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

/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the  BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
 * Version:	$Id: tcp_output.c,v 1.129 2000/11/28 17:04:10 davem Exp $
 *
 * Authors:	Ross Biro, <[e-mail]>
 *		Fred N. van Kempen, <[e-mail]>
 *		Mark Evans, <[e-mail]>
 *		Corey Minyard <wf-rch![e-mail]>
 *		Florian La Roche, <[e-mail]>
 *		Charles Hedrick, <[e-mail]>
 *		Linus Torvalds, <[e-mail]>
 *		Alan Cox, <[e-mail]>
 *		Matthew Dillon, <[e-mail]>
 *		Arnt Gulbrandsen, <[e-mail]>
 *		Jorge Cwik, <[e-mail]>
 */

/*
 * Changes:	Pedro Roque	:	Retransmit queue handled by TCP.
 *				:	Fragmentation on mtu decrease
 *				:	Segment collapse on retransmit
 *				:	AF independence
 *
 *		Linus Torvalds	:	send_delayed_ack
 *		David S. Miller	:	Charge memory using the right skb
 *					during syn/ack processing.
 *		David S. Miller :	Output engine completely rewritten.
 *		Andrea Arcangeli:	SYNACK carry ts_recent in tsecr.
 *		Cacophonix Gaul :	draft-minshall-nagle-01
 *		J Hadi Salim	:	ECN support
 *
 */

#include <net/tcp.h>

#include <linux/smp_lock.h>

/* People can turn this off for buggy TCP's found in printers etc. */
int sysctl_tcp_retrans_collapse = 1;

static __inline__
void update_send_head(struct sock *sk, struct tcp_opt *tp, struct sk_buff *skb)
{
	tp->send_head = skb->next;
	if (tp->send_head == (struct sk_buff *) &sk->write_queue)
		tp->send_head = NULL;
	tp->snd_nxt = TCP_SKB_CB(skb)->end_seq;
	if (tp->packets_out++ == 0)
		tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
}

/* SND.NXT, if window was not shrunk.
 * If window has been shrunk, what should we make? It is not clear at all.
 * Using SND.UNA we will fail to open window, SND.NXT is out of window. :-(
 * Anything in between SND.UNA...SND.UNA+SND.WND also can be already
 * invalid. OK, let's make this for now:
 */
static __inline__ __u32 tcp_acceptable_seq(struct sock *sk, struct tcp_opt *tp)
{
	if (!before(tp->snd_una+tp->snd_wnd, tp->snd_nxt))
		return tp->snd_nxt;
	else
		return tp->snd_una+tp->snd_wnd;
}

/* Calculate mss to advertise in SYN segment.
 * RFC1122, RFC1063, draft-ietf-tcpimpl-pmtud-01 state that:
 *
 * 1. It is independent of path mtu.
 * 2. Ideally, it is maximal possible segment size i.e. 65535-40.
 * 3. For IPv4 it is reasonable to calculate it from maximal MTU of
 *    attached devices, because some buggy hosts are confused by
 *    large MSS.
 * 4. We do not make 3, we advertise MSS, calculated from first
 *    hop device mtu, but allow to raise it to ip_rt_min_advmss.
 *    This may be overriden via information stored in routing table.
 * 5. Value 65535 for MSS is valid in IPv6 and means "as large as possible,
 *    probably even Jumbo".
 */
static __u16 tcp_advertise_mss(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	struct dst_entry *dst = __sk_dst_get(sk);
	int mss = tp->advmss;

	if (dst && dst->advmss < mss) {
		mss = dst->advmss;
		tp->advmss = mss;
	}

	return (__u16)mss;
}

/* RFC2861. Reset CWND after idle period longer RTO to "restart window".
 * This is the first part of cwnd validation mechanism. */
static void tcp_cwnd_restart(struct tcp_opt *tp)
{
	s32 delta = tcp_time_stamp - tp->lsndtime;
	u32 restart_cwnd = tcp_init_cwnd(tp);
	u32 cwnd = tp->snd_cwnd;

	tp->snd_ssthresh = tcp_current_ssthresh(tp);
	restart_cwnd = min(restart_cwnd, cwnd);

	while ((delta -= tp->rto) > 0 && cwnd > restart_cwnd)
		cwnd >>= 1;
	tp->snd_cwnd = max(cwnd, restart_cwnd);
	tp->snd_cwnd_stamp = tcp_time_stamp;
	tp->snd_cwnd_used = 0;
}

static __inline__ void tcp_event_data_sent(struct tcp_opt *tp, struct sk_buff *skb)
{
	u32 now = tcp_time_stamp;

	if (!tp->packets_out && (s32)(now - tp->lsndtime) > tp->rto)
		tcp_cwnd_restart(tp);

	tp->lsndtime = now;

