Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Linux's GPLv2 licence is routinely violated (2015)

devever.net

Post reply on HN
Print view View original

131–140 of 202 posts

Re: Linux's GPLv2 licence is routinely violated (2015)

#131

Post by tsimionescu »

belorn wrote: post #129
tsimionescu wrote: post #128

Earlier quoted context omitted.

Let's imagine the following scenario. I am creating and distributing a program that, when run on a user's system, will dynamically link with a copy of libc. I am not distributing this copy, but my program will fail to run if no such copy exists. Furthermore, I tell my customers that the only lib C I officially support is FictitiousLibC, released under the full GPL (not the LGPL like the real GNU Libc). The question i…

Let's remove the license and just look at what copyright says in order to understand your example. You have some copyrighted work and you want to make copies, and you do not have a copyright license from the author of said work. It this legal? Answer is mostly going to be no, especially if you are going to sell the copies. Commercially exploiting someones else's copyrighted work without a license is very often illega…

The main point of my argument, and the main concern I feel with GPL-style licenses, is the notion of "do I have to distribute my work under the terms of the GPL even if I don't distribute any GPL code directly; in other words, when is my program a derived work of a GPL work?". If instead I am distributing GPL code directly, the question is more clear - as you say, copyright law makes few exceptions there.

But again, if my program doesn't directly include any code at all from a GPL work, but it does depend on such code to exist on the end-user system, knowing in what circumstances my code is a derived work and thus can only be distributed under the terms of the GPL is much harder.

This is exactly the case of the kernel modules: the people distributing the kernel modules are often NOT distributing Linux itself. But, can their modules be considered derived works of the Linux kernel?

Note that this kind of question can also arise in traditional copyright. I can make a play that very much resembles a novel, and depending on many ultimately subjective factors my play may or may not be considered a derived work of that novel. If I am directly including substantial parts of the novel, then the answer is pretty clear, but a work that doesn't have a single word of the original can still be a derived work (most obviously, a translation is a derived work, but it likely literally doesn't include a single word in common with the original work).

Re: Linux's GPLv2 licence is routinely violated (2015)

#132

Post by tsimionescu »

jarpadat wrote: post #94
Borealid wrote: post #73

Earlier quoted context omitted.

I find it difficult to imagine a situation in which you could use the library without copying the library. How would you obtain the code to execute it? Someone must distribute it to you, and/or you must copy it. Those are the governed actions. EDIT: note that the GPL broadly doesn't impose restrictions on interacting with a program over a network ("using" a server), which is why the AGPL exists.

So, one answer (not necessarily the right one!) to how you can use software without copying it is that "the license says you can". In other words the license definitely views "running" and "copying" as distinct, regardless of you, me, or federal law. The other, maybe more familiar idea to lawyers, and maybe more plausible to you, is the one in MAI v. Peak, that running is copying by definition. (The argument is based…

> The other, maybe more familiar idea to lawyers, and maybe more plausible to you, is the one in MAI v. Peak, that running is copying by definition.

That is a bit over-simplified. The important part in MAI v. Peak was that Peak was a third party to the license between MAI and their client. The court recognized that MAI's client had the right to load/copy the program that they had acquired from MAI into memory without any additional license from MAI - there was already an explicit exemption in copyright law for this. However, they decided that this right can't be extended to a 3rd party (Peak, who as acting as a support technician), even if on the same machines.

Re: Linux's GPLv2 licence is routinely violated (2015)

#133

Post by tsimionescu »

Borealid wrote: post #76
Dylan16807 wrote: post #64

Earlier quoted context omitted.

I don't see anywhere in the GPL that defines deriving, except the part where it mentions "derivative work under copyright law".

The relevant clauses read as follows: b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. [...] These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can…

That is not necessarily the part I was thinking of - I had indeed made some confusion between what the GPL says and what the LGPL and other exceptions-to-the-GPL licenses say.

I have detailed this more in another post, but basically my argument is that the very existence of the LGPL makes sense only if we take the GPL's definition of derived works to include programs which dynamically link with the GPL work (and this is indeed what e.g. Stallman believes).

There are also some similar exemptions in the GPL license for GCC, to make sure that the binary resulting from compiling a program with GCC does not have to be distributed under GCC's license terms even if it might be a derived work of GCC itself (e.g. when the compiler includes bits of code that it does consider copyrighted, such as replacing calls to the library function memcpy with an optimized version that is defined in GCC's source code).

