*BSD News Article 27785


Return to BSD News archive

Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!elroy.jpl.nasa.gov!swrinde!cs.utexas.edu!howland.reston.ans.net!EU.net!news.funet.fi!nntp.hut.fi!nntp!sja
From: sja@snakemail.hut.fi (Sakari Jalovaara)
Newsgroups: comp.os.386bsd.development
Subject: Re: Notes on the *new* FreeBSD V1.1 VM system
Date: 23 Feb 94 15:14:01 GMT
Organization: Helsinki University of Technology, Finland
Lines: 16
Distribution: inet
Message-ID: <SJA.94Feb23171401@gamma.hut.fi>
References: <BcxpGux.dysonj@delphi.com> <MYCROFT.94Feb20102534@duality.gnu.ai.mit.edu> <CLL9J6.FCF@endicor.com>
NNTP-Posting-Host: gamma.hut.fi
In-reply-to: tsarna@endicor.com's message of 21 Feb 94 19:16:17 GMT

> AIX implemented the killing behavior,
> and there were so many complaints from customers (database servers
> getting killed and corrupting the database, etc) that I believe they
> finally changed it.

Probably the main reason why an operating system designer might allow
memory overbooking is Fortran code.  In Fortran-77 the only way to go
is to allocate static arrays that are "large enough" and then use only
small parts of those arrays, depending on the size of the input data.
Hence lots of allocated memory pages that never get touched.

For mostly C environments, overbooking is usually a Bad Thing.
IBM/AIX users are not necessarily "mostly C" by a long shot.
(Kernel hacker's solution: make a separate a.out type for Fortran
programs.  And damn the kernel size.)
									++sja