tcp_output.c

来自「eCos操作系统源码」· C语言 代码 · 共 949 行 · 第 1/2 页

//==========================================================================
//
//      src/sys/netinet/tcp_output.c
//
//==========================================================================
//####BSDCOPYRIGHTBEGIN####
//
// -------------------------------------------
//
// Portions of this software may have been derived from OpenBSD, 
// FreeBSD or other sources, and are covered by the appropriate
// copyright disclaimers included herein.
//
// Portions created by Red Hat are
// Copyright (C) 2002 Red Hat, Inc. All Rights Reserved.
//
// -------------------------------------------
//
//####BSDCOPYRIGHTEND####
//==========================================================================

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
 *	The Regents of the University of California.  All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 * 3. All advertising materials mentioning features or use of this software
 *    must display the following acknowledgement:
 *	This product includes software developed by the University of
 *	California, Berkeley and its contributors.
 * 4. Neither the name of the University nor the names of its contributors
 *    may be used to endorse or promote products derived from this software
 *    without specific prior written permission.
 *
 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 *
 *	@(#)tcp_output.c	8.4 (Berkeley) 5/24/95
 * $FreeBSD: src/sys/netinet/tcp_output.c,v 1.39.2.10 2001/07/07 04:30:38 silby Exp $
 */

#include <sys/param.h>
#include <sys/sysctl.h>
#include <sys/mbuf.h>
#include <sys/domain.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>

#include <net/route.h>

#include <netinet/in.h>
#include <netinet/in_systm.h>
#include <netinet/ip.h>
#include <netinet/in_pcb.h>
#include <netinet/ip_var.h>
#ifdef INET6
#include <netinet6/in6_pcb.h>
#include <netinet/ip6.h>
#include <netinet6/ip6_var.h>
#endif
#include <netinet/tcp.h>
#define	TCPOUTFLAGS
#include <netinet/tcp_fsm.h>
#include <netinet/tcp_seq.h>
#include <netinet/tcp_timer.h>
#include <netinet/tcp_var.h>
#include <netinet/tcpip.h>
#ifdef TCPDEBUG
#include <netinet/tcp_debug.h>
#endif

#ifdef IPSEC
#include <netinet6/ipsec.h>
#endif /*IPSEC*/

#ifdef notyet
extern struct mbuf *m_copypack();
#endif

static int path_mtu_discovery = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, path_mtu_discovery, CTLFLAG_RW,
	&path_mtu_discovery, 1, "Enable Path MTU Discovery");

int ss_fltsz = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, slowstart_flightsize, CTLFLAG_RW,
	&ss_fltsz, 1, "Slow start flight size");

int ss_fltsz_local = TCP_MAXWIN;               /* something large */
SYSCTL_INT(_net_inet_tcp, OID_AUTO, local_slowstart_flightsize, CTLFLAG_RW,
	&ss_fltsz_local, 1, "Slow start flight size for local networks");

int     tcp_do_newreno = 1;
SYSCTL_INT(_net_inet_tcp, OID_AUTO, newreno, CTLFLAG_RW, &tcp_do_newreno,
        0, "Enable NewReno Algorithms");
/*
 * Tcp output routine: figure out what should be sent and send it.
 */
int
tcp_output(tp)
	register struct tcpcb *tp;
{
	register struct socket *so = tp->t_inpcb->inp_socket;
	register long len, win;
	int off, flags, error;
	register struct mbuf *m;
	struct ip *ip = NULL;
	register struct ipovly *ipov = NULL;
#ifdef INET6
	struct ip6_hdr *ip6 = NULL;
#endif /* INET6 */
	register struct tcphdr *th;
	u_char opt[TCP_MAXOLEN];
	unsigned ipoptlen, optlen, hdrlen;
	int idle, sendalot;
	int maxburst = TCP_MAXBURST;
	struct rmxp_tao *taop;
	struct rmxp_tao tao_noncached;
#ifdef INET6
	int isipv6;
#endif

