Return to BSD News archive
Path: sserve!newshost.anu.edu.au!munnari.oz.au!news.Hawaii.Edu!ames!olivea!decwrl!sony!vixie!vixie
From: vixie@vix.com (Paul A Vixie)
Newsgroups: comp.os.386bsd.development
Subject: Re: Notes on the *new* FreeBSD V1.1 VM system
Date: 22 Feb 94 09:42:35
Organization: Vixie Enterprises
Lines: 18
Message-ID: <VIXIE.94Feb22094235@office.home.vix.com>
References: <BcxpGux.dysonj@delphi.com> <MYCROFT.94Feb20102534@duality.gnu.ai.mit.edu>
<CLL9J6.FCF@endicor.com>
NNTP-Posting-Host: office.home.vix.com
In-reply-to: tsarna@endicor.com's message of 21 Feb 94 19:16:17 GMT
>Is there any reason why memory allocation can't simply fail when there
>isn't any more to allocate?
Sure. At the point in the kernel when you know you've run out, you don't
know what process is responsible. In fact at that point it could be the
expansion of a dynamic kernel data structure rather than of a user address
space that has caused the shortage. Rearchitecting this to avoid the
asynchrony is a ``hard problem'' according to the folks I know who do VM.
(I beat my head against PMAP during a year-long, aborted VAX port; I can
from personal experience and opinion that PMAP is a horribly bad thing. It
causes itself a lot worse problems than the above, its just that on a Mach
system the above problem doesn't bite you as badly.)
--
Paul Vixie
Redwood City, CA Also: <comp-sources-unix@uunet.uu.net>, <vixie@bsdi.com>,
decwrl!vixie!paul <ftpmail-admin@pa.dec.com>, <vixie@sony.com>,
<paul@vix.com> <{bind-workers,objectivism}-request@vix.com>