[qemu.git] / slirp / tcp_output.c

/*
 * 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. 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
 * tcp_output.c,v 1.3 1994/09/15 10:36:55 davidg Exp
 */

/*
 * Changes and additions relating to SLiRP
 * Copyright (c) 1995 Danny Gasparovski.
 *
 * Please read the file COPYRIGHT for the
 * terms and conditions of the copyright.
 */

#include <slirp.h>

static const u_char  tcp_outflags[TCP_NSTATES] = {
	TH_RST|TH_ACK, 0,      TH_SYN,        TH_SYN|TH_ACK,
	TH_ACK,        TH_ACK, TH_FIN|TH_ACK, TH_FIN|TH_ACK,
	TH_FIN|TH_ACK, TH_ACK, TH_ACK,
};


#define MAX_TCPOPTLEN	32	/* max # bytes that go in options */

/*
 * Tcp output routine: figure out what should be sent and send it.
 */
int
tcp_output(struct tcpcb *tp)
{
	register struct socket *so = tp->t_socket;
	register long len, win;
	int off, flags, error;
	register struct mbuf *m;
	register struct tcpiphdr *ti;
	u_char opt[MAX_TCPOPTLEN];
	unsigned optlen, hdrlen;
	int idle, sendalot;

	DEBUG_CALL("tcp_output");
	DEBUG_ARG("tp = %lx", (long )tp);

	/*
	 * 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:
	sendalot = 0;
	off = tp->snd_nxt - tp->snd_una;
	win = min(tp->snd_wnd, tp->snd_cwnd);

	flags = tcp_outflags[tp->t_state];

	DEBUG_MISC((dfd, " --- tcp_output flags = 0x%x\n",flags));

	/*
	 * 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;
		}
	}

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

	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 ((1 || 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;
	}

	/*
	 * 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_datalen)
			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;

	/*
	 * 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 (struct tcpiphdr) + optlen <= MHLEN
	 */
	optlen = 0;
	hdrlen = sizeof (struct tcpiphdr);
	if (flags & TH_SYN) {
		tp->snd_nxt = tp->iss;
		if ((tp->t_flags & TF_NOOPT) == 0) {
			uint16_t mss;

			opt[0] = TCPOPT_MAXSEG;
			opt[1] = 4;
			mss = htons((uint16_t) tcp_mss(tp, 0));
			memcpy((caddr_t)(opt + 2), (caddr_t)&mss, sizeof(mss));
			optlen = 4;
		}
 	}

 	hdrlen += optlen;

	/*
	 * Adjust data length if insertion of options will
	 * bump the packet length beyond the t_maxseg length.
	 */
	 if (len > tp->t_maxseg - optlen) {
		len = tp->t_maxseg - optlen;
		sendalot = 1;
	 }

	/*
	 * Grab a header mbuf, attaching a copy of data to
	 * be transmitted, and initialize the header from
	 * the template for sends on this connection.
	 */
	if (len) {
		m = m_get(so->slirp);
		if (m == NULL) {
			error = 1;
			goto out;
		}
		m->m_data += IF_MAXLINKHDR;
		m->m_len = hdrlen;

		sbcopy(&so->so_snd, off, (int) len, mtod(m, caddr_t) + hdrlen);
		m->m_len += len;

		/*
		 * If we're sending everything we've got, set PUSH.
		 * (This will keep happy those implementations which only
		 * give data to the user when a buffer fills or
		 * a PUSH comes in.)
		 */
		if (off + len == so->so_snd.sb_cc)
			flags |= TH_PUSH;
	} else {
		m = m_get(so->slirp);
		if (m == NULL) {
			error = 1;
			goto out;
		}
		m->m_data += IF_MAXLINKHDR;
		m->m_len = hdrlen;
	}

	ti = mtod(m, struct tcpiphdr *);

	memcpy((caddr_t)ti, &tp->t_template, sizeof (struct tcpiphdr));

