At 10:44 Uhr +0100 06.11.1999, Ignatios Souvatzis wrote:
>On Fri, Nov 05, 1999 at 10:55:41PM -0800, Space Case wrote:

>> What happens is that I'll get the system connected, and start a job, sup
>> for example.  It'll run for some period of time, say five minutes or an
>> hour, then traffic stops going over the modem.  About a minute later, the
>> modem disconnects.

>> Nov  5 22:29:13 c610 /netbsd: zstty0: 22 silo overflows, 3 ibuf floods

ibuf floods, even, and on a Q650... what port speed are you using? You can
bump the serial buffer size somewhere in dev/zs.c, IIRC.

>to me, this sounds like a handshaking problem.
>
>You probably know that, but you have to
>
>- check that your pppd uses crtscts, (or maybe it is cdtrcts for your sort
>  of Mac?)

You want cdtrcts -- which, IMO, is a bad misnomer, because on mac68k it
delivers what drtscts promises to do. This is causing all sorts of havoc
when you try to run off-the-shelf unix comm applications on mac68k (try to
run uucp with hardware flow control =8( ).

>- check that you have 8 wires running in between the two: lines 2,3,4,5,6,7,20
>  on a 25pin connector.

That^s the usual Macintosh "hardware handshake" cable issue.

I can add one more: Make sure you have your modem configured to ignore DTR
(that^s ""at &d0", usually). The Mac has only one outgoing handshake line
that you can either use for hanging up the modem (that would be DTR
semantics) or for hardware flow control (that would be RTS). "Hardware
handshake" cables link the Sub-D25 pins 4 (RTS) and 20 (DTR) with HSKo
(Mini-DIN pin 1).

	hauke


--
"It's never straight up and down"     (DEVO)