From mboxrd@z Thu Jan 1 00:00:00 1970 From: Hannes Frederic Sowa Subject: Re: ipv6: Should inet6_addr_del consider IFA_F_MANAGETEMPADDR? Date: Tue, 8 Apr 2014 22:14:30 +0200 Message-ID: <20140408201430.GD27255@order.stressinduktion.org> References: <53443F5D.6030509@web.de> Mime-Version: 1.0 Content-Type: text/plain; charset=utf-8 Cc: "netdev@vger.kernel.org" , jiri@resnulli.us, thaller@redhat.com To: Heiner Kallweit Return-path: Received: from order.stressinduktion.org ([87.106.68.36]:59077 "EHLO order.stressinduktion.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1756551AbaDHUOb (ORCPT ); Tue, 8 Apr 2014 16:14:31 -0400 Content-Disposition: inline In-Reply-To: <53443F5D.6030509@web.de> Sender: netdev-owner@vger.kernel.org List-ID: [cc Jiri + Thomas] Hi! On Tue, Apr 08, 2014 at 08:26:37PM +0200, Heiner Kallweit wrote: > I stumbled accross the fact that inet6_addr_add as well as inet6_addr_modify > consider IFA_F_TEMPORARY whilst inet6_addr_del does not. > > My understanding of IFA_F_MANAGETEMPADDR is that it allows userspace applications > to deal with temporary addresses w/o having to create / track (and possibly delete) > each and every temporary address. > > This works fine when creating / modifying a global address. However when a > global address is deleted the orphaned temporary addresses remain. > Shouldn't the userspace be able to set IFA_F_MANAGETEMPADDR also for RTM_DELADDR > to signalize that the kernel should delete all related temporary addresses as well? Sure, that is possible. I guess it was not done already because inet6_addr_del did not handle IFA_FLAGS at all. But it should be a very tiny change to pass that flag down. Please submit an iproute patch along with your changes. > It would be a rather small extension to manage_tempaddrs to add a "delete" parameter > and set valid_lft=prefered_lft=0 for each temporary address if set. I think valid_lft=prefered_lft=0; manage_tempaddr(...); addrconf_verify_rtnl(); is enough to clean up the old temporary address. No need for a further flag. > Sorry if this was discussed before and I miss something. I don't remember a discussion about that. ;) Bye, Hannes