	/*
	 * Fill in fields, remembering maximum advertised
	 * window for use in delaying messages about window sizes.
	 * If resending a FIN, be sure not to use a new sequence number.
	 */
	if (flags & TH_FIN && tp->t_flags & TF_SENTFIN &&
	    tp->snd_nxt == tp->snd_max)
		tp->snd_nxt--;
	/*
	 * If we are doing retransmissions, then snd_nxt will
	 * not reflect the first unsent octet.  For ACK only
	 * packets, we do not want the sequence number of the
	 * retransmitted packet, we want the sequence number
	 * of the next unsent octet.  So, if there is no data
	 * (and no SYN or FIN), use snd_max instead of snd_nxt
	 * when filling in ti_seq.  But if we are in persist
	 * state, snd_max might reflect one byte beyond the
	 * right edge of the window, so use snd_nxt in that
	 * case, since we know we aren't doing a retransmission.
	 * (retransmit and persist are mutually exclusive...)
	 */
	if (len || (flags & (TH_SYN|TH_FIN)) || tp->t_timer[TCPT_PERSIST])
		ti->ti_seq = htonl(tp->snd_nxt);
	else
		ti->ti_seq = htonl(tp->snd_max);
	ti->ti_ack = htonl(tp->rcv_nxt);
	if (optlen) {
		memcpy((caddr_t)(ti + 1), (caddr_t)opt, optlen);
		ti->ti_off = (sizeof (struct tcphdr) + optlen) >> 2;
	}
	ti->ti_flags = flags;
	/*
	 * Calculate receive window.  Don't shrink window,
	 * but avoid silly window syndrome.
	 */
	if (win < (long)(so->so_rcv.sb_datalen / 4) && win < (long)tp->t_maxseg)
		win = 0;
	if (win > (long)TCP_MAXWIN << tp->rcv_scale)
		win = (long)TCP_MAXWIN << tp->rcv_scale;
	if (win < (long)(tp->rcv_adv - tp->rcv_nxt))
		win = (long)(tp->rcv_adv - tp->rcv_nxt);
	ti->ti_win = htons((uint16_t) (win>>tp->rcv_scale));

	if (SEQ_GT(tp->snd_up, tp->snd_una)) {
		ti->ti_urp = htons((uint16_t)(tp->snd_up - ntohl(ti->ti_seq)));
		ti->ti_flags |= TH_URG;
	} else
		/*
		 * If no urgent pointer to send, then we pull
		 * the urgent pointer to the left edge of the send window
		 * so that it doesn't drift into the send window on sequence
		 * number wraparound.
		 */
		tp->snd_up = tp->snd_una;		/* drag it along */

	/*
	 * Put TCP length in extended header, and then
	 * checksum extended header and data.
	 */
	if (len + optlen)
		ti->ti_len = htons((uint16_t)(sizeof (struct tcphdr) +
		    optlen + len));
	ti->ti_sum = cksum(m, (int)(hdrlen + len));

	/*
	 * In transmit state, time the transmission and arrange for
	 * the retransmit.  In persist state, just set snd_max.
	 */
	if (tp->t_force == 0 || tp->t_timer[TCPT_PERSIST] == 0) {
		tcp_seq startseq = tp->snd_nxt;

		/*
		 * Advance snd_nxt over sequence space of this segment.
		 */
		if (flags & (TH_SYN|TH_FIN)) {
			if (flags & TH_SYN)
				tp->snd_nxt++;
			if (flags & TH_FIN) {
				tp->snd_nxt++;
				tp->t_flags |= TF_SENTFIN;
			}
		}
		tp->snd_nxt += len;
		if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
			tp->snd_max = tp->snd_nxt;
			/*
			 * Time this transmission if not a retransmission and
			 * not currently timing anything.
			 */
			if (tp->t_rtt == 0) {
				tp->t_rtt = 1;
				tp->t_rtseq = startseq;
			}
		}

		/*
		 * Set retransmit timer if not currently set,
		 * and not doing an ack or a keep-alive probe.
		 * Initial value for retransmit timer is smoothed
		 * round-trip time + 2 * round-trip time variance.
		 * Initialize shift counter which is used for backoff
		 * of retransmit time.
		 */
		if (tp->t_timer[TCPT_REXMT] == 0 &&
		    tp->snd_nxt != tp->snd_una) {
			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
			if (tp->t_timer[TCPT_PERSIST]) {
				tp->t_timer[TCPT_PERSIST] = 0;
				tp->t_rxtshift = 0;
			}
		}
	} else
		if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
			tp->snd_max = tp->snd_nxt + len;

	/*
	 * Fill in IP length and desired time to live and
	 * send to IP level.  There should be a better way
	 * to handle ttl and tos; we could keep them in
	 * the template, but need a way to checksum without them.
	 */
	m->m_len = hdrlen + len; /* XXX Needed? m_len should be correct */

    {

	((struct ip *)ti)->ip_len = m->m_len;

