Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264064AbTKSNUV (ORCPT ); Wed, 19 Nov 2003 08:20:21 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264073AbTKSNUV (ORCPT ); Wed, 19 Nov 2003 08:20:21 -0500 Received: from hermine.idb.hist.no ([158.38.50.15]:58895 "HELO hermine.idb.hist.no") by vger.kernel.org with SMTP id S264064AbTKSNUL (ORCPT ); Wed, 19 Nov 2003 08:20:11 -0500 Message-ID: <3FBB7075.7060406@aitel.hist.no> Date: Wed, 19 Nov 2003 14:30:29 +0100 From: Helge Hafting Organization: AITeL, HiST User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.4) Gecko/20031010 Debian/1.4-6 X-Accept-Language: no, en MIME-Version: 1.0 To: "Robert L. Harris" CC: Linux-Kernel Subject: Re: Missing /dev/pts? References: <20031119125533.GN3351@rdlg.net> In-Reply-To: <20031119125533.GN3351@rdlg.net> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 3105 Lines: 68 Robert L. Harris wrote: > > Had to rebuild a machine due to a dead deathstar. It's 99% there at > this point I think except I can't log in via xterm, eterm or ssh. > Poking around, the most obvious factor is that /dev/pts is empty. > > I can see that /dev and /dev/pts mounted by using "df -k /dev /dev/pts" > which returns: > > Filesystem 1K-blocks Used Available Use% Mounted on > - 0 0 0 - /dev > devpts 0 0 0 - /dev/pts > > whish is identicle to the machine I'm currently logged into and is > working fine. /dev and /dev/pts are both compiled into the kernel and > are reported by "cat /proc/filesystms". Devfsd is up and running with a > command line of "/sbin/devfsd /dev". > > Anyone have any idea why this is malfunctioning? Both devfs and devpts can be troublesome in several ways. First, the obvious: Make sure you didn't mount devpts on /dev/pts first and then devfs on /dev later. That will hide the devpts mount completely. (Assuming there is a pts directory in /dev before devfs gets mounted. Many people haven't actually removed anything from the oldfashioned /dev after installing devfs because it isn't necessary.) You don't say what kernel version you're using. In 2.6, this problem exists when using devfs+devpts: Devfs documentation claims there is a problem using devpts with devfs. The same docs also claim that you don't need devpts because devfs does the same job and does it without those problems. This used to be true. Devfs is becoming obsolete though, and someone did a "cleanup" patch that removed the pts support from devfs, forcing everybody to use devpts wether they use devfs or not. Unfortunately, the bad devfs/devpts interactions remained. I recently got rid of devfs for this reason, I got tired of xterms that 1. didn't appear, often crashing the x server instead 2. appeared after 20 -60 seconds or so. And then I got lots because I clicked the xterm menu over and over. :-( This doesn't work, and won't likely be fixed ever now that devfs is being obsoleted. I recommend going back to oldfashioned /dev with plain device nodes while waiting for the udev stuff to become useable. Udev will probably achieve many of the same goals as devfs, such as: 1. showing device nodes only for those devices that actually have a driver in the running kernel, avoiding clutter and wondering which cdrom driver is the "right" one. 2. showing a device node for every driver in the kernel, no need to remember to use MAKEDEV or even mknod just because you got some experimental hardware that your distro vendor hasn't heard about yet. Udev will probably not get rid of the silly major/minor numbers though. The devfs approach didn't really need them, now they are being extended instead. At least we won't ever have to look them up and type them in if udev gets going. :-/ Helge Hafting - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/