	/* If it is a reply for ato after last received
	 * packet, enter pingpong mode.
	 */
	if ((u32)(now - tp->ack.lrcvtime) < tp->ack.ato)
		tp->ack.pingpong = 1;
}

static __inline__ void tcp_event_ack_sent(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	tcp_dec_quickack_mode(tp);
	tcp_clear_xmit_timer(sk, TCP_TIME_DACK);
}

/* Chose a new window to advertise, update state in tcp_opt for the
 * socket, and return result with RFC1323 scaling applied.  The return
 * value can be stuffed directly into th->window for an outgoing
 * frame.
 */
static __inline__ u16 tcp_select_window(struct sock *sk)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
	u32 cur_win = tcp_receive_window(tp);
	u32 new_win = __tcp_select_window(sk);

	/* Never shrink the offered window */
	if(new_win < cur_win) {
		/* Danger Will Robinson!
		 * Don't update rcv_wup/rcv_wnd here or else
		 * we will not be able to advertise a zero
		 * window in time.  --DaveM
		 *
		 * Relax Will Robinson.
		 */
		new_win = cur_win;
	}
	tp->rcv_wnd = new_win;
	tp->rcv_wup = tp->rcv_nxt;

	/* RFC1323 scaling applied */
	new_win >>= tp->rcv_wscale;

#ifdef TCP_FORMAL_WINDOW
	if (new_win == 0) {
		/* If we advertise zero window, disable fast path. */
		tp->pred_flags = 0;
	} else if (cur_win == 0 && tp->pred_flags == 0 &&
		   skb_queue_len(&tp->out_of_order_queue) == 0 &&
		   !tp->urg_data) {
		/* If we open zero window, enable fast path.
		   Without this it will be open by the first data packet,
		   it is too late to merge checksumming to copy.
		 */
		tcp_fast_path_on(tp);
	}
#endif

	return new_win;
}


/* This routine actually transmits TCP packets queued in by
 * tcp_do_sendmsg().  This is used by both the initial
 * transmission and possible later retransmissions.
 * All SKB's seen here are completely headerless.  It is our
 * job to build the TCP header, and pass the packet down to
 * IP so it can do the same plus pass the packet off to the
 * device.
 *
 * We are working here with either a clone of the original
 * SKB, or a fresh unique copy made by the retransmit engine.
 */
int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb)
{
	if(skb != NULL) {
		struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);
		struct tcp_skb_cb *tcb = TCP_SKB_CB(skb);
		int tcp_header_size = tp->tcp_header_len;
		struct tcphdr *th;
		int sysctl_flags;
		int err;

#define SYSCTL_FLAG_TSTAMPS	0x1
#define SYSCTL_FLAG_WSCALE	0x2
#define SYSCTL_FLAG_SACK	0x4

		sysctl_flags = 0;
		if (tcb->flags & TCPCB_FLAG_SYN) {
			tcp_header_size = sizeof(struct tcphdr) + TCPOLEN_MSS;
			if(sysctl_tcp_timestamps) {
				tcp_header_size += TCPOLEN_TSTAMP_ALIGNED;
				sysctl_flags |= SYSCTL_FLAG_TSTAMPS;
			}
			if(sysctl_tcp_window_scaling) {
				tcp_header_size += TCPOLEN_WSCALE_ALIGNED;
				sysctl_flags |= SYSCTL_FLAG_WSCALE;
			}
			if(sysctl_tcp_sack) {
				sysctl_flags |= SYSCTL_FLAG_SACK;
				if(!(sysctl_flags & SYSCTL_FLAG_TSTAMPS))
					tcp_header_size += TCPOLEN_SACKPERM_ALIGNED;
			}
		} else if (tp->eff_sacks) {
			/* A SACK is 2 pad bytes, a 2 byte header, plus
			 * 2 32-bit sequence numbers for each SACK block.
			 */
			tcp_header_size += (TCPOLEN_SACK_BASE_ALIGNED +
					    (tp->eff_sacks * TCPOLEN_SACK_PERBLOCK));
		}
		th = (struct tcphdr *) skb_push(skb, tcp_header_size);
		skb->h.th = th;
		skb_set_owner_w(skb, sk);