#ifdef INET6
	isipv6 = (tp->t_inpcb->inp_vflag & INP_IPV6) != 0;
#endif

	/*
	 * Determine length of data that should be transmitted,
	 * and flags that will be used.
	 * If there is some data or critical controls (SYN, RST)
	 * to send, then transmit; otherwise, investigate further.
	 */
	idle = (tp->snd_max == tp->snd_una);
	if (idle && (ticks - tp->t_rcvtime) >= tp->t_rxtcur) {
		/*
		 * We have been idle for "a while" and no acks are
		 * expected to clock out any data we send --
		 * slow start to get ack "clock" running again.
		 *       
		 * Set the slow-start flight size depending on whether
		 * this is a local network or not.
		 */      
		if (
#ifdef INET6
		    (isipv6 && in6_localaddr(&tp->t_inpcb->in6p_faddr)) ||
		    (!isipv6 &&
#endif
		     in_localaddr(tp->t_inpcb->inp_faddr)
#ifdef INET6
		     )
#endif
		    )
			tp->snd_cwnd = tp->t_maxseg * ss_fltsz_local;
		else     
			tp->snd_cwnd = tp->t_maxseg * ss_fltsz;
	}
again:
	sendalot = 0;
	off = tp->snd_nxt - tp->snd_una;
	win = min(tp->snd_wnd, tp->snd_cwnd);

	flags = tcp_outflags[tp->t_state];
	/*
	 * Get standard flags, and add SYN or FIN if requested by 'hidden'
	 * state flags.
	 */
	if (tp->t_flags & TF_NEEDFIN)
		flags |= TH_FIN;
	if (tp->t_flags & TF_NEEDSYN)
		flags |= TH_SYN;

	/*
	 * If in persist timeout with window of 0, send 1 byte.
	 * Otherwise, if window is small but nonzero
	 * and timer expired, we will send what we can
	 * and go to transmit state.
	 */
	if (tp->t_force) {
		if (win == 0) {
			/*
			 * If we still have some data to send, then
			 * clear the FIN bit.  Usually this would
			 * happen below when it realizes that we
			 * aren't sending all the data.  However,
			 * if we have exactly 1 byte of unsent data,
			 * then it won't clear the FIN bit below,
			 * and if we are in persist state, we wind
			 * up sending the packet without recording
			 * that we sent the FIN bit.
			 *
			 * We can't just blindly clear the FIN bit,
			 * because if we don't have any more data
			 * to send then the probe will be the FIN
			 * itself.
			 */
			if (off < so->so_snd.sb_cc)
				flags &= ~TH_FIN;
			win = 1;
		} else {
			callout_stop(tp->tt_persist);
			tp->t_rxtshift = 0;
		}
	}

	len = (long)ulmin(so->so_snd.sb_cc, win) - off;

	if ((taop = tcp_gettaocache(tp->t_inpcb)) == NULL) {
		taop = &tao_noncached;
		bzero(taop, sizeof(*taop));
	}

	/*
	 * Lop off SYN bit if it has already been sent.  However, if this
	 * is SYN-SENT state and if segment contains data and if we don't
	 * know that foreign host supports TAO, suppress sending segment.
	 */
	if ((flags & TH_SYN) && SEQ_GT(tp->snd_nxt, tp->snd_una)) {
		flags &= ~TH_SYN;
		off--, len++;
		if (len > 0 && tp->t_state == TCPS_SYN_SENT &&
		    taop->tao_ccsent == 0)
			return 0;
	}

	/*
	 * Be careful not to send data and/or FIN on SYN segments
	 * in cases when no CC option will be sent.
	 * This measure is needed to prevent interoperability problems
	 * with not fully conformant TCP implementations.
	 */
	if ((flags & TH_SYN) &&
	    ((tp->t_flags & TF_NOOPT) || !(tp->t_flags & TF_REQ_CC) ||
	     ((flags & TH_ACK) && !(tp->t_flags & TF_RCVD_CC)))) {
		len = 0;
		flags &= ~TH_FIN;
	}

