[PATCH 3/3] ASoC: AMD: add AMD ASoC ACP-I2S driver [v4]

[Alex Deucher]

On Thu, Aug 20, 2015 at 7:45 PM, Mark Brown <broonie at kernel.org> wrote:
> On Thu, Aug 20, 2015 at 07:11:22PM -0400, Alex Deucher wrote:
>> On Thu, Aug 20, 2015 at 6:31 PM, Mark Brown <broonie at kernel.org> wrote:
>
>> > DRM is a special case here since there has always been work to share
>> > bits of the code with other operating systems, the licensing for DRM is
>> > very unusual within the kernel.  A quick sampling of drivers suggests
>> > that this license is not universally used there either.
>
>> > There is also the issue I raised with the fact that your non-GPL license
>> > statement does not appear to correspond to the MODULE_LICENSE() that
>> > you've included which claims the code is GPLed.
>
>> Once again this is exactly what the other drm drivers and support
>> modules do.  AMD generally prefers to contribute code licensed under
>> an MIT/X11 style licence which is why we've done this.  IANAL, but I
>> think that is compatible with "GPL and additional rights".
>
> This is not a DRM driver, it is an ASoC driver, and like I say DRM is a
> bit special.  I do note that AMD has contributed at least its CCP and
> cpufreq drivers (indeed all their non-DRM code I could find with
> explicit license statements) under a normal kernel license.  As far as I
> can tell this licensing is entirely confined to the DRM drivers.
>
> I am not readily able to convince myself that this is compatible with
> the intent of exporting the ASoC APIs _GPL(), this looks like it can be
> used as the basis for a shim layer for non-GPL code (the licensing
> strategy seems very similar).  I'd need to think through this carefully.

If you have strong concerns, I can double check with our lawyers and
see about changing it, I'm just not sure how long that will take :(
The licensing tends to vary based on the teams involved.  The ACP
audio and i2s was developed by the GPU teams and we've mostly worked
on drm and Xorg stuff until now.

Alex