Re: [dpdk-dev] [dpdk-stable] [PATCH] net/dpaa2: fix build error about timesync functions

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

On 9/17/2020 4:40 PM, Hemant Agrawal wrote:
> On 9/17/2020 5:29 PM, Ferruh Yigit wrote:
>> On 9/17/2020 12:50 PM, Hemant Agrawal wrote:
>>>
>>> On 9/17/2020 5:08 PM, Ferruh Yigit wrote:
>>>> On 9/17/2020 3:03 AM, Di, ChenxuX wrote:
>>>>> Hi,
>>>>>
>>>>>> -----Original Message-----
>>>>>> From: Ferruh Yigit <ferruh.yigit@intel.com>
>>>>>> Sent: Wednesday, September 16, 2020 11:29 PM
>>>>>> To: Di, ChenxuX <chenxux.di@intel.com>; hemant.agrawal@nxp.com
>>>>>> Cc: sachin.saxena@nxp.com; stable@dpdk.org; dev@dpdk.org; Richardson,
>>>>>> Bruce <bruce.richardson@intel.com>
>>>>>> Subject: Re: [dpdk-stable] [PATCH] net/dpaa2: fix build error about timesync
>>>>>> functions
>>>>>>
>>>>>> On 9/15/2020 3:40 AM, Chenxu Di wrote:
>>>>>>> When the build option has '-DRTE_LIBRTE_IEEE1588=1', the announce of
>>>>>>> timesync functions will be build.
>>>>>>> However the dpdk_conf doesn't hav RTE_LIBRTE_IEEE1588 so that the file
>>>>>>> dpaa2_ptp.c will not be build.
>>>>>>> It cause the build error.
>>>>>>> This patch fixes it by adding set for dpdk_conf.
>>>>>>>
>>>>>>> Fixes: 184c39d16568 ("net/dpaa2: add DPRTC sub-module")
>>>>>>> Cc: stable@dpdk.org
>>>>>>>
>>>>>>> Signed-off-by: Chenxu Di <chenxux.di@intel.com>
>>>>>>> ---
>>>>>>>    drivers/net/dpaa2/meson.build | 4 ++++
>>>>>>>    1 file changed, 4 insertions(+)
>>>>>>>
>>>>>>> diff --git a/drivers/net/dpaa2/meson.build
>>>>>>> b/drivers/net/dpaa2/meson.build index 6dd0eb274..d9aadfdae 100644
>>>>>>> --- a/drivers/net/dpaa2/meson.build
>>>>>>> +++ b/drivers/net/dpaa2/meson.build
>>>>>>> @@ -17,6 +17,10 @@ sources = files('base/dpaa2_hw_dpni.c',
>>>>>>>            'mc/dpdmux.c',
>>>>>>>            'mc/dpni.c')
>>>>>>>
>>>>>>> +if '-DRTE_LIBRTE_IEEE1588=1' in get_option('c_args')
>>>>>>
>>>>>> The "RTE_LIBRTE_IEEE1588=1" can fail,
>>>>>> all places looking for "#ifdef RTE_LIBRTE_IEEE1588", so a "-Dc_args=-
>>>>>> DRTE_LIBRTE_IEEE1588" is more likely, but why not "-Dc_args=-
>>>>>> DRTE_LIBRTE_IEEE1588=666"
>>>>>>
>>>>>
>>>>> Yes, I will change it
>>>>>
>>>>>>> + dpdk_conf.set('RTE_LIBRTE_IEEE1588', 1) endif
>>>>>>> +
>>>>>>>    if dpdk_conf.has('RTE_LIBRTE_IEEE1588')
>>>>>>>        sources += files('mc/dprtc.c')
>>>>>>>        sources += files('dpaa2_ptp.c')
>>>>>>>
>>>>>>
>>>>>> Can't we just remove the conditional build:
>>>>>>
>>>>>>    -if dpdk_conf.has('RTE_LIBRTE_IEEE1588')
>>>>>>    -       sources += files('mc/dprtc.c')
>>>>>>    -       sources += files('dpaa2_ptp.c')
>>>>>>    -endif
>>>>>>    +sources += files('mc/dprtc.c')
>>>>>>    +sources += files('dpaa2_ptp.c')
>>>>>
>>>>> The announce of timesync functions are in the #define DRTE_LIBRTE_IEEE1588
>>>>> While the define of the functions are in the file 'dpaa2_ptp.c'.
>>>>> So they should be both build or not build by whether the build option 
>>>>> -DRTE_LIBRTE_IEEE1588=1 or not.
>>>>> So it seems not a good idea that remove the conditional.
>>>>>
>>>>
>>>> timesyncs_* dev_ops functions defined but not used is not big problem, only 
>>>> can increase the library size.
>>>>
>>>> I believe more concern is on:
>>>> 'RTE_PMD_REGISTER_DPAA2_OBJECT(dprtc, rte_dpaa2_dprtc_obj);'
>>>> which looks like register function to run in constructor when 'dpaa2_ptp.c' 
>>>> is compiled, but not sure affect of it.
>>>> It can be possible to wrap that call with 'RTE_LIBRTE_IEEE1588' ifdef, that 
>>>> should work.
>>>>
>>>> It is preferred to remove the compile time flag instead of finding ways to 
>>>> make it work.
>>>>
>>>> Let's wait for the dpaa2 maintainers' response, perhaps they can come with a 
>>>> smart way to remove the compile time flag.
>>>
>>> Hi Ferruh,
>>>
>>>     enabling  IEEE1588 causes some performance drop in the dpaa2 performance. 
>>> That is the reason, we have kept this code in compile time flag.
>>>
>>>
>>> However, we will work in future to make it run-time configurable but that 
>>> will require some code restructuring and it will be a moderate size work.
>>>
>>
>> OK, thanks Hemant.
>>
>> At least can it be possible to remove it from the build files, what do you 
>> think about wrapping those two files (or their relevant parts) with 
>> 'RTE_LIBRTE_IEEE1588' ifdef and remove the checks from meson file?
>>
> ok.

Hi Hemant, Chenxu,

Can you send a new version of the patch as discussed above?

Thanks,
ferruh