tcp_output.c

MIPS处理器的bootloader,龙芯就是用的修改过的PMON2

/*	$OpenBSD: tcp_output.c,v 1.30 2000/02/21 21:42:13 provos Exp $	*/
/*	$NetBSD: tcp_output.c,v 1.16 1997/06/03 16:17:09 kml Exp $	*/

/*
 * Copyright (c) 1982, 1986, 1988, 1990, 1993
 *	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	8.3 (Berkeley) 12/30/93
 */

/*
%%% portions-copyright-nrl-95
Portions of this software are Copyright 1995-1998 by Randall Atkinson,
Ronald Lee, Daniel McDonald, Bao Phan, and Chris Winters. All Rights
Reserved. All rights under this copyright have been assigned to the US
Naval Research Laboratory (NRL). The NRL Copyright Notice and License
Agreement Version 1.1 (January 17, 1995) applies to these portions of the
software.
You should have received a copy of the license with this software. If you
didn't get a copy, you may request one from <[e-mail]>.
*/

#include <sys/param.h>
#include <sys/systm.h>
#include <sys/malloc.h>
#include <sys/mbuf.h>
#include <sys/protosw.h>
#include <sys/socket.h>
#include <sys/socketvar.h>
#include <sys/errno.h>
#include <sys/domain.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>
#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>
#include <netinet/tcp_debug.h>

#ifdef TUBA
#include <netiso/iso.h>
#include <netiso/tuba_table.h>
#endif

#ifdef INET6
#include <netinet6/tcpipv6.h>
#endif /* INET6 */

#ifdef TCP_SIGNATURE
#include <sys/md5k.h>
#endif /* TCP_SIGNATURE */

#ifdef PMON
#include <pmon.h>
#endif /* PMON */

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

#ifdef TCP_SACK
extern int tcprexmtthresh;
#endif

#ifdef TCP_SACK
#ifdef TCP_SACK_DEBUG
void
tcp_print_holes(tp)
struct tcpcb *tp;
{
	struct sackhole *p = tp->snd_holes;
	if (p == 0)
		return;
	printf("Hole report: start--end dups rxmit\n");
	while (p) {
		printf("%x--%x d %d r %x\n",  p->start, p->end, p->dups,
                    p->rxmit);
		p = p->next;
	}
	printf("\n");
}
#endif /* TCP_SACK_DEBUG */

/*
 * Returns pointer to a sackhole if there are any pending retransmissions;
 * NULL otherwise.
 */
struct sackhole *
tcp_sack_output(tp)
register struct tcpcb *tp;
{
	struct sackhole *p;
	if (tp->sack_disable)
		return 0;
	p = tp->snd_holes;
	while (p) {
		if (p->dups >= tcprexmtthresh && SEQ_LT(p->rxmit, p->end)) {
			if (SEQ_LT(p->rxmit, tp->snd_una)) {/* old SACK hole */
				p = p->next;
				continue;
			}
#ifdef TCP_SACK_DEBUG
			if (p)
				tcp_print_holes(tp);
#endif
			return p;
		}
        	p = p->next;
	}
	return 0;
}

/*
 * After a timeout, the SACK list may be rebuilt.  This SACK information
 * should be used to avoid retransmitting SACKed data.  This function
 * traverses the SACK list to see if snd_nxt should be moved forward.
 */
void
tcp_sack_adjust(tp)
	struct tcpcb *tp;
{
	struct sackhole *cur = tp->snd_holes;
	if (cur == 0)
		return; /* No holes */
	if (SEQ_GEQ(tp->snd_nxt, tp->rcv_lastsack))
		return; /* We're already beyond any SACKed blocks */
	/* 
	 * Two cases for which we want to advance snd_nxt:  
	 * i) snd_nxt lies between end of one hole and beginning of another
	 * ii) snd_nxt lies between end of last hole and rcv_lastsack
	 */
	while (cur->next) {
		if (SEQ_LT(tp->snd_nxt, cur->end))
			return;
		if (SEQ_GEQ(tp->snd_nxt, cur->next->start)) 
			cur = cur->next;
		else {
			tp->snd_nxt = cur->next->start;
			return;
		}
	}
	if (SEQ_LT(tp->snd_nxt, cur->end))
		return;
	tp->snd_nxt = tp->rcv_lastsack;
	return;
}
#endif /* TCP_SACK */

/*
 * 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;
	register struct tcphdr *th;
	u_char opt[MAX_TCPOPTLEN];
	unsigned int optlen, hdrlen;
	int idle, sendalot = 0;
#ifdef TCP_SACK
	int i, sack_rxmit = 0;
	struct sackhole *p;
#endif
#if defined(TCP_SACK)
	int maxburst = TCP_MAXBURST;
#endif
#ifdef TCP_SIGNATURE
	unsigned int sigoff;
#endif /* TCP_SIGNATURE */

#if defined(TCP_SACK) && defined(TCP_SIGNATURE) && defined(DIAGNOSTIC)
	if (!tp->sack_disable && (tp->t_flags & TF_SIGNATURE))
		return (EINVAL);
#endif /* defined(TCP_SACK) && defined(TCP_SIGNATURE) && defined(DIAGNOSTIC) */

	/*
	 * 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 && tp->t_idle >= 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.
		 */
		tp->snd_cwnd = tp->t_maxseg;
again:
#ifdef TCP_SACK
	/*
	 * If we've recently taken a timeout, snd_max will be greater than
	 * snd_nxt.  There may be SACK information that allows us to avoid
	 * resending already delivered data.  Adjust snd_nxt accordingly.
	 */
	if (!tp->sack_disable && SEQ_LT(tp->snd_nxt, tp->snd_max))
		tcp_sack_adjust(tp);
#endif
	off = tp->snd_nxt - tp->snd_una;
	win = ulmin(tp->snd_wnd, tp->snd_cwnd);

	flags = tcp_outflags[tp->t_state];