	((struct ip *)ti)->ip_ttl = IPDEFTTL;
	((struct ip *)ti)->ip_tos = so->so_iptos;

	error = ip_output(so, m);
    }
	if (error) {
out:
		return (error);
	}

	/*
	 * Data sent (as far as we can tell).
	 * If this advertises a larger window than any other segment,
	 * then remember the size of the advertised window.
	 * Any pending ACK has now been sent.
	 */
	if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv))
		tp->rcv_adv = tp->rcv_nxt + win;
	tp->last_ack_sent = tp->rcv_nxt;
	tp->t_flags &= ~(TF_ACKNOW|TF_DELACK);
	if (sendalot)
		goto again;

	return (0);
}

void
tcp_setpersist(struct tcpcb *tp)
{
    int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;

	/*
	 * Start/restart persistence timer.
	 */
	TCPT_RANGESET(tp->t_timer[TCPT_PERSIST],
	    t * tcp_backoff[tp->t_rxtshift],
	    TCPTV_PERSMIN, TCPTV_PERSMAX);
	if (tp->t_rxtshift < TCP_MAXRXTSHIFT)
		tp->t_rxtshift++;
}
Commit	Line	Data
f0cbd3ec FB	1	/*
	2	* Copyright (c) 1982, 1986, 1988, 1990, 1993
	3	* The Regents of the University of California. All rights reserved.
	4	*
	5	* Redistribution and use in source and binary forms, with or without
	6	* modification, are permitted provided that the following conditions
	7	* are met:
	8	* 1. Redistributions of source code must retain the above copyright
	9	* notice, this list of conditions and the following disclaimer.
	10	* 2. Redistributions in binary form must reproduce the above copyright
	11	* notice, this list of conditions and the following disclaimer in the
	12	* documentation and/or other materials provided with the distribution.
2f5f8996	13	* 3. Neither the name of the University nor the names of its contributors
f0cbd3ec FB	14	* may be used to endorse or promote products derived from this software
	15	* without specific prior written permission.
	16	*
	17	* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
	18	* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	19	* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	20	* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
	21	* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	22	* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	23	* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	24	* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	25	* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	26	* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	27	* SUCH DAMAGE.
	28	*
	29	* @(#)tcp_output.c 8.3 (Berkeley) 12/30/93
	30	* tcp_output.c,v 1.3 1994/09/15 10:36:55 davidg Exp
	31	*/
	32
	33	/*
	34	* Changes and additions relating to SLiRP
	35	* Copyright (c) 1995 Danny Gasparovski.
5fafdf24 TS	36	*
5fafdf24 TS	37	* Please read the file COPYRIGHT for the
f0cbd3ec FB	38	* terms and conditions of the copyright.
	39	*/
	40
	41	#include <slirp.h>
	42
9634d903	43	static const u_char tcp_outflags[TCP_NSTATES] = {
f0cbd3ec	44	TH_RST\|TH_ACK, 0, TH_SYN, TH_SYN\|TH_ACK,
5fafdf24	45	TH_ACK, TH_ACK, TH_FIN\|TH_ACK, TH_FIN\|TH_ACK,
f0cbd3ec FB	46	TH_FIN\|TH_ACK, TH_ACK, TH_ACK,
	47	};
	48
	49
	50	#define MAX_TCPOPTLEN 32 /* max # bytes that go in options */
	51
	52	/*
	53	* Tcp output routine: figure out what should be sent and send it.
	54	*/
	55	int
511d2b14	56	tcp_output(struct tcpcb *tp)
f0cbd3ec FB	57	{
	58	register struct socket *so = tp->t_socket;
	59	register long len, win;
	60	int off, flags, error;
	61	register struct mbuf *m;
	62	register struct tcpiphdr *ti;
	63	u_char opt[MAX_TCPOPTLEN];
