X-Git-Url: https://git.subgeniuskitty.com/unix-history/.git/blobdiff_plain/02c1608bc36b1ae9b88347a4e120d7be4f3334df..05db08ca40dcf52f15be693f25a96a593409f690:/usr/src/sys/netinet/tcp_output.c

diff --git a/usr/src/sys/netinet/tcp_output.c b/usr/src/sys/netinet/tcp_output.c
index 64c3f2b3f4..4d150a48a6 100644
--- a/usr/src/sys/netinet/tcp_output.c
+++ b/usr/src/sys/netinet/tcp_output.c
@@ -1,39 +1,39 @@
-/*	tcp_output.c	4.43	82/08/02	*/
+/*
+ * Copyright (c) 1982, 1986 Regents of the University of California.
+ * All rights reserved.  The Berkeley software License Agreement
+ * specifies the terms and conditions for redistribution.
+ *
+ *	@(#)tcp_output.c	7.10 (Berkeley) %G%
+ */
+
+#include "param.h"
+#include "systm.h"
+#include "mbuf.h"
+#include "protosw.h"
+#include "socket.h"
+#include "socketvar.h"
+#include "errno.h"
 
-#include "../h/param.h"
-#include "../h/systm.h"
-#include "../h/mbuf.h"
-#include "../h/protosw.h"
-#include "../h/socket.h"
-#include "../h/socketvar.h"
-#include "../net/in.h"
 #include "../net/route.h"
-#include "../net/in_pcb.h"
-#include "../net/in_systm.h"
-#include "../net/ip.h"
-#include "../net/ip_var.h"
-#include "../net/tcp.h"
-#define	TCPOUTFLAGS
-#include "../net/tcp_fsm.h"
-#include "../net/tcp_seq.h"
-#include "../net/tcp_timer.h"
-#include "../net/tcp_var.h"
-#include "../net/tcpip.h"
-#include "../net/tcp_debug.h"
-#include <errno.h>
 
-char *tcpstates[]; /* XXX */
+#include "in.h"
+#include "in_pcb.h"
+#include "in_systm.h"
+#include "ip.h"
+#include "ip_var.h"
+#include "tcp.h"
+#define	TCPOUTFLAGS
+#include "tcp_fsm.h"
+#include "tcp_seq.h"
+#include "tcp_timer.h"
+#include "tcp_var.h"
+#include "tcpip.h"
+#include "tcp_debug.h"
 
 /*
- * Initial options: indicate max segment length 1/2 of space
- * allocated for receive; if TCPTRUEOOB is defined, indicate
- * willingness to do true out-of-band.
+ * Initial options.
  */
-#ifndef TCPTRUEOOB
 u_char	tcp_initopt[4] = { TCPOPT_MAXSEG, 4, 0x0, 0x0, };
-#else
-u_char	tcp_initopt[6] = { TCPOPT_MAXSEG, 4, 0x0, 0x0, TCPOPT_WILLOOB, 2 };
-#endif
 
 /*
  * Tcp output routine: figure out what should be sent and send it.
@@ -42,15 +42,14 @@ tcp_output(tp)
 	register struct tcpcb *tp;
 {
 	register struct socket *so = tp->t_inpcb->inp_socket;
-	register int len;
+	register int len, win;
 	struct mbuf *m0;
-	int off, flags, win, error;
+	int off, flags, error;
 	register struct mbuf *m;
 	register struct tcpiphdr *ti;
 	u_char *opt;
 	unsigned optlen = 0;
-	int sendalot;
-
+	int idle, sendalot;
 