	if (len < 0) {
		/*
		 * If FIN has been sent but not acked,
		 * but we haven't been called to retransmit,
		 * len will be -1.  Otherwise, window shrank
		 * after we sent into it.  If window shrank to 0,
		 * cancel pending retransmit, pull snd_nxt back
		 * to (closed) window, and set the persist timer
		 * if it isn't already going.  If the window didn't
		 * close completely, just wait for an ACK.
		 */
		len = 0;
		if (win == 0) {
			callout_stop(tp->tt_rexmt);
			tp->t_rxtshift = 0;
			tp->snd_nxt = tp->snd_una;
			if (!callout_active(tp->tt_persist))
				tcp_setpersist(tp);
		}
	}
	if (len > tp->t_maxseg) {
		len = tp->t_maxseg;
		sendalot = 1;
	}
	if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
		flags &= ~TH_FIN;

	win = sbspace(&so->so_rcv);

	/*
	 * Sender silly window avoidance.  If connection is idle
	 * and can send all data, a maximum segment,
	 * at least a maximum default-size segment do it,
	 * or are forced, do it; otherwise don't bother.
	 * If peer's buffer is tiny, then send
	 * when window is at least half open.
	 * If retransmitting (possibly after persist timer forced us
	 * to send into a small window), then must resend.
	 */
	if (len) {
		if (len == tp->t_maxseg)
			goto send;
		if (!(tp->t_flags & TF_MORETOCOME) &&
		    (idle || tp->t_flags & TF_NODELAY) &&
		    (tp->t_flags & TF_NOPUSH) == 0 &&
		    len + off >= so->so_snd.sb_cc)
			goto send;
		if (tp->t_force)
			goto send;
		if (len >= tp->max_sndwnd / 2 && tp->max_sndwnd > 0)
			goto send;
		if (SEQ_LT(tp->snd_nxt, tp->snd_max))
			goto send;
	}

	/*
	 * Compare available window to amount of window
	 * known to peer (as advertised window less
	 * next expected input).  If the difference is at least two
	 * max size segments, or at least 50% of the maximum possible
	 * window, then want to send a window update to peer.
	 */
	if (win > 0) {
		/*
		 * "adv" is the amount we can increase the window,
		 * taking into account that we are limited by
		 * TCP_MAXWIN << tp->rcv_scale.
		 */
		long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) -
			(tp->rcv_adv - tp->rcv_nxt);

		if (adv >= (long) (2 * tp->t_maxseg))
			goto send;
		if (2 * adv >= (long) so->so_rcv.sb_hiwat)
			goto send;
	}

	/*
	 * Send if we owe peer an ACK.
	 */
	if (tp->t_flags & TF_ACKNOW)
		goto send;
	if ((flags & TH_RST) ||
	    ((flags & TH_SYN) && (tp->t_flags & TF_NEEDSYN) == 0))
		goto send;
	if (SEQ_GT(tp->snd_up, tp->snd_una))
		goto send;
	/*
	 * If our state indicates that FIN should be sent
	 * and we have not yet done so, or we're retransmitting the FIN,
	 * then we need to send.
	 */
	if (flags & TH_FIN &&
	    ((tp->t_flags & TF_SENTFIN) == 0 || tp->snd_nxt == tp->snd_una))
		goto send;

	/*
	 * TCP window updates are not reliable, rather a polling protocol
	 * using ``persist'' packets is used to insure receipt of window
	 * updates.  The three ``states'' for the output side are:
	 *	idle			not doing retransmits or persists
	 *	persisting		to move a small or zero window
	 *	(re)transmitting	and thereby not persisting
	 *
	 * callout_active(tp->tt_persist)
	 *	is true when we are in persist state.
	 * tp->t_force
	 *	is set when we are called to send a persist packet.
	 * callout_active(tp->tt_rexmt)
	 *	is set when we are retransmitting
	 * The output side is idle when both timers are zero.
	 *
	 * If send window is too small, there is data to transmit, and no
	 * retransmit or persist is pending, then go to persist state.
	 * If nothing happens soon, send when timer expires:
	 * if window is nonzero, transmit what we can,
	 * otherwise force out a byte.
	 */
	if (so->so_snd.sb_cc && !callout_active(tp->tt_rexmt) &&
	    !callout_active(tp->tt_persist)) {
		tp->t_rxtshift = 0;
		tcp_setpersist(tp);
	}

