[ovs-dev] [PATCH] mac-learning: Implement per-port MAC learning fairness.

[Ethan Jackson]

Solution seems clean, I'm happy with this as well.

Acked-by: Ethan Jackson <ethan at nicira.com>


On Thu, Feb 12, 2015 at 11:02 AM, Alex Wang <alexw at nicira.com> wrote:
> Looks good to me,
>
>  /* A MAC learning table entry.
>> - * Guarded by owning 'mac_learning''s rwlock */
>> + * Guarded by owning 'mac_learning''s rwlock. */
>>  struct mac_entry {
>>      struct hmap_node hmap_node; /* Node in a mac_learning hmap. */
>>      time_t expires;             /* Expiration time. */
>> @@ -47,14 +108,30 @@ struct mac_entry {
>>      uint16_t vlan;              /* VLAN tag. */
>>
>>      /* The following are marked guarded to prevent users from iterating
>> over or
>> -     * accessing a mac_entry without hodling the parent mac_learning
>> rwlock. */
>> +     * accessing a mac_entry without holding the parent mac_learning
>> rwlock. */
>>      struct ovs_list lru_node OVS_GUARDED; /* Element in 'lrus' list. */
>>
>> -    /* Learned port. */
>> -    union {
>> -        void *p;
>> -        ofp_port_t ofp_port;
>> -    } port OVS_GUARDED;
>> +    /* Learned port.
>> +     *
>> +     * The client-specified data is mlport->port. */
>> +    struct mac_learning_port *mlport;
>>
>
>
> Simple C question, why don't we need to forward declare the struct
> 'mac_learning_port'?
> _______________________________________________
> dev mailing list
> dev at openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev