From owner-freebsd-hackers Sun Jul 30 12:48: 5 2000 Delivered-To: freebsd-hackers@freebsd.org Received: from localhost (localhost [127.0.0.1]) by hub.freebsd.org (Postfix) with ESMTP id F29B637B76F; Sun, 30 Jul 2000 12:47:35 -0700 (PDT) (envelope-from green@FreeBSD.org) Date: Sun, 30 Jul 2000 15:47:19 -0400 (EDT) From: Brian Fundakowski Feldman X-Sender: green@green.dyndns.org To: Adam Cc: Doug Barton , freebsd-hackers@FreeBSD.ORG, dillon@FreeBSD.ORG Subject: Re: /tmp on a ramdisk? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-hackers@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG The issue is that mount_mfs is simply newfs with a catch: it constructs the new filesystem completely in memory and lives on as the storage for the mounted filesystem. If you view the processes on a system using MFS, you will notice that one of them is the original mount_mfs, having become a daemon. Yes, things are stored twice in memory: once in the buffer cache and once in the MFS process. Yes, they are also copied multiple times. MFS simply can't perform as well as you might expect. The malloc disk device can because it simply creates a kernel-memory backing store. The disadvantage here is that it's wired memory and can't get swapped out like mount_mfs can. -- Brian Fundakowski Feldman \ FreeBSD: The Power to Serve! / green@FreeBSD.org `------------------------------' To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-hackers" in the body of the message