 	/*
 	 * Determine length of data that should be transmitted,
@@ -58,70 +57,126 @@ tcp_output(tp)
 	 * If there is some data or critical controls (SYN, RST)
 	 * to send, then transmit; otherwise, investigate further.
 	 */
+	idle = (tp->snd_max == tp->snd_una);
 again:
 	sendalot = 0;
 	off = tp->snd_nxt - tp->snd_una;
-	len = MIN(so->so_snd.sb_cc, tp->snd_wnd+tp->t_force) - off;
-	if (len < 0)
-		return (0);	/* ??? */	/* past FIN */
+	win = MIN(tp->snd_wnd, tp->snd_cwnd);
+
+	/*
+	 * 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)
+			win = 1;
+		else {
+			tp->t_timer[TCPT_PERSIST] = 0;
+			tp->t_rxtshift = 0;
+		}
+	}
+
+	len = MIN(so->so_snd.sb_cc, win) - off;
+	flags = tcp_outflags[tp->t_state];
+
+	if (len < 0) {
+		/*
+		 * If FIN has been sent but not acked,
+		 * but we haven't been called to retransmit,
+		 * len will be -1; transmit if acking, otherwise no need.
+		 * 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.
+		 */
+		if (flags & TH_FIN) {
+			if (tp->t_flags & TF_ACKNOW)
+				len = 0;
+			else
+				return (0);
+		} else if (win == 0) {
+			tp->t_timer[TCPT_REXMT] = 0;
+			tp->snd_nxt = tp->snd_una;
+			len = 0;
+		} else
+			return (0);
+	}
 	if (len > tp->t_maxseg) {
 		len = tp->t_maxseg;
 		sendalot = 1;
 	}
-
-	flags = tcp_outflags[tp->t_state];
-	if (tp->snd_nxt + len < tp->snd_una + so->so_snd.sb_cc)
+	if (SEQ_LT(tp->snd_nxt + len, tp->snd_una + so->so_snd.sb_cc))
 		flags &= ~TH_FIN;
-	if (flags & (TH_SYN|TH_RST|TH_FIN))
+	win = sbspace(&so->so_rcv);
+
+
+	/*
+	 * 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;
+	/*
+	 * 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;
 
 	/*
-	 * Sender silly window avoidance.  If can send all data,
-	 * a maximum segment, at least 1/4 of window do it,
+	 * 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 || off+len >= so->so_snd.sb_cc)
+		if (len == tp->t_maxseg)
 			goto send;
-		if (len * 4 >= tp->snd_wnd)		/* a lot */
+		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)
+			goto send;
+		if (SEQ_LT(tp->snd_nxt, tp->snd_max))
+			goto send;
 	}
 
 	/*
-	 * Send if we owe peer an ACK.
+	 * Compare available window to amount of window
+	 * known to peer (as advertised window less
+	 * next expected input.)  If the difference is 35% or more of the
+	 * maximum possible window, then want to send a window update to peer.
 	 */
-	if (tp->t_flags&TF_ACKNOW)
-		goto send;
-
-#ifdef TCPTRUEOOB
-	/*
-	 * Send if an out of band data or ack should be transmitted.
-	 */
-	if (tp->t_oobflags&(TCPOOB_OWEACK|TCPOOB_NEEDACK)))
-		goto send;
-#endif
+	if (win > 0) {
+		int adv = win - (tp->rcv_adv - tp->rcv_nxt);
 