#ifdef TCP_SACK
	/* 
	 * Send any SACK-generated retransmissions.  If we're explicitly trying
	 * to send out new data (when sendalot is 1), bypass this function.
	 * If we retransmit in fast recovery mode, decrement snd_cwnd, since
	 * we're replacing a (future) new transmission with a retransmission 
	 * now, and we previously incremented snd_cwnd in tcp_input().
	 */
	if (!tp->sack_disable && !sendalot) {
		if ((p = tcp_sack_output(tp))) {
			off = p->rxmit - tp->snd_una;
			sack_rxmit = 1;
#if 0
			/* Coalesce holes into a single retransmission */
#endif
			len = min(tp->t_maxseg, p->end - p->rxmit);
#ifndef TCP_FACK
			/* in FACK, hold snd_cwnd constant during recovery */
			if (SEQ_LT(tp->snd_una, tp->snd_last))
				tp->snd_cwnd -= tp->t_maxseg;
#endif
    		}
	}
#endif /* TCP_SACK */

	sendalot = 0;
	/*
	 * 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 unset 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 {
			tp->t_timer[TCPT_PERSIST] = 0;
			tp->t_rxtshift = 0;
		}
	}

#ifdef TCP_SACK
	if (!sack_rxmit) {
#endif
	len = ulmin(so->so_snd.sb_cc, win) - off;

#if defined(TCP_SACK) && defined(TCP_FACK)
	/* 
	 * If we're in fast recovery (SEQ_GT(tp->snd_last, tp->snd_una)), and 
	 * amount of outstanding data (snd_awnd) is >= snd_cwnd, then
	 * do not send data (like zero window conditions)
	 */
	if (!tp->sack_disable && len && SEQ_GT(tp->snd_last, tp->snd_una) && 
	    (tp->snd_awnd >= tp->snd_cwnd)) 
		len = 0;
#endif /* TCP_FACK */
#ifdef TCP_SACK
	}
#endif

	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 and pull snd_nxt
		 * back to (closed) window.  We will enter persist
		 * state below.  If the window didn't close completely,
		 * just wait for an ACK.
		 */
		len = 0;
		if (win == 0) {
			tp->t_timer[TCPT_REXMT] = 0;
			tp->snd_nxt = tp->snd_una;
		}
	}
	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 ((idle || tp->t_flags & TF_NODELAY) &&
		    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;
#ifdef TCP_SACK
		if (sack_rxmit)
			goto send;
#endif
	}

	/*
	 * 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 = lmin(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_SYN|TH_RST))
		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;
#ifdef TCP_SACK
	/*
	 * In SACK, it is possible for tcp_output to fail to send a segment 
	 * after the retransmission timer has been turned off.  Make sure
	 * that the retransmission timer is set.
	 */
	if (SEQ_GT(tp->snd_max, tp->snd_una) &&
	    tp->t_timer[TCPT_REXMT] == 0 &&
	    tp->t_timer[TCPT_PERSIST] == 0) {
		tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
		return (0);
	}
#endif /* TCP_SACK */

	/*
	 * 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
	 *
	 * tp->t_timer[TCPT_PERSIST]
	 *	is set when we are in persist state.
	 * tp->t_force
	 *	is set when we are called to send a persist packet.
	 * tp->t_timer[TCPT_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 && tp->t_timer[TCPT_REXMT] == 0 &&
	    tp->t_timer[TCPT_PERSIST] == 0) {
		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(network header) + sizeof(struct tcphdr +
	 * 		optlen <= MHLEN
	 */
	optlen = 0;

	switch (tp->pf) {
	case 0:	/*default to PF_INET*/
#ifdef INET
	case PF_INET:
		hdrlen = sizeof(struct ip) + sizeof(struct tcphdr);
		break;
#endif /* INET */
#ifdef INET6
	case PF_INET6:
		hdrlen = sizeof(struct ip6_hdr) + sizeof(struct tcphdr);
		break;
#endif /* INET6 */
	default:
		return (EPFNOSUPPORT);
	}

	if (flags & TH_SYN) {
		tp->snd_nxt = tp->iss;
		if ((tp->t_flags & TF_NOOPT) == 0) {
			u_int16_t mss;

			opt[0] = TCPOPT_MAXSEG;
			opt[1] = 4;
			mss = htons((u_int16_t) tcp_mss(tp, 0));
			bcopy((caddr_t)&mss, (caddr_t)(opt + 2), sizeof(mss));
			optlen = 4;
#ifdef TCP_SACK
			/* 
			 * If this is the first SYN of connection (not a SYN 
			 * ACK), include SACK_PERMIT_HDR option.  If this is a 
			 * SYN ACK, include SACK_PERMIT_HDR option if peer has 
			 * already done so.
			 */
			if (!tp->sack_disable && ((flags & TH_ACK) == 0 ||
			    (tp->t_flags & TF_SACK_PERMIT))) {
				*((u_int32_t *) (opt + optlen)) =
				    htonl(TCPOPT_SACK_PERMIT_HDR);
				optlen += 4;
			}
#endif
	 
			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_SYN|TH_ACK)) == TH_SYN ||
	     (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(tcp_now);
		*lp   = htonl(tp->ts_recent);
		optlen += TCPOLEN_TSTAMP_APPA;
	}

#ifdef TCP_SIGNATURE
	if (tp->t_flags & TF_SIGNATURE) {
		u_int8_t *bp = (u_int8_t *)(opt + optlen);

		/* Send signature option */
		*(bp++) = TCPOPT_SIGNATURE;
		*(bp++) = TCPOLEN_SIGNATURE;