[dpdk-dev] [PATCH] eal/linux: fix return after alarm registration failure

Burakov, Anatoly anatoly.burakov at intel.com
Wed Jun 26 14:52:17 CEST 2019


On 26-Jun-19 1:36 PM, Bruce Richardson wrote:
> On Wed, Jun 26, 2019 at 01:55:53PM +0200, Thomas Monjalon wrote:
>> 26/06/2019 13:43, Burakov, Anatoly:
>>> On 26-Jun-19 12:39 PM, David Marchand wrote:
>>>> On Wed, Jun 26, 2019 at 1:36 PM Thomas Monjalon <thomas at monjalon.net> wrote:
>>>>
>>>>> 26/06/2019 13:20, David Marchand:
>>>>>> On Wed, Jun 26, 2019 at 12:41 PM Thomas Monjalon <thomas at monjalon.net>
>>>>>> wrote:
>>>>>>
>>>>>>> When adding an alarm, if an error happen when registering
>>>>>>> the common alarm callback, it is not considered as a major failure.
>>>>>>> The alarm is then inserted in the list.
>>>>>>> However it was returning an error code after inserting the alarm.
>>>>>>>
>>>>>>> The error code is reset to 0 so the behaviour and the return code
>>>>>>> are consistent.
>>>>>>> Other return code related lines are cleaned up for easier
>>>>> understanding.
>>>>>>>
>>>>> [...]
>>>>>>> --- a/lib/librte_eal/linux/eal/eal_alarm.c
>>>>>>> +++ b/lib/librte_eal/linux/eal/eal_alarm.c
>>>>>>>           if (!handler_registered) {
>>>>>>> -               ret |= rte_intr_callback_register(&intr_handle,
>>>>>>> +               ret = rte_intr_callback_register(&intr_handle,
>>>>>>>                                   eal_alarm_callback, NULL);
>>>>>>> -               handler_registered = (ret == 0) ? 1 : 0;
>>>>>>> +               if (ret == 0)
>>>>>>> +                       handler_registered = 1;
>>>>>>> +               else
>>>>>>> +                       /* not fatal, callback can be registered later
>>>>> */
>>>>>>> +                       ret = 0;
>>>>>>>           }
>>>>>>
>>>>>> Well, then it means that you don't want to touch ret at all.
>>>>>> How about:
>>>>>> if (rte_intr_callback_register(&intr_handle,
>>>>>>                                  eal_alarm_callback, NULL) == 0)
>>>>>>           handler_registered = 1;
>>>>>>
>>>>>> ?
>>>>>
>>>>> Too much simple :)
>>>>>
>>>>> I think we try to avoid calling a function in a "if"
>>>>> per coding style.
>>>>> And my proposal has the benefit of offering a comment
>>>>> about the non-fatal error.
>>>>>
>>>>
>>>> /* not fatal, callback can be registered later */
>>>> if (rte_intr_callback_register(&intr_handle,
>>>>                                 eal_alarm_callback, NULL) == 0)
>>>>          handler_registered = 1;
>>>>
>>>
>>> I prefer the original. It's more explicit and conveys the intention
>>> better. Did i break the tie? :)
>>
>> I was going to send a v2 with David's suggestion.
>> Now I'm confused.
>>
> I always tend to prefer shorter versions, so +1 for v2 (does that make it a
> v3? :-) )
> 
> /Bruce
> 

OK, but then the suggested comment needs to be fixed. It makes it seem 
like registering the handler is the "non fatal" part. Perhaps something 
like:

/* failed register is not a fatal error - callback can be registered 
later */

-- 
Thanks,
Anatoly


More information about the dev mailing list