tcp_input.c

GNU Hurd 源代码

		return;

	sp->start_seq = TCP_SKB_CB(skb)->end_seq;
	if(!before(sp->start_seq, sp->end_seq)) {
		/* Zap this SACK, by moving forward any other SACKS. */
		for(this_sack += 1; this_sack < num_sacks; this_sack++, sp++) {
			struct tcp_sack_block *next = (sp + 1);
			sp->start_seq = next->start_seq;
			sp->end_seq = next->end_seq;
		}
		tp->num_sacks--;
	}
}

static void tcp_sack_extend(struct tcp_opt *tp, struct sk_buff *old_skb, struct sk_buff *new_skb)
{
	struct tcp_sack_block *sp = &tp->selective_acks[0];
	int num_sacks = tp->num_sacks;
	int this_sack;

	for(this_sack = 0; this_sack < num_sacks; this_sack++, sp++) {
		if(sp->end_seq == TCP_SKB_CB(old_skb)->end_seq)
			break;
	}
	if(this_sack >= num_sacks)
		return;
	sp->end_seq = TCP_SKB_CB(new_skb)->end_seq;
}

/* This one checks to see if we can put data from the
 * out_of_order queue into the receive_queue.
 */
static void tcp_ofo_queue(struct sock *sk)
{
	struct sk_buff *skb;
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	while ((skb = skb_peek(&tp->out_of_order_queue))) {
		if (after(TCP_SKB_CB(skb)->seq, tp->rcv_nxt))
			break;

		if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) {
			SOCK_DEBUG(sk, "ofo packet was already received \n");
			__skb_unlink(skb, skb->list);
			kfree_skb(skb);
			continue;
		}
		SOCK_DEBUG(sk, "ofo requeuing : rcv_next %X seq %X - %X\n",
			   tp->rcv_nxt, TCP_SKB_CB(skb)->seq,
			   TCP_SKB_CB(skb)->end_seq);

		if(tp->sack_ok)
			tcp_sack_remove_skb(tp, skb);
		__skb_unlink(skb, skb->list);
		__skb_queue_tail(&sk->receive_queue, skb);
		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
		if(skb->h.th->fin)
			tcp_fin(skb, sk, skb->h.th);
	}
}

static void tcp_data_queue(struct sock *sk, struct sk_buff *skb)
{
	struct sk_buff *skb1;
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	/*  Queue data for delivery to the user.
	 *  Packets in sequence go to the receive queue.
	 *  Out of sequence packets to the out_of_order_queue.
	 */
	if (TCP_SKB_CB(skb)->seq == tp->rcv_nxt) {
		/* Ok. In sequence. */
	queue_and_out:
		dst_confirm(sk->dst_cache);
		__skb_queue_tail(&sk->receive_queue, skb);
		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
		if(skb->h.th->fin) {
			tcp_fin(skb, sk, skb->h.th);
		} else {
			tcp_remember_ack(tp, skb->h.th, skb); 
		}
		/* This may have eaten into a SACK block. */
		if(tp->sack_ok && tp->num_sacks)
			tcp_sack_remove_skb(tp, skb);
		tcp_ofo_queue(sk);

		/* Turn on fast path. */ 
		if (skb_queue_len(&tp->out_of_order_queue) == 0)
			tp->pred_flags = htonl(((tp->tcp_header_len >> 2) << 28) |
					       (0x10 << 16) |
					       tp->snd_wnd);
		return;
	}
	
	/* An old packet, either a retransmit or some packet got lost. */
	if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) {
		/* A retransmit, 2nd most common case.  Force an imediate ack. */
		SOCK_DEBUG(sk, "retransmit received: seq %X\n", TCP_SKB_CB(skb)->seq);
		tcp_enter_quickack_mode(tp);
		kfree_skb(skb);
		return;
	}

	if (before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) {
		/* Partial packet, seq < rcv_next < end_seq */
		SOCK_DEBUG(sk, "partial packet: rcv_next %X seq %X - %X\n",
			   tp->rcv_nxt, TCP_SKB_CB(skb)->seq,
			   TCP_SKB_CB(skb)->end_seq);

		goto queue_and_out;
	}

	/* Ok. This is an out_of_order segment, force an ack. */
	tp->delayed_acks++;
	tcp_enter_quickack_mode(tp);

	/* Disable header prediction. */
	tp->pred_flags = 0;