-	/*
-	 * Calculate available window in i, and also amount
-	 * of window known to peer (as advertised window less
-	 * next expected input.)  If this is 35% or more of the
-	 * maximum possible window, then want to send a segment to peer.
-	 */
-	win = sbspace(&so->so_rcv);
-	if (win > 0 &&
-	    ((100*(win-(tp->rcv_adv-tp->rcv_nxt))/so->so_rcv.sb_hiwat) >= 35))
-		goto send;
+		if (100 * adv / so->so_rcv.sb_hiwat >= 35)
+			goto send;
+		if (adv >= 2 * tp->t_maxseg && so->so_rcv.sb_cc == 0)
+			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 zero window
+	 *	persisting		to move a small or zero window
 	 *	(re)transmitting	and thereby not persisting
 	 *
 	 * tp->t_timer[TCPT_PERSIST]
@@ -132,13 +187,14 @@ again:
 	 *	is set when we are retransmitting
 	 * The output side is idle when both timers are zero.
 	 *
-	 * If send window is closed, there is data to transmit, and no
-	 * retransmit or persist is pending, then go to persist state,
-	 * arranging to force out a byte to get more current window information
-	 * if nothing happens soon.
+	 * 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 (tp->snd_wnd == 0 && so->so_snd.sb_cc &&
-	    tp->t_timer[TCPT_REXMT] == 0 && tp->t_timer[TCPT_PERSIST] == 0) {
+	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);
 	}
@@ -154,16 +210,33 @@ send:
 	 * be transmitted, and initialize the header from
 	 * the template for sends on this connection.
 	 */
-	MGET(m, 0);
-	if (m == 0)
+	MGET(m, M_DONTWAIT, MT_HEADER);
+	if (m == NULL)
 		return (ENOBUFS);
 	m->m_off = MMAXOFF - sizeof (struct tcpiphdr);
 	m->m_len = sizeof (struct tcpiphdr);
 	if (len) {
+		if (tp->t_force && len == 1)
+			tcpstat.tcps_sndprobe++;
+		else if (SEQ_LT(tp->snd_nxt, tp->snd_max)) {
+			tcpstat.tcps_sndrexmitpack++;
+			tcpstat.tcps_sndrexmitbyte += len;
+		} else {
+			tcpstat.tcps_sndpack++;
+			tcpstat.tcps_sndbyte += len;
+		}
 		m->m_next = m_copy(so->so_snd.sb_mb, off, len);
 		if (m->m_next == 0)
 			len = 0;
-	}
+	} else if (tp->t_flags & TF_ACKNOW)
+		tcpstat.tcps_sndacks++;
+	else if (flags & (TH_SYN|TH_FIN|TH_RST))
+		tcpstat.tcps_sndctrl++;
+	else if (SEQ_GT(tp->snd_up, tp->snd_una))
+		tcpstat.tcps_sndurg++;
+	else
+		tcpstat.tcps_sndwinup++;
+
 	ti = mtod(m, struct tcpiphdr *);
 	if (tp->t_template == 0)
 		panic("tcp_output");
@@ -172,40 +245,31 @@ send:
 	/*
 	 * 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.
 	 */
-	ti->ti_seq = tp->snd_nxt;
-	ti->ti_ack = tp->rcv_nxt;
-#if vax
-	ti->ti_seq = htonl(ti->ti_seq);
-	ti->ti_ack = htonl(ti->ti_ack);
-#endif
+	if (flags & TH_FIN && tp->t_flags & TF_SENTFIN && 
+	    tp->snd_nxt == tp->snd_max)
+		tp->snd_nxt--;
+	ti->ti_seq = htonl(tp->snd_nxt);
+	ti->ti_ack = htonl(tp->rcv_nxt);
 	/*
 	 * Before ESTABLISHED, force sending of initial options
 	 * unless TCP set to not do any options.
 	 */
-	if (tp->t_state < TCPS_ESTABLISHED) {
-		if (tp->t_flags&TF_NOOPT)
-			goto noopt;
-		opt = tcp_initopt;
-		optlen = sizeof (tcp_initopt);
-		*(u_short *)(opt + 2) = so->so_rcv.sb_hiwat / 2;
-#if vax
-		*(u_short *)(opt + 2) = htons(*(u_short *)(opt + 2));
-#endif
-	} else {
