From owner-freebsd-arch@FreeBSD.ORG Wed Jun 28 22:06:33 2006 Return-Path: X-Original-To: freebsd-arch@freebsd.org Delivered-To: freebsd-arch@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 3F6CE16A407; Wed, 28 Jun 2006 22:06:33 +0000 (UTC) (envelope-from andrea@webcom.it) Received: from www.webcom.it (gen053.n002.c03.escapebox.net [213.73.82.53]) by mx1.FreeBSD.org (Postfix) with ESMTP id 4FC0343D8D; Wed, 28 Jun 2006 22:06:31 +0000 (GMT) (envelope-from andrea@webcom.it) Received: from andrea by webcom.it with local (Exim 3.36 #1) id 1FviB3-000BrU-00; Wed, 28 Jun 2006 22:06:29 +0000 Date: Wed, 28 Jun 2006 22:06:29 +0000 From: Andrea Campi To: Garance A Drosehn Message-ID: <20060628220628.GA12488@webcom.it> References: <62426.1151433799@critter.freebsd.dk> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.5.11 Sender: Andrea Campi Cc: Poul-Henning Kamp , pjd@FreeBSD.org, freebsd-arch@FreeBSD.org Subject: Re: Accessing disks via their serial numbers. X-BeenThere: freebsd-arch@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussion related to FreeBSD architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 28 Jun 2006 22:06:33 -0000 On Tue, Jun 27, 2006 at 04:38:23PM -0400, Garance A Drosehn wrote: > This strikes me as a worthwhile idea. Leave the device- > entries in /dev as they are now, but then create some > sub-directories which would hold the more arbitrary (or > "non-sanitized") names. > /dev/info/disk/serial-num > or /dev/info/geom/whatever > > etc. That way there's only one new entry in /dev, and > people could just de-permit that directory (or turn the > feature off) if they didn't want or need to have that > extra info available. Picking a semi-random message to reply to, I seem to recall Solaris has a separate /devices filesystem for this: all entries are dynamic, so the filesystem is strictly readonly from a userland point of view. In our case, we would have quite a lot of overlap with devinfo, but it could still be something to keep in mind. Andrea -- Loose bits sink chips.