	64	unsigned optlen, hdrlen;
	65	int idle, sendalot;
5fafdf24	66
f0cbd3ec FB	67	DEBUG_CALL("tcp_output");
f0cbd3ec FB	68	DEBUG_ARG("tp = %lx", (long )tp);
5fafdf24	69
f0cbd3ec FB	70	/*
	71	* Determine length of data that should be transmitted,
	72	* and flags that will be used.
	73	* If there is some data or critical controls (SYN, RST)
	74	* to send, then transmit; otherwise, investigate further.
	75	*/
	76	idle = (tp->snd_max == tp->snd_una);
	77	if (idle && tp->t_idle >= tp->t_rxtcur)
	78	/*
	79	* We have been idle for "a while" and no acks are
	80	* expected to clock out any data we send --
	81	* slow start to get ack "clock" running again.
	82	*/
	83	tp->snd_cwnd = tp->t_maxseg;
	84	again:
	85	sendalot = 0;
	86	off = tp->snd_nxt - tp->snd_una;
	87	win = min(tp->snd_wnd, tp->snd_cwnd);
	88
	89	flags = tcp_outflags[tp->t_state];
5fafdf24	90
f0cbd3ec	91	DEBUG_MISC((dfd, " --- tcp_output flags = 0x%x\n",flags));
5fafdf24	92
f0cbd3ec FB	93	/*
	94	* If in persist timeout with window of 0, send 1 byte.
	95	* Otherwise, if window is small but nonzero
	96	* and timer expired, we will send what we can
	97	* and go to transmit state.
	98	*/
	99	if (tp->t_force) {
	100	if (win == 0) {
	101	/*
	102	* If we still have some data to send, then
	103	* clear the FIN bit. Usually this would
	104	* happen below when it realizes that we
	105	* aren't sending all the data. However,
	106	* if we have exactly 1 byte of unset data,
	107	* then it won't clear the FIN bit below,
	108	* and if we are in persist state, we wind
	109	* up sending the packet without recording
	110	* that we sent the FIN bit.
	111	*
	112	* We can't just blindly clear the FIN bit,
	113	* because if we don't have any more data
	114	* to send then the probe will be the FIN
	115	* itself.
	116	*/
	117	if (off < so->so_snd.sb_cc)
	118	flags &= ~TH_FIN;
	119	win = 1;
	120	} else {
	121	tp->t_timer[TCPT_PERSIST] = 0;
	122	tp->t_rxtshift = 0;
	123	}
	124	}
	125
	126	len = min(so->so_snd.sb_cc, win) - off;
	127
	128	if (len < 0) {
	129	/*
	130	* If FIN has been sent but not acked,
	131	* but we haven't been called to retransmit,
	132	* len will be -1. Otherwise, window shrank
	133	* after we sent into it. If window shrank to 0,
	134	* cancel pending retransmit and pull snd_nxt
	135	* back to (closed) window. We will enter persist
	136	* state below. If the window didn't close completely,
	137	* just wait for an ACK.
	138	*/
	139	len = 0;
	140	if (win == 0) {
	141	tp->t_timer[TCPT_REXMT] = 0;
	142	tp->snd_nxt = tp->snd_una;
	143	}
	144	}
5fafdf24	145
f0cbd3ec FB	146	if (len > tp->t_maxseg) {
	147	len = tp->t_maxseg;
	148	sendalot = 1;
	149	}
	150	if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
	151	flags &= ~TH_FIN;
	152
	153	win = sbspace(&so->so_rcv);
	154
	155	/*
	156	* Sender silly window avoidance. If connection is idle
	157	* and can send all data, a maximum segment,
	158	* at least a maximum default-size segment do it,
	159	* or are forced, do it; otherwise don't bother.
	160	* If peer's buffer is tiny, then send
	161	* when window is at least half open.
	162	* If retransmitting (possibly after persist timer forced us
	163	* to send into a small window), then must resend.
	164	*/
	165	if (len) {
	166	if (len == tp->t_maxseg)
	167	goto send;
	168	if ((1 \|\| idle \|\| tp->t_flags & TF_NODELAY) &&
	169	len + off >= so->so_snd.sb_cc)
	170	goto send;
	171	if (tp->t_force)
	172	goto send;
	173	if (len >= tp->max_sndwnd / 2 && tp->max_sndwnd > 0)
