From owner-freebsd-current  Thu Dec  7 05:17:14 1995
Return-Path: owner-current
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.3/8.7.3) id FAA11031
          for current-outgoing; Thu, 7 Dec 1995 05:17:14 -0800 (PST)
Received: from ref.tfs.com (ref.tfs.com [140.145.254.251])
          by freefall.freebsd.org (8.7.3/8.7.3) with SMTP id FAA11025
          for <current@freebsd.org>; Thu, 7 Dec 1995 05:17:12 -0800 (PST)
Received: (from julian@localhost) by ref.tfs.com (8.6.12/8.6.9) id EAA18999; Thu, 7 Dec 1995 04:58:01 -0800
From: Julian Elischer <julian@ref.tfs.com>
Message-Id: <199512071258.EAA18999@ref.tfs.com>
Subject: Re: sysctl status right now, and plea for testing.
To: phk@critter.tfs.com (Poul-Henning Kamp)
Date: Thu, 7 Dec 1995 04:58:01 -0800 (PST)
Cc: se@ZPR.Uni-Koeln.DE, wollman@lcs.mit.edu, current@freebsd.org
In-Reply-To: <2910.818337607@critter.tfs.com> from "Poul-Henning Kamp" at Dec 7, 95 01:00:07 pm
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-current@freebsd.org
Precedence: bulk

> 
> > > It's not like it has ever been documented to a level where too many
> > > people would claim to know how it's supposed to work though... :-(
> > I really need this now..
> 
> And I do for PCCARD and Stefan does for PCI...
> 
> > 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..
> 
> Why does it need to put in devsw before probe ?
It's a SCOPE thing..
devsw entries are  PER DRIVER thing..

it's part of linking the driver to the system

devfs is a Per 'found' instance..

While it could be argued that you only bother puting in the devsw entries
on finding the first instance, it some how seems wrong to me..

not to worry though.. it could be done either way..
(once the code is in to do the work, it just depends on where
you call it from..)

remember that eventually devsw entries will go away anyhow :)



> 
> How about:
> 
> 	probe
> 		(may examine sysctl for hints)
> 
> 	attach
> 		(may examine sysctl for hints)
> 		devsw reg.
> 		sysctl reg.
> 		devfs reg.
sounds fair but the devsw reg is already done.. 
means we treat all devices the same 
whether or not they have h/w to probe..
(e.g. /dev/mem)