dave@onfcanim.UUCP (12/01/87)
Hayes-type modems have an "escape sequence" that lets you get back into command mode after the modem is online - usually three plus signs with a 1-second pause before and after the group. I have both Telebit Trailblazer and Hayes V-series Smartmodem 2400's with this feature. With both modems, the escape sequence gets me into command mode OK, and I can issue commands there, but when I try to get back online (via ATo), I get no response. The modems on both ends of the connection are still there, and have carrier up. A "pstat -t" reveals nothing unusual about the tty port I am going into. Yet I get no echo, nor any response of any sort. With the Telebit, at least, I can also no longer get back to command mode, and the only way to get out of this hung state is to turn off the modem. The fact that two very different modems do this suggests it's a UNIX problem or a modem control lead problem rather than something in the modems themselves. However, I don't have a serial communications analyzer to look at the connection between the dialin modem and the VAX. Has anyone else seen this, or understand what's going on. The machine is a VAX /780 running 4.3BSD UNIX with all the ucb-fixes patches installed (but not much else). The serial ports are on an Emulex CS21 emulating a DMF32. The modems are all new within the last few weeks, so their ROMS should be current.
david@ms.uky.edu (David Herron -- Resident E-mail Hack) (12/03/87)
In article <15491@onfcanim.UUCP> dave@onfcanim.UUCP (Dave Martindale) writes: >I have both Telebit Trailblazer and Hayes V-series Smartmodem 2400's >with this feature. With both modems, the escape sequence gets me into >command mode OK, and I can issue commands there, but when I try to >get back online (via ATo), I get no response. Ok. I happen to be hooked in through a Telebit right now. Let's see what happens. Just a second please ... Ok. I thought so ... I went back and forth without problems. HOWEVER, I had experienced a problem like that before with the Telebit. If I'd popped to the modem's command level and issued a command to the remote modem using AT%something, then the remote modem would sort-of be in it's command mode. To get connected back correctly I had to say "at%o" before saying "ato". Do the V-series modems have something like this? Ok. I can get it to repeat with "at%&r", after doing that I do "ato" and can't do anything. "+++" then to "AT%o" then "ato" and all the control-r's I'd done the first time I'd gone back happen plus the one I typed the second time I'd gone back. If you want to know details of the telebit configuration, send me mail. -- <---- David Herron -- The E-Mail guy <david@ms.uky.edu> <---- or: {rutgers,uunet,cbosgd}!ukma!david, david@UKMA.BITNET <---- <---- Classic beer bellies aren't born! They're made!
wtm@neoucom.UUCP (12/06/87)
<< can't go back on line after command escape>>
I've had my Telebit do the same thing. Type +++, do some modem
command at your end, then ATO {return} to go back. Now, wait
forever.
I think ATO is acting more like ATDT. It wants to hear a dial
tone, ringing, then carrier. ATO really ought to so straight away
to trying to sync with the carrier from the other end.
I have run into this problem when I have to originate a call from
my Trailblazer which I dialed by hand to get to the modem extension
through a human PBX operater. On a Hayes, you can ATO and it will
work.
The solution is to pick a mode in the Telebit setup that forgoes
call progress detection. I haven't tried this yet, but I think
that will do the trick.
--Billdave@onfcanim.UUCP (12/06/87)
Michael T. Sullivan (richp1!mike) provided the solution. In retrospect, it's so obvious I'm embarassed.... I had complained that most of the time I used the "escape sequence" to put my local modem into command mode, my connection to UNIX was apparently hung when I attempted to go online again. What happens is that as I type the three plus signs in rapid succession, the local modem sends them to UNIX where they are echoed back to my terminal. Just as my local modem sees its "escape sequence" and goes into command mode, if the UNIX-end modem is a Hayes-compatible it also goes into command mode and ignores further characters from the phone line. Since the UNIX-end modem is not listening to the line, and UNIX is not likely to send the string "ATO\r" on its own, the connection is now hung until I hang up the phone. The fix is simple - turn off the escape sequence on the modems at the UNIX end (s2=255 on the Hayes, s55=3 on the Telebit). These modems are used by uucico in originate mode as well as being dialin lines, but uucp never really needs to talk to the modem once it has established a connection. (The stock Hayes dialer routines use the escape sequence to tell the modem to reset at the end of a call, but they can easily be changed to drop DTR to hang up the phone first, putting the modem back into command mode that way.) Anyone that wants to use the modems for tip AND wants to use the escape sequence will have to change the appropriate s-register before dialing. Too bad.
dave@onfcanim.UUCP (12/08/87)
In article <15500@onfcanim.UUCP> dave@onfcanim.UUCP (Dave Martindale) writes: > >The fix is simple - turn off the >escape sequence on the modems at the UNIX end (s2=255 on the Hayes, >s55=3 on the Telebit). Oops. It turns out that the Telebit has at least two ways of disabling escape sequences - setting s55 as above, or setting s2 to something illegal. Bravo for Telebit. The Hayes, on the other hand, doesn't have anything equivalent to s55, and it doesn't store the s2 setting in the EEPROM either. There seems to be *no way* to permanently disable the escape sequence (without completely disabling command recognition, which is too drastic). Why is it that Hayes always manages to miss what you want by just a little bit? Another example is the &D parameter - you can have DTR disable auto-answer when it is low, or you can have the fall of DTR reset the modem, but you can't have both - so either your modem answers when the machine is down, or you accept the fact that you can't reset the modem each connection. Bah humbug.
wtm@neoucom.UUCP (Bill Mayhew) (12/11/87)
I really hate to give out free advertising for any product, but the Telebit Trailblazer really is in a class by itself-- especially when the Trailblazer's features are compared against other modems in its price catagory (especially if you got in on the half price deal). I run Sys V plain vanilla uucp on my 3b1, and the Trailblazer is really the only modem I've been able to find that is intelligent enough to work around uucp's lack of sophistication in dealing with modems. The Trailblazer's s55 register is really nifty; I can use the thing to support both dial in and dial out. Does anybody out there know if the idosyncracies of the hayes 1200 and 2400 buad modems have been done away with in the 9600? I've always been puzzled that there was no decent way to control command echoing with real Hayes modems. The big question is which modem is going to get to be the defacto standard. I'd be willing to settle for the telebit, especially since the money is spent. I was surprised to see the ISO standards get frozen at a pretty low level of performace. I do agree that it would be nice to have full rate bidirectional communication. For the moment most of the available protocols are really half duplex oriented (uucp for instance). Of course appropriate hardware would change that eventaully. I have used the Trailblazer to drive a terminal at 9600 baud. The fact that the connection is simulated full duplex isn't all that noticable. The vi editor does get a bit cranky, though, dealing with the turn-around delays. --Bill