On 1/17/2018 10:19 PM, Thomas Monjalon wrote:
> 17/01/2018 23:09, Thomas Monjalon:
>> 17/01/2018 22:57, Ferruh Yigit:
>>> port_id is now 16bits, update function parameter according.
>>>
>>> Fixes: 4c270218aa26 ("ethdev: support security APIs")
>>> Cc: sta...@dpdk.org
>>> Cc: declan.dohe...@intel.com
>>>
>>> Signed-off-by: Ferruh Yigit <ferruh.yi...@intel.com>
>>> Acked-by: Hemant Agrawal <hemant.agra...@nxp.com>
>>
>> Obviously,
>> Acked-by: Thomas Monjalon <tho...@monjalon.net>
>
> Not so obvious actually.
> It is a good fix, but an API change.
It is more like API fix.
While whole library is using 16bits for port_id, I think we shouldn't deliver
release with this specific API uses 8bits.
I am for getting this because what it does is wrong.
> This function was not declared experimental.
> It must wait 18.05.
>
> And the function has no doxygen!
There was a request to add it [1] but not received the patch yet.
[1]
https://dpdk.org/ml/archives/dev/2017-December/082824.html
> And the function was placed randomly in the middle of struct declarations!
> One more proof of the poor quality of rte_security stuff.
>
