I was *only* talking about Linus' position, which I admit was being
selective in that context.
My real point was this: It appears to me that NVIDIA have gone as far as
they can in releasing the code to their driver. It has certainly been my
own policy to do so with various code, and the result was not GPL because
of legal constraints.
Punishing a company who have, with goodwill, opened up their code as far as
they were allowed by preexisting agreements for license issues is not a
smart move, and will only hurt the free software community in the long run.
And to those who say 'well, just release the specs': Quite likely NVIDIA
did not design all the subsystems of their chips, but instead bought 'IP
block' licenses from someone else. The license NVIDIA have access to those
under probably will not allow that release, whether NVIDIA would like to
release that information or not.
Effectively, the binary part of the driver can be viewed as part of the
hardware, just as much as it can be viewed as part of the kernel. It is
constrained in hardware-like ways, not much like software at all.
My view, for what it's worth, is that if binary modules are not allowed by
the kernel being GPL, then it is worth going to some trouble to allow
binary hardware drivers by some other mechanism than a module, since it is
effectively impossible to change the license on the kernel now, as you
correctly point out. Even if they want to, many hardware vendors will not
be able to release full specifications or GPL code for quite some time, and
it is better to allow those that are motivated to to open up as much as
they can, than to require only that hardware for which full information or
GPL-able code is available to be used with Linux. And saying that the
vendor then has to assume all the maintenance trouble keeps the pressure on
them to evolve toward openness.
--On Saturday, January 04, 2003 02:12:09 -0500 Ryan Anderson
> On Sat, Jan 04, 2003 at 12:56:53PM +1300, Andrew McGregor wrote:
>> Linus has made it quite clear in the past that his position on binary
>> modules is that they are explicitly allowed, but that the maintainers of
>> such a thing 'get everything they deserve' in terms of maintenance
> I *really* think you need to do some searches on this list to verify
> this statement.
> Let me summarize what I remember from past discussions of this nature.
> Linus put his code under the GPL. Contributions came in, under the same
> license. At some point, the first binary only module showed up. When
> asked about the legality, Linus said something to the effect of, "I
> think they're ok."
> Note the lack of clarification from the other (miriad) copyright
> In summary - If you want to write binary only modules, you need to talk
> to a lawyer that understands the issues involved. "Linus said they were
> ok" doesn't even begin to encompass the number of copyright holders
> Ryan Anderson
> sometimes Pug Majere
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to firstname.lastname@example.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/