> On Aug 6, 2014, at 2:55 PM, Justin Pettit <jpet...@nicira.com> wrote:
> 
> A future commit will add more L4 state fields that will be part of the
> L3 section.
> 
> Signed-off-by: Justin Pettit <jpet...@nicira.com>
> ---
> lib/flow.h |    5 ++++-
> 1 files changed, 4 insertions(+), 1 deletions(-)
> 
> diff --git a/lib/flow.h b/lib/flow.h
> index 3b8d24d..250a10e 100644
> --- a/lib/flow.h
> +++ b/lib/flow.h
> @@ -121,7 +121,10 @@ struct flow {
>     uint8_t arp_sha[6];         /* ARP/ND source hardware address. */
>     uint8_t arp_tha[6];         /* ARP/ND target hardware address. */
>     struct in6_addr nd_target;  /* IPv6 neighbor discovery (ND) target. */
> -    ovs_be16 tcp_flags;         /* TCP flags. With L3 to avoid matching L4. 
> */
> +    /* These fields are associated with L4 state.  Rules are often
> +     * written over the L3 fields and the L4 state (without referencing
> +     * the L4 headers, so we lump these with the L3 section. */

How about:

"These fields are associated with L4 state, but keeping them away from the L4 
stage in lookups results in better datapath wildcarding."

> +    ovs_be16 tcp_flags;         /* TCP flags. */
>     ovs_be16 pad;               /* Padding. */
> 
>     /* L4 */
> -- 
> 1.7.5.4
> 
> _______________________________________________
> dev mailing list
> dev@openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev

Reply via email to