	174	goto send;
	175	if (SEQ_LT(tp->snd_nxt, tp->snd_max))
	176	goto send;
	177	}
	178
	179	/*
	180	* Compare available window to amount of window
	181	* known to peer (as advertised window less
	182	* next expected input). If the difference is at least two
	183	* max size segments, or at least 50% of the maximum possible
	184	* window, then want to send a window update to peer.
	185	*/
	186	if (win > 0) {
5fafdf24	187	/*
f0cbd3ec FB	188	* "adv" is the amount we can increase the window,
	189	* taking into account that we are limited by
	190	* TCP_MAXWIN << tp->rcv_scale.
	191	*/
	192	long adv = min(win, (long)TCP_MAXWIN << tp->rcv_scale) -
	193	(tp->rcv_adv - tp->rcv_nxt);
	194
	195	if (adv >= (long) (2 * tp->t_maxseg))
	196	goto send;
	197	if (2 * adv >= (long) so->so_rcv.sb_datalen)
	198	goto send;
	199	}
	200
	201	/*
	202	* Send if we owe peer an ACK.
	203	*/
	204	if (tp->t_flags & TF_ACKNOW)
	205	goto send;
	206	if (flags & (TH_SYN\|TH_RST))
	207	goto send;
	208	if (SEQ_GT(tp->snd_up, tp->snd_una))
	209	goto send;
	210	/*
	211	* If our state indicates that FIN should be sent
	212	* and we have not yet done so, or we're retransmitting the FIN,
	213	* then we need to send.
	214	*/
	215	if (flags & TH_FIN &&
	216	((tp->t_flags & TF_SENTFIN) == 0 \|\| tp->snd_nxt == tp->snd_una))
	217	goto send;
	218
	219	/*
	220	* TCP window updates are not reliable, rather a polling protocol
	221	* using ``persist'' packets is used to insure receipt of window
	222	* updates. The three ``states'' for the output side are:
	223	* idle not doing retransmits or persists
	224	* persisting to move a small or zero window
	225	* (re)transmitting and thereby not persisting
	226	*
	227	* tp->t_timer[TCPT_PERSIST]
	228	* is set when we are in persist state.
	229	* tp->t_force
	230	* is set when we are called to send a persist packet.
	231	* tp->t_timer[TCPT_REXMT]
	232	* is set when we are retransmitting
	233	* The output side is idle when both timers are zero.
	234	*
	235	* If send window is too small, there is data to transmit, and no
	236	* retransmit or persist is pending, then go to persist state.
	237	* If nothing happens soon, send when timer expires:
	238	* if window is nonzero, transmit what we can,
	239	* otherwise force out a byte.
	240	*/
	241	if (so->so_snd.sb_cc && tp->t_timer[TCPT_REXMT] == 0 &&
	242	tp->t_timer[TCPT_PERSIST] == 0) {
	243	tp->t_rxtshift = 0;
	244	tcp_setpersist(tp);
	245	}
	246
	247	/*
	248	* No reason to send a segment, just return.
	249	*/
f0cbd3ec FB	250	return (0);
	251
	252	send:
	253	/*
	254	* Before ESTABLISHED, force sending of initial options
	255	* unless TCP set not to do any options.
	256	* NOTE: we assume that the IP/TCP header plus TCP options
	257	* always fit in a single mbuf, leaving room for a maximum
	258	* link header, i.e.
	259	* max_linkhdr + sizeof (struct tcpiphdr) + optlen <= MHLEN
	260	*/
	261	optlen = 0;
	262	hdrlen = sizeof (struct tcpiphdr);
	263	if (flags & TH_SYN) {
	264	tp->snd_nxt = tp->iss;
	265	if ((tp->t_flags & TF_NOOPT) == 0) {
b6dce92e	266	uint16_t mss;
f0cbd3ec FB	267
	268	opt[0] = TCPOPT_MAXSEG;
	269	opt[1] = 4;
b6dce92e	270	mss = htons((uint16_t) tcp_mss(tp, 0));
f0cbd3ec FB	271	memcpy((caddr_t)(opt + 2), (caddr_t)&mss, sizeof(mss));
f0cbd3ec FB	272	optlen = 4;
f0cbd3ec FB	273	}
f0cbd3ec FB	274	}
5fafdf24	275
f0cbd3ec	276	hdrlen += optlen;
5fafdf24	277
f0cbd3ec FB	278	/*
	279	* Adjust data length if insertion of options will
