On 23 October 2014 11:21, Joe Stringer wrote:
> Hi Jarno,
>
> On 7 October 2014 14:43, Jarno Rajahalme wrote:
>
>> flow inserts and removals are simplified:
>>
>> - No need for classifier internal mutex, as dpif-netdev already has a
>> 'flow_mutex'.
>> - Number of memory allocations/frees can
Hi Jarno,
On 7 October 2014 14:43, Jarno Rajahalme wrote:
> flow inserts and removals are simplified:
>
> - No need for classifier internal mutex, as dpif-netdev already has a
> 'flow_mutex'.
> - Number of memory allocations/frees can be halved.
>
> Lookup code path is a bit more effcient as w
On Oct 16, 2014, at 2:09 PM, Jarno Rajahalme wrote:
> Thanks for the reviews, Alex!
>
> On Oct 14, 2014, at 3:31 PM, Alex Wang wrote:
>>
>> +/* Insert 'rule' into 'cls'. */
>> +static void
>> +dpcls_insert(struct dpcls *cls, struct dpcls_rule *rule,
>> + const struct netdev_flow_k
Thanks for the reviews, Alex!
On Oct 14, 2014, at 3:31 PM, Alex Wang wrote:
> static inline struct netdev_flow_key *
> -miniflow_to_netdev_flow_key(const struct miniflow *mf)
> +miniflow_to_netdev_flow_key(const struct miniflow *miniflow)
> {
> -return (struct netdev_flow_key *) CONST_CAST(
On Oct 14, 2014, at 4:22 PM, Alex Wang wrote:
> the output of function 'netdev_flow_key_hash()' seems not got used,
> for dpcls_lookup(), we calculate the hash inside the function,
> for emc_lookup(), we use 'dpif_netdev_packet_get_dp_hash()'
> for flow_table(), we use the 'flow_hash(&flow->flow
>
> @ -2697,19 +2845,28 @@ fast_path_processing(struct dp_netdev_pmd_thread
>> *pmd,
>> ? &put_actions
>> : &actions;
>>
>> -ovs_mutex_lock(&dp->flow_mutex);
>> -/* XXX: There's a brief race where this flow could have
>> already
>> -
>
> static inline struct netdev_flow_key *
> -miniflow_to_netdev_flow_key(const struct miniflow *mf)
> +miniflow_to_netdev_flow_key(const struct miniflow *miniflow)
> {
> -return (struct netdev_flow_key *) CONST_CAST(struct miniflow *, mf);
> +struct netdev_flow_key *key;
> +
> +INIT_C
On Mon, Oct 13, 2014 at 6:51 PM, Alex Wang wrote:
> Hey Jarno,
>
> I have some high-level comments/questions as follow:
>
> On Tue, Oct 7, 2014 at 2:43 PM, Jarno Rajahalme
> wrote:
>
>> flow inserts and removals are simplified:
>>
>> - No need for classifier internal mutex, as dpif-netdev alread
Hey Jarno,
I have some high-level comments/questions as follow:
On Tue, Oct 7, 2014 at 2:43 PM, Jarno Rajahalme
wrote:
> flow inserts and removals are simplified:
>
> - No need for classifier internal mutex, as dpif-netdev already has a
> 'flow_mutex'.
>
I'm thinking maybe we should keep the
flow inserts and removals are simplified:
- No need for classifier internal mutex, as dpif-netdev already has a
'flow_mutex'.
- Number of memory allocations/frees can be halved.
Lookup code path is a bit more effcient as well, as we can rely on
netdev_flow_key always having inline data.
This w
10 matches
Mail list logo
