tag to S.
 *
 * It is pleasant to note, that state diagram turns out to be commutative,
 * so that we are allowed not to be bothered by order of our actions,
 * when multiple events arrive simultaneously. (see the function below).
 *
 * Reordering detection.
 * --------------------
 * Reordering metric is maximal distance, which a packet can be displaced
 * in packet stream. With SACKs we can estimate it:
 *
 * 1. SACK fills old hole and the corresponding segment was not
 *    ever retransmitted -> reordering. Alas, we cannot use it
 *    when segment was retransmitted.
 * 2. The last flaw is solved with D-SACK. D-SACK arrives
 *    for retransmitted and already SACKed segment -> reordering..
 * Both of these heuristics are not used in Loss state, when we cannot
 * account for retransmits accurately.
 */
static int
tcp_sacktag_write_queue(struct sock *sk, struct sk_buff *ack_skb, u32 prior_snd_una)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	unsigned char *ptr = ack_skb->h.raw + TCP_SKB_CB(ack_skb)->sacked;
	struct tcp_sack_block *sp = (struct tcp_sack_block *)(ptr+2);
	int num_sacks = (ptr[1] - TCPOLEN_SACK_BASE)>>3;
	int reord = tp->packets_out;
	int prior_fackets;
	u32 lost_retrans = 0;
	int flag = 0;
	int i;

	if (!tp->sacked_out)
		tp->fackets_out = 0;
	prior_fackets = tp->fackets_out;

	for (i=0; i<num_sacks; i++, sp++) {
		struct sk_buff *skb;
		__u32 start_seq = ntohl(sp->start_seq);
		__u32 end_seq = ntohl(sp->end_seq);
		int fack_count = 0;
		int dup_sack = 0;

		/* Check for D-SACK. */
		if (i == 0) {
			u32 ack = TCP_SKB_CB(ack_skb)->ack_seq;

			if (before(start_seq, ack)) {
				dup_sack = 1;
				tp->sack_ok |= 4;
				NET_INC_STATS_BH(TCPDSACKRecv);
			} else if (num_sacks > 1 &&
				   !after(end_seq, ntohl(sp[1].end_seq)) &&
				   !before(start_seq, ntohl(sp[1].start_seq))) {
				dup_sack = 1;
				tp->sack_ok |= 4;
				NET_INC_STATS_BH(TCPDSACKOfoRecv);
			}

			/* D-SACK for already forgotten data...
			 * Do dumb counting. */
			if (dup_sack &&
			    !after(end_seq, prior_snd_una) &&
			    after(end_seq, tp->undo_marker))
				tp->undo_retrans--;

			/* Eliminate too old ACKs, but take into
			 * account more or less fresh ones, they can
			 * contain valid SACK info.
			 */
			if (before(ack, prior_snd_una-tp->max_window))
				return 0;
		}

		/* Event "B" in the comment above. */
		if (after(end_seq, tp->high_seq))
			flag |= FLAG_DATA_LOST;

		for_retrans_queue(skb, sk, tp) {
			u8 sacked = TCP_SKB_CB(skb)->sacked;
			int in_sack;

			/* The retransmission queue is always in order, so
			 * we can short-circuit the walk early.
			 */
			if(!before(TCP_SKB_CB(skb)->seq, end_seq))
				break;

			fack_count++;

			in_sack = !after(start_seq, TCP_SKB_CB(skb)->seq) &&
				!before(end_seq, TCP_SKB_CB(skb)->end_seq);

			/* Account D-SACK for retransmitted packet. */
			if ((dup_sack && in_sack) &&
			    (sacked & TCPCB_RETRANS) &&
			    after(TCP_SKB_CB(skb)->end_seq, tp->undo_marker))
				tp->undo_retrans--;

			/* The frame is ACKed. */
			if (!after(TCP_SKB_CB(skb)->end_seq, tp->snd_una)) {
				if (sacked&TCPCB_RETRANS) {
					if ((dup_sack && in_sack) &&
					    (sacked&TCPCB_SACKED_ACKED))
						reord = min(fack_count, reord);
				} else {
					/* If it was in a hole, we detected reordering. */
					if (fack_count < prior_fackets &&
					    !(sacked&TCPCB_SACKED_ACKED))
						reord = min(fack_count, reord);
				}

				/* Nothing to do; acked frame is about to be dropped. */
				continue;
			}

			if ((sacked&TCPCB_SACKED_RETRANS) &&
			    after(end_seq, TCP_SKB_CB(skb)->ack_seq) &&
			    (!lost_retrans || after(end_seq, lost_retrans)))
				lost_retrans = end_seq;

			if (!in_sack)
				continue;

			if (!(sacked&TCPCB_SACKED_ACKED)) {
				if (sacked & TCPCB_SACKED_RETRANS) {
					/* If the segment is not tagged as lost,
					 * we do not clear RETRANS, believing
					 * that retransmission is still in flight.
					 */
					if (sacked & TCPCB_LOST) {
						TCP_SKB_CB(skb)->sacked &= ~(TCPCB_LOST|TCPCB_SACKED_RETRANS);
						tp->lost_out--;
						tp->retrans_out--;
					}
				} else {
					/* New sack for not retransmitted frame,
					 * which was in hole. It is reordering.
					 */
					if (!(sacked & TCPCB_RETRANS) &&
					    fack_count < prior_fackets)