	280	* bump the packet length beyond the t_maxseg length.
	281	*/
	282	if (len > tp->t_maxseg - optlen) {
	283	len = tp->t_maxseg - optlen;
	284	sendalot = 1;
	285	}
	286
	287	/*
	288	* Grab a header mbuf, attaching a copy of data to
	289	* be transmitted, and initialize the header from
	290	* the template for sends on this connection.
	291	*/
	292	if (len) {
460fec67	293	m = m_get(so->slirp);
f0cbd3ec	294	if (m == NULL) {
f0cbd3ec FB	295	error = 1;
	296	goto out;
	297	}
9634d903	298	m->m_data += IF_MAXLINKHDR;
f0cbd3ec	299	m->m_len = hdrlen;
3b46e624	300
0d62c4cf JK	301	sbcopy(&so->so_snd, off, (int) len, mtod(m, caddr_t) + hdrlen);
0d62c4cf JK	302	m->m_len += len;
f0cbd3ec	303
f0cbd3ec FB	304	/*
	305	* If we're sending everything we've got, set PUSH.
	306	* (This will keep happy those implementations which only
	307	* give data to the user when a buffer fills or
	308	* a PUSH comes in.)
	309	*/
	310	if (off + len == so->so_snd.sb_cc)
	311	flags \|= TH_PUSH;
	312	} else {
460fec67	313	m = m_get(so->slirp);
f0cbd3ec	314	if (m == NULL) {
f0cbd3ec FB	315	error = 1;
	316	goto out;
	317	}
9634d903	318	m->m_data += IF_MAXLINKHDR;
f0cbd3ec FB	319	m->m_len = hdrlen;
	320	}
	321
	322	ti = mtod(m, struct tcpiphdr *);
5fafdf24	323
f0cbd3ec FB	324	memcpy((caddr_t)ti, &tp->t_template, sizeof (struct tcpiphdr));
	325
	326	/*
	327	* Fill in fields, remembering maximum advertised
	328	* window for use in delaying messages about window sizes.
	329	* If resending a FIN, be sure not to use a new sequence number.
	330	*/
5fafdf24	331	if (flags & TH_FIN && tp->t_flags & TF_SENTFIN &&
f0cbd3ec FB	332	tp->snd_nxt == tp->snd_max)
	333	tp->snd_nxt--;
	334	/*
	335	* If we are doing retransmissions, then snd_nxt will
	336	* not reflect the first unsent octet. For ACK only
	337	* packets, we do not want the sequence number of the
	338	* retransmitted packet, we want the sequence number
	339	* of the next unsent octet. So, if there is no data
	340	* (and no SYN or FIN), use snd_max instead of snd_nxt
	341	* when filling in ti_seq. But if we are in persist
	342	* state, snd_max might reflect one byte beyond the
	343	* right edge of the window, so use snd_nxt in that
	344	* case, since we know we aren't doing a retransmission.
	345	* (retransmit and persist are mutually exclusive...)
	346	*/
	347	if (len \|\| (flags & (TH_SYN\|TH_FIN)) \|\| tp->t_timer[TCPT_PERSIST])
	348	ti->ti_seq = htonl(tp->snd_nxt);
	349	else
	350	ti->ti_seq = htonl(tp->snd_max);
	351	ti->ti_ack = htonl(tp->rcv_nxt);
	352	if (optlen) {
	353	memcpy((caddr_t)(ti + 1), (caddr_t)opt, optlen);
	354	ti->ti_off = (sizeof (struct tcphdr) + optlen) >> 2;
	355	}
	356	ti->ti_flags = flags;
	357	/*
	358	* Calculate receive window. Don't shrink window,
	359	* but avoid silly window syndrome.
	360	*/
	361	if (win < (long)(so->so_rcv.sb_datalen / 4) && win < (long)tp->t_maxseg)
	362	win = 0;
	363	if (win > (long)TCP_MAXWIN << tp->rcv_scale)
	364	win = (long)TCP_MAXWIN << tp->rcv_scale;
	365	if (win < (long)(tp->rcv_adv - tp->rcv_nxt))
	366	win = (long)(tp->rcv_adv - tp->rcv_nxt);
b6dce92e	367	ti->ti_win = htons((uint16_t) (win>>tp->rcv_scale));
5fafdf24	368
f0cbd3ec	369	if (SEQ_GT(tp->snd_up, tp->snd_una)) {
b6dce92e	370	ti->ti_urp = htons((uint16_t)(tp->snd_up - ntohl(ti->ti_seq)));
f0cbd3ec FB	371	ti->ti_flags \|= TH_URG;
	372	} else
	373	/*
	374	* If no urgent pointer to send, then we pull
	375	* the urgent pointer to the left edge of the send window
	376	* so that it doesn't drift into the send window on sequence
