[alsa-devel] [RFC PATCH 8/9] soundwire: intel: remove platform devices and provide new interface

[Pierre-Louis Bossart]

On 9/18/19 7:06 AM, Greg KH wrote:
> On Tue, Sep 17, 2019 at 09:29:52AM -0500, Pierre-Louis Bossart wrote:
>> On 9/17/19 12:55 AM, Greg KH wrote:
>>> On Mon, Sep 16, 2019 at 04:23:41PM -0500, Pierre-Louis Bossart wrote:
>>>> +/**
>>>> + * sdw_intel_probe() - SoundWire Intel probe routine
>>>> + * @parent_handle: ACPI parent handle
>>>> + * @res: resource data
>>>> + *
>>>> + * This creates SoundWire Master and Slave devices below the controller.
>>>> + * All the information necessary is stored in the context, and the res
>>>> + * argument pointer can be freed after this step.
>>>> + */
>>>> +struct sdw_intel_ctx
>>>> +*sdw_intel_probe(struct sdw_intel_res *res)
>>>> +{
>>>> +	return sdw_intel_probe_controller(res);
>>>> +}
>>>> +EXPORT_SYMBOL(sdw_intel_probe);
>>>> +
>>>> +/**
>>>> + * sdw_intel_startup() - SoundWire Intel startup
>>>> + * @ctx: SoundWire context allocated in the probe
>>>> + *
>>>> + */
>>>> +int sdw_intel_startup(struct sdw_intel_ctx *ctx)
>>>> +{
>>>> +	return sdw_intel_startup_controller(ctx);
>>>> +}
>>>> +EXPORT_SYMBOL(sdw_intel_startup);
>>>
>>> Why are you exporting these functions if no one calls them?
>>
>> They are used in the next series, see '[RFC PATCH 04/12] ASoC: SOF: Intel:
>> add SoundWire configuration interface'
> 
> That wasn't obvious :)
> 
> Also, why not EXPORT_SYMBOL_GPL()?  :)

Since the beginning of this SoundWire work, the intent what that the 
code could be reused in non-GPL open-source circles, hence the dual 
license and EXPORT_SYMBOL.
That said, there are cases where the code only makes sense for Linux, or 
relies on symbols that are exported with EXPORT_SYMBOL_GPL, in those 
cases we rely on GPLv2 and EXPORT_SYMBOL_GPL. For this series I added a 
disclaimer in the cover letter that those parts need to be reviewed 
further to make sure there are no conflicts with GPL.
This is an RFC-level contribution to check if my understanding of the 
bus/device/driver model is aligned with recommendations. I've already 
made local improvements by fixing bisect issues, removing warnings, 
improved some sequences, and that GPL question will be revisited before 
I send a formal patch.