Path: utzoo!utgpu!jarvis.csri.toronto.edu!cs.utexas.edu!wuarchive!mit-eddie!uw-beaver!Teknowledge.COM!unix!hplabs!hpfcso!hpfcdc!perry
From: perry@hpfcdc.HP.COM (Perry Scott)
Newsgroups: comp.sys.hp
Subject: Re: 2 serial port questions (HP 9000/835 HP-UX)
Message-ID: <5570381@hpfcdc.HP.COM>
Date: 22 Feb 90 16:04:42 GMT
References: <4685@aplcen.apl.jhu.edu>
Organization: HP Ft. Collins, Co.
Lines: 24

>We have XON/XOFF trouble with the 6 port mux on the 9000/550.  As I
>understand it, this card for the 800 and 500 series is nearly the same.

True.  Both are CIO MUX cards.  Same firmware.

>Upon receipt of an XOFF character, the 6 port mux (27140) takes longer
>than the 8 port mux (27130) to stop its output stream.  This has caused
>data overflow on some of our terminals (Z-19 with Northwest Digital
>graphics boards).  The terminals evidently have almost no buffer space
>left when they send the XOFF to the mux.  This problem occurs at 9600
>bps with the 6 port mux but not with the 8 port mux.

When DEC first came out with XON/XOFF, they specified that at least 32
bytes of buffer should be available after XOFF is sent.  This has been
widely ignored by the industry.  In fact, the HP RS232 ThinkJet has only
16 bytes left when XOFF is sent, so even HP is guilty.

>We have no problem running a LaserWriter NTX-II on an 8 port mux
>at 19,200 bps using XON/XOFF flow control.

After the ThinkJet flap, we made sure the peripheral divisions clearly
understood XON/XOFF protocol.  :-)

Perry Scott