	SOCK_DEBUG(sk, "out of order segment: rcv_next %X seq %X - %X\n",
		   tp->rcv_nxt, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq);

	if (skb_peek(&tp->out_of_order_queue) == NULL) {
		/* Initial out of order segment, build 1 SACK. */
		if(tp->sack_ok) {
			tp->num_sacks = 1;
			tp->selective_acks[0].start_seq = TCP_SKB_CB(skb)->seq;
			tp->selective_acks[0].end_seq = TCP_SKB_CB(skb)->end_seq;
		}
		__skb_queue_head(&tp->out_of_order_queue,skb);
	} else {
		for(skb1=tp->out_of_order_queue.prev; ; skb1 = skb1->prev) {
			/* Already there. */
			if (TCP_SKB_CB(skb)->seq == TCP_SKB_CB(skb1)->seq) {
				if (skb->len >= skb1->len) {
					if(tp->sack_ok)
						tcp_sack_extend(tp, skb1, skb);
					__skb_append(skb1, skb);
					__skb_unlink(skb1, skb1->list);
					kfree_skb(skb1);
				} else {
					/* A duplicate, smaller than what is in the
					 * out-of-order queue right now, toss it.
					 */
					kfree_skb(skb);
				}
				break;
			}
			
			if (after(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb1)->seq)) {
				__skb_append(skb1, skb);
				if(tp->sack_ok)
					tcp_sack_new_ofo_skb(sk, skb);
				break;
			}

                        /* See if we've hit the start. If so insert. */
			if (skb1 == skb_peek(&tp->out_of_order_queue)) {
				__skb_queue_head(&tp->out_of_order_queue,skb);
				if(tp->sack_ok)
					tcp_sack_new_ofo_skb(sk, skb);
				break;
			}
		}
	}
}


/*
 *	This routine handles the data.  If there is room in the buffer,
 *	it will be have already been moved into it.  If there is no
 *	room, then we will just have to discard the packet.
 */

static int tcp_data(struct sk_buff *skb, struct sock *sk, unsigned int len)
{
	struct tcphdr *th;
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	th = skb->h.th;
	skb_pull(skb, th->doff*4);
	skb_trim(skb, len - (th->doff*4));

        if (skb->len == 0 && !th->fin)
		return(0);

	/* 
	 *	If our receive queue has grown past its limits shrink it.
	 *	Make sure to do this before moving snd_nxt, otherwise
	 *	data might be acked for that we don't have enough room.
	 */
	if (atomic_read(&sk->rmem_alloc) > sk->rcvbuf) { 
		if (prune_queue(sk) < 0) { 
			/* Still not enough room. That can happen when
			 * skb->true_size differs significantly from skb->len.
			 */
			return 0;
		}
	}

	tcp_data_queue(sk, skb);

	if (before(tp->rcv_nxt, tp->copied_seq)) {
		printk(KERN_DEBUG "*** tcp.c:tcp_data bug acked < copied\n");
		tp->rcv_nxt = tp->copied_seq;
	}

	/* Above, tcp_data_queue() increments delayed_acks appropriately.
	 * Now tell the user we may have some data.
	 */
	if (!sk->dead) {
		sk->data_ready(sk,0);
	}
	return(1);
}

static void __tcp_data_snd_check(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	if (!after(TCP_SKB_CB(skb)->end_seq, tp->snd_una + tp->snd_wnd) &&
	    tcp_packets_in_flight(tp) < tp->snd_cwnd) {
		/* Put more data onto the wire. */
		tcp_write_xmit(sk);
	} else if (tp->packets_out == 0 && !tp->pending) {
		/* Start probing the receivers window. */
		tcp_reset_xmit_timer(sk, TIME_PROBE0, tp->rto);
	}
}

static __inline__ void tcp_data_snd_check(struct sock *sk)
{
	struct sk_buff *skb = sk->tp_pinfo.af_tcp.send_head;

	if (skb != NULL)
		__tcp_data_snd_check(sk, skb); 
}

/* 
 * Adapt the MSS value used to make delayed ack decision to the 
 * real world. 
 */ 
static __inline__ void tcp_measure_rcv_mss(struct sock *sk, struct sk_buff *skb)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	unsigned int len = skb->len, lss; 

	if (len > tp->rcv_mss) 
		tp->rcv_mss = len; 
	lss = tp->last_seg_size; 
	tp->last_seg_size = 0; 
	if (len >= 536) {
		if (len == lss) 
			tp->rcv_mss = len; 
		tp->last_seg_size = len; 
	}
}

/*
 * Check if sending an ack is needed.
 */
static __inline__ void __tcp_ack_snd_check(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	/* This also takes care of updating the window.
	 * This if statement needs to be simplified.
	 *
	 * Rules for delaying an ack:
	 *      - delay time <= 0.5 HZ
	 *      - we don't have a window update to send
	 *      - must send at least every 2 full sized packets
	 *	- must send an ACK if we have any out of order data
	 *
	 * With an extra heuristic to handle loss of packet
	 * situations and also helping the sender leave slow
	 * start in an expediant manner.
	 */

	    /* Two full frames received or... */
	if (((tp->rcv_nxt - tp->rcv_wup) >= tp->rcv_mss * MAX_DELAY_ACK) ||
	    /* We will update the window "significantly" or... */
	    tcp_raise_window(sk) ||
	    /* We entered "quick ACK" mode or... */
	    tcp_in_quickack_mode(tp) ||
	    /* We have out of order data */
	    (skb_peek(&tp->out_of_order_queue) != NULL)) {
		/* Then ack it now */
		tcp_send_ack(sk);
	} else {
		/* Else, send delayed ack. */
		tcp_send_delayed_ack(tp, HZ/2);
	}
}

static __inline__ void tcp_ack_snd_check(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	if (tp->delayed_acks == 0) {
		/* We sent a data segment already. */
		return;
	}
	__tcp_ack_snd_check(sk);
}


