To: netbsd-help@NetBSD.ORG
Date: Fri, 29 Sep 1995 19:22:21 GMT
From: christos@deshaw.com (Christos Zoulas)
Message-Id: <DFoLtA.DGA@deshaw.com>
Organization: D. E. Shaw & Co.
References: <Pine.SGI.3.91.950928173648.8792B-100000@win31.nas.nasa.gov>, <Pine.A32.3.91.950929090505.20145A-100000@stalin.ts.umu.se>
Subject: Re: CPU loop problem
Sender: owner-netbsd-help@NetBSD.ORG
Precedence: list

In article <Pine.A32.3.91.950929090505.20145A-100000@stalin.ts.umu.se> sillen@ts.umu.se (Daniel Silen) writes:
>On Thu, 28 Sep 1995, George E. Navas wrote:
>
>>=20
>> I'm currently running NETBSD 1.0 and it works great. However, at times
>> it appears that if a telnet connection is dropped and you are in the
>> middle of vi, the vi command goes into a loop consuming all of the CPU.
>> I was wondering if anybody else encountered a similar problem or has
>> any info on why this happens.=20
>>=20
>> Thanks in Advance
>>=20
>Vi? That way a new one...
>We are running AIX on a couple of RS6000, and has the same problem mainly=
>=20
>with elm and pine.
>
>I think the cause simply is that the program looses connection with the=20
>terminal, but don't "understand it".
>It simply keep trying to get userdata from the tty that not is=20
>responding, thus ending up in an endless loop.
>
>If there is a solution to the problem, I'm unaware of it. We simply=20
>solved it by creating a crontab-job, that looks for pine / elm -
>processeses that uses more than 20% of the CPU.
>
>Hope someone else knows better :-)

For most of these ailments it is the shell who is responsible for not
propagating SIGHUP to the foreground process group. I don't remember
what the 1.0 {c,}sh used to do, but I think all that is cured by now.

christos