		/* Build TCP header and checksum it. */
		th->source		= sk->sport;
		th->dest		= sk->dport;
		th->seq			= htonl(tcb->seq);
		th->ack_seq		= htonl(tp->rcv_nxt);
		*(((__u16 *)th) + 6)	= htons(((tcp_header_size >> 2) << 12) | tcb->flags);
		if (tcb->flags & TCPCB_FLAG_SYN) {
			/* RFC1323: The window in SYN & SYN/ACK segments
			 * is never scaled.
			 */
			th->window	= htons(tp->rcv_wnd);
		} else {
			th->window	= htons(tcp_select_window(sk));
		}
		th->check		= 0;
		th->urg_ptr		= 0;

		if (tp->urg_mode &&
		    between(tp->snd_up, tcb->seq+1, tcb->seq+0xFFFF)) {
			th->urg_ptr		= htons(tp->snd_up-tcb->seq);
			th->urg			= 1;
		}

		if (tcb->flags & TCPCB_FLAG_SYN) {
			tcp_syn_build_options((__u32 *)(th + 1),
					      tcp_advertise_mss(sk),
					      (sysctl_flags & SYSCTL_FLAG_TSTAMPS),
					      (sysctl_flags & SYSCTL_FLAG_SACK),
					      (sysctl_flags & SYSCTL_FLAG_WSCALE),
					      tp->rcv_wscale,
					      tcb->when,
		      			      tp->ts_recent);
		} else {
			tcp_build_and_update_options((__u32 *)(th + 1),
						     tp, tcb->when);

			TCP_ECN_send(sk, tp, skb, tcp_header_size);
		}
		tp->af_specific->send_check(sk, th, skb->len, skb);

		if (tcb->flags & TCPCB_FLAG_ACK)
			tcp_event_ack_sent(sk);

		if (skb->len != tcp_header_size)
			tcp_event_data_sent(tp, skb);

		TCP_INC_STATS(TcpOutSegs);

		err = tp->af_specific->queue_xmit(skb);
		if (err <= 0)
			return err;

		tcp_enter_cwr(tp);

		/* NET_XMIT_CN is special. It does not guarantee,
		 * that this packet is lost. It tells that device
		 * is about to start to drop packets or already
		 * drops some packets of the same priority and
		 * invokes us to send less aggressively.
		 */
		return err == NET_XMIT_CN ? 0 : err;
	}
	return -ENOBUFS;
#undef SYSCTL_FLAG_TSTAMPS
#undef SYSCTL_FLAG_WSCALE
#undef SYSCTL_FLAG_SACK
}


/* This is the main buffer sending routine. We queue the buffer
 * and decide whether to queue or transmit now.
 *
 * NOTE: probe0 timer is not checked, do not forget tcp_push_pending_frames,
 * otherwise socket can stall.
 */
void tcp_send_skb(struct sock *sk, struct sk_buff *skb, int force_queue, unsigned cur_mss)
{
	struct tcp_opt *tp = &(sk->tp_pinfo.af_tcp);

	/* Advance write_seq and place onto the write_queue. */
	tp->write_seq = TCP_SKB_CB(skb)->end_seq;
	__skb_queue_tail(&sk->write_queue, skb);
	tcp_charge_skb(sk, skb);

	if (!force_queue && tp->send_head == NULL && tcp_snd_test(tp, skb, cur_mss, tp->nonagle)) {
		/* Send it out now. */
		TCP_SKB_CB(skb)->when = tcp_time_stamp;
		if (tcp_transmit_skb(sk, skb_clone(skb, sk->allocation)) == 0) {
			tp->snd_nxt = TCP_SKB_CB(skb)->end_seq;
			tcp_minshall_update(tp, cur_mss, skb);
			if (tp->packets_out++ == 0)
				tcp_reset_xmit_timer(sk, TCP_TIME_RETRANS, tp->rto);
			return;
		}
	}
	/* Queue it, remembering where we must start sending. */
	if (tp->send_head == NULL)
		tp->send_head = skb;
}

/* Function to create two new TCP segments.  Shrinks the given segment
 * to the specified size and appends a new segment with the rest of the
 * packet to the list.  This won't be called frequently, I hope. 
 * Remember, these are still headerless SKBs at this point.
 */
static int tcp_fragment(struct sock *sk, struct sk_buff *skb, u32 len)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
	struct sk_buff *buff;
	int nsize = skb->len - len;
	u16 flags;

