Return-Path: From: Alexander Aring To: linux-wpan@vger.kernel.org Cc: kernel@pengutronix.de, linux-bluetooth@vger.kernel.org, lukasz.duda@nordicsemi.no, jukka.rissanen@linux.intel.com, Alexander Aring Subject: [RFCv2 bluetooth-next 0/2] 6lowpan: introduce generic 6lowpan private data Date: Fri, 31 Jul 2015 14:38:06 +0200 Message-Id: <1438346288-14546-1-git-send-email-alex.aring@gmail.com> Sender: linux-wpan-owner@vger.kernel.org List-ID: Hi, I already thought many times to introduce something like that. Here is a draft for introducing the generic 6lowpan private data into each lowpan interface. For the beginning I introduced the enum "ll_type" which contains the LL type of a lowpan interface. Use cases for such feature (LL types): - If we do in upper layers (6LoWPAN/IPv6) a evaluation of dev->type then the value is on all lowpan interfaces "APRHRD_6LOWPAN". If we checked on "dev->type" and it's ARPHRD_6LOWPAN we can safe use lowpan_priv to get 6LoWPAN generic private data. With this data we can check the LL type which can be currently BTLE or IEEE802154. This could be useful to make different handling in iphc compress/decompress and evaluating LL private data of skb control block "skb->cb". Example (802.15.4 has different address handling functionality): switch (lowpan_priv(dev)->ll_type) { case LOWPAN_LL_TYPE_BTLE: /* do EUI64 btle handling */ break; case LOWPAN_LL_TYPE_IEEE802154: /* do complicated short/extended address handling */ /* we can surely call skb->cb to some other private data from LL which was set before iphc compress/decompress function call */ break; } The handling is currently for 802.15.4 is in generic 6lowpan code currently not quite. This should be handled by private data. At the moment btle use the same handling like for 802.15.4 extended address. Nevertheless is we can cleanup some handling then in generic iphc functionality. This also possible in layers like IPv6, just doing a check on APRHRD_6LOWPAN before calling lowpan_priv(dev)->ll_type, then we are sure that the private data of the interface is a 6LoWPAN interface and we can cast it to lowpan_priv. Then we can do 6LoWPAN generic stuff OR 6lowpan specific LL stuff. - In Lukasz Duda RFC for introducing stateful address compression, I saw some behaviour which such functionality is also useful. Currently we don't have a allocated space for a 6LoWPAN generic space which is assigned to a lowpan interface. When LL layers (BTLE, IEEE802154) calls iphc compress/decompress Lukasz had no change to access some room which is needed for storing the context information into a table. The workaround for this feature was to add allocate a static data room for the table and doing a lookup/match of netdev name. Which a generic 6lowpan private data room per each lowpan interface we can put this table according to the netdev private room, which means in short: no lookup is needed anymore, just dereferencing lowpan_priv. In upper layer like IPv6 Lukasz could use that to get the stateful context table information as an example for doing 6LoWPAN generic stuff in upper layers. Note about the implementation for LL 6LoWPAN private pointer: I stole some mechanism from wireless for that. See [0], the vif struct is part of sdata which is also part of private data of a wireless interface (mac80211). I hope I can just adapt this behaviour. - Alex [0] http://lxr.free-electrons.com/source/include/net/mac80211.h#L1366 changes since v2: - rename L2/l2 to LL/ll. I think the usually word in ietf for L2 is LL which means "link layer". Also ipv6 stack use ll for variables like lladdr -> link layer address. - rebase to current bluetooth-next/master - remove some brackets in function lowpan_priv. Alexander Aring (2): bluetooth: 6lowpan: change netdev_priv to lowpan_dev 6lowpan: add generic 6lowpan netdev private data include/net/6lowpan.h | 18 ++++++++++++++++++ net/bluetooth/6lowpan.c | 9 ++++++--- net/ieee802154/6lowpan/6lowpan_i.h | 3 ++- net/ieee802154/6lowpan/core.c | 5 ++++- 4 files changed, 30 insertions(+), 5 deletions(-) -- 2.5.0