From mboxrd@z Thu Jan 1 00:00:00 1970 From: David Miller Subject: Re: [PATCH] Expose netdevice dev_id through sysfs Date: Sat, 19 Apr 2008 22:32:41 -0700 (PDT) Message-ID: <20080419.223241.252348841.davem@davemloft.net> References: <1208176322.31695.73.camel@pmac.infradead.org> <20080419.183341.46125500.davem@davemloft.net> <200804200921.09072.arvidjaar@mail.ru> Mime-Version: 1.0 Content-Type: Text/Plain; charset=us-ascii Content-Transfer-Encoding: 7bit Cc: linux-hotplug@vger.kernel.org, dwmw2@infradead.org, kay.sievers@vrfy.org, md@linux.it, harald@redhat.com, netdev@vger.kernel.org, schwidefsky@de.ibm.com To: arvidjaar@mail.ru Return-path: Received: from 74-93-104-97-Washington.hfc.comcastbusiness.net ([74.93.104.97]:37605 "EHLO sunset.davemloft.net" rhost-flags-OK-FAIL-OK-OK) by vger.kernel.org with ESMTP id S1750933AbYDTFcj (ORCPT ); Sun, 20 Apr 2008 01:32:39 -0400 In-Reply-To: <200804200921.09072.arvidjaar@mail.ru> Sender: netdev-owner@vger.kernel.org List-ID: From: Andrey Borzenkov Date: Sun, 20 Apr 2008 09:21:03 +0400 > On Sparc system we also have global device tree which provides unique > and persistent reference to every device. Solaris has no problems with > having same MAC for all interfaces. SunOS/Solaris names devices geographically on the I/O bus. So if you move the card from one slot to another, it's a different device and you have to change your configuration, which is pretty bogus. At least under Linux the identity follows the MAC address, which I definitely see as a superior behavior. The only problem is handling the multiple devices with same MAC case. > What is exactly wrong with using device topology path? This should > exist on any system, it is unique and it is persistent. Because if you move the device to a different PCI slot it should still show up as the same device, using the MAC address as the key. The Solaris behavior isn't so intelligent.