Re: [PATCH] ethdev: add function name to log message

[Ferruh Yigit <ferruh.yigit@intel.com>]

On 10/17/2018 10:26 AM, Gaëtan Rivet wrote:
> Hi Ferruh,
> 
> On Tue, Oct 16, 2018 at 04:55:43PM +0100, Ferruh Yigit wrote:
>> On 10/12/2018 3:54 PM, Stephen Hemminger wrote:
>>> On Fri, 12 Oct 2018 14:43:57 +0200
>>> Adrien Mazarguil <adrien.mazarguil@6wind.com> wrote:
>>>
>>>> On Fri, Oct 12, 2018 at 11:45:01AM +0100, Ferruh Yigit wrote:
>>>>> On 10/12/2018 11:42 AM, Ferruh Yigit wrote:  
>>>>>> On 10/11/2018 6:59 PM, Stephen Hemminger wrote:  
>>>>>>> @@ -161,8 +161,9 @@ extern "C" {
>>>>>>>  
>>>>>>>  extern int rte_eth_dev_logtype;
>>>>>>>  
>>>>>>> -#define RTE_ETHDEV_LOG(level, ...) \
>>>>>>> -	rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, "" __VA_ARGS__)
>>>>>>> +#define RTE_ETHDEV_LOG(level, fmt, ...)		\
>>>>>>> +	rte_log(RTE_LOG_ ## level, rte_eth_dev_logtype, \
>>>>>>> +		"%s():" fmt, __func__, ## __VA_ARGS__)  
>>>>>>
>>>>>> +1 to adding function name, but
>>>>>>
>>>>>> failsafe is giving build error [1] with clang because of ## usage [2], that is
>>>>>> why I add this as ` "" __VA_ARGS__` at first place but you can't do this trick
>>>>>> if __VA_ARGS__ used after fmt.
>>>>>>
>>>>>> I am not aware of a solution for this, __VA_OPT__(,) also didn't worked with clang.  
>>>>>
>>>>> +cc Adrien & Gaetan,
>>>>>
>>>>> I saw Adrien put some "workaround" to this for mlx5  
>>>>
>>>> Yes, through RTE_FMT() (rte_common.h). Something like this:
>>>>
>>>>  #define RTE_ETHDEV_LOG(level, fmt, ...) \
>>>>      rte_log(RTE_LOG_ ## level, \
>>>>              rte_eth_dev_logtype, \
>>>>              "%s():" fmt, \
>>>>              __func__, \
>>>>              ## __VA_ARGS__)
>>>>
>>>> Can be rewritten like that:
>>>>
>>>>  #define RTE_ETHDEV_LOG(level, ...) \
>>>>      rte_log(RTE_LOG_ ## level, \
>>>>              rte_eth_dev_logtype, \
>>>>              RTE_FMT("%s():" RTE_FMT_HEAD(__VA_ARGS__,), \
>>>>                      __func__, \
>>>>                      RTE_FMT_TAIL(__VA_ARGS__,)))
>>>>
>>>> Although not too pretty and convenient, it does the job. In short:
>>>>
>>>> - Remove "fmt" argument from prototype.
>>>> - Enclose format string and its arguments in RTE_FMT().
>>>> - Replace "fmt" with RTE_FMT_HEAD(__VA_ARGS__,).
>>>> - Replace "## __VA_ARGS__" with RTE_FMT_TAIL(__VA_ARGS__,).
>>>> - Yes, trailing commas are mandatory in RTE_FMT_(HEAD|TAIL)().
>>>> - Note it quietly appends a dummy "%.0s" argument to the format string.
>>>>
>>>>>> [1]
>>>>>> .../build/include/rte_ethdev.h:166:26: error: token pasting of ',' and
>>>>>> __VA_ARGS__ is a GNU extension [-Werror,-Wgnu-zero-variadic-macro-arguments]
>>>>>>                 "%s():" fmt, __func__, ## __VA_ARGS__)
>>>>>>                                        ^
>>>>>>
>>>>>> [2]
>>>>>> This seems because of "-pedantic" argument driver uses, and other PMDs using
>>>>>> "-pedantic", like mlx,  will have same error although they are disable by
>>>>>> default and error not observed in default build.
>>>>>>   
>>>>>   
>>>>
>>>
>>> Since zero varadic macros is a GNU extension, maybe just adding GNU source?
>>
>> I think `-pedantic` is preventing using extension whether we have `_GNU_SOURCE`
>> or not.
>>
>> Gaetan,
>>
>> Why we have `-pedantic` option enabled for failsafe?
> 
> Because I wanted to enforce it in failsafe, and core DPDK headers were
> compatible so it wasn't too much of a hassle.
> 
> If rte_ethdev.h is meant to lose this compatibility, then pedantic will
> be disabled upon including it. But I guess if it was kept compatible
> until now, then it was a deliberate effort?

I think it is deliberate effort, I remember Adrien put some patches in the past
to make DPDK headers pedantic compatible by adding __extension__ keywords.

I tried to use __extension__ for ## but not able to make it work, is there a way
to do this?

And I see the reasoning of making DPDK headers pedantic compatible, I think we
should keep them compatible.