Path: utzoo!mnetor!uunet!seismo!sundc!pitstop!texsun!texsun.central-relay.sun.com!convex!trsvax!authorplaceholder From: uhclem@trsvax.UUCP Newsgroups: comp.unix.xenix Subject: Re: tty error] Message-ID: <196500012@trsvax> Date: 1 Apr 88 14:49:00 GMT References: <884@udccvax1.acs.udel.EDU> Lines: 56 Nf-ID: #R:udccvax1.acs.udel.EDU:-88400:trsvax:196500012:000:2497 Nf-From: trsvax.UUCP!uhclem Apr 1 09:49:00 1988 <> Base notes wants help on message: B> init:/dev/ttyx05: getty keeps dying. There may be a problem As one of the responses has noted, this message may have come from init for having to spin-off too many gettys in a short time. If this is a real IBM AT serial/parallel adapter or one that is an exact clone (used same RS232-C line receivers) then this information may assist. The line driver IBM used the 75154 line receiver which is sensitive to noise, more so than the 1489 which more systems use for RS232-C line receivers. Anyway, we found if you had an IBM AT adapter in your machine and a non-ribbon cable (6ft) other than IBM's connected and it was not connected to anything, if you enabled the line, the earlier releases of XENIX-286 would bog down or die after a while. The above message appears to part of a way of keeping that from happening in the new releases. It does come from /etc/init. We did something similar in one of our own XENIX releases. What causes it? We found that when the line was enabled, the CD line would be floating high enough to cause a getty to be started. The banner would be transmitted and these transitions would cause the CD line to swing low and high, very minute, but enough to generate 8 to 10 interrupts per character requesting the getty by killed and another started. The real, sold by big-blue IBM serial cable is shielded and does not exhibit this problem unless you extend it with a lot of unshielded stuff. Perhaps a better word for the message might have been "thrashing". We also found that ribbon cable did not seem to have this problem - apparently enough distance between 8 and 2/3 to reduce induction. Adapter cards from other vendors we tried all used 1489's and did not display this problem. The moral of all this was not to enable a line that did not have a computer, modem or terminal attached. Hope that helps. "Thank you, Uh Clem." Frank Durda IV @ ...decvax!microsoft!trsvax!uhclem ...convex!infoswx!hal6000!trsvax!uhclem