back to article VMware survives GPL breach case, but plaintiff promises appeal

Linux kernel developer Christoph Hellwig's bid to have VMware's knuckles rapped for breaching the GNU General Public Licence (GPL) has failed, for now, after the Landgericht Hamburg found in Virtzilla's favour. The Software Freedom Conservancy backed Hellwig when he alleged that some of his contributions to the Linux kernel …

  1. Anonymous Coward
    Anonymous Coward

    Given that a lot of the code in question seems to be related to SCSI, I wonder how much the shape of such code is effectively mandated by the SCSI standards, and how much is left up to the imagination of the programmer?

    After all, if the standards effectively say "Do this, this and this" it would hardly be surprising if different implementations end up looking similar.

    It's like that with electronic components. The data sheets say "lay out your board like this". Guess what - every board using that component ends up looking very similar.

    1. Hans 1
      Headmaster

      Court item 50:

      A linux patch to the Linux scsi device driver was taken over to 99% by vmware in vmklinux. I think that this probably means that vmkernel must implement "very" similar calls to Linux, but that is just "wild speculation" on my part (IANAL) ...

      Sadly, just saying that vmkernel might contain Linux source code is not enough for the court, they want exact lines, however, since vmkernel is proprietary, you cannot say without the source.

      1. Anonymous Coward
        Holmes

        >They want exact lines, however, since vmkernel is proprietary, you cannot say without the source.

        Isn't that what court orders, injunctions, subpoenas, warrants, summonses and all that sort of stuff are supposed to be for?

        Sounds to me like the court simply didn't want to know.

        (Just in case you can't tell: IANALE)

  2. Mk4

    How to know when a corp. has pinched your code?

    Frivolous cases aside, could the court order an independent and private review of proprietary code to compare it with the open source code it's supposed to have pinched from? Hang on a minute! I might be wrong but it's just possible this could be automated! :-D

  3. davenewman

    Inappropriate gavels

    German courts, like English courts do not use a gavel. That is an Americanism. So don't use that misleading stock photo.See http://inappropriategavels.tumblr.com/

    1. no-one in particular

      Re: Inappropriate gavels

      I assumed it meant that this case is an auction, deepest pockets win...

  4. jms222

    Much GPL software is likewise pretty much a copy of what was already available under other licenses (BSD) and the "writers" just made something functionally equivalent with a different license.

    1. Hans 1
      Headmaster

      > is likewise pretty much a copy

      No, it is NOT, not at all ... I am pretty sure you cannot copy a scsi driver patch from Linux and apply it to a BSD, changing only 1% of the source code.

      > The "writers" just made something functionally equivalent with a different license.

      BSD explicitly allows copying code, GNU/LINUX does NOT!!!!

      Why did VMWare not use BSD iso Linux, that is the question ...

      1. Anonymous Coward
        Anonymous Coward

        > Much GPL software is...

        That's a pretty huge over-generalization. A lot of code is re-implemented, meaning folks had to re-write the code to implement the same behavior, on the BSD side of things since these are the folks without the "must use same license" clause and can't just "borrow" it.

        > ...changing only 1% of the source code.

        That actually is a "depends upon the bits being lifted" since if it doesn't involve the kernel structures, then it is possible in some places. Albeit rather small places places.

        > ...not use BSD...

        Because the point was to create a Linux derived object, not a BSD one.

        ...

        Just clarifying and not weighing in on the case.

      2. Anonymous Coward
        Anonymous Coward

        @Hans 1,

        Er, I think you have misread jms222's comment. He's suggesting that a lot of GNU'd code was inspired by BSD'd code which, as BSD is more permissive, is probably fine.

        Which begs the question as to whether or not such GPL code is appropriately licensed. Some guy lightly alters some BSD code and then whacks GPL all over it. That's a bit disingenuous, even if permitted by BSD. For the avoidance of future doubt such code should at least come with some statement of origin.

        In this particular case, if the relevant sections of VMWare's code also has strong similarity to chunks of BSD code then I don't really see how Hellwig will be able to sustain his case. The SCF analysis compared the entire Linux code base with the entire FreeBSD code base; what would also be interesting is if they repeated the analysis of the lines Hellwig wrote for Linux with the equivalent in FreeBSD, a much smaller and more focused comparison.