/*
 *	This routine is only called when we have urgent data
 *	signalled. Its the 'slow' part of tcp_urg. It could be
 *	moved inline now as tcp_urg is only called from one
 *	place. We handle URGent data wrong. We have to - as
 *	BSD still doesn't use the correction from RFC961.
 *	For 1003.1g we should support a new option TCP_STDURG to permit
 *	either form (or just set the sysctl tcp_stdurg).
 */
 
static void tcp_check_urg(struct sock * sk, struct tcphdr * th)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	u32 ptr = ntohs(th->urg_ptr);

	if (ptr && !sysctl_tcp_stdurg)
		ptr--;
	ptr += ntohl(th->seq);

	/* Ignore urgent data that we've already seen and read. */
	if (after(tp->copied_seq, ptr))
		return;

	/* Do we already have a newer (or duplicate) urgent pointer? */
	if (tp->urg_data && !after(ptr, tp->urg_seq))
		return;

	/* Tell the world about our new urgent pointer. */
	if (sk->proc != 0) {
		if (sk->proc > 0)
			kill_proc(sk->proc, SIGURG, 1);
		else
			kill_pg(-sk->proc, SIGURG, 1);
	}

	/* We may be adding urgent data when the last byte read was
	 * urgent. To do this requires some care. We cannot just ignore
	 * tp->copied_seq since we would read the last urgent byte again
	 * as data, nor can we alter copied_seq until this data arrives
	 * or we break the sematics of SIOCATMARK (and thus sockatmark())
	 */
	if (tp->urg_seq == tp->copied_seq)
		tp->copied_seq++;	/* Move the copied sequence on correctly */
	tp->urg_data = URG_NOTYET;
	tp->urg_seq = ptr;

	/* Disable header prediction. */
	tp->pred_flags = 0;
}

/* This is the 'fast' part of urgent handling. */
static inline void tcp_urg(struct sock *sk, struct tcphdr *th, unsigned long len)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	/* Check if we get a new urgent pointer - normally not. */
	if (th->urg)
		tcp_check_urg(sk,th);

	/* Do we wait for any urgent data? - normally not... */
	if (tp->urg_data == URG_NOTYET) {
		u32 ptr = tp->urg_seq - ntohl(th->seq) + (th->doff*4);

		/* Is the urgent pointer pointing into this packet? */	 
		if (ptr < len) {
			tp->urg_data = URG_VALID | *(ptr + (unsigned char *) th);
			if (!sk->dead)
				sk->data_ready(sk,0);
		}
	}
}

/* Clean the out_of_order queue if we can, trying to get
 * the socket within its memory limits again.
 *
 * Return less than zero if we should start dropping frames
 * until the socket owning process reads some of the data
 * to stabilize the situation.
 */
static int prune_queue(struct sock *sk)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp; 
	struct sk_buff * skb;

	SOCK_DEBUG(sk, "prune_queue: c=%x\n", tp->copied_seq);

	net_statistics.PruneCalled++; 

	/* First, purge the out_of_order queue. */
	skb = __skb_dequeue_tail(&tp->out_of_order_queue);
	if(skb != NULL) {
		/* Free it all. */
		do {	net_statistics.OfoPruned += skb->len; 
			kfree_skb(skb);
			skb = __skb_dequeue_tail(&tp->out_of_order_queue);
		} while(skb != NULL);

		/* Reset SACK state.  A conforming SACK implementation will
		 * do the same at a timeout based retransmit.  When a connection
		 * is in a sad state like this, we care only about integrity
		 * of the connection not performance.
		 */
		if(tp->sack_ok)
			tp->num_sacks = 0;
	}
	
	/* If we are really being abused, tell the caller to silently
	 * drop receive data on the floor.  It will get retransmitted
	 * and hopefully then we'll have sufficient space.
	 *
	 * We used to try to purge the in-order packets too, but that
	 * turns out to be deadly and fraught with races.  Consider:
	 *
	 * 1) If we acked the data, we absolutely cannot drop the
	 *    packet.  This data would then never be retransmitted.
	 * 2) It is possible, with a proper sequence of events involving
	 *    delayed acks and backlog queue handling, to have the user
	 *    read the data before it gets acked.  The previous code
	 *    here got this wrong, and it lead to data corruption.
	 * 3) Too much state changes happen when the FIN arrives, so once
	 *    we've seen that we can't remove any in-order data safely.
	 *
	 * The net result is that removing in-order receive data is too
	 * complex for anyones sanity.  So we don't do it anymore.  But
	 * if we are really having our buffer space abused we stop accepting
	 * new receive data.
	 */
	if(atomic_read(&sk->rmem_alloc) < (sk->rcvbuf << 1))
		return 0;

	/* Massive buffer overcommit. */
	return -1;
}

/*
 *	TCP receive function for the ESTABLISHED state. 
 *
 *	It is split into a fast path and a slow path. The fast path is