From owner-freebsd-hackers  Wed Oct 18 00:19:11 1995
Return-Path: owner-hackers
Received: (from root@localhost)
          by freefall.freebsd.org (8.6.12/8.6.6) id AAA02432
          for hackers-outgoing; Wed, 18 Oct 1995 00:19:11 -0700
Received: from godzilla.zeta.org.au (godzilla.zeta.org.au [203.2.228.19])
          by freefall.freebsd.org (8.6.12/8.6.6) with ESMTP id AAA02421
          for <hackers@freebsd.org>; Wed, 18 Oct 1995 00:19:04 -0700
Received: (from bde@localhost) by godzilla.zeta.org.au (8.6.9/8.6.9) id RAA16864; Wed, 18 Oct 1995 17:14:55 +1000
Date: Wed, 18 Oct 1995 17:14:55 +1000
From: Bruce Evans <bde@zeta.org.au>
Message-Id: <199510180714.RAA16864@godzilla.zeta.org.au>
To: davidg@root.com, dennis@etinc.com
Subject: Re: Shutdown Procedure
Cc: hackers@freebsd.org
Sender: owner-hackers@freebsd.org
Precedence: bulk

>I noticed that if_ed.c calls ed_registerdev() at the beginning of the probe
>routine rather than after if_attach() as in previous releases and as some
>other drivers do. I've also noticed that the device is in the table even if
>it is disabled. What is the reason for the change?

So that devices are registered even if they aren't configured.  Perhaps
there should be more states to distinguish probed nonexistent devices
from probed existent ones and attached devices from unattached ones;
currently all probed existent devices are attached so there would be
no devices in the new states.

>Who/what uses the kdc_state information, and what is the impact of a device
>being IDLE or BUSY?

Currently, mainly lsdev.  Users looking at lsdev output would be
confused if the device state is reported incorrectly.

Bruce