[PATCH v3 1/2] examples/ptpclient: add the check for PTP capability

[lihuisong (C)]

Hi Hemant and Ferruh,

在 2023/9/21 19:22, Hemant Agrawal 写道:
>> On 9/21/2023 10:18 AM, lihuisong (C) wrote:
>>> 在 2023/9/16 1:29, Ferruh Yigit 写道:
>>>> On 8/17/2023 9:42 AM, Huisong Li wrote:
>>>>> If a port doesn't support PTP, there is no need to keep running app.
>>>>> So this patch adds the check for PTP capability.
>>>>>
>>>>> Fixes: 7a04a4f67dca ("examples/ptpclient: enable Rx timestamp
>>>>> offload")
>>>>> Cc: stable at dpdk.org
>>>>>
>>>>> Signed-off-by: Huisong Li <lihuisong at huawei.com>
>>>>> Signed-off-by: Dongdong Liu <liudongdong3 at huawei.com>
>>>>> ---
>>>>>    examples/ptpclient/ptpclient.c | 5 +++++
>>>>>    1 file changed, 5 insertions(+)
>>>>>
>>>>> diff --git a/examples/ptpclient/ptpclient.c
>>>>> b/examples/ptpclient/ptpclient.c index cdf2da64df..181d8fb357 100644
>>>>> --- a/examples/ptpclient/ptpclient.c
>>>>> +++ b/examples/ptpclient/ptpclient.c
>>>>> @@ -196,6 +196,11 @@ port_init(uint16_t port, struct rte_mempool
>>>>> *mbuf_pool)
>>>>>          if (dev_info.rx_offload_capa & RTE_ETH_RX_OFFLOAD_TIMESTAMP)
>>>>>            port_conf.rxmode.offloads |=
>> RTE_ETH_RX_OFFLOAD_TIMESTAMP;
>>>>> +    else {
>>>>> +        printf("port(%u) doesn't support PTP: %s\n", port,
>>>>> +               strerror(-retval));
>>>>> +        return -ENOTSUP;
>>>>> +    }
>>>>>
>>>> I am not sure why TIMESTAMP offload is required for PTP, I think
>>>> there is a confusion.
>>> If TIMESTAMP offload is not required for PTP, there isn't PTP offload
>>> in ethdev lib.
>>>
>> What do you mean with "PTP offload"?
>>
>> If you check the ptpclient sample app, it parses ptp packets in the application.
>>
>>>>
>>>> Gagandeep, Hemant,
>>>> Can you please clarify why TIMESTAMP offload is enabled?
>>> looking forward to your reply.
> [Hemant] as explained in other mail, it is a requirement for dpaa2. So, we are just passing the offload argument.
>
> Well, currently there is no such offload to know HW PTP support in DPDK. It can be introduced.
I agree with you, Heman.
>
> And I agree the above else should not be there atleast w.r.t TIMESTAMP OFFLOAD.

Ack.

@Ferruh, I wonder what you think. Looking forward to your reply.

>
>>>> .