Re: [PATCHv4 bluetooth-next 3/8] ieee802154: 6lowpan: fix intra pan id check

[Alexander Aring <aar@pengutronix.de>]

Hi,

On 07/06/2016 11:32 PM, Alexander Aring wrote:
> The RIOT-OS stack does send intra-pan frames but don't set the intra pan
> flag inside the mac header. It seems this is valid frame addressing but
> inefficient. Anyway this patch adds a new function for intra pan
> addressing, doesn't matter if intra pan flag or source and destination
> are the same. The newly introduction function will be used to check on
> intra pan addressing for 6lowpan.
> 
> Signed-off-by: Alexander Aring <aar@pengutronix.de>
> ---
>  include/net/mac802154.h     | 19 +++++++++++++++++++
>  net/ieee802154/6lowpan/rx.c |  2 +-
>  2 files changed, 20 insertions(+), 1 deletion(-)
> 
> diff --git a/include/net/mac802154.h b/include/net/mac802154.h
> index ec01b35..d757edd 100644
> --- a/include/net/mac802154.h
> +++ b/include/net/mac802154.h
> @@ -346,6 +346,25 @@ static inline unsigned char *ieee802154_skb_src_pan(__le16 fc,
>  }
>  
>  /**
> + * ieee802154_skb_is_intra_pan_addressing - checks whenever the mac addressing
> + *	is an intra pan communication
> + * @fc: mac header frame control field
> + * @skb: skb where the source and destination pan should be get from
> + */
> +static inline bool ieee802154_skb_is_intra_pan_addressing(__le16 fc,
> +							  const struct sk_buff *skb)
> +{
> +	unsigned char *dst_pan = ieee802154_skb_dst_pan(fc, skb),
> +		      *src_pan = ieee802154_skb_src_pan(fc, skb);
> +
> +	/* if one is NULL is no intra pan addressing */
> +	if (!dst_pan || !src_pan)
> +		return false;
> +
> +	return !memcmp(dst_pan, src_pan, IEEE802154_PAN_ID_LEN);
> +}
> +
> +/**
>   * ieee802154_be64_to_le64 - copies and convert be64 to le64
>   * @le64_dst: le64 destination pointer
>   * @be64_src: be64 source pointer
> diff --git a/net/ieee802154/6lowpan/rx.c b/net/ieee802154/6lowpan/rx.c
> index ef185dd..649e7d45 100644
> --- a/net/ieee802154/6lowpan/rx.c
> +++ b/net/ieee802154/6lowpan/rx.c
> @@ -262,7 +262,7 @@ static inline bool lowpan_rx_h_check(struct sk_buff *skb)
>  
>  	/* check on ieee802154 conform 6LoWPAN header */
>  	if (!ieee802154_is_data(fc) ||
> -	    !ieee802154_is_intra_pan(fc))
> +	    !ieee802154_skb_is_intra_pan_addressing(fc, skb))
>  		return false;
>  

I had some dicussion at IETF that the destination pan id could be
different than the source pan. So we need to change it "again" to remove
it. :-)

On rx side:
 - dst pan id can be everything (also 0xfff?)
 - src pan cannot be 0xffff, which should be filtered before reaching
   packet layer.

On tx side:
 - We some mechanism to tell the destination pan, we need anyway a
   6lowpan netlink api, so we could add 802.15.4 6lowpan option on it,
   to change the destination panid per interface.

   Or we do what vlan does:

# ip link add link eth0 name eth0.100 type vlan id 100

e.g.

# ip link add link wpan0 name lowpan0 type lowpan pan id 0xbeef

which can be handled by ip plugin command stuff.

For such reason mutliple lowpan should be possible to setup. Currently
the UAPI is made for that, but we don't allow it.

- Alex