A limited patent grant from Nokia [LWN.net]

A limited patent grant from Nokia

Posted May 25, 2005 23:07 UTC (Wed) by mjr (guest, #6979)
In reply to: A limited patent grant from Nokia by xtifr
Parent article: A limited patent grant from Nokia

You misunderstand. Not promising to sue is certainly not a GPL violation, however, if a case is made that Linux actually needs this patent pledge for some spesific thing, section 7 would prohibit distribution if that thing is included, therefore making the whole pledge pretty much null and void (except perhaps for pure end users, who'd then have no rights to distribute the software anymore).

Of course, a common interpretation is that if a patent isn't actively enforced, then people can just go ahead and do what they will. But what if the patent is something that is enforced (by Nokia), just not for the Linux kernel? This is exactly the sort of trap that section 7 is supposed to prevent.

Now, to clarify, I'm not saying it'd be Nokia who'd be violating the GPL in this case (after all, they might be distributing only kernels without the spesific feature). It'd be everyone who distributed a kernel encompassing said feature (at least if knowing that Nokia has and is asserting patent rights over it; I shan't speculate how much due diligence would be required of the distributors).

Hope this clarifies the issue.

to post comments

A limited patent grant from Nokia

Posted May 25, 2005 23:58 UTC (Wed) by JoeBuck (subscriber, #2330) [Link] (4 responses)

But Nokia only has to promise not to encumber what it distributes.

Let's take an example. IBM distributes GCC and the rest of the C development toolchain to some of its customers. IBM has also licensed some relevant patents, for example having to do with register allocation, for use in all GPL programs. But there are many other patents they own that haven't been donated. In principle, someone could extend the GNU compiler, assembler, and linker to clone the function of Purify, which IBM owns patents for (specifically, object code insertion for the purpose of detecting memory faults). IBM has not promised not to sue over such a hypothetical event. In fact, Red Hat seems worried enough about this patent that they do not distribute valgrind, which might be considered as infringing, so this patent is having a real effect. Do you therefore claim that IBM is violating the GPL?

Remember, all Nokia has done is mention that they are not promising not to sue some over future hypothetical Linux kernel that contains functionality not in the current kernel and that infringes one of their patents.

A limited patent grant from Nokia

Posted May 26, 2005 0:12 UTC (Thu) by pgb (guest, #30022) [Link]

Actually as of Fedora Core 3 Red Hat does distribute Valgrind. Behind the scenes apparently the patent situation got resolved somehow. But your basic point does stand; they didn't distribute Valgrind because of patent concerns for quite a while, and there are other similar situations.

Analogy to recent W3C situation

Posted May 26, 2005 7:16 UTC (Thu) by mjr (guest, #6979) [Link]

Do you therefore claim that IBM is violating the GPL?

You appear to have missed the entire point of my post, even though I spesifically said that it wouldn't be Nokia who'd be (necessarily) violating the GPL. Please read it again more carefully.

By the by, this situation is somewhat analogous to the W3C situation of recent years, where allowing "field of use" restrictions on patent grants was considered. See the FSF position which explains why this is incompatible with section 7. Then consider that if a field of use restriction is incompatible, restricting the grant to particular versions of a particular program as published by a particular party surely is too.

A limited patent grant from Nokia

Posted May 26, 2005 14:41 UTC (Thu) by madscientist (subscriber, #16861) [Link]

> In fact, Red Hat seems worried enough about this patent that they do not
> distribute valgrind, which might be considered as infringing,

I don't see how Valgrind can infringe Purify. Purify is a link-time tool that modifies the object code emitted by the linker by adding instructions to perform memory analysis. Valgrind has always been a run-time tool and (at least the 2.x versions) doesn't instrument the code: as far as I understand it Valgrind implements a virtual machine to run the user's code as-is, with no instrumentation, where the memory analysis is done in the VM.

If someone is claiming that violates Purify patents that has got to be one BROAD patent grant!!

A limited patent grant from Nokia

Posted May 26, 2005 15:44 UTC (Thu) by Ross (guest, #4065) [Link]

Valgrind doesn't annotate code like Purify does it? I hadn't heard of any
patent concerns about it before, though that doesn't mean much.