From mboxrd@z Thu Jan 1 00:00:00 1970 From: Sowmini Varadhan Subject: Re: [PATCH RFC net-next 1/3] RDS-TCP: Make RDS-TCP work correctly when it is set up in a netns other than init_net Date: Thu, 30 Jul 2015 19:58:57 +0200 Message-ID: <20150730175857.GA20426@oracle.com> References: <89cdf26d7506a1d2d1a2d6cfda2fab9a5fc84ad6.1438245718.git.sowmini.varadhan@oracle.com> <55BA58D1.3050108@gmail.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: netdev@vger.kernel.org, cwang@twopensource.com, ajaykumar.hotchandani@oracle.com To: David Ahern Return-path: Received: from aserp1040.oracle.com ([141.146.126.69]:18289 "EHLO aserp1040.oracle.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754689AbbG3R7G (ORCPT ); Thu, 30 Jul 2015 13:59:06 -0400 Content-Disposition: inline In-Reply-To: <55BA58D1.3050108@gmail.com> Sender: netdev-owner@vger.kernel.org List-ID: On (07/30/15 11:03), David Ahern wrote: > >+ write_pnet(&conn->c_net, net); > > these are typically in wrappers like sock_net and sock_net_set : > >+ conn = rds_conn_create(&init_net, dp->dp_daddr, dp->dp_saddr, > >+ &rds_ib_transport, GFP_KERNEL); > > I forget what connection this is -- control channel? this is IB. It should/will eventually also use any net than init_net, but for the moment, I'd like to figure out the bigger issue of pernet vs per_subsys, which is harder to fix than the above (and which I'll happily fix later). I suspect that the right solution may be to have some notifier callbacks in rds_tcp that listen for ifdown and tear down the sockets, rather than wait for keepalive timeout. > Ditto here. yes, me too. --Sowmini