	/*
	 * No reason to send a segment, just return.
	 */
	return (0);

send:
	/*
	 * Before ESTABLISHED, force sending of initial options
	 * unless TCP set not to do any options.
	 * NOTE: we assume that the IP/TCP header plus TCP options
	 * always fit in a single mbuf, leaving room for a maximum
	 * link header, i.e.
	 *	max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MCLBYTES
	 */
	optlen = 0;
#ifdef INET6
	if (isipv6)
		hdrlen = sizeof (struct ip6_hdr) + sizeof (struct tcphdr);
	else
#endif
	hdrlen = sizeof (struct tcpiphdr);
	if (flags & TH_SYN) {
		tp->snd_nxt = tp->iss;
		if ((tp->t_flags & TF_NOOPT) == 0) {
			u_short mss;

			opt[0] = TCPOPT_MAXSEG;
			opt[1] = TCPOLEN_MAXSEG;
			mss = htons((u_short) tcp_mssopt(tp));
			(void)memcpy(opt + 2, &mss, sizeof(mss));
			optlen = TCPOLEN_MAXSEG;

			if ((tp->t_flags & TF_REQ_SCALE) &&
			    ((flags & TH_ACK) == 0 ||
			    (tp->t_flags & TF_RCVD_SCALE))) {
				*((u_int32_t *)(opt + optlen)) = htonl(
					TCPOPT_NOP << 24 |
					TCPOPT_WINDOW << 16 |
					TCPOLEN_WINDOW << 8 |
					tp->request_r_scale);
				optlen += 4;
			}
		}
 	}

 	/*
	 * Send a timestamp and echo-reply if this is a SYN and our side
	 * wants to use timestamps (TF_REQ_TSTMP is set) or both our side
	 * and our peer have sent timestamps in our SYN's.
 	 */
 	if ((tp->t_flags & (TF_REQ_TSTMP|TF_NOOPT)) == TF_REQ_TSTMP &&
 	    (flags & TH_RST) == 0 &&
	    ((flags & TH_ACK) == 0 ||
	     (tp->t_flags & TF_RCVD_TSTMP))) {
		u_int32_t *lp = (u_int32_t *)(opt + optlen);

 		/* Form timestamp option as shown in appendix A of RFC 1323. */
 		*lp++ = htonl(TCPOPT_TSTAMP_HDR);
 		*lp++ = htonl(ticks);
 		*lp   = htonl(tp->ts_recent);
 		optlen += TCPOLEN_TSTAMP_APPA;
 	}

 	/*
	 * Send `CC-family' options if our side wants to use them (TF_REQ_CC),
	 * options are allowed (!TF_NOOPT) and it's not a RST.
 	 */
 	if ((tp->t_flags & (TF_REQ_CC|TF_NOOPT)) == TF_REQ_CC &&
 	     (flags & TH_RST) == 0) {
		switch (flags & (TH_SYN|TH_ACK)) {
		/*
		 * This is a normal ACK, send CC if we received CC before
		 * from our peer.
		 */
		case TH_ACK:
			if (!(tp->t_flags & TF_RCVD_CC))
				break;
			/*FALLTHROUGH*/

		/*
		 * We can only get here in T/TCP's SYN_SENT* state, when
		 * we're a sending a non-SYN segment without waiting for
		 * the ACK of our SYN.  A check above assures that we only
		 * do this if our peer understands T/TCP.
		 */
		case 0:
			opt[optlen++] = TCPOPT_NOP;
			opt[optlen++] = TCPOPT_NOP;
			opt[optlen++] = TCPOPT_CC;
			opt[optlen++] = TCPOLEN_CC;
			*(u_int32_t *)&opt[optlen] = htonl(tp->cc_send);

			optlen += 4;
			break;

		/*
		 * This is our initial SYN, check whether we have to use
		 * CC or CC.new.