From owner-freebsd-new-bus@FreeBSD.ORG Thu Jan 29 11:59:07 2004 Return-Path: Delivered-To: freebsd-new-bus@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 76D8316A4CE; Thu, 29 Jan 2004 11:59:07 -0800 (PST) Received: from feith1.FEITH.COM (feith1.FEITH.COM [192.251.93.1]) by mx1.FreeBSD.org (Postfix) with ESMTP id 5A3DD43D31; Thu, 29 Jan 2004 11:59:06 -0800 (PST) (envelope-from john@feith.com) Received: from jwlab.FEITH.COM (jwlab.FEITH.COM [192.251.93.16]) by feith1.FEITH.COM (8.12.10+Sun/8.12.9) with ESMTP id i0TJx3AL000159; Thu, 29 Jan 2004 14:59:03 -0500 (EST) Received: (from john@localhost) by jwlab.FEITH.COM (8.11.7+Sun/8.11.7) id i0TJx2m25103; Thu, 29 Jan 2004 14:59:02 -0500 (EST) Date: Thu, 29 Jan 2004 14:59:02 -0500 (EST) From: John Wehle Message-Id: <200401291959.i0TJx2m25103@jwlab.FEITH.COM> To: imp@bsdimp.com Content-Type: text X-Scanned-By: MIMEDefang 2.39 X-Archived: cashew.FEITH.COM cc: new-bus@FreeBSD.org cc: mdodd@FreeBSD.org cc: jhb@FreeBSD.org Subject: Re: nasty device_delete_child interaction X-BeenThere: freebsd-new-bus@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: FreeBSD's new-bus architecture List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 29 Jan 2004 19:59:07 -0000 > Don't do that. You are duplicating the storage of children in two > places. If you need to cache a copy of a child, that's fine. > However, don't delete it explicitly in xxx_detach. > > I'd say that these drivers are wrong and should be fixed. What's the correct approach? Currently (at least in FreeBSD 4.9) if_xl.c uses device_add_child in the attach routine to add on miibus. It then calls device_delete_child in the detach routine to delete miibus (creating a nice symmetry). Should if_xl.c still call device_add_child in the attach and simply not call device_delete_child? Then who's responsible for deleting miibus when if_xl is unloaded? >: > It seems to me that any driver which calls device_delete_child >: > as part of detaching must also implement something like: > > No. They should avoid the problem by using newbus correctly. This > sort of solution just adds code to no good purpose. If the driver has cached a copy of a child, then doesn't xxx_child_detached still need to be implemented so the driver knows when the cached copy is invalid? -- John ------------------------------------------------------------------------- | Feith Systems | Voice: 1-215-646-8000 | Email: john@feith.com | | John Wehle | Fax: 1-215-540-5495 | | -------------------------------------------------------------------------