        On the topic of the SCF's comparison of various functions, the fact that there is no consistency to the ratios of similarities suggests to me a lack of whole sale copying. A straightforward copy and obfuscation would surely result in more evenness between the various functions they've analysed. Whereas in fact some are barely similar at all. Also, after all that work, where's the side by side comparison for us all to see?!

        1. Hans 1
          Holmes

          @AC

          >Er, I think you have misread jms222's comment. He's suggesting that a lot of GNU'd code was inspired by BSD'd code which, as BSD is more permissive, is probably fine.

          No, no, no, no AND NO ?

          1. BSD was, at the time the Linux kernel was written, still in legal disputes on the matter of the copyright of its code.

          2. Linux was written from scratch, mostly ... the kernel, I mean. The userland tools as well ... now, you will certainly find this or that, such as the zfs implementation ...

          >Which begs the question as to whether or not such GPL code is appropriately licensed. Some guy lightly alters some BSD code and then whacks GPL all over it. That's a bit disingenuous, even if permitted by BSD. For the avoidance of future doubt such code should at least come with some statement of origin.

          WTF ? Seriously, WTF? A great number of proprietary software houses abuse BSD's "openness" and never give anything back ... look at the worst bloatware purveyor on this planet, namely Microsoft, they have repeatedly taken stuff from BSD, most notably, its TCP/IP stack.

          Now, how is this any worse than what GPL types do ? I mean, GPL'd code is freely available, provided you stick to the license - it is not. GPL is there so that proprietary competitors do not use the work of a gazillion devs in their proprietary BS.

          1. Anonymous Coward
            Anonymous Coward

            @Hans1,

            WTF ? Seriously, WTF? A great number of proprietary software houses abuse BSD's "openness" and never give anything back ... look at the worst bloatware purveyor on this planet, namely Microsoft, they have repeatedly taken stuff from BSD, most notably, its TCP/IP stack.

            Well, clearly the BSD guys are cool with that, otherwise their license wouldn't be written the way it is.

            And as you mention Microsoft and the BSD stack in particular, you are being a ignorantly harsh on them. I have glanced at the Darwin source code, which is available from Apple, and is based on FreeBSD. And there, right in the heart of the stack source code, is a whole load of comments related to Windows.

            Yes that's right. MS have taken the stack code, made some changes, and then pushed them back to the community where they have been accepted (at least by some other parties). So shove that rabid anti-MS all-commercial-is-bad attitude back in your head an let it fester there.

            1. Hans 1
              Windows

              >And as you mention Microsoft and the BSD stack in particular, you are being a ignorantly harsh on them. I have glanced at the Darwin source code, which is available from Apple, and is based on FreeBSD. And there, right in the heart of the stack source code, is a whole load of comments related to Windows.

              Yes, indeed, PPTP, SMB, CIFS ... of course there is ;-)

              Now, I do not have all the source code handily available, but I downloaded network_cmds from OS X 10.0 (closest to BSD, imho), and found a reference to PPTP with Microsoft, not sure MS wrote that, however, if they did, thanks ... but no thanks, nobody wants to use that security sieve.

              As for all the many, many, many references to Windows, those are for clarifying what a file extension or DLL is (for the Windows lusers) ... just saying ... if you have proof to the contrary, feel free to share with me ... I can also be wrong, sometimes ...

              >So shove that rabid anti-MS all-commercial-is-bad attitude back in your head an let it fester there.

              Commercial is bad, as you have no control over what is running on your system. An installer might be installing a security update or a start menu replacement ... that is, if your are lucky ... it might also install nagware to get you to update to the latest version of Windows, telemetry software or, heaven forbid, a browser toolbar with search hijacker, who knows ? None of that shit in Linux or FreeBSD, where all the software I need is in the repos, with source handily available ... the non-free stuff is in the non-free repo, there you have similar issues, however, those binaries are somewhat vetted, and optional, disabled by default (on my distribution of choice).

              You are all learning this the hard way and I know it hurts your feelings. Let out your wrath, you know where the down-vote button is, don't you ?