I absolutely agree with you that, when distributing a combined work including GPL code very directly (say, a CD with Linux + some proprietary software), the GPL may well impose extra restrictions even on parts that are not considered derived works under copyright law. You are free to not accept those extra restrictions, but then you have no right to distribute the GPL works at all (though you can still distribute the parts that are not derived works independently).

Re: Linux's GPLv2 licence is routinely violated (2015)

#134

Post by phendrenad2 »

Related reading: https://lwn.net/Articles/860262/

Key takeaway:

> purveyors of proprietary modules have engaged in a number of tricks to get around GPL-only exports

There are many technical workarounds possible still. You could scan the kernel binary for byte patterns and build your own export list. How will the kernel maintainers plug that hole? (Maybe they can implement compile-time binary obfuscation, and sacrifice performance to stop proprietary code, wouldn't that be funny?)

I think that these efforts to prevent (what the kernel developers see as) "license infringement" through technical means sort of reveals that they don't really think that the line of reasoning would hold up in court.

Re: Linux's GPLv2 licence is routinely violated (2015)

#135

Post by tsimionescu »

throwawaylinux wrote: post #127
pjmlp wrote: post #126

Earlier quoted context omitted.

In Europe unless the law has changed, in most jurisdictions the typical US style licenses are considered void unless the consumer of such software has been given and agreed to the license before acquisition. That is why most ship wrapped software started putting a reference to the license on the box.

If the license is void then copyright law applies.

Yes, but copying a piece of software from a CD into program memory or onto my own hard-drive for archival purposes has been established as a fair-use exemption. So, if I legally acquire a copy of a program (say, I buy a CD), it can't impose extra restrictions after-the-fact through an EULA that only appears when I run it, and I am well within my rights to use my copy (e.g. by copying it into RAM/CPU icache to run it) even if I don't abide by the EULA, according to European Union jurisprudence.

For example, many game EULAs (used to?) impose a restriction on lending the CD to others (you had to seek explicit permission from the copyright holder to do so). If this clause is only knowable after buying the game, then in the EU it is null and void, and I can still use the software and lend it to others per regular copyright law.

If, however, the license must be accepted before purchase/download, my purchase/download can be considered acceptance of this clause and I then have no right to use the software if I don't abide by the license agreement, since the copy was distributed to me only under specific conditions.

This is just like if I buy a book, I have every right to read it, lend it or re-sell it; but we can also sign a contract under which you give me the book but with no right to read past page 87, and no right to lend it to others.

Re: Linux's GPLv2 licence is routinely violated (2015)

#136

Post by phendrenad2 »

kmeisthax wrote: post #115
phendrenad2 wrote: post #68

Earlier quoted context omitted.

> Linus is a copyright holder...So...the court will listen to Linus to determine where the boundary is This seems completely insane. If I become a programmer on Monday, and download the Linux source on Tuesday, and read the license file on Wednesday, and publish code on Thursday, and Linus sues me on Friday, am I going to be penalized by the judge because I didn't read random LKML posts by Linus? Surely the license i…

In this particular case, we're talking about a license clarification that gives you more rights than you otherwise could. So, if you hadn't read Linus's interpretation of his own license, you would think that you always had to GPL your LKMs. So there's no point in which Linus could actually sue you for not reading his LKML rants - they strictly increase what you are allowed to do under the license. If this was the op…

Actually, if you only read the license files included with the kernel source, you'll be very confused and unsure of what you can actually do. The GPL is painfully vague, and there have been endless debates about how it applies to kernel modules, programs, etc. (such as this very one we're having now). (And I'm taking into account the clarification files in the source). Linus' LKML clarifications are post-hoc and not linked in the kernel files anywhere. I do agree with you that the GPL_ONLY "code is law" theory has some merit - and certainly a judge could use that as part of a ruling to uphold that interpretation. But I don't think that this aspect of the kernel license is going to be tested in court any time soon, so it's all speculation at this point.

Re: Linux's GPLv2 licence is routinely violated (2015)

#137

Post by tsimionescu »

Dylan16807 wrote: post #59
rtpg wrote: post #57

Earlier quoted context omitted.

