*BSD News Article 18302


Return to BSD News archive

Xref: sserve comp.os.linux:47672 comp.os.386bsd.questions:3751 comp.windows.x.i386unix:2466
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!cronkite.cisco.com!cisco.com!vandys
From: vandys@cisco.com (Andrew Valencia)
Newsgroups: comp.os.linux,comp.os.386bsd.questions,comp.windows.x.i386unix
Subject: Re: SUMMARY: 486DX2/66 for Unix conclusions (fairly long)
Date: 12 Jul 93 17:26:02 GMT
Organization: cisco Systems
Lines: 16
Message-ID: <vandys.742497962@cisco.com>
References: <21k903$3q4@GRAPEVINE.LCS.MIT.EDU> <PCG.93Jul12003233@decb.aber.ac.uk> <JOHNSONM.93Jul12091953@calypso.oit.unc.edu> <1993Jul12.182304@informatik.uni-kl.de>
NNTP-Posting-Host: glare.cisco.com

>And if you want to get *really* pedantic, you could say that, yes,
>Linux *does* occasionally swap -- if all the pages of an executable
>are paged out to disk, then the application is technically swapped
>out, no?

"Classic" UNIX reclaimed the memory for your U area and kernel stack
(some claim the kernel stack is a part of the U area, so the previous
statement may contain a redundancy.)  BSD flavors also reclaimed the
memory for the PTEs.  Linux doubtless has analogs to some of these, and
it sounds like their memory is only reclaimed on process exit.

					Andy Valencia
-- 

|		    Andy Valencia: vandys@cisco.com		      |
| Disclaimer: I speak only for myself, certainly not for my employer! |