tcp_output.c

Linux内核源代码 为压缩文件 是<<Linux内核>>一书中的源代码

	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	unsigned int mss_now;

	/* If we are closed, the bytes will have to remain here.
	 * In time closedown will finish, we empty the write queue and all
	 * will be happy.
	 */
	if(sk->state != TCP_CLOSE) {
		struct sk_buff *skb;
		int sent_pkts = 0;

		/* Account for SACKS, we may need to fragment due to this.
		 * It is just like the real MSS changing on us midstream.
		 * We also handle things correctly when the user adds some
		 * IP options mid-stream.  Silly to do, but cover it.
		 */
		mss_now = tcp_current_mss(sk); 

		while((skb = tp->send_head) &&
		      tcp_snd_test(tp, skb, mss_now, tcp_skb_is_last(sk, skb) ? tp->nonagle : 1)) {
			if (skb->len > mss_now) {
				if (tcp_fragment(sk, skb, mss_now))
					break;
			}

			TCP_SKB_CB(skb)->when = tcp_time_stamp;
			if (tcp_transmit_skb(sk, skb_clone(skb, GFP_ATOMIC)))
				break;
			/* Advance the send_head.  This one is sent out. */
			update_send_head(sk, tp, skb);
			tcp_minshall_update(tp, mss_now, skb);
			sent_pkts = 1;
		}

		if (sent_pkts) {
			tcp_cwnd_validate(sk, tp);
			return 0;
		}

		return !tp->packets_out && tp->send_head;
	}
	return 0;
}

/* This function returns the amount that we can raise the
 * usable window based on the following constraints
 *  
 * 1. The window can never be shrunk once it is offered (RFC 793)
 * 2. We limit memory per socket
 *
 * RFC 1122:
 * "the suggested [SWS] avoidance algorithm for the receiver is to keep
 *  RECV.NEXT + RCV.WIN fixed until:
 *  RCV.BUFF - RCV.USER - RCV.WINDOW >= min(1/2 RCV.BUFF, MSS)"
 *
 * i.e. don't raise the right edge of the window until you can raise
 * it at least MSS bytes.
 *
 * Unfortunately, the recommended algorithm breaks header prediction,
 * since header prediction assumes th->window stays fixed.
 *
 * Strictly speaking, keeping th->window fixed violates the receiver
 * side SWS prevention criteria. The problem is that under this rule
 * a stream of single byte packets will cause the right side of the
 * window to always advance by a single byte.
 * 
 * Of course, if the sender implements sender side SWS prevention
 * then this will not be a problem.
 * 
 * BSD seems to make the following compromise:
 * 
 *	If the free space is less than the 1/4 of the maximum
 *	space available and the free space is less than 1/2 mss,
 *	then set the window to 0.
 *	[ Actually, bsd uses MSS and 1/4 of maximal _window_ ]
 *	Otherwise, just prevent the window from shrinking
 *	and from being larger than the largest representable value.
 *
 * This prevents incremental opening of the window in the regime
 * where TCP is limited by the speed of the reader side taking
 * data out of the TCP receive queue. It does nothing about
 * those cases where the window is constrained on the sender side
 * because the pipeline is full.
 *
 * BSD also seems to "accidentally" limit itself to windows that are a
 * multiple of MSS, at least until the free space gets quite small.
 * This would appear to be a side effect of the mbuf implementation.
 * Combining these two algorithms results in the observed behavior
 * of having a fixed window size at almost all times.
 *
 * Below we obtain similar behavior by forcing the offered window to
 * a multiple of the mss when it is feasible to do so.
 *
 * Note, we don't "adjust" for TIMESTAMP or SACK option bytes.
 * Regular options like TIMESTAMP are taken into account.
 */
u32 __tcp_select_window(struct sock *sk)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
	/* MSS for the peer's data.  Previous verions used mss_clamp
	 * here.  I don't know if the value based on our guesses
	 * of peer's MSS is better for the performance.  It's more correct
	 * but may be worse for the performance because of rcv_mss
	 * fluctuations.  --SAW  1998/11/1
	 */
	unsigned int mss = tp->ack.rcv_mss;
	int free_space;
	u32 window;

