From owner-freebsd-hackers Thu Aug 31 10:26:21 1995 Return-Path: hackers-owner Received: (from majordom@localhost) by freefall.FreeBSD.org (8.6.11/8.6.6) id KAA14096 for hackers-outgoing; Thu, 31 Aug 1995 10:26:21 -0700 Received: from plains.nodak.edu (plains.NoDak.edu [134.129.111.64]) by freefall.FreeBSD.org (8.6.11/8.6.6) with ESMTP id KAA14090 for ; Thu, 31 Aug 1995 10:26:20 -0700 Received: (from tinguely@localhost) by plains.nodak.edu (8.6.11/8.6.10) id MAA07284; Thu, 31 Aug 1995 12:25:42 -0500 Date: Thu, 31 Aug 1995 12:25:42 -0500 From: Mark Tinguely Message-Id: <199508311725.MAA07284@plains.nodak.edu> To: dyson@freefall.FreeBSD.org Subject: Re: problems with programs being swaped out!! Cc: gurney_j@efn.org, hackers@freebsd.org Content-Length: 1077 Sender: hackers-owner@freebsd.org Precedence: bulk > Probably the best approach is for the system to free-up swap space for > pages that are resident when the system is low on memory. I agree. In the earlier post, I should have made it more clear that hardware concerns in the 386BSD 0.x days are different from those today. Today, if your machine is pushed to backstore starvation, your performance goes bad anyway, I agree, that would a good time to switch strategies to free pages in RAM. to patronize the other readers and to make sure I am not saying anything critical, anytime a person changes the philosophy of the VM, the whole system is affected. For My Information: Wouldn't an additional mask be required to tell which pages are backstored but also resident or are you going to run down processes and flush out already backstored pages leaving the other pages that have never been backstored resident (because there is not place to put them because we are backstore starved)? or are you going remove the backstore pages for the resident pages and let the pageout alorithm act normally? or, ... or, ... --mark.