From mboxrd@z Thu Jan 1 00:00:00 1970 From: ebiederm@xmission.com (Eric W. Biederman) Subject: Re: [net-2.6 patch 1/2][NETNS] fix device renaming for sysfs Date: Wed, 04 Jun 2008 04:59:53 -0700 Message-ID: References: <20080430213113.410541118@localhost.localdomain> <20080430213320.825349005@localhost.localdomain> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: davem@davemloft.net, netdev@vger.kernel.org To: Daniel Lezcano Return-path: Received: from out02.mta.xmission.com ([166.70.13.232]:60585 "EHLO out02.mta.xmission.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751996AbYFDMAZ (ORCPT ); Wed, 4 Jun 2008 08:00:25 -0400 In-Reply-To: <20080430213320.825349005@localhost.localdomain> (Daniel Lezcano's message of "Wed, 30 Apr 2008 23:31:14 +0200") Sender: netdev-owner@vger.kernel.org List-ID: Daniel Lezcano writes: > When a netdev is moved across namespaces with the 'dev_change_net_namespace' > function, the 'device_rename' function is used to fixup kobject and refresh > the sysfs tree. The device_rename function will call kobject_rename and this > one will check if there is an object with the same name and this is the case > because we are renaming the object with the same name. > > The use of 'device_rename' seems for me wrong because we usually don't > rename it but just move it across namespaces. As we just want to do a > mini "netdev_[un]register", IMO the functions 'netdev_[un]register_kobject' > should be used instead, like an usual network device [un]registering. device_rename is correct. Because doing device_del; device_add removes any driver specific sysfs attributes. Causing your patch to introduce subtle driver breakage only when we move a network device between namespaces. Ouch! To support this operation we need something like device_rename that can preserve driver specific attributes, and generally handle movement between namespaces. Since device_rename already does 99% of what we need it to do I found it easier to extend device_rename to handle this case then to introduce yet another function into the kobject layer. > This patch replace device_rename by netdev_unregister_kobject, followed by > netdev_register_kobject. And sysfs comes a' tumblin' down. My apologies for not replying sooner I just spotted this. Eric