article on GPLv3, Linux kernel, and Devices Rigged to Malfunction

Mark Brown broonie at
Tue Oct 24 12:02:23 UTC 2006

On Mon, Oct 23, 2006 at 02:19:49PM +0100, Alex Hudson wrote:
> On Mon, 2006-10-23 at 12:50 +0100, Mark Brown wrote:

> > There's also things like EXPORT_SYMBOL_GPL, kernel tainting when
> > proprietary modules are loaded and an advertised (and regularly used)
> > lack of stability in internal kernel APIs.

> I would say all those things are driven mostly by technical concerns
> rather than legal ones necessarily, and that in general they do not
> impact those who are writing non-free drivers, nor do they constitute
> "action" per se. 

Especially in the case of EXPORT_SYMBOL_GPL that's not really true: the
only technical effect is to impose a requirement on the licensing of any
kernel module that wishes to use the API provided.  This is all about
keeping things working in that area free (one could draw a comparison
with the FSF asking people making new libraries to use the GPL rather
than LGPL).

> For example, a number of companies have made a case for a consistent API
> if not ABI, and that has generally been rejected on the grounds of
> freedom to code, not freedom of code (which I think is a concise
> description of the difference between open source-thinking people and
> free software-thinking people).

In the case of APIs that's really the only approach that covers all
bases - third party free software modules are in the same position as
proprietary ones if they're kept out of tree.  It's also not unknown for
the people presenting these arguments to actually care about free
software but present a practical argument in order to better reach their
audience ("Even if I were to accept everything you say it still wouldn't
be a good idea to do what you ask because...").

"You grabbed my hand and we fell into it, like a daydream - or a fever."

More information about the Discussion mailing list