	/* Sometimes free_space can be < 0. */
	free_space = tcp_space(sk); 
	if (tp->window_clamp < mss)
		mss = tp->window_clamp; 

	if (free_space < (int)min(tp->window_clamp, tcp_full_space(sk)) / 2) {
		tp->ack.quick = 0;

		if (tcp_memory_pressure)
			tp->rcv_ssthresh = min(tp->rcv_ssthresh, 4*tp->advmss);

		if (free_space < ((int)mss))
			return 0;
	}

	if (free_space > tp->rcv_ssthresh)
		free_space = tp->rcv_ssthresh;

	/* Get the largest window that is a nice multiple of mss.
	 * Window clamp already applied above.
	 * If our current window offering is within 1 mss of the
	 * free space we just keep it. This prevents the divide
	 * and multiply from happening most of the time.
	 * We also don't do any window rounding when the free space
	 * is too small.
	 */
	window = tp->rcv_wnd;
	if ((((int) window) <= (free_space - ((int) mss))) ||
	    (((int) window) > free_space))
		window = (((unsigned int) free_space)/mss)*mss;

	return window;
}

/* Attempt to collapse two adjacent SKB's during retransmission. */
static void tcp_retrans_try_collapse(struct sock *sk, struct sk_buff *skb, int mss_now)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
	struct sk_buff *next_skb = skb->next;

	/* The first test we must make is that neither of these two
	 * SKB's are still referenced by someone else.
	 */
	if(!skb_cloned(skb) && !skb_cloned(next_skb)) {
		int skb_size = skb->len, next_skb_size = next_skb->len;
		u16 flags = TCP_SKB_CB(skb)->flags;

		/* Also punt if next skb has been SACK'd. */
		if(TCP_SKB_CB(next_skb)->sacked & TCPCB_SACKED_ACKED)
			return;

		/* Next skb is out of window. */
		if (after(TCP_SKB_CB(next_skb)->end_seq, tp->snd_una+tp->snd_wnd))
			return;

		/* Punt if not enough space exists in the first SKB for
		 * the data in the second, or the total combined payload
		 * would exceed the MSS.
		 */
		if ((next_skb_size > skb_tailroom(skb)) ||
		    ((skb_size + next_skb_size) > mss_now))
			return;

		/* Ok.  We will be able to collapse the packet. */
		__skb_unlink(next_skb, next_skb->list);

		if(skb->len % 4) {
			/* Must copy and rechecksum all data. */
			memcpy(skb_put(skb, next_skb_size), next_skb->data, next_skb_size);
			skb->csum = csum_partial(skb->data, skb->len, 0);
		} else {
			/* Optimize, actually we could also combine next_skb->csum
			 * to skb->csum using a single add w/carry operation too.
			 */
			skb->csum = csum_partial_copy_nocheck(next_skb->data,
							      skb_put(skb, next_skb_size),
							      next_skb_size, skb->csum);
		}
	
		/* Update sequence range on original skb. */
		TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(next_skb)->end_seq;

		/* Merge over control information. */
		flags |= TCP_SKB_CB(next_skb)->flags; /* This moves PSH/FIN etc. over */
		TCP_SKB_CB(skb)->flags = flags;

		/* All done, get rid of second SKB and account for it so
		 * packet counting does not break.
		 */
		TCP_SKB_CB(skb)->sacked |= TCP_SKB_CB(next_skb)->sacked&(TCPCB_EVER_RETRANS|TCPCB_AT_TAIL);
		if (TCP_SKB_CB(next_skb)->sacked&TCPCB_SACKED_RETRANS)
			tp->retrans_out--;
		if (TCP_SKB_CB(next_skb)->sacked&TCPCB_LOST) {
			tp->lost_out--;
			tp->left_out--;
		}
		if (!tp->sack_ok && tp->sacked_out) {
			/* Reno case is special. Sigh... */
			tp->sacked_out--;
			tp->left_out--;
		}
		/* Not quite right: it can be > snd.fack, but
		 * it is better to underestimate fackets.
		 */
		if (tp->fackets_out)
			tp->fackets_out--;
		tcp_free_skb(sk, next_skb);
		tp->packets_out--;
	}
}