          2. bazza Silver badge

            @Hans1,

            1. BSD was, at the time the Linux kernel was written, still in legal disputes on the matter of the copyright of its code.

            FreeBSD came into being in 1993, about 2 years after Linux first hit the servers. Linux was massively incomplete (compared to today) at that stage, and both have grown up more or less in parallel. FreeBSD itself has earlier origins, 386BSD, etc, which go all the way back to 1976; Linux was just 6 years old at the time and, gifted though he is, I doubt he was writing Linux back then.

            2. Linux was written from scratch, mostly ... the kernel, I mean. The userland tools as well ... now, you will certainly find this or that, such as the zfs implementation ...

            What, there's no BSD-inspired code in there at all? Not one single line? I don't really care, but I bet you cannot prove that.

            Now, how is this any worse than what GPL types do ? I mean, GPL'd code is freely available, provided you stick to the license - it is not. GPL is there so that proprietary competitors do not use the work of a gazillion devs in their proprietary BS.

            Are you crazy? GPL is cool with usage in proprietary systems so long as the terms of the license are adhered to. There's no "For non-commercial uses only" clause like there is in some proprietary licenses (eg VMWARE Player, a great proprietary gift to the world). Linus even resisted transfer of Linux from GPL2 to GPL3 to ensure that use of Linux didn't drop off as a result.

            1. Hans 1
              Holmes

              >BSD-inspired code

              Define "BSD-inspired code" ... very loose term. We are talking code-lifting here, as in, I copy 1000 lines of code, change a few variable names, change indentation, and claim the lot is my brain-child. Not GNU cat, for example, which was inspired by UNIX cat. At the time, there were copyright disputes for the BSD flavor of UNIX, so they could not lift the code.

              >I don't really care, but I bet you cannot prove that.

              There is, ex ZFS, but most is original work.

              The GPL is there so that a company does not use Linux without releasing the code of the Linux kernel they are using. If they make changes to the kernel, they MUST share them with the world. VMWare does not do this although they appear to be using substantial parts of the kernel (according to the dev), they'll lose if the dev is right, eventually.

              1. bazza Silver badge

                Define "BSD-inspired code" ... very loose term.

                Use of some one else's source code to tell one how a device should be driven, rather than plough through some boring over long data sheet or standards document to learn the same information. Of course I'm not suggesting that Hellwig or VMWARE have done that here.

                And regardless of that possibility, if both Hellwig and VMWARE based their code on reading the SCSI standards then it's not entirely impossible for the two code bases to end up looking similar, guided as they are by the SCSI standards itself. That's the thing about standards, they are the result of someone else already having done a lot of the thinking on behalf of the software implementer.

                I think there'd be something to learn from the SCF repeating their analysis, but for VMWARE's code vs, say, FreeBSD's SCSI code (they did whole kernel vs whole kernel, not just the SCSI code) to see what similarity scores that generates. If they're wildly different then it's possible Hellwig has a point (though as I say above there's still plenty of scope for common solutions arising from separate readings of the SCSI standard). If they're not so different to the scores for VMWARE vs Linux, then Hellwig definitely does not have a point.

                We are talking code-lifting here, as in, I copy 1000 lines of code, change a few variable names, change indentation, and claim the lot is my brain-child.

                By the SCF's own analysis that's not what's happened here. There's some similarity between some of the functions related to SCSI, whilst others are very different, judging from their stated "ratio of similarity". Not that they're saying what 99% or 14% means: I notice that the SCF haven't put up the two pieces of code side by side for all to see.

                At the time, there were copyright disputes for the BSD flavor of UNIX, so they could not lift the code.

                By the time FreeBSD came along it was already almost entirely free from AT&T code, most of the work having been done in 1989/1990.

        2. Named coward

          Even the most permissive BSD license requires acknowledgement (reproduction of the copyright notice). So if the GNU'd code was "inspired" by BSD it should be appropriately acknowledged.

      3. Anonymous Coward
        Anonymous Coward

        Drivers...

        > Why did VMWare not use BSD iso Linux, that is the question ...

