Software Freedom Conservancy: Distributing Linux With ZFS Is Illegal (phoronix.com)

LichtSpektren writes: Phoronix reports that Bradley M. Kuhn and Karen M. Sandler at the Software Freedom Conservancy (SFS) have posted a blog post today arguing that Canonical's plan to distribute Ubuntu 16.04 LTS "Xenial Xerus" with support for the ZFS file system violates the Linux kernel's GPLv2 license.

On February 18, Dustin Kirkland at Canonical wrote on his blog: "We at Canonical have conducted a legal review, including discussion with the industry's leading software freedom legal counsel, of the licenses that apply to the Linux kernel and to ZFS. And in doing so, we have concluded that we are acting within the rights granted and in compliance with their terms of both of those licenses...The CDDL cannot apply to the Linux kernel because zfs.ko is a self-contained file system module — the kernel itself is quite obviously not a derivative work of this new file system. And zfs.ko, as a self-contained file system module, is clearly not a derivative work of the Linux kernel but rather quite obviously a derivative work of OpenZFS and OpenSolaris. Equivalent exceptions have existed for many years, for various other stand alone, self-contained, non-GPL kernel modules. Our conclusion is good for Ubuntu users, good for Linux, and good for all of free and open source software."

The SFS's blog post of today states: "We are sympathetic to Canonical's frustration in this desire to easily support more features for their users. However, as set out below, we have concluded that their distribution of zfs.ko violates the GPL."

Comments

No winners here.

[Anonymous Coward]

I'm not sure what anyone would stand to gain by arguing that Ubuntu is violating the GPLv2 license. It seems kind of ridiculous.

Re:No winners here.

[epine]

Having the user compile them is just another means of distributing a binary.

You're the reason the rest of us have to constantly write IANAL as if that's the only thing that makes it permissible to so much as clear your throat, when all that really needs to be clear is that we're capable of distinguishing A from B to any necessary degree.

"Having people cook their own pseudoephedrine is just another means of distributing methamphetamine."

Actually, no. What you actually distribute, and the end result of what people are perhaps likely to do with it are two entirely different things. This is not Hair Splitting 101 with a laser scalpel. This is retiring your flexible plastic picnic knife. There exist, of course, various ways to spin language to link these things together, if that is indeed desired. But then we should be analyzing that extra language, not wishing it into existence by waving our hands and muttering "kids there days".

Digression: Lawyers might make better guesses about the "necessary degree", but in practice lawyers are about as useful as political pundits empanelled on ISIS Today in speculating about the future will of the court. Maybe we should just all write CTLNKS instead (concerning the law, nobody knows shit). The main difference between critical thought at large and your lawyer is mainly that your lawyer can get into heaps of trouble by giving you a failing grade of shit. Pretty much the only thing I really want from intelligent agents of xmas future is to see the profession of law refactored, with all of their muddles and uncertainties hopelessly exposed to eternal ridicule, until we actually find in necessary to draft legislation to a sane standard (including an automated regression test suite, an unintended consequence verification gauntlet, and some Valgrind fuzz testing).

Back to the matter at hand, I completely disagree that this analysis should start with the semantics of static/dynamic linkage.

The real question should be this: if zfs.ko is erased from your system, does what remains function 99.99% the same way, minus only those features that zfs.ko provides?

With static linkage you can't do this, and thus you can't ask this question, either.

The key issue here is separability. What natural line exists between dynamic linkage and a ZeroMQ socket layer? And what about Joyent's "double hull" virtualization where ZFS is native to the underlying OS which provides a Linux API to the client OS? "Linking" is just indirection with more established history. And we all know you can solve every problem in computer science with another layer of indirection, except for too many layers of indirection.

I think my "separability" question above is far more likely to be legally productive, but then IANAL YMMV & CTLNKS.

SFS? More like FFS...

[Bearhouse]

Why oh why do we have to keep shooting ourselves in the foot?
OK I'm a BSD user so, well, stones and glass houses, but even so the open source community's continuing ability to why things should not be allowed is depressing...most people in our crowd want our stuff to be USED by as many people as possible...