-		if (tp->t_tcpopt == 0)
-			goto noopt;
-		opt = mtod(tp->t_tcpopt, u_char *);
-		optlen = tp->t_tcpopt->m_len;
+	opt = NULL;
+	if (flags & TH_SYN && (tp->t_flags & TF_NOOPT) == 0) {
+		u_short mss;
+
+		mss = MIN(so->so_rcv.sb_hiwat / 2, tcp_mss(tp));
+		if (mss > IP_MSS - sizeof(struct tcpiphdr)) {
+			opt = tcp_initopt;
+			optlen = sizeof (tcp_initopt);
+			*(u_short *)(opt + 2) = htons(mss);
+		}
 	}
-#ifndef TCPTRUEOOB
-	if (opt)
-#else
-	if (opt || (tp->t_oobflags&(TCPOOB_OWEACK|TCPOOB_NEEDACK)))
-#endif
-	{
+	if (opt) {
 		m0 = m->m_next;
-		m->m_next = m_get(M_DONTWAIT);
+		m->m_next = m_get(M_DONTWAIT, MT_DATA);
 		if (m->m_next == 0) {
 			(void) m_free(m);
 			m_freem(m0);
@@ -213,35 +277,9 @@ send:
 		}
 		m->m_next->m_next = m0;
 		m0 = m->m_next;
-		m0->m_off = MMINOFF;
 		m0->m_len = optlen;
 		bcopy((caddr_t)opt, mtod(m0, caddr_t), optlen);
 		opt = (u_char *)(mtod(m0, caddr_t) + optlen);
-#ifdef TCPTRUEOOB
-		if (tp->t_oobflags&TCPOOB_OWEACK) {
-printf("tp %x send OOBACK for %x\n", tp->t_iobseq);
-			*opt++ = TCPOPT_OOBACK;
-			*opt++ = 3;
-			*opt++ = tp->t_iobseq;
-			m0->m_len += 3;
-			tp->t_oobflags &= ~TCPOOB_OWEACK;
-			/* sender should rexmt oob to force ack repeat */
-		}
-		if (tp->t_oobflags&TCPOOB_NEEDACK) {
-printf("tp %x send OOBDATA seq %x data %x\n", tp->t_oobseq, tp->t_oobc);
-			*opt++ = TCPOPT_OOBDATA;
-			*opt++ = 8;
-			*opt++ = tp->t_oobseq;
-			*opt++ = tp->t_oobc;
-			*(tcp_seq *)opt = tp->t_oobmark - tp->snd_nxt;
-#ifdef vax
-			*(tcp_seq *)opt = htonl((unsigned)*(tcp_seq *)opt);
-#endif
-			m0->m_len += 8;
-			TCPT_RANGESET(tp->t_timer[TCPT_OOBREXMT],
-			    tcp_beta * tp->t_srtt, TCPTV_MIN, TCPTV_MAX);
-		}
-#endif
 		while (m0->m_len & 0x3) {
 			*opt++ = TCPOPT_EOL;
 			m0->m_len++;
@@ -249,22 +287,18 @@ printf("tp %x send OOBDATA seq %x data %x\n", tp->t_oobseq, tp->t_oobc);
 		optlen = m0->m_len;
 		ti->ti_off = (sizeof (struct tcphdr) + optlen) >> 2;
 	}
-noopt:
 	ti->ti_flags = flags;
-	win = sbspace(&so->so_rcv);
-	if (win < so->so_rcv.sb_hiwat / 4)	/* avoid silly window */
+	/*
+	 * Calculate receive window.  Don't shrink window,
+	 * but avoid silly window syndrome.
+	 */
+	if (win < so->so_rcv.sb_hiwat / 4 && win < tp->t_maxseg)
 		win = 0;
-	if (win > 0)
-#if vax
-		ti->ti_win = htons((u_short)win);
-#else
-		ti->ti_win = win;
-#endif
+	if (win < (int)(tp->rcv_adv - tp->rcv_nxt))
+		win = (int)(tp->rcv_adv - tp->rcv_nxt);
+	ti->ti_win = htons((u_short)win);
 	if (SEQ_GT(tp->snd_up, tp->snd_nxt)) {
-		ti->ti_urp = tp->snd_up - tp->snd_nxt;
-#if vax
-		ti->ti_urp = htons(ti->ti_urp);
-#endif
+		ti->ti_urp = htons((u_short)(tp->snd_up - tp->snd_nxt));
 		ti->ti_flags |= TH_URG;
 	} else
 		/*
@@ -277,9 +311,8 @@ noopt:
 	/*
 	 * If anything to send and we can send it all, set PUSH.
 	 * (This will keep happy those implementations which only
-	 * give data to the user when a buffer fills or a PUSH comes in.
+	 * give data to the user when a buffer fills or a PUSH comes in.)
 	 */
-/*	if (len && (ti->ti_flags & (TH_FIN|TH_RST|TH_SYN)) == 0) */
 	if (len && off+len == so->so_snd.sb_cc)
 		ti->ti_flags |= TH_PUSH;
 
