[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: How long is it acceptable to leave *undistributable* files in the kernel package?

Brian Thomas Sniffen wrote:
> Thiemo Seufer <ica2_ts@csv.ica.uni-stuttgart.de> writes:
> > The firmware typically wasn't patched, and nothing is derived from it.
> Isn't the kernel containing the firmware derivative of it?

AFAICS it contains not a derivative in the legal sense but the
original in a different representation.

> If not,
> why can't I put some GPL-incompatible x86 code into the kernel, load
> it into a device in my system -- the main memory -- and then issue a
> command to the processor to execute it?

You surely can, nothing prevents you from doing so. Distributing that
combination is the problematic part.

> That is, doesn't your interpretation allow arbitrary linking of GPL'd
> programs?

>From the copyright law POV, you most likely need to derive some call
wrapper for this, which needs some compatible licence for further
distribution. For an requirement to GPL the binary you have to prove
it includes important parts of/ideas from a GPL'ed work.

> as to why the GPL prohibits
> distributing linkages of GPL'd and GPL-incompatible code.

It doesn't. If some work includes a GPL'ed work and is distributed,
then the whole work must be GPL compatible. This doesn't extend to a
collection of works.

For some binary driver, you have to show why a bunch of kernel modules
is included in the kernel, and why the fact you can drop them easily
from a compilation doesn't reinforce the idea it's a collection of

For some firmware, you have to explain how this provision becomes
relevant, that is why the definition of "source" makes sense for it,
why the presence of this firmware constitutes an inclusion when it's
only use is to be loaded in a separate device, and how exactly
moving it to a userland file makes a difference.

> I would be much more convinced if I saw an argument from the
> GPL-incompatible-firmware-is-OK side

I don't say "GPL incompatible" firmware is OK. I say I can't prove it's
_not_ OK, while its copyright holder claims it is. Without proof to the
contrary, I'd rather follow his idea of what his work actually is.


Reply to: