Re: [PATCHv2 1/2] kernel/workqueue: Make schedule_on_each_cpu as EXPORT_SYMBOL_GPL

[Vlastimil Babka <vbabka@suse.cz>]

On 8/24/21 04:06, Dennis Zhou wrote:
> Hello,
> 
> On Tue, Aug 24, 2021 at 02:12:29AM +0530, Ritesh Harjani wrote:
>> Make schedule_on_each_cpu as EXPORT_SYMBOL_GPL
>> 
>> Signed-off-by: Ritesh Harjani <riteshh@linux.ibm.com>
>> ---
>> [v1 -> v2]: Use EXPORT_SYMBOL_GPL instead of EXPORT_SYMBOL
>> 
>>  kernel/workqueue.c | 1 +
>>  1 file changed, 1 insertion(+)
>> 
>> diff --git a/kernel/workqueue.c b/kernel/workqueue.c
>> index f148eacda55a..993f8983186d 100644
>> --- a/kernel/workqueue.c
>> +++ b/kernel/workqueue.c
>> @@ -3309,6 +3309,7 @@ int schedule_on_each_cpu(work_func_t func)
>>  	free_percpu(works);
>>  	return 0;
>>  }
>> +EXPORT_SYMBOL_GPL(schedule_on_each_cpu);
>> 
>>  /**
>>   * execute_in_process_context - reliably execute the routine with user context
>> --
>> 2.31.1
>> 
> 
> I think you missed the point of Christoph's comment. I agree with him
> and don't think a test justifies exporting of this particular function.

Could symbol namespaces help in these cases? It feels wrong to limit in-tree
test modules this way.
https://www.kernel.org/doc/html/latest/core-api/symbol-namespaces.html

> Thanks,
> Dennis
>