Path: utzoo!news-server.csri.toronto.edu!cs.utexas.edu!usc!zaphod.mps.ohio-state.edu!think.com!mintaka!spdcc!tauxersvilli!alphalpha!nazgul
From: nazgul@alphalpha.com (Kee Hinckley)
Newsgroups: comp.sys.apollo
Subject: Re: 4500, 10.3, BSD, 2400 modem?
Message-ID: <1991Mar8.233359.13061@alphalpha.com>
Date: 8 Mar 91 23:33:59 GMT
References: <1991Mar8.073658.12083@world.std.com> <9103081115.aa05182@concour.cs.concordia.ca>
Organization: asi
Lines: 98

In article <9103081115.aa05182@concour.cs.concordia.ca> goldfish@CONCOUR.CS.CONCORDIA.CA writes:
>            *   APOLLO SERIAL PORTS DON'T WORK   *
If anyone cares, it's because the person doing the SR10 work for them
left in the middle and noone ever finished it.  Did you care?  I thought not.

>	1)  handshaking:  the DSR-DTR interaction is broken.  HP knows
>	    it is broken, Apollo knew it was broken before the merger.
For what it's worth, I think CTS works but RTS doesn't, or some such.

>	1a) The computer will drop the DTR occasionally and hang up
>	    your MODEM.
Ah.  That explains a lot.

>	1b) Upon hanging up the MODEM, it will not terminate the
>	    session so the next caller gets your session already
>	    logged in.  (caution boys and girls, don't try this
See my source below.

>	2)  BAUD switching: The word I got from HP (semi-official) was
>	    that BAUD switching was not supported in the Domain-OS.
Ditto.

>	3)  The SIO and TTY drivers do NOT maintain consistent
>	    internal structures.  (can you say "probably written in
>	    unstructured COBOL" everyone?) 
No, that's almost certainly due to having it dropped in the middle.

>	3)  talk to a single speed MODEM if you are careful
We drive a Telebit T2500 succesfully, although we do so by having
the following entry in /etc/ttys:

tty02   "/etc/fakegetty std.9600"       dialin

It all seems to work pretty well, although of course none of it
should be necessary.  Apollo proper does actually have dialin
lines, and they were constantly screwing up until they got a
terminal concentrator - but they aren't heavily used, so things
never got fixed.  Terminals are obsolete, you see...

Here's fakegetty.c:


#include <sys/file.h>
#include <sgtty.h>

main(argc,argv)
int argc;
char **argv;
{
	int fd;
	char dpn[100];
	struct sgttyb sgbuf;

	strcpy(dpn, "/dev/");
	strcat(dpn, argv[2]);

	fd = open(dpn, O_RDWR|O_NDELAY);
	if (fd < 0) {
		perror("dpn");
		exit(1);
	}

	if (ioctl(fd, TIOCGETP, &sgbuf) < 0) {
		perror("getp");
		exit(1);
	}
	
	sgbuf.sg_flags &= ~ECHO;
	sgbuf.sg_ispeed = B2400;
	sgbuf.sg_ospeed = B2400;

	if (ioctl(fd, TIOCSETN, &sgbuf) < 0) {
		perror("setn");
		exit(1);
	}

	sleep(2);
	write(fd, "+++", 3);
	sleep(2);
	write(fd, "ATZ\r", 4);
	sleep(2);
	if (ioctl(fd, TIOCNOTTY, 0) < 0) {
		perror("notty");
		exit(1);
	}
	close(fd);

	execv("/etc/uugetty", argv);
}

-- 
Alfalfa Software, Inc.          |       Poste:  The EMail for Unix
nazgul@alfalfa.com              |       Send Anything... Anywhere
617/646-7703 (voice/fax)        |       info@alfalfa.com

I'm not sure which upsets me more: that people are so unwilling to accept
responsibility for their own actions, or that they are so eager to regulate
everyone else's.