On 03 Feb 1999 21:48:34 -0800 
 Michael Graff <explorer@flame.org> wrote:

 > The comment and the code in tcp_input.c don't quite match, I believe.
 > If the comment is right, the code is wrong.  This change makes the
 > window update only occur when the TH_ACK bit is set in a packet.
 > The question is, is it right?

I believe you are correct.  MAtt Thomas found this and mentioned it,
at some point, as did cgd (for Matt).  I think it got botched when
doing the -Wall changes.  Please commit this fix.

 > 
 > Index: sys/netinet/tcp_input.c
 > ===================================================================
 > RCS file: /cvsroot/src/sys/netinet/tcp_input.c,v
 > retrieving revision 1.75
 > diff -u -r1.75 tcp_input.c
 > --- tcp_input.c	1999/01/24 01:19:28	1.75
 > +++ tcp_input.c	1999/02/04 03:44:36
 > @@ -1379,9 +1379,9 @@
 >  	 * Update window information.
 >  	 * Don't look at window if no ACK: TAC's send garbage on first SYN.
 >  	 */
 > -	if (((tiflags & TH_ACK) && SEQ_LT(tp->snd_wl1, ti->ti_seq)) ||
 > +	if ((tiflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, ti->ti_seq) ||
 >  	    (tp->snd_wl1 == ti->ti_seq && SEQ_LT(tp->snd_wl2, ti->ti_ack)) ||
 > -	    (tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd)) {
 > +	    (tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd))) {
 >  		/* keep track of pure window updates */
 >  		if (ti->ti_len == 0 &&
 >  		    tp->snd_wl2 == ti->ti_ack && tiwin > tp->snd_wnd)

        -- Jason R. Thorpe <thorpej@nas.nasa.gov>