[dpdk-dev] [PATCH 1/2] mbuf: new NSH packet type

[Olivier Matz]

Hi Jingjing,

On 05/03/2016 07:51 AM, Jingjing Wu wrote:
> Signed-off-by: Jingjing Wu <jingjing.wu at intel.com>
> ---
>  lib/librte_mbuf/rte_mbuf.h | 7 +++++++
>  1 file changed, 7 insertions(+)
> 
> diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> index 529debb..79edae3 100644
> --- a/lib/librte_mbuf/rte_mbuf.h
> +++ b/lib/librte_mbuf/rte_mbuf.h
> @@ -274,6 +274,13 @@ extern "C" {
>   */
>  #define RTE_PTYPE_L2_ETHER_LLDP             0x00000004
>  /**
> + * NSH (Network Service Header) packet type.
> + *
> + * Packet format:
> + * <'ether type'=0x894F>
> + */
> +#define RTE_PTYPE_L2_ETHER_NSH              0x00000005
> +/**
>   * Mask of layer 2 packet types.
>   * It is used for outer packet for tunneling cases.
>   */
> 

Acked-by: Olivier Matz <olivier.matz at 6wind.com>


I have no objection for this patch, but it makes me think about
2 things:

- we have the room for 16 types for each layer, maybe we should
  start to be careful about which types should be supported to
  avoid running out of types in the future.

- The types supported in outer and inner have diverged. It would
  have been better to have something like:

  #define RTE_PTYPE_INNER_$type (RTE_PTYPE_$type << 16)

  Because it would make the software using the packet types
  simpler.

It's maybe a bit late now because it would break the ABI, but
this is something we could keep in mind in case we change the
ABI for another reason.

Regards,
Olivier