On 17 Mar 2022, at 2:01, Chris Mi wrote:
> On 2022-03-11 8:53 PM, Eelco Chaudron wrote:
>>
<SNIP>
>>>>>
>>>>> @@ -449,6 +462,7 @@ dpif_close(struct dpif *dpif)
>>>>> if (dpif) {
>>>>> struct registered_dpif_class *rc;
>>>>>
>>>>> + dpif_offload_close(dpif);
>>>> ** Not sure I understand, but why are we destroying the offload dpif class
>>>> here, it can be used by another dpif type.
>>>>
>>>> ** I guess this is all because your design has a 1:1 mapping? Guess it
>>>> should be two dpif_types that could share the same offload class type.
>>> Now it is moved to dpif_netlink_close().
>>>
>>> Except the 1:1 mapping comment which I think need Ilya's feedback, I have
>>> addressed your other comments.
>>> Thanks for your comments. The dpif-offload for dummy is not needed and
>>> removed.
>>> If needed, I can send v21.
>>
>> Thanks for taking care of the questions and fixing them in your sandbox.
>> I would prefer for you to not send any more revisions until we have a clear
>> answer from Ilya.
> Since Ilya didn't reply, I'll send a new version to reflect the latest change.
Well, my goal was to not do any more reviews until Ilya would reply, as every
review cycle takes up quite some time.
However I could not get v21 to apply cleanly, so I will hold off on any further
reviews of this series until we have a clear direction from Ilya.
//Eelco
_______________________________________________
dev mailing list
d...@openvswitch.org
https://mail.openvswitch.org/mailman/listinfo/ovs-dev
