From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262410AbTEFHEk (ORCPT ); Tue, 6 May 2003 03:04:40 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262412AbTEFHEk (ORCPT ); Tue, 6 May 2003 03:04:40 -0400 Received: from mail.zmailer.org ([62.240.94.4]:11976 "EHLO mail.zmailer.org") by vger.kernel.org with ESMTP id S262410AbTEFHEj (ORCPT ); Tue, 6 May 2003 03:04:39 -0400 Date: Tue, 6 May 2003 10:17:08 +0300 From: Matti Aarnio To: Sumit Narayan Cc: linux-kernel@vger.kernel.org Subject: Re: Write file in EXT2 Message-ID: <20030506071708.GM24892@mea-ext.zmailer.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Mon, May 05, 2003 at 11:14:46PM -0400, Sumit Narayan wrote: > Hi, > > I would like to create a log file containing the reads and writes made > on a disk, by adding a function in the kernel. And once this log table > reaches a limit, say 10,000 records, I would like it to be written on > hard disk automatically. I am unable to do this, since I dont know how > to write to a file, while in the kernel. I tried System Calls, but they > dont seem to work. Could someone tell me what is the list of functions > that I need to use to do this job. I think I have to play with > super-blocks and inodes. But I dont know how to do that. :) Please help > me. Considering how to do that log writing: Kernel contains several codes that are writing data to disk for various "logging" tasks. Most promimnent example of them is: kernel/acct.c It keeps kernel internal file descriptor ("filp") for its internal use. It has code that opens a file for writing to it, actual writer (one smallish block at the time, but that is merely size parameter issue), and it also closes the file when wanted (e.g. under administrator control). All that completely independent of target filesystem. Oh, and of course it has management interface, so that sysadmin can tell to it: - when to activate / deactivate - into which file to log In your application there is a danger of snaring yourself: disk activity must not stop at logging something, when the log buffer is full and flushing it is under way. Otherwise you are in danger of halting the log-flush, and then you have a dead machine. > Thanks. > Sumit > > p.s. I am using Kernel 2.4.20 and want this in EXT2 FS /Matti Aarnio