[dpdk-dev] [RFC] Kernel Control Path (KCP)
Wiles, Keith
keith.wiles at intel.com
Tue Jun 13 20:18:15 CEST 2017
> On Jun 13, 2017, at 1:04 PM, Dumitrescu, Cristian <cristian.dumitrescu at intel.com> wrote:
>
>
>
>> -----Original Message-----
>> From: dev [mailto:dev-bounces at dpdk.org] On Behalf Of Jay Rolette
>> Sent: Tuesday, June 13, 2017 7:00 PM
>> To: Yigit, Ferruh <ferruh.yigit at intel.com>
>> Cc: Thomas Monjalon <thomas at monjalon.net>; DPDK <dev at dpdk.org>
>> Subject: Re: [dpdk-dev] [RFC] Kernel Control Path (KCP)
>>
>> On Tue, Jun 13, 2017 at 12:21 PM, Ferruh Yigit <ferruh.yigit at intel.com>
>> wrote:
>>
>>> On 5/30/2017 11:55 AM, Thomas Monjalon wrote:
>>>> 26/05/2017 18:52, Ferruh Yigit:
>>>>> We are looking for re-sending [1] the Kernel Control Path (KCP)
>>>>> with some updates [2].
>>>>>
>>>>> Mainly this is an usability improvement for DPDK.
>>>>>
>>>>> And a quick reminder about what KCP is:
>>>>>
>>>>> "KCP is Linux virtual network interface that can control DPDK ports".
>>>>>
>>>>> So DPDK interfaces, somehow will be visible and it will be possible to
>>>>> use common Linux tools on DPDK interfaces.
>>>>
>>>> Reminder: the Mellanox PMDs live with their upstream kernel modules,
>>>> allowing such features.
>>>>
>>>> The best model would be to have control path in kernel for every PMDs.
>>>
>>> That is the intention with this feature.
>>>
>>>>
>>>> Anyway, do you think KCP (or NCI) could be upstreamed in any way?
>>>
>>> Unfortunately I believe the answer is same, it may not be possible to
>>> upsteam this kernel module. Should this fact block the feature?
>>>
>>
>> Upstream is better, but KCP is a nice quality-of-life feature that I'd like
>> to see go in regardless. Anything that helps make DPDK less "foreign" to
>> normal port configuration and status tools is goodness.
>>
>> Jay
>
> +1
+1
>
> Let's not block this feature based on Linux kernel upstream reasons.
Regards,
Keith
More information about the dev
mailing list