Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!usc!sdd.hp.com!wuarchive!udel!haven!decuac!shlump.nac.dec.com!energy.irc.cbm.dec.com!mamros
From: mamros@energy.irc.cbm.dec.com (Shawn Mamros)
Newsgroups: comp.unix.ultrix
Subject: Re: Process Table Corrupt?
Keywords: ps, kmem, 4.0, 6310
Message-ID: <1990Nov8.132959@energy.irc.cbm.dec.com>
Date: 8 Nov 90 18:29:59 GMT
References: <3752@vela.acs.oakland.edu>
Sender: news@shlump.nac.dec.com
Reply-To: mamros@crl.dec.com (Shawn Mamros)
Distribution: usa
Organization: Digital Equipment Corporation
Lines: 21

The way ps works, it has to open /dev/kmem and read the proc structs
in kernel memory.  However, it's not an atomic operation.  If you
happen to catch the system at a point where it's moving things around,
it can result in some strange output from ps.  Usually not as strange
as what you're seeing, but it's possible.

Is the error a repeatable one?  If, say, you type ps once and see weird
process ownerships, what happens when you do a ps a few seconds afterward?
Is everything back to normal?  If so, I'd say it's just a temporary glitch
with ps catching the proc structs in mid-transition.

And if not... does the system ever correct itself?  Or do things stay
messed up until reboot time?

I must admit I've never seen ps behave that way before, but then again
I haven't used ULTRIX V4.0 heavily on VAXen either.

-Shawn Mamros
E-mail to: mamros@crl.dec.com
Disclaimer: I am neither an "official" representative nor an ULTRIX developer.
Just another ULTRIX "customer" ;-)