[nycbug-talk] kernels

Roland C. Dowdeswell elric
Thu Jun 3 14:46:38 EDT 2004


On 1086285136 seconds since the Beginning of the UNIX epoch
Bob Ippolito wrote:
>

>> As far as I know, GPL means you have to release the source, and since
>> there is no source for NVidia, they must not be GPL. I'm guessing, so
>> you might be right. They might be LGPL or whatever, to allow such
>> drivers to link against the kernel. I don't know how it works. I know
>> there is no source for NV module driver from NVidia.
>
>No, the Linux kernel is GPL, but does allow for *runtime* linking of 
>closed code.  There was a big dispute over whether or not this was 
>allowed by the GPL, but Linus decreed that vendors should be allowed to 
>do this, so they can with certain limitations.

Actually, unless Linus demands that all copyrights be assigned to
him of all contributed code like the FSF does, he is not in a
position to make any such statement.  Any contributor to Linux
could in principal sue NVidia for violating _his_ GPLed code and
Linus would not have anything [legally] to say about it.  He might
be called as a witness, and perhaps NVidia could make the case that
by contributing to Linux under the GPL one has tacitly accepted
Linus' interpretation of the GPL (which is not the same as RMS').
But that's not necessarily going to help.  Also, no one actually
runs linux.  People run Red Hat or SuSe or Yellow Cat or whatnot.
All of these take the kernel as distributed from Linus and apply
tens of MB worth of patches on it before shipping it.  So, there
is even more confusion about who owns what and who can sue whom
about GPL violations.  One presumes that the people whose code ends
up in these patchballs have not even tacitly agreed with Linus'
decrees about what the GPL actually means.

Effectively though, the GPL hasn't held up in court yet and so all
of this is just a risk as far as companies such as NVidia are
concerned.  The ``GPL wrappers'' are most likely their way of
convincing themselves that they are ameliorating the risk adequately.

--
    Roland Dowdeswell                      http://www.Imrryr.ORG/~elric/




More information about the talk mailing list