[PATCH] examples/distributor: update dynamic configuration

[Hunt, David]

Hi Ömer,

On 27/06/2022 17:28, Omer Yamac wrote:
> Hi David,
>
> Thank you for your review. I have two questions. The first one is 
> about new release. As I remember new DPDK realize will be published in 
> a short time and my previous fix in that release. Therefore, Should I 
> wait for that release to submit patch?
>

Yes, You can wait, or you can submit to the mailing list now and mark 
the patch as "Deferred" in patchwork. Once 22.07 is released it will get 
marked as "New", and under consideration for 22.11.


> The other question is below,
>
> On 27.06.2022 18:51, Hunt, David wrote:
>> Hi Ömer,
>>
>> I've a few comments:
>>
>> On 21/06/2022 21:15, Abdullah Ömer Yamaç wrote:
>>> In this patch,
>>>     * It is possible to switch the running mode of the distributor
>>> using the command line argument.
>>>     * With "-c" parameter, you can run RX and Distributor
>>> on the same core.
>>>     * Without "-c" parameter, you can run RX and Distributor
>>> on the different core.
>>>     * Syntax error of the single RX and distributor core is fixed.
>>>     * When "-c" parameter is active, the wasted distributor core is
>>> also deactivated in the main function.
>>>
>>> Fixes: 4a7f40c0ff9a ("examples/distributor: add dedicated core")
>>> Cc: stable at dpdk.org
>>>
>>> Signed-off-by: Abdullah Ömer Yamaç <omer.yamac at ceng.metu.edu.tr>
>>>
>>> --- 


--snip--


>>>                   "1 lcore for stats (can be core 0)\n"
>>> @@ -733,6 +808,15 @@ main(int argc, char *argv[])
>>>                   "1 lcore for packet TX\n"
>>>                   "and at least 1 lcore for worker threads\n");
>>>   +    if (rte_lcore_count() < 4 && enable_lcore_rx_distributor)
>>> +        rte_exit(EXIT_FAILURE, "Error, This application needs at "
>>> +                "least 4 logical cores to run:\n"
>>> +                "1 lcore for stats (can be core 0)\n"
>>> +                "1 lcore for packet RX and distribution\n"
>>> +                "1 lcore for packet TX\n"
>>> +                "and at least 1 lcore for worker threads\n");
>>> +
>>
>>
>> the two checks above could be replaced with something like:
>>
>> min_cores = 4 + enable_lcore_rx_distributor;
>> if (rte_lcore_count() < min_cores)
>>                 rte_exit(EXIT_FAILURE, "Error, This application needs 
>> at "
>>                                 "least %d logical cores to run:\n"
>>                                 "1 lcore for stats (can be core 0)\n"
>>                                 "1 lcore for packet RX\n"
>>                                 "1 lcore for distribution\n"
>>                                 "1 lcore for packet TX\n"
>>                                 "and at least 1 lcore for worker 
>> threads\n",
>>                                 min_cores);
>>
> Is it okay, if I change the error string such that:
> "1 or 2 lcore for packet RX and distribution"
>

Sure, that's fine.



--snip--


Thanks,

Dave.