Re: [PATCH 4/7] net/packet: Ask driver for protocol if not provided by user

[Willem de Bruijn <willemdebruijn.kernel@gmail.com>]

On Mon, Jan 14, 2019 at 8:20 AM Maxim Mikityanskiy <maximmi@mellanox.com> wrote:
>
> If a socket was created with socket(AF_PACKET, SOCK_RAW, 0), the
> protocol number is unavailable. Try to ask the driver to extract it from
> the L2 header in order for skb_try_probe_transport_header to succeed.
>
> Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
> ---
>  net/packet/af_packet.c | 15 ++++++++++++---
>  1 file changed, 12 insertions(+), 3 deletions(-)
>
> diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
> index 8fc76e68777a..d1d89749a17a 100644
> --- a/net/packet/af_packet.c
> +++ b/net/packet/af_packet.c
> @@ -1850,6 +1850,15 @@ static int packet_rcv_spkt(struct sk_buff *skb, struct net_device *dev,
>         return 0;
>  }
>
> +static void packet_parse_headers(struct sk_buff *skb, struct socket *sock)
> +{
> +       if (!skb->protocol && sock->type == SOCK_RAW) {
> +               skb_reset_mac_header(skb);
> +               skb->protocol = dev_parse_header_protocol(skb);
> +       }
> +
> +       skb_try_probe_transport_header(skb);
> +}


In relation to the discussion at

  af_packet: fix raw sockets over 6in4 tunnel
  http://patchwork.ozlabs.org/patch/1023623/

if adding a new header_ops callback to parse link layer headers,
please have it return both protocol and link layer header length.

also, this information may be needed earlier in (t)packet_snd than the
transport header, so I would suggest not combining the two in a new
packet_parse_headers wrapper function.