/* Do a simple retransmit without using the backoff mechanisms in
 * tcp_timer. This is used for path mtu discovery. 
 * The socket is already locked here.
 */ 
void tcp_simple_retransmit(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct sk_buff *skb;
	unsigned int mss = tcp_current_mss(sk);
	int lost = 0;

	for_retrans_queue(skb, sk, tp) {
		if (skb->len > mss && 
		    !(TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_ACKED)) {
			if (TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_RETRANS) {
				TCP_SKB_CB(skb)->sacked &= ~TCPCB_SACKED_RETRANS;
				tp->retrans_out--;
			}
			if (!(TCP_SKB_CB(skb)->sacked&TCPCB_LOST)) {
				TCP_SKB_CB(skb)->sacked |= TCPCB_LOST;
				tp->lost_out++;
				lost = 1;
			}
		}
	}

	if (!lost)
		return;

	tp->left_out = tp->sacked_out + tp->lost_out;

 	/* Don't muck with the congestion window here.
	 * Reason is that we do not increase amount of _data_
	 * in network, but units changed and effective
	 * cwnd/ssthresh really reduced now.
	 */
	if (tp->ca_state != TCP_CA_Loss) {
		tp->high_seq = tp->snd_nxt;
		tp->snd_ssthresh = tcp_current_ssthresh(tp);
		tp->prior_ssthresh = 0;
		tp->undo_marker = 0;
		tp->ca_state = TCP_CA_Loss;
	}
	tcp_xmit_retransmit_queue(sk);
}

/* This retransmits one SKB.  Policy decisions and retransmit queue
 * state updates are done by the caller.  Returns non-zero if an
 * error occurred which prevented the send.
 */
int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	unsigned int cur_mss = tcp_current_mss(sk);
	int err;

	/* Do not sent more than we queued. 1/4 is reserved for possible
	 * copying overhead: frgagmentation, tunneling, mangling etc.
	 */
	if (atomic_read(&sk->wmem_alloc) > min(sk->wmem_queued+(sk->wmem_queued>>2),sk->sndbuf))
		return -EAGAIN;

	if(skb->len > cur_mss) {
		if(tcp_fragment(sk, skb, cur_mss))
			return -ENOMEM; /* We'll try again later. */

		/* New SKB created, account for it. */
		tp->packets_out++;
	}

	/* Collapse two adjacent packets if worthwhile and we can. */
	if(!(TCP_SKB_CB(skb)->flags & TCPCB_FLAG_SYN) &&
	   (skb->len < (cur_mss >> 1)) &&
	   (skb->next != tp->send_head) &&
	   (skb->next != (struct sk_buff *)&sk->write_queue) &&
	   (sysctl_tcp_retrans_collapse != 0))
		tcp_retrans_try_collapse(sk, skb, cur_mss);

	if(tp->af_specific->rebuild_header(sk))
		return -EHOSTUNREACH; /* Routing failure or similar. */

	/* Some Solaris stacks overoptimize and ignore the FIN on a
	 * retransmit when old data is attached.  So strip it off
	 * since it is cheap to do so and saves bytes on the network.
	 */
	if(skb->len > 0 &&
	   (TCP_SKB_CB(skb)->flags & TCPCB_FLAG_FIN) &&
	   tp->snd_una == (TCP_SKB_CB(skb)->end_seq - 1)) {
		TCP_SKB_CB(skb)->seq = TCP_SKB_CB(skb)->end_seq - 1;
		skb_trim(skb, 0);
		skb->csum = 0;
	}

	/* Make a copy, if the first transmission SKB clone we made
	 * is still in somebody's hands, else make a clone.
	 */
	TCP_SKB_CB(skb)->when = tcp_time_stamp;

	err = tcp_transmit_skb(sk, (skb_cloned(skb) ?
				    skb_copy(skb, GFP_ATOMIC):
				    skb_clone(skb, GFP_ATOMIC)));

	if (err == 0) {
		/* Update global TCP statistics. */
		TCP_INC_STATS(TcpRetransSegs);

#if FASTRETRANS_DEBUG > 0
		if (TCP_SKB_CB(skb)->sacked&TCPCB_SACKED_RETRANS) {
			if (net_ratelimit())
				printk(KERN_DEBUG "retrans_out leaked.\n");
		}
#endif
		TCP_SKB_CB(skb)->sacked |= TCPCB_RETRANS;
		tp->retrans_out++;

