[dpdk-dev] [PATCH v3 00/20] DPDK PMD for ThunderX NIC device

Ferruh Yigit ferruh.yigit at intel.com
Wed Jun 8 15:15:18 CEST 2016


On 6/8/2016 1:43 PM, Jerin Jacob wrote:
> On Wed, Jun 08, 2016 at 01:30:28PM +0100, Ferruh Yigit wrote:
>> On 6/7/2016 5:40 PM, Jerin Jacob wrote:
>>> Jerin Jacob (20):
>>>   thunderx/nicvf/base: add hardware API for ThunderX nicvf inbuilt NIC
>>>   thunderx/nicvf: add pmd skeleton
>>>   thunderx/nicvf: add link status and link update support
>>>   thunderx/nicvf: add get_reg and get_reg_length support
>>>   thunderx/nicvf: add dev_configure support
>>>   thunderx/nicvf: add dev_infos_get support
>>>   thunderx/nicvf: add rx_queue_setup/release support
>>>   thunderx/nicvf: add tx_queue_setup/release support
>>>   thunderx/nicvf: add rss and reta query and update support
>>>   thunderx/nicvf: add mtu_set and promiscuous_enable support
>>>   thunderx/nicvf: add stats support
>>>   thunderx/nicvf: add single and multi segment tx functions
>>>   thunderx/nicvf: add single and multi segment rx functions
>>>   thunderx/nicvf: add dev_supported_ptypes_get and rx_queue_count
>>>     support
>>>   thunderx/nicvf: add rx queue start and stop support
>>>   thunderx/nicvf: add tx queue start and stop support
>>>   thunderx/nicvf: add device start,stop and close support
>>>   thunderx/config: set max numa node to two
>>>   thunderx/nicvf: updated driver documentation and release notes
>>>   maintainers: claim responsibility for the ThunderX nicvf PMD
>>>
>>
>> Hi Jerin,
>>
>> In patch subject, as tag, other drivers are using only driver name, and
>> Intel drivers also has "driver/base", since base code has some special
>> case. For thunderx, what do you think about keeping subject as:
>>  "thunderx: ...."
>>
> 
> Hi Ferruh,
> 
> We may add crypto or other builtin ThunderX HW accelerated block drivers
> in future to DPDK.
> So that is the reason why I thought of keeping the subject as thunderx/nicvf.
> If you don't have any objection then I would like to keep it as
> thunderx/nicvf or just nicvf.
> 

Ring has similar problem, but we are using same tag "ring:" for both
ring_pmd and ring library.

For this case perhaps we can use net/thunderx, crypto/thunderx ?

I am not aware of any defined convention for the case.

Thanks,
ferruh





More information about the dev mailing list