        I'm guessing the answer is why they use vmklinux at all: DRIVERS! They know that a lot of companies produce Linux drivers, but very few produce drivers for other OSs (except for Windows, of course).

        So using Linux was an "easy win" for VMware.

        1. Hans 1

          Re: Drivers...

          > I'm guessing the answer is why they use vmklinux at all: DRIVERS!

          FreeBSD, for example, has a whack of drivers, too, not as many as Linux, but most enterprise kit should be covered.

          The whole point is, Linux probably had more drivers but Linux comes with a burden which is the GPL, FreeBSD have no such burden. If you choose Linux, then you must release the code, if you choose BSD, do as you please.

          They thought they could get away with releasing the code to the kernel module, only, without the kernel ... the dev is arguing that the module and kernel are one piece ... VMWare will be paying up, here. The Hamburger court (sorry) already acknowledged that we have a case, here ...

          All the dev needs to do now is get the police to investigate code-lifting ... and get more kernel devs into the party ...

    2. Nick Kew

      Red herring

      Much GPL software is likewise pretty much a copy

      If that claim is true, and if it applies to code in question here (two big assumptions), then it would be a defence for VMWare to say "no, we didn't copy the GPL code, we copied the similar BSD code".

      Was that defence used? If yes, then I'd expect the report to comment on it. If no then your claim is utterly irrelevant.

  5. thames

    It's how much does he own

    My understanding of the case is that the problem is not whether VMWare was distributing software in violation of the license terms. It was whether Hellwig could demonstrate that enough of the unlicensed code had copyrights held by him personally as opposed to other software developers. Without demonstrating that a sufficient threshold of unlicensed copying of software owned by him was reached, under German law Hellwig does not have a sufficient complaint against VMWare to proceed.

    The court is not saying that the software in question wasn't Hellwig's. They're just saying that Hellwig hasn't shown to their satisfaction which of the code in question was his, versus which of the code in question belonged to other software developers who are not taking part in the lawsuit. Hellwig has to show that he has enough code in question to make a lawsuit worth the court's time.

    Where Hellwig went wrong was to not have all his legal ducks in a row before launching the case. If he has a second go at it later, he may be better prepared. He has to present the right evidence at the right time, and he has to do it in the format which the court wants to see. It's a technicality, but that's what lawyers get paid for.

    It's cases like this by the way which are why the FSF requires assignment of copyright to them for any projects which they own. That way they don't have any problems proving they have sufficient interest in the software when enforcing the copyrights.

    With Linux kernel development on the other hand, the copyrights are distributed over a very large number of parties. That means that getting enough copyright holders together to agree on enforcing the license terms can be difficult.

    1. Anonymous Coward
      Anonymous Coward

      Linux kernel copyrights

      I wonder if that choice of letting everyone hold individual copyrights instead of signing them over was by design from Linus. He has never seemed to be overly bothered by people copying code from the Linux kernel into proprietary products, at least not the degree that Hellwig and certain others are, let alone the FSF.

      Linux might not have reached where it was today if copyright had been assigned to the FSF, and there had been a few high profile lawsuits that made large companies back off. Even if they weren't intending to misuse the code, just the worry about accidental misuse causing exposure of their proprietary IP would be enough to dissuade some. Who knows, maybe Android would have ended up based on FreeBSD. Not that I think that would really make any difference to Android, but getting Linux on a billion devices worldwide certainly has helped increase its reach far beyond what anyone would have thought possible 15 years ago.

      1. thames

        Re: Linux kernel copyrights

        @DougS - There are two ways that GPL licensed projects tend to be run. One is copyright held by a single party, the other is distributed copyright (everyone retains individual copyrights). The latter is the more common one these days.

        The FSF got into the game fairly early when "open source" was still considered to be a fairly novel idea, as was downloading software for free off the Internet. They also didn't have big corporate backing to finance lawsuits or a long list of favourable legal precedents, so they wanted to make any court cases as cut and dried as possible. Hellwig's current problem is exactly the sort of thing they were trying to avoid.

        The FSF does not like to sue people. They simply want compliance with the terms of their software license. And we should remember that it's Hellwig with Linux kernel code who is suing here, not the FSF.

        The case for distributed copyