Path: utzoo!attcan!utgpu!jarvis.csri.toronto.edu!mailrus!cwjcc!gatech!hubcap!ncrcae!wescott
From: wescott@ncrcae.Columbia.NCR.COM (Mike Wescott)
Newsgroups: comp.unix.questions
Subject: Re: Clist problem
Keywords: clist
Message-ID: <4922@ncrcae.Columbia.NCR.COM>
Date: 17 Aug 89 13:31:09 GMT
References: <134@jassys.UUCP> <1026@virtech.UUCP>
Reply-To: wescott@ncrcae.Columbia.NCR.COM (Mike Wescott)
Distribution: usa
Organization: NCR Corp., Engineering & Manufacturing - Columbia, SC
Lines: 26

In article <1026@virtech.UUCP> cpcahil@virtech.UUCP (Conor P. Cahill) writes:
> In article <134@jassys.UUCP>, tony@jassys.UUCP (Tony Holden) writes:
> > Question.  What exactly is clist (I know unix but not that deep).  CCI
> > wants us to reboot every 3 days until they fix the problem.  Bogus?
> [...] I would guess that
> the CCI kernel has a bug that causes it to loose track of some clists over
> a period of time, hence the request to reboot every once in a while.

> This cannot be a big time bug (I'm sure you won't agree :-) because
> with any moderate amount of tty i/o, you would run out of clists in a very
> short period of time.

Actually there is a nasty little bug in SysVr2 clist manipulation code
(fixed in SysVr3) in canon() in io/tty.c by which clist blocks can be lost.
It is fairly rare since it requires two readers of the tty with ICANON not
set.  The most common situation in which this occurs is running HDB uucp to
dial out on a port that has uugetty running.

This may not be the problem with the CCI machine, I'm not familiar with it.
I am, however, rather familiar with the bug and kind of fond of it, since
it got me sent to England to investigate a customer's complaint.


-- 
	-Mike Wescott
	 mike.wescott@ncrcae.Columbia.NCR.COM