From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mail-wm0-f45.google.com ([74.125.82.45]:35402 "EHLO mail-wm0-f45.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932447AbbKMKyn (ORCPT ); Fri, 13 Nov 2015 05:54:43 -0500 Received: by wmdw130 with SMTP id w130so23772863wmd.0 for ; Fri, 13 Nov 2015 02:54:41 -0800 (PST) Date: Fri, 13 Nov 2015 11:54:35 +0100 From: Alexander Aring Subject: Re: 6lowpan socket Message-ID: <20151113105430.GA2306@omega> References: <20151112085333.GA3045@omega> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: Sender: linux-wpan-owner@vger.kernel.org List-ID: To: Baptiste Clenet Cc: Christopher Friedt , linux-wpan@vger.kernel.org On Fri, Nov 13, 2015 at 11:29:36AM +0100, Baptiste Clenet wrote: > 2015-11-12 9:53 GMT+01:00 Alexander Aring : > > Hi Baptiste, > > > > On Thu, Nov 12, 2015 at 01:14:48AM +0100, Baptiste Clenet wrote: > >> From Christopher: > >> > >> Oh - sorry. 6LowPAN uses IPv6. So it's equivalent to setting up an IPv6 socket. > >> > >> Just remember to set up routes as appropriate and it should work fine. > >> > >> Unfortunately, I think vger bounces email from android devices, so my > >> message won't make it out on the list. > >> > >> © > >> > >> Sent from my Android > >> > >> Yeah sure, it's equivalent, it's what I thought. I will search deeper > > > > I see you as well on the RIOT mailinglist. Do you has these issues with > > RIOT? > > > > Yes I'm in both mailing list. > The problem was not with Riot but openwrt. It works now, thanks you. > ok. > > You can check it with fakelb and bsd-netcat and create/connect udp v6 socket, > > if it works with a linux<->linux connection, otherwise something is > > wrong with interoperability (which you can might debug via sniffing). > > > > I know they have recently added the rfc6282 udp nhc compression, see > > [0]. We have nhc_udp enabled by default, (receive handling should have > > a always possible strategy). > > > > Usually you can disable UDP NHC by unloading the "nhc_udp" module if you > > built it as module. But then you can also not receive anymore other udp > > nhc iphc headers. > > > > To make a fine granularity setting to the nhc strategy, we need a > > userspace tool and interface for that. > > > > - Alex > > > > [0] https://github.com/RIOT-OS/RIOT/pull/4189 > > > Ok for nhc compression, I let the default configuration and it worked. > You should look for "default configuration" a look into rfc6282. There is a special port-range which can be compressed. Maybe you want to use that. It will be used automatically. > I'm wondering something else about IPV6, how to change the default > IPV6 that the at86rf2XX will trigger? > If we change the IPV6 of lowpan0 (with ip -6 addr del ... for > instance) and then add the new one, it doesn't affect the reg > IEEE_ADDR_0,1,2.... mac address != ip address. with ip -6 addr you mean == ip address. with ip link set dev xx address yyy == mac address. > So I wanted to use iwpan but I can only change the short_addr if I'm not wrong. short address is part of mac address which is currently not supported by 802.15.4 6lowpan. This is a bigger issue because the whole ipv6 stack/net core api does know only _ONE_ type of mac address and this has only _ONE_ length. There exists few ideas to handle that, but we need to talk about which is the right one... > Which way is the cleanest to edit the IPV6 address of the transceiver? > (and it will impact lowpan0) I thought about edit it in the at86rf23XX > driver. > There is a bug one some kernel versions, see [0]. If you want a new mac address, ifdown both wpan/lowpan interface -> then change mac address via ip link set ... (extended address). Then ifup both interfaces again. - Alex [0] https://git.kernel.org/cgit/linux/kernel/git/torvalds/linux.git/commit/net/mac802154/iface.c?id=09095fdc9e5d5438051fc4e92867f1aff764cd21