Re: [PATCH] mac80211: keep mesh ifaces in allmulti mode

[Johannes Berg <johannes@sipsolutions.net>]

[-- Attachment #1: Type: text/plain, Size: 1558 bytes --]

On Wed, 2008-08-06 at 13:17 +0200, Luis Carlos Cobo wrote:
> Currently a mesh node will not forward a multicast frame if it is not subscribed
> to the specific multicast address. This patch addresses the issue and fixes mesh
> multicast forwarding.
> 
> Signed-off-by: Luis Carlos Cobo <luisca@cozybit.com>

Acked-by: Johannes Berg <johannes@sipsolutions.net>

> ---
>  net/mac80211/main.c |    8 +++++++-
>  1 files changed, 7 insertions(+), 1 deletions(-)
> 
> diff --git a/net/mac80211/main.c b/net/mac80211/main.c
> index 0c02c47..aa5a191 100644
> --- a/net/mac80211/main.c
> +++ b/net/mac80211/main.c
> @@ -245,10 +245,13 @@ static int ieee80211_open(struct net_device *dev)
>  	case IEEE80211_IF_TYPE_AP:
>  		sdata->bss = &sdata->u.ap;
>  		break;
> +	case IEEE80211_IF_TYPE_MESH_POINT:
> +		/* mesh ifaces must set allmulti to forward mcast traffic */
> +		atomic_inc(&local->iff_allmultis);
> +		break;
>  	case IEEE80211_IF_TYPE_STA:
>  	case IEEE80211_IF_TYPE_MNTR:
>  	case IEEE80211_IF_TYPE_IBSS:
> -	case IEEE80211_IF_TYPE_MESH_POINT:
>  		/* no special treatment */
>  		break;
>  	case IEEE80211_IF_TYPE_INVALID:
> @@ -495,6 +498,9 @@ static int ieee80211_stop(struct net_device *dev)
>  		netif_addr_unlock_bh(local->mdev);
>  		break;
>  	case IEEE80211_IF_TYPE_MESH_POINT:
> +		/* allmulti is always set on mesh ifaces */
> +		atomic_dec(&local->iff_allmultis);
> +		/* fall through */
>  	case IEEE80211_IF_TYPE_STA:
>  	case IEEE80211_IF_TYPE_IBSS:
>  		sdata->u.sta.state = IEEE80211_DISABLED;

[-- Attachment #2: This is a digitally signed message part --]
[-- Type: application/pgp-signature, Size: 836 bytes --]