From owner-freebsd-current  Wed Oct  9 19:39:27 1996
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id TAA02641
          for current-outgoing; Wed, 9 Oct 1996 19:39:27 -0700 (PDT)
Received: from dyson.iquest.net (dyson.iquest.net [198.70.144.127])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id TAA02632
          for <current@FreeBSD.org>; Wed, 9 Oct 1996 19:39:23 -0700 (PDT)
Received: (from root@localhost) by dyson.iquest.net (8.7.5/8.6.9) id VAA09727; Wed, 9 Oct 1996 21:38:53 -0500 (EST)
From: "John S. Dyson" <toor@dyson.iquest.net>
Message-Id: <199610100238.VAA09727@dyson.iquest.net>
Subject: Re: 'dead' binary stays 'dead'?
To: msmith@atrad.adelaide.edu.au (Michael Smith)
Date: Wed, 9 Oct 1996 21:38:53 -0500 (EST)
Cc: current@FreeBSD.org
In-Reply-To: <199610100143.LAA16437@genesis.atrad.adelaide.edu.au> from "Michael Smith" at Oct 10, 96 11:13:30 am
Reply-To: dyson@FreeBSD.org
X-Mailer: ELM [version 2.4 PL24 ME8]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: owner-current@FreeBSD.org
X-Loop: FreeBSD.org
Precedence: bulk

> 
> Obviously, machines with serious memory errors deserve to loose
> infinitely, but this box doesn't fall into that category.  It's
> survived numerous 'make world' cycles faultlessly, and stands up all
> day to the excessive pounding that the physics geeks here subject it
> to.
> 
> Comments? Refutations? Merciless laughter?
> 
You are describing a problem that I know *can* happen, but I don't know
why.  In essence, once you have a copy of a programs .text, .data in
memory, it will continue to be cached until the memory is reclaimed.
If any part of that image gets modified, then it will stay modified.
If you remove the file that has it's in-memory image broken, that
brokenness will go away.  However, if you try to copy the
file like:  cp ls ls.new, it is likely that ls.new will also be broken
because the same image that is executed is also in the buffer cache
(the cache and the image are pretty much one in the same.)  The
best way to make the problem go-away is to fill memory and cause the
broken binary to disappear.  Now, the complicated part is why it happened.

I just don't know why...

Are you using NFS?  Are you using the most recent -current (snap)?...
You know the typical questions :-).

John