[ovs-dev] [PATCH] learning-switch: Wildcard more fields to avoid normalization warnings.

Justin Pettit jpettit at nicira.com
Wed Jun 1 18:13:28 UTC 2011


Looks correct to me.

--Justin


On Jun 1, 2011, at 9:34 AM, Ben Pfaff wrote:

> FWW_NW_TOS and some other L3 fields weren't wildcarded, but should have
> been.  It's better to start with all fields wildcarded and then unwildcard
> the ones that shouldn't be.
> 
> Reported-by: Jean Tourrilhes <jt at hpl.hp.com>
> ---
> lib/learning-switch.c |   11 ++++++-----
> 1 files changed, 6 insertions(+), 5 deletions(-)
> 
> diff --git a/lib/learning-switch.c b/lib/learning-switch.c
> index dc9af77..77791a5 100644
> --- a/lib/learning-switch.c
> +++ b/lib/learning-switch.c
> @@ -99,15 +99,16 @@ lswitch_create(struct rconn *rconn, const struct lswitch_config *cfg)
>     sw->ml = cfg->mode == LSW_LEARN ? mac_learning_create() : NULL;
>     sw->action_normal = cfg->mode == LSW_NORMAL;
> 
> -    flow_wildcards_init_exact(&sw->wc);
>     if (!cfg->exact_flows) {
>         /* We cannot wildcard all fields.
>          * We need in_port to detect moves.
>          * We need both SA and DA to do learning. */
> -        sw->wc.wildcards = (FWW_DL_TYPE | FWW_NW_PROTO
> -                            | FWW_TP_SRC | FWW_TP_DST);
> -        sw->wc.nw_src_mask = htonl(0);
> -        sw->wc.nw_dst_mask = htonl(0);
> +        flow_wildcards_init_catchall(&sw->wc);
> +        sw->wc.wildcards &= ~(FWW_IN_PORT | FWW_DL_SRC | FWW_DL_DST
> +                              | FWW_ETH_MCAST);
> +        sw->wc.vlan_tci_mask = htons(VLAN_CFI | VLAN_VID_MASK);
> +    } else {
> +        flow_wildcards_init_exact(&sw->wc);
>     }
> 
>     sw->default_queue = cfg->default_queue;
> -- 
> 1.7.4.4
> 
> _______________________________________________
> dev mailing list
> dev at openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev




More information about the dev mailing list