Linux Ethernet Bridging - Aug 2020 - [Bridge] [RFC PATCH bpf-next 2/3] bpf: Add helper to do forwarding lookups in kernel FDB table

On 7/30/20 10:44 PM, Yoshiki Komachi wrote:
> diff --git a/net/core/filter.c b/net/core/filter.c
> index 654c346b7d91..68800d1b8cd5 100644
> --- a/net/core/filter.c
> +++ b/net/core/filter.c
> @@ -5084,6 +5085,46 @@ static const struct bpf_func_proto
bpf_skb_fib_lookup_proto = {
>  	.arg4_type	= ARG_ANYTHING,
>  };
>  
> +#if IS_ENABLED(CONFIG_BRIDGE)
> +BPF_CALL_4(bpf_xdp_fdb_lookup, struct xdp_buff *, ctx,
> +	   struct bpf_fdb_lookup *, params, int, plen, u32, flags)
> +{
> +	struct net_device *src, *dst;
> +	struct net *net;
> +
> +	if (plen < sizeof(*params))
> +		return -EINVAL;
I need to look at the details more closely, but on first reading 2
things caught me eye:
1. you need to make sure flags is 0 since there are no supported flags
at the moment, and
> +
> +	net = dev_net(ctx->rxq->dev);
> +
> +	if (is_multicast_ether_addr(params->addr) ||
> +	    is_broadcast_ether_addr(params->addr))
> +		return BPF_FDB_LKUP_RET_NOENT;
> +
> +	src = dev_get_by_index_rcu(net, params->ifindex);
> +	if (unlikely(!src))
> +		return -ENODEV;
> +
> +	dst = br_fdb_find_port_xdp(src, params->addr, params->vlan_id);
2. this needs to be done via netdev ops to avoid referencing bridge code
which can be compiled as a module. I suspect the build robots will id
this part soon.

> 2020/08/01 2:15?David Ahern <dsahern at gmail.com>????:
> 
> On 7/30/20 10:44 PM, Yoshiki Komachi wrote:
>> diff --git a/net/core/filter.c b/net/core/filter.c
>> index 654c346b7d91..68800d1b8cd5 100644
>> --- a/net/core/filter.c
>> +++ b/net/core/filter.c
>> @@ -5084,6 +5085,46 @@ static const struct bpf_func_proto
bpf_skb_fib_lookup_proto = {
>> 	.arg4_type	= ARG_ANYTHING,
>> };
>> 
>> +#if IS_ENABLED(CONFIG_BRIDGE)
>> +BPF_CALL_4(bpf_xdp_fdb_lookup, struct xdp_buff *, ctx,
>> +	   struct bpf_fdb_lookup *, params, int, plen, u32, flags)
>> +{
>> +	struct net_device *src, *dst;
>> +	struct net *net;
>> +
>> +	if (plen < sizeof(*params))
>> +		return -EINVAL;
> 
> I need to look at the details more closely, but on first reading 2
> things caught me eye:
> 1. you need to make sure flags is 0 since there are no supported flags
> at the moment, and
Thanks for your initial comments!

I will make sure whether this flag is required or not.
>> +
>> +	net = dev_net(ctx->rxq->dev);
>> +
>> +	if (is_multicast_ether_addr(params->addr) ||
>> +	    is_broadcast_ether_addr(params->addr))
>> +		return BPF_FDB_LKUP_RET_NOENT;
>> +
>> +	src = dev_get_by_index_rcu(net, params->ifindex);
>> +	if (unlikely(!src))
>> +		return -ENODEV;
>> +
>> +	dst = br_fdb_find_port_xdp(src, params->addr, params->vlan_id);
> 
> 2. this needs to be done via netdev ops to avoid referencing bridge code
> which can be compiled as a module. I suspect the build robots will id
> this part soon.
I guess that no build errors will occur because the API is allowed when
CONFIG_BRIDGE is enabled.

I successfully build my kernel applying this patch, and I don?t receive any
messages from build robots for now.

Thanks & Best regards,


?
Yoshiki Komachi
komachi.yoshiki at gmail.com