@@ -287,56 +320,60 @@ noopt:
 	 * Put TCP length in extended header, and then
 	 * checksum extended header and data.
 	 */
-	if (len + optlen) {
-		ti->ti_len = sizeof (struct tcphdr) + optlen + len;
-#if vax
-		ti->ti_len = htons((u_short)ti->ti_len);
-#endif
-	}
+	if (len + optlen)
+		ti->ti_len = htons((u_short)(sizeof(struct tcphdr) +
+		    optlen + len));
 	ti->ti_sum = in_cksum(m, sizeof (struct tcpiphdr) + (int)optlen + len);
 
 	/*
 	 * In transmit state, time the transmission and arrange for
-	 * the retransmit.  In persist state, reset persist time for
-	 * next persist.
+	 * the retransmit.  In persist state, just set snd_max.
 	 */
-	if (tp->t_force == 0) {
+	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))
+		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 (SEQ_GT(tp->snd_nxt, tp->snd_max) && tp->t_rtt == 0) {
-			tp->t_rtt = 1;
-			tp->t_rtseq = tp->snd_nxt - len;
+			/*
+			 * 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;
+				tcpstat.tcps_segstimed++;
+			}
 		}
 
 		/*
-		 * Set retransmit timer if not currently set.
-		 * Initial value for retransmit timer to tcp_beta*tp->t_srtt.
-		 * Initialize shift counter which is used for exponential
-		 * backoff of retransmit time.
+		 * 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) {
-			TCPT_RANGESET(tp->t_timer[TCPT_REXMT],
-			    tcp_beta * tp->t_srtt, TCPTV_MIN, TCPTV_MAX);
-			tp->t_rtt = 0;
-			tp->t_rxtshift = 0;
+			tp->t_timer[TCPT_REXMT] = tp->t_rxtcur;
+			if (tp->t_timer[TCPT_PERSIST]) {
+				tp->t_timer[TCPT_PERSIST] = 0;
+				tp->t_rxtshift = 0;
+			}
 		}
-		tp->t_timer[TCPT_PERSIST] = 0;
-	} else {
-		if (SEQ_GT(tp->snd_una+1, tp->snd_max))
-			tp->snd_max = tp->snd_una+1;
-	}
+	} else
+		if (SEQ_GT(tp->snd_nxt + len, tp->snd_max))
+			tp->snd_max = tp->snd_nxt + len;
 
 	/*
 	 * Trace.
@@ -350,20 +387,22 @@ noopt:
 	 */
 	((struct ip *)ti)->ip_len = sizeof (struct tcpiphdr) + optlen + len;
 	((struct ip *)ti)->ip_ttl = TCP_TTL;
-	if (error = ip_output(m, tp->t_ipopt, (so->so_options & SO_DONTROUTE) ?
-	    &routetoif : &tp->t_inpcb->inp_route, 0))
+	error = ip_output(m, tp->t_inpcb->inp_options, &tp->t_inpcb->inp_route,
+	    so->so_options & SO_DONTROUTE);
+	if (error)
 		return (error);
+	tcpstat.tcps_sndtotal++;
 
 	/*
 	 * 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.
-	 * Drop send for purpose of ACK requirements.
+	 * 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->t_flags &= ~(TF_ACKNOW|TF_DELACK);
-	if (sendalot && tp->t_force == 0)
+	if (sendalot)
 		goto again;
 	return (0);
 }
@@ -371,6 +410,7 @@ noopt:
 tcp_setpersist(tp)
 	register struct tcpcb *tp;
 {
+	register t = ((tp->t_srtt >> 2) + tp->t_rttvar) >> 1;
 
 	if (tp->t_timer[TCPT_REXMT])
 		panic("tcp_output REXMT");
@@ -378,9 +418,8 @@ tcp_setpersist(tp)
 	 * Start/restart persistance timer.
 	 */
 	TCPT_RANGESET(tp->t_timer[TCPT_PERSIST],
-	    ((int)(tcp_beta * tp->t_srtt)) << tp->t_rxtshift,
-	    TCPTV_PERSMIN, TCPTV_MAX);
-	tp->t_rxtshift++;
-	if (tp->t_rxtshift >= TCP_MAXRXTSHIFT)
-		tp->t_rxtshift = 0;
+	    t * tcp_backoff[tp->t_rxtshift],
+	    TCPTV_PERSMIN, TCPTV_PERSMAX);
+	if (tp->t_rxtshift < TCP_MAXRXTSHIFT)
+		tp->t_rxtshift++;
 }