	377	* number wraparound.
	378	*/
	379	tp->snd_up = tp->snd_una; /* drag it along */
	380
	381	/*
	382	* Put TCP length in extended header, and then
	383	* checksum extended header and data.
	384	*/
	385	if (len + optlen)
b6dce92e	386	ti->ti_len = htons((uint16_t)(sizeof (struct tcphdr) +
f0cbd3ec FB	387	optlen + len));
	388	ti->ti_sum = cksum(m, (int)(hdrlen + len));
	389
	390	/*
	391	* In transmit state, time the transmission and arrange for
	392	* the retransmit. In persist state, just set snd_max.
	393	*/
	394	if (tp->t_force == 0 \|\| tp->t_timer[TCPT_PERSIST] == 0) {
	395	tcp_seq startseq = tp->snd_nxt;
	396
	397	/*
	398	* Advance snd_nxt over sequence space of this segment.
	399	*/
	400	if (flags & (TH_SYN\|TH_FIN)) {
	401	if (flags & TH_SYN)
	402	tp->snd_nxt++;
	403	if (flags & TH_FIN) {
	404	tp->snd_nxt++;
	405	tp->t_flags \|= TF_SENTFIN;
	406	}
	407	}
	408	tp->snd_nxt += len;
	409	if (SEQ_GT(tp->snd_nxt, tp->snd_max)) {
	410	tp->snd_max = tp->snd_nxt;
	411	/*
	412	* Time this transmission if not a retransmission and
	413	* not currently timing anything.
	414	*/
	415	if (tp->t_rtt == 0) {
	416	tp->t_rtt = 1;
	417	tp->t_rtseq = startseq;
f0cbd3ec FB	418	}
	419	}
	420
	421	/*
	422	* Set retransmit timer if not currently set,
	423	* and not doing an ack or a keep-alive probe.
	424	* Initial value for retransmit timer is smoothed
	425	* round-trip time + 2 * round-trip time variance.
	426	* Initialize shift counter which is used for backoff
	427	* of retransmit time.
	428	*/
	429	if (tp->t_timer[TCPT_REXMT] == 0 &&
	430	tp->snd_nxt != tp->snd_una) {
	431	tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
	432	if (tp->t_timer[TCPT_PERSIST]) {
	433	tp->t_timer[TCPT_PERSIST] = 0;
	434	tp->t_rxtshift = 0;
	435	}
	436	}
	437	} else
	438	if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
	439	tp->snd_max = tp->snd_nxt + len;
	440
	441	/*
	442	* Fill in IP length and desired time to live and
	443	* send to IP level. There should be a better way
	444	* to handle ttl and tos; we could keep them in
	445	* the template, but need a way to checksum without them.
	446	*/
	447	m->m_len = hdrlen + len; /* XXX Needed? m_len should be correct */
5fafdf24	448
f0cbd3ec	449	{
3b46e624	450
f0cbd3ec FB	451	((struct ip *)ti)->ip_len = m->m_len;
f0cbd3ec FB	452
9634d903	453	((struct ip *)ti)->ip_ttl = IPDEFTTL;
f0cbd3ec	454	((struct ip *)ti)->ip_tos = so->so_iptos;
3b46e624	455
f0cbd3ec	456	error = ip_output(so, m);
f0cbd3ec FB	457	}
	458	if (error) {
	459	out:
f0cbd3ec FB	460	return (error);
f0cbd3ec FB	461	}
f0cbd3ec FB	462
	463	/*
	464	* Data sent (as far as we can tell).
	465	* If this advertises a larger window than any other segment,
	466	* then remember the size of the advertised window.
	467	* Any pending ACK has now been sent.
	468	*/
	469	if (win > 0 && SEQ_GT(tp->rcv_nxt+win, tp->rcv_adv))
	470	tp->rcv_adv = tp->rcv_nxt + win;
	471	tp->last_ack_sent = tp->rcv_nxt;
	472	tp->t_flags &= ~(TF_ACKNOW\|TF_DELACK);
	473	if (sendalot)
	474	goto again;
	475
	476	return (0);
	477	}
	478
	479	void
511d2b14	480	tcp_setpersist(struct tcpcb *tp)
f0cbd3ec FB	481	{
	482	int t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;
	483
f0cbd3ec FB	484	/*
	485	* Start/restart persistence timer.
	486	*/
	487	TCPT_RANGESET(tp->t_timer[TCPT_PERSIST],
	488	t * tcp_backoff[tp->t_rxtshift],
	489	TCPTV_PERSMIN, TCPTV_PERSMAX);
	490	if (tp->t_rxtshift < TCP_MAXRXTSHIFT)
	491	tp->t_rxtshift++;
	492	}