Re: [PATCH -next] hid: fix hid-steelseries kconfig/build

[simon@mungewell.org]

> On 05/01/13 12:27, simon@mungewell.org wrote:
>> Hi Randy and all,
>> Seems like you found a problem... but the relevant sections of
>> 'hid-steelseries.c' already have
>> --
>> #if defined(CONFIG_LEDS_CLASS) || defined(CONFIG_LEDS_CLASS_MODULE)
>
> which should be replaced with:
>
> #include <linux/kconfig.h>
>
> #if IS_ENABLED(LEDS_CLASS)
>
>
>> ...
>> #endif
>> --
>>
>> Shouldn't this prevent the module having calls to register/unregister if
>> the LED_CLASS is not enabled?
>
> Please read the patch description.  CONFIG_LEDS_CLASS=m but
> CONFIG_HID_STEELSEREIS=y.
> A builtin driver cannot make calls to a modular driver if the modular
> driver
> is not loaded.  Is there a way to require that the modular driver is
> loaded
> for hid-steelseries?  Maybe so, but I don't know.

I had a look around what other drivers do, but there is not much 'case
law' to follow (or copy from).

'drivers/media/radio/radio-shark.c' does it like this:
--
#if defined(CONFIG_LEDS_CLASS) || \
    (defined(CONFIG_LEDS_CLASS_MODULE) && defined(CONFIG_RADIO_SHARK_MODULE))
...
#endif
--

Which I think is another solution, and perhaps a little more
readable/understandable (ie. you have led control if LED_CLASS is
built-in, or both the driver _and_ LED_CLASS are modules).


If there is not an objection to this style I'll prepare and test a patch
in the next couple of days.

Simon.

PS. 'drivers/platform/x86/fujitsu-laptop.c' might have the same issues as
hid-steelseries....

>
>
>> Does forcing a 'depends on LED_CLASS' in Kconfig prevent the
>> hid-steelseries module being built on systems without LEDs, or is this
>> simply a way to ensure that the LED_CLASS module gets loaded first?
>
> It prevents the driver from being built.  I think that a satisfactory
> solution
> would be this small change:
>
> 	depends on LEDS_CLASS || LEDS_CLASS=n
>
> That makes HID_STEELSERIES depend on LEDS_CLASS when it is enabled but
> still
> allows the driver to build when LEDS_CLASS=n.  Makes sense?
>
>
>
> Oh, and please don't top-post.
>
> Thanks.
>
>>
>> Simon.
>>
>>> From: Randy Dunlap <rdunlap@infradead.org>
>>>
>>> Fix hid-steelseries build by making it depends on LEDS_CLASS.
>>> Build errors happen when LEDS_CLASS=m  and HID_STEELSERIES=y.
>>>
>>> drivers/built-in.o: In function `steelseries_srws1_remove':
>>> hid-steelseries.c:(.text+0x3b97a1): undefined reference to
>>> `led_classdev_unregister'
>>> drivers/built-in.o: In function `steelseries_srws1_probe':
>>> hid-steelseries.c:(.text+0x3b9c51): undefined reference to
>>> `led_classdev_register'
>>> hid-steelseries.c:(.text+0x3b9ce5): undefined reference to
>>> `led_classdev_register'
>>> hid-steelseries.c:(.text+0x3b9d4b): undefined reference to
>>> `led_classdev_unregister'
>>>
>>> Signed-off-by: Randy Dunlap <rdunlap@infradead.org>
>>> ---
>>>  drivers/hid/Kconfig |    1 +
>>>  1 file changed, 1 insertion(+)
>>>
>>> --- linux-next-20130501.orig/drivers/hid/Kconfig
>>> +++ linux-next-20130501/drivers/hid/Kconfig
>>> @@ -610,6 +610,7 @@ config HID_SPEEDLINK
>>>  config HID_STEELSERIES
>>>  	tristate "Steelseries SRW-S1 steering wheel support"
>>>  	depends on HID
>>> +	depends on LEDS_CLASS
>>>  	---help---
>>>  	Support for Steelseries SRW-S1 steering wheel
>>>
>>> --
>
>
> --
> ~Randy
> --
> To unsubscribe from this list: send the line "unsubscribe linux-input" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>