From owner-freebsd-current Thu Dec 7 10:17:33 1995 Return-Path: owner-current Received: (from root@localhost) by freefall.freebsd.org (8.7.3/8.7.3) id KAA04787 for current-outgoing; Thu, 7 Dec 1995 10:17:33 -0800 (PST) Received: from kitten.mcs.com (Kitten.mcs.com [192.160.127.90]) by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id KAA04782 for ; Thu, 7 Dec 1995 10:17:29 -0800 (PST) Received: from mailbox.mcs.com (Mailbox.mcs.com [192.160.127.87]) by kitten.mcs.com (8.6.10/8.6.9) with SMTP id MAA02012; Thu, 7 Dec 1995 12:17:24 -0600 Received: by mailbox.mcs.com (/\==/\ Smail3.1.28.1 #28.5) id ; Thu, 7 Dec 95 12:17 CST Received: by mercury.mcs.com (/\==/\ Smail3.1.28.1 #28.5) id ; Thu, 7 Dec 95 12:17 CST Message-Id: Subject: Re: sysctl status right now, and plea for testing. To: julian@ref.tfs.com (Julian Elischer) Date: Thu, 7 Dec 1995 12:17:21 -0600 (CST) From: "Lars Fredriksen" Cc: phk@critter.tfs.com, se@ZPR.Uni-Koeln.DE, wollman@lcs.mit.edu, current@FreeBSD.org In-Reply-To: <199512071135.DAA18582@ref.tfs.com> from "Julian Elischer" at Dec 7, 95 03:35:22 am X-Mailer: ELM [version 2.4 PL24] Content-Type: text Sender: owner-current@FreeBSD.org Precedence: bulk Julian Elischer writes: > [stuff deleted] > I'm 80% through a few big changes to how devices are done.. > the JREMOD changes are now unconditional in my system here, > so every driver is called by the SYSINIT code at startup, well before > it get's asked to probe. At that time. it hooks itself into the > devsw tables etc. and for devices for which there is no more > probing to be done, (e.g. /dev/mem) it also > puts itself into the devfs at this time.. > This brings up the point.. > Should we have a single time drivers are called for initialisation, etc.? > I can see the driver being called once on loading or booting > to add devsw entried > once to probe > once to put sysctl stuff in > etc.. > I wouldn't mind making sure we co-ordinate on this.. > particularly, devfs and devconf look like they might be duplicating > some structure.. Hi Julian, I have a quick question here. Is there now a possibility that controller to name mapping change dynamically upon a failure or removal of controller/device? I am looking to avoid things like this: configuration: ed0 ed1 default route through ed1 and which is the way to get to the NFS server. Action: remove ed0 reboot new configuration: ed0 <--- this should still be ed1 IBM has that problem with their device to name configuration and it causes no end of troubles. Lars > > > > > > Well, if there is a user interface to query system configuration > > > variables (i.e. sysctl), why not use it. > > exactly. > I was thinking of having an 'info' page for each device in devfs > # cat /dev/fd0.info > Floppy Disk > IRQ=6 > DMA=4 > capacity=1.44MB > driver version 1.45 > # > sounds like sysctl is trying to do the same thing.. > :) > I don't quite understand how devconf and sysctl go together.. > do they? > > julian > > -- ------------------------------------------------------------------- Lars Fredriksen fredriks@mcs.com (home) lars@fredriks.pr.mcs.net (home-home) fredriks@asiago.cs.wisc.edu