From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262016AbULPU62 (ORCPT ); Thu, 16 Dec 2004 15:58:28 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262014AbULPU61 (ORCPT ); Thu, 16 Dec 2004 15:58:27 -0500 Received: from host62-24-231-113.dsl.vispa.com ([62.24.231.113]:36514 "EHLO cenedra.walrond.org") by vger.kernel.org with ESMTP id S262016AbULPU5o (ORCPT ); Thu, 16 Dec 2004 15:57:44 -0500 From: Andrew Walrond To: linux-kernel@vger.kernel.org Subject: Re: Linux 2.6.10-rc2 start_udev very slow Date: Thu, 16 Dec 2004 20:57:25 +0000 User-Agent: KMail/1.7.1 Cc: Greg KH References: <200411171932.47163.andrew@walrond.org> <20041216155643.GB27421@kroah.com> In-Reply-To: <20041216155643.GB27421@kroah.com> MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit Content-Disposition: inline Message-Id: <200412162057.25244.andrew@walrond.org> X-Spam-Score: 0.0 (/) Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thursday 16 Dec 2004 15:56, Greg KH wrote: > On Wed, Nov 17, 2004 at 07:32:47PM +0000, Andrew Walrond wrote: > > I noticed that when upgrading from 2.6.8.1 to rc2, start_udev now takes > > 10-15s after printing > > > > "Creating initial udev device nodes:" > > udevstart should be used instead of start_udev. It goes a lot faster > and fixes odd startup dependancies that are needed. I'm using 048 at the moment. Works great, but if I replace start_udev with udevstart in my init scripts as you suggest, it all goes horribly wrong... udevstart is just a symlink to udev, but start_udev is a script which: - mounts ramfs - runs udevstart - makes some extra nodes not exported by sysfs (stdin/out/err) So I guess I need to migrate this functionality to my init system before I can call udevstart directly. Is that list of 'extra nodes not exported by sysfs likely to change?' Andrew Walrond