From owner-freebsd-questions Thu Oct 5 17:10:19 1995 Return-Path: owner-questions Received: (from root@localhost) by freefall.freebsd.org (8.6.12/8.6.6) id RAA28957 for questions-outgoing; Thu, 5 Oct 1995 17:10:19 -0700 Received: from phaeton.artisoft.com (phaeton.Artisoft.COM [198.17.250.211]) by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id RAA28950 for ; Thu, 5 Oct 1995 17:10:16 -0700 Received: (from terry@localhost) by phaeton.artisoft.com (8.6.11/8.6.9) id RAA05011; Thu, 5 Oct 1995 17:09:04 -0700 From: Terry Lambert Message-Id: <199510060009.RAA05011@phaeton.artisoft.com> Subject: Re: tunefs To: bmk@dtr.com Date: Thu, 5 Oct 1995 17:09:04 -0700 (MST) Cc: serges@umr.edu, freebsd-questions@freefall.freebsd.org In-Reply-To: <199510052330.QAA00567@everest> from "bmk@dtr.com" at Oct 5, 95 04:30:51 pm X-Mailer: ELM [version 2.4 PL24] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Content-Length: 1494 Sender: owner-questions@FreeBSD.org Precedence: bulk > > Is there a way under FBSD to 'tunefs' the / filesystem, WITHOUT > > creating another filesystem and without making an existing filesystem > > bootable? > > (This is all to the best of my knowledge. I may be wrong. You've been > warned. :) > > You must can only tunefs a dismounted filesystem - so you must do one of > a few things: Oh heck; I forgot that I had hacked that. The main problem is getting the tuning changes into the in-core copy of the struct so that it effects subsequent operations on the FS in question. The -current code can't do this, so that's why the mount/remount. The fix is to replace the read/write interface with an in/out ioctl() interface. You could just update the in core copy without too much problem, since it's synced to disk on unmount. I did this because I was trying to find optimal tuning parameters for a particular benchmark and got tired of the mount/unmount. If it's all the same with you, I'd like to hack on the logical device management using Julians new devfs once that's all the way up before rolling changes in -- that way I won't have to hack them multiple times after they're committed. I've been hacking logical volume management for partitions, extended partitions and BSD extended partitions (ie: the results of disklabel) for a port, and I'd like to get it all rolled together first. Terry Lambert terry@lambert.org --- Any opinions in this posting are my own and not those of my present or previous employers.