		/* Save stamp of the first retransmit. */
		if (!tp->retrans_stamp)
			tp->retrans_stamp = TCP_SKB_CB(skb)->when;

		tp->undo_retrans++;

		/* snd_nxt is stored to detect loss of retransmitted segment,
		 * see tcp_input.c tcp_sacktag_write_queue().
		 */
		TCP_SKB_CB(skb)->ack_seq = tp->snd_nxt;
	}
	return err;
}

/* This gets called after a retransmit timeout, and the initially
 * retransmitted data is acknowledged.  It tries to continue
 * resending the rest of the retransmit queue, until either
 * we've sent it all or the congestion window limit is reached.
 * If doing SACK, the first ACK which comes back for a timeout
 * based retransmit packet might feed us FACK information again.
 * If so, we use it to avoid unnecessarily retransmissions.
 */
void tcp_xmit_retransmit_queue(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct sk_buff *skb;
	int packet_cnt = tp->lost_out;

	/* First pass: retransmit lost packets. */
	if (packet_cnt) {
		for_retrans_queue(skb, sk, tp) {
			__u8 sacked = TCP_SKB_CB(skb)->sacked;

			if (tcp_packets_in_flight(tp) >= tp->snd_cwnd)
				return;

			if (sacked&TCPCB_LOST) {
				if (!(sacked&(TCPCB_SACKED_ACKED|TCPCB_SACKED_RETRANS))) {
					if (tcp_retransmit_skb(sk, skb))
						return;
					if (tp->ca_state != TCP_CA_Loss)
						NET_INC_STATS_BH(TCPFastRetrans);
					else
						NET_INC_STATS_BH(TCPSlowStartRetrans);

					if (skb == skb_peek(&sk->write_queue))
						tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
				}

				if (--packet_cnt <= 0)
					break;
			}
		}
	}

	/* OK, demanded retransmission is finished. */

	/* Forward retransmissions are possible only during Recovery. */
	if (tp->ca_state != TCP_CA_Recovery)
		return;

	/* No forward retransmissions in Reno are possible. */
	if (!tp->sack_ok)
		return;

	/* Yeah, we have to make difficult choice between forward transmission
	 * and retransmission... Both ways have their merits...
	 *
	 * For now we do not retrnamsit anything, while we have some new
	 * segments to send.
	 */

	if (tcp_may_send_now(sk, tp))
		return;

	packet_cnt = 0;

	for_retrans_queue(skb, sk, tp) {
		if(++packet_cnt > tp->fackets_out)
			break;

		if (tcp_packets_in_flight(tp) >= tp->snd_cwnd)
			break;

		if(TCP_SKB_CB(skb)->sacked & TCPCB_TAGBITS)
			continue;

		/* Ok, retransmit it. */
		if(tcp_retransmit_skb(sk, skb))
			break;

		if (skb == skb_peek(&sk->write_queue))
			tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);

		NET_INC_STATS_BH(TCPForwardRetrans);
	}
}


/* Send a fin.  The caller locks the socket for us.  This cannot be
 * allowed to fail queueing a FIN frame under any circumstances.
 */
void tcp_send_fin(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);	
	struct sk_buff *skb = skb_peek_tail(&sk->write_queue);
	unsigned int mss_now;
	
	/* Optimization, tack on the FIN if we have a queue of
	 * unsent frames.  But be careful about outgoing SACKS
	 * and IP options.
	 */
	mss_now = tcp_current_mss(sk); 

	/* Please, find seven differences of 2.3.33 and loook
	 * what I broke here. 8) --ANK
	 */

	if(tp->send_head != NULL) {
		/* tcp_write_xmit() takes care of the rest. */
		TCP_SKB_CB(skb)->flags |= TCPCB_FLAG_FIN;
		TCP_SKB_CB(skb)->end_seq++;
		tp->write_seq++;

		/* Special case to avoid Nagle bogosity.  If this
		 * segment is the last segment, and it was queued
		 * due to Nagle/SWS-avoidance, send it out now.
		 */
		if(tp->send_head == skb &&
		   !after(tp->write_seq, tp->snd_una + tp->snd_wnd)) {