It isn't features, it is stability + features

[poet]

Linux does not have a stable file-system and volume manager that can come anywhere near the performance and stability of ZFS.

SFS/FSF does not get to rule on GPL

[Anonymous Coward]

The Linux Foundation, as the copyright holder of the Linux kernel, gets to determine what is and is not allowed.

This is matter does not involve Software Freedom Conservancy (SFS) or Free Software Foundation (FSF), they are are not even parties to the copyright license used for the Linux kernel.

In my professional opinion, Canonical can feel free to tell SFS to go pound sand.

Re:SFS/FSF does not get to rule on GPL

[jedidiah]

Agreed. The SFC is not the owner of either of the works in question and is not the author of the license in question. They have no stake or standing in the matter.

As authors of the GPL itself, I would say that Stallman or the FSF do have some say in the matter.

The SFC can go pound sand...

Re:SFS/FSF does not get to rule on GPL

[slashping]

SFC only advises and warns, just like the FSF could do. Ultimately it's the copyright holders that can sue in court, and then it will be a judge or jury that decide the matter.

Re:SFS/FSF does not get to rule on GPL

[edtice1559]

IANAL but it's not required to assign the copyright to the Linux foundation. Many people contribute to the Kernel. It's GPL so you could fork it. And the Linux Foundation would definitely *not* automatically get copyright to your version. It's ultimately the Linux Foundation who could prove copyright to large swaths and take action bu the foundation could not take the kernel as is and relicense it under GPLv3 for example. They'd have to rewrite all the parts for which they don't hold the copyright.

Re:Software Freedom?

[slashping]

There's no such thing as unlimited freedom. You have to pick the freedoms you care about, and then forbid those that would violate it.

Linus doesn't object to binary kernel modules

[edtice1559]

Regardless of the legal situation, the holder of the copyright to the vast majority of the kernel doesn't object to binary modules. There are those who would argue that even certain user-space binaries are a GPL violation. Fortunately those voices haven't prevailed. I guess, at some point, to squelch any doubts, ZFS will have to get turned into something akin to a hybrid driver where code to load interchangeable binary blobs with a certain interface is contributed to the kernel under GPL and the blobs come with their own license. In this case it will be CDDLv1. This is somewhat of a PITA but it's actually good in the medium-term because it means that users face less ambiguity in terms of knowing under which license they are receiving code.

Hilarious

[Anonymous Coward]

>The FSF, stewards of the GPL, have stated many times over the past decades that they believe there is no legal distinction between dynamic and static linking of a C program and we agree.

The FSF is almost certainly wrong in this. Linking isn't going to even be considered in a court room. What they *will* consider is whether or not zfs.ko contains derivative work from Linux. Headers are now subject to copyright (as is my understanding) so it's very possible that it does. However, if they use "api-only" headers, such as posix stuff and it does not rely on being compiled against the GPL-ed version of the headers, then they are most likely going to be free and clear of any issues.

Personally, the SFC can go eat a bag of dicks and die in a fire.

Re:Software Freedom?

[drew_kime]

So, Software Freedom means that you can't do what you want with your software? Is this one of those Richard Stalin - I mean, Stallman - groups or something?

If you didn't write it, it's not your software. You can acquire the right to use it in exchange for your agreement to the terms of the license. The freedom part comes in where the terms of the license say that you get to modify and distribute it without anyone's permission. Again, in exchange for your agreement to the terms of the license.

Software Freedom Conspiracy

[wjcofkc]

I've read both arguments, and since I know everyone posting here has also (right?), I won't bother with too much of a critique.

I have to side with Canonical on this one. Their short and sweet post pretty much puts to bed the legal ramblings of the Software Freedom Conservancy. The SFS article reads like they started out trying to make a point, then has to fallback on an academic lesson on GPL Incompatibility and Combined/Derivative works. I would go so far as to say their verbosity defeats their own argument, while managing to inset the letters "ZFS" wherever they can find room.

Past that, as someone who uses ZFS on FreeBSD, ZFS is pretty fucking useful. Canonical's legal inclusion of ZFS as self-contained file system module is a big step forward for Linux.