Subject: Re: GPL code contamination?
To: Oliver Korpilla <>
From: Greg Troxel <>
List: tech-kern
Date: 04/24/2005 08:26:07
"Oliver Korpilla" <> writes:

> > On Sat, Apr 23, 2005 at 07:30:00PM +0100, Richard Earnshaw wrote:
> > [gdt is not the least bit sure the above attribution is correct]
> >
> > You're right that that could happen (someone put the GPL on code covered 
> > by a patent for which they do not have the right to distribute patent 
> > access). It would be a very unusual legal situation, as the code would not
> > be legally GPL-able. :-)

Strictly, the distribution of such code would not be permitted.
> This is all very contrived, though IANAL, too.

This could easily happen accidentally, e.g. if the author was not
aware of the patent.

> What could realistically happen in respect to patents and the GPL, that a
> patent holder gave a GPL-compatible patent license or the patent holder
> itself released the GPL code, which makes him automatically give a
> GPL-compatible patent license away.

I have seen no support for such automatic licenses under the GPL in
patent or copyright law, which is a "bare license" rather than a
"contract license".  I suggest you read the book I pointed out earlier
if you would like to understand these issues.

> But this license would only cover that
> program the license is for, and the code could not be used as reference
> material for any other program. Even that is contrived.

The use as "reference material" is irrelevant under patent law.  The
only question is whether the new program practices an invention
without a patent license.  Whether the author looked at other code, or
independently reinvented doesn't matter (unless one wants to challenge
on obviousness grounds).

> My real issue is really: Is it okay, from an ethical and legal standpoint,
> to look at Linux scheduler code for inspiration for NetBSD scheduler code,
> especially if this would not be the only inspiration? If you additionally
> read papers, excerpts from books, and FreeBSD's SCHED_ULE? 

This is a long-standing tradition, and it's fine.  A problem arises if
and only if you either copy code without following copyright licenses
or the resulting code infringes a patent.

> And, if you'd try to implement something similar and explicitly state it
> embodies some principles embodied in Linux code or reference it with a
> credit in the code, would that be accessible to the NetBSD community???

I believe it would be.  Giving credit for inspiration is not legally
required, but it is the polite and decent thing to do, and there's a
long tradition of that too.

I think you are worrying about this aspect more than you should be.
The question deserving attention (or not - seek your own legal advice
:-) is whether the new code infringes any patents.

An interesting question for NetBSD would be patents such as those from
Sun or IBM that are licensed for "open source use".  While their use
in NetBSD itself would likely be covered by those license grants,
licenses would probably not extend to proprietary derivative works.

        Greg Troxel <>