	/* Get a new skb... force flag on. */
	buff = tcp_alloc_skb(sk, nsize + MAX_TCP_HEADER, GFP_ATOMIC);
	if (buff == NULL)
		return -ENOMEM; /* We'll just try again later. */
	tcp_charge_skb(sk, buff);

	/* Reserve space for headers. */
	skb_reserve(buff, MAX_TCP_HEADER);
		
	/* Correct the sequence numbers. */
	TCP_SKB_CB(buff)->seq = TCP_SKB_CB(skb)->seq + len;
	TCP_SKB_CB(buff)->end_seq = TCP_SKB_CB(skb)->end_seq;
	
	/* PSH and FIN should only be set in the second packet. */
	flags = TCP_SKB_CB(skb)->flags;
	TCP_SKB_CB(skb)->flags = flags & ~(TCPCB_FLAG_FIN|TCPCB_FLAG_PSH);
	TCP_SKB_CB(buff)->flags = flags;
	TCP_SKB_CB(buff)->sacked = TCP_SKB_CB(skb)->sacked&(TCPCB_LOST|TCPCB_EVER_RETRANS|TCPCB_AT_TAIL);
	if (TCP_SKB_CB(buff)->sacked&TCPCB_LOST) {
		tp->lost_out++;
		tp->left_out++;
	}
	TCP_SKB_CB(buff)->sacked &= ~TCPCB_AT_TAIL;

	/* Copy and checksum data tail into the new buffer. */
	buff->csum = csum_partial_copy_nocheck(skb->data + len, skb_put(buff, nsize),
					       nsize, 0);

	/* This takes care of the FIN sequence number too. */
	TCP_SKB_CB(skb)->end_seq = TCP_SKB_CB(buff)->seq;
	skb_trim(skb, len);

	/* Rechecksum original buffer. */
	skb->csum = csum_partial(skb->data, skb->len, 0);

	/* Looks stupid, but our code really uses when of
	 * skbs, which it never sent before. --ANK
	 */
	TCP_SKB_CB(buff)->when = TCP_SKB_CB(skb)->when;

	/* Link BUFF into the send queue. */
	__skb_append(skb, buff);

	return 0;
}

/* This function synchronize snd mss to current pmtu/exthdr set.

   tp->user_mss is mss set by user by TCP_MAXSEG. It does NOT counts
   for TCP options, but includes only bare TCP header.

   tp->mss_clamp is mss negotiated at connection setup.
   It is minumum of user_mss and mss received with SYN.
   It also does not include TCP options.

   tp->pmtu_cookie is last pmtu, seen by this function.

   tp->mss_cache is current effective sending mss, including
   all tcp options except for SACKs. It is evaluated,
   taking into account current pmtu, but never exceeds
   tp->mss_clamp.

   NOTE1. rfc1122 clearly states that advertised MSS
   DOES NOT include either tcp or ip options.

   NOTE2. tp->pmtu_cookie and tp->mss_cache are READ ONLY outside
   this function.			--ANK (980731)
 */

int tcp_sync_mss(struct sock *sk, u32 pmtu)
{
	struct tcp_opt *tp = &sk->tp_pinfo.af_tcp;
	int mss_now;

	/* Calculate base mss without TCP options:
	   It is MMS_S - sizeof(tcphdr) of rfc1122
	 */

	mss_now = pmtu - tp->af_specific->net_header_len - sizeof(struct tcphdr);

	/* Clamp it (mss_clamp does not include tcp options) */
	if (mss_now > tp->mss_clamp)
		mss_now = tp->mss_clamp;

	/* Now subtract optional transport overhead */
	mss_now -= tp->ext_header_len;

	/* Then reserve room for full set of TCP options and 8 bytes of data */
	if (mss_now < 48)
		mss_now = 48;

	/* Now subtract TCP options size, not including SACKs */
	mss_now -= tp->tcp_header_len - sizeof(struct tcphdr);

	/* Bound mss with half of window */
	if (tp->max_window && mss_now > (tp->max_window>>1))
		mss_now = max((tp->max_window>>1), 68 - tp->tcp_header_len);

	/* And store cached results */
	tp->pmtu_cookie = pmtu;
	tp->mss_cache = mss_now;
	return mss_now;
}


/* This routine writes packets to the network.  It advances the
 * send_head.  This happens as incoming acks open up the remote
 * window for us.
 *
 * Returns 1, if no segments are in flight and we have queued segments, but
 * cannot send anything now because of SWS or another problem.
 */
int tcp_write_xmit(struct sock *sk)
{