From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262040AbULPVrP (ORCPT ); Thu, 16 Dec 2004 16:47:15 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262038AbULPVrO (ORCPT ); Thu, 16 Dec 2004 16:47:14 -0500 Received: from mail.kroah.org ([69.55.234.183]:16537 "EHLO perch.kroah.org") by vger.kernel.org with ESMTP id S262034AbULPVqz (ORCPT ); Thu, 16 Dec 2004 16:46:55 -0500 Date: Thu, 16 Dec 2004 13:46:19 -0800 From: Greg KH To: Andrew Walrond Cc: linux-kernel@vger.kernel.org Subject: Re: Linux 2.6.10-rc2 start_udev very slow Message-ID: <20041216214619.GA9827@kroah.com> References: <200412162057.25244.andrew@walrond.org> <20041216211137.GA9475@kroah.com> <200412162120.33995.andrew@walrond.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <200412162120.33995.andrew@walrond.org> User-Agent: Mutt/1.5.6i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Thu, Dec 16, 2004 at 09:20:33PM +0000, Andrew Walrond wrote: > On Thursday 16 Dec 2004 21:11, Greg KH wrote: > > > > Then I don't really know what to recommend. As the udev startup logic > > is very tightly tied to how the distro is set up, I recommend using > > whatever they do, and ignore what I say :) > > They is me; My distro is Rubyx :) Heh, ok. Then stick with what you are doing, as long as you rely on udevstart to create your device nodes, and not try to do it with a bash script (which is what the original start_udev did), you should be fine. > > > Is that list of 'extra nodes not exported by sysfs likely to change?' > > > > What does that list contain? > > make_extra_nodes () { > # there are a few things that sysfs does not export for us. > # these things go here (and remember to remove them in > # remove_extra_nodes() > # > # Thanks to Gentoo for the initial list of these. > ln -snf /proc/self/fd $udev_root/fd > ln -snf /proc/self/fd/0 $udev_root/stdin > ln -snf /proc/self/fd/1 $udev_root/stdout > ln -snf /proc/self/fd/2 $udev_root/stderr > ln -snf /proc/kcore $udev_root/core Those aren't nodes, they are symlinks. No way for udev to know about them :) > mkdir $udev_root/pts > mkdir $udev_root/shm Subdirs for mounting file systems on, again, no way udev can know about them. So, it looks like udev is really creating every device node you need. thanks, greg k-h