Re: [PATCH] HID: uclogic: avoid linking common code into multiple modules

[Arnd Bergmann]

On Sat, Jun 8, 2024, at 20:28, José Expósito wrote:
> On Wed, May 29, 2024 at 11:48:05AM +0200, Arnd Bergmann wrote:

>> @@ -154,10 +152,8 @@ obj-$(CONFIG_HID_WINWING)	+= hid-winwing.o
>>  obj-$(CONFIG_HID_SENSOR_HUB)	+= hid-sensor-hub.o
>>  obj-$(CONFIG_HID_SENSOR_CUSTOM_SENSOR)	+= hid-sensor-custom.o
>>  
>> -hid-uclogic-test-objs		:= hid-uclogic-rdesc.o \
>> -				   hid-uclogic-params.o \
>> -				   hid-uclogic-rdesc-test.o
>> -obj-$(CONFIG_HID_KUNIT_TEST)	+= hid-uclogic-test.o
>> +hid-uclogic-test-objs		:= hid-uclogic-rdesc-test.o
>> +obj-$(CONFIG_HID_KUNIT_TEST)	+= hid-uclogic-test.o hid-uclogic-params.o hid-uclogic-params.o
>>  
>>  obj-$(CONFIG_USB_HID)		+= usbhid/
>>  obj-$(CONFIG_USB_MOUSE)		+= usbhid/
>
> I tested your patch with:
>
> 	hid-uclogic-objs		:= hid-uclogic-core.o \
> 					   hid-uclogic-rdesc.o \
> 					   hid-uclogic-params.o
> 	obj-$(CONFIG_HID_UCLOGIC)	+= hid-uclogic.o
> 	[...]
> 	hid-uclogic-test-objs		:= hid-uclogic-rdesc-test.o
> 	obj-$(CONFIG_HID_KUNIT_TEST)	+= hid-uclogic.o hid-uclogic-test.o
>
> And I think it is a bit more clear and it looks like it does the trick
> removing the warning.

Right, that seems fine.

> Also, with that change only "EXPORT_SYMBOL_GPL(uclogic_rdesc_template_apply);"
> is required. The other EXPORT_SYMBOL_GPL can be removed.
>
> However, I'm not sure about what are the best practices using EXPORT_SYMBOL_GPL
> and if it should be used for each function/data in the .h file. Maybe that's
> why you added them.

No, having only the single export is better here, you should
have as few of them as possible. I did picked the more complicated
approach as I wasn't sure if loading the entire driver from the
test module caused any problems. Let's use your simpler patch
then.

     Arnd