Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!henry
From: henry@utzoo.UUCP (Henry Spencer)
Newsgroups: hacknews
Subject: Re: dialin/dialout operational
Message-ID: <5187@utzoo.UUCP>
Date: Thu, 7-Mar-85 16:49:38 EST
Article-I.D.: utzoo.5187
Posted: Thu Mar  7 16:49:38 1985
Date-Received: Thu, 7-Mar-85 16:49:38 EST
References: <5186@utzoo.UUCP>
Organization: U of Toronto Zoology
Lines: 22

The "one lingering problem" alluded to in the previous message is now
understood, and a fixed kernel will be tested Monday.  It shouldn't
affect normal operation of 5121.  The problem was that any modem-status
interrupt after a dialout device was opened but before it made a call
successfully would abort the call, IF there wasn't a getty waiting on
the same line.  Roughly speaking.  The modem-interrupt code noticed
that carrier was off and nobody was waiting for it to be on, so it
turned the line off (dropping DTR etc. to the modem).  The real fun is
that there is guaranteed to be a modem-status interrupt just after the
dialout device is opened, since opening it raises RTS and the modem
soon replies by raising CTS.  So an outbound call got blown away almost
instantly, unless there was also a getty waiting for carrier.  The fix
isn't hard, the modem-interrupt code just needs to know whether there's
an uncompleted call in progress on the line.

P.S. An overdue acknowledgement:  Ron Gomes was the one who originally
	wrote the DH dialin/out code, although we've modified it a
	little.  The bugs we've found have been our mistakes or the
	result of different circumstances.
-- 
				Henry Spencer @ U of Toronto Zoology
				{allegra,ihnp4,linus,decvax}!utzoo!henry