That is not what the word "permissive" habitually means in the English language, especially if you consider a more broad view of users including those who are writing the software and integrating it into their products. Not to say GPL is bad, just that the VAT-style "only the last user counts" view of things is playing with semantics.

> writing the software and integrating it into their products That's distribution, not use. (Except for when they're testing it or running local instances, and GPL doesn't get in the way of those things.) > Not to say GPL is bad, just that the VAT-style "only the last user counts" view of things is playing with semantics. By my understanding VAT doesn't work that way. Once a user pays VAT and starts using something,…

> By my understanding VAT doesn't work that way. Once a user pays VAT and starts using something, that's it. The taxes are done with, and VAT doesn't pass on to later users.

VAT works something like this: when you buy X, you pay VAT on X. If you then sell X, you reclaim the VAT on X from the government. The last "user", the one that didn't sell X to anyone else (or integrate it into a larger product that they sell to anyone else) is the only one that effectively pays any tax once you draw the line.

Re: Linux's GPLv2 licence is routinely violated (2015)

#138

Post by Jiro »

Borealid wrote: post #39
tsimionescu wrote: post #25

> The userspace ABI is stable and intended not to cause anything targeting it to become a derived work. But once again, this seems to be a way in which the kernel project seems to think it has the power to interpret the GPL. In their minds, targeting the userspace ABI doesn't make a derived work, but writing a module does, unless it only targets GPL_ONLY symbols, in which case for some reason it isn't. I think this w…

I think there's some confusion in the above post. The relevant "derived work" definition for the purpose of linking isn't one stemming from copyright law, it's the one in the GPL itself. This is because the GPL defines what licensees are permitted to do with the licensed work, and it imposes restrictions on those rights. Copyright law by default lets you make very little use of the software - the GPL broadens that ou…

>Said differently, the GPL doesn't try to apply itself to "derived works" because of some contralegal dictum they don't have a different creative origin. It merely says "you have two choices - license the things you link with OUR stuff under OUR terms, or you don't have the right to use our stuff".

The GPL doesn't restrict use, and says so. When people say that the GPL only lets you use the work under the author's terms, "use" means things that you'd need copyright licenses for. If you aren't creating something that is a derivative work according to the legal definition, you may not be using (in this sense) the author's work at all.

This also fails for interpretations like "you can link to GPL_ONLY symbols". This isn't directly written in the license, and any kernel contributor can deny it whenever they want.

Re: Linux's GPLv2 licence is routinely violated (2015)

#139

Post by cozzyd »

grzm wrote: post #72
drjasonharrison wrote: post #71

Earlier quoted context omitted.

What does SCO mean? Couldn't find it in Wikipedia or lists of abbreviations. I think it is/was a company...

> "The SCO–Linux disputes were a series of legal and public disputes between the software company SCO Group (SCO) and various Linux vendors and users. The SCO Group alleged that its license agreements with IBM meant that source code IBM wrote and donated to be incorporated into Linux was added in violation of SCO's contractual rights. Members of the Linux community disagreed with SCO's claims; IBM, Novell and Red Hat…

Someone should write a play or something about the SCO saga. I would watch that...

Re: Linux's GPLv2 licence is routinely violated (2015)

#140

Post by xorcist »

tsimionescu wrote: post #128
Borealid wrote: post #39

Earlier quoted context omitted.

I think there's some confusion in the above post. The relevant "derived work" definition for the purpose of linking isn't one stemming from copyright law, it's the one in the GPL itself. This is because the GPL defines what licensees are permitted to do with the licensed work, and it imposes restrictions on those rights. Copyright law by default lets you make very little use of the software - the GPL broadens that ou…

Let's imagine the following scenario. I am creating and distributing a program that, when run on a user's system, will dynamically link with a copy of libc. I am not distributing this copy, but my program will fail to run if no such copy exists. Furthermore, I tell my customers that the only lib C I officially support is FictitiousLibC, released under the full GPL (not the LGPL like the real GNU Libc). The question i…

> According to the FSF, [...] then YES, I am breaking the copyright

The FSF have actual copyright lawyers on hand. Where have they said this?

The example is not very easy to understand. In general copyright law isn't really applicable to your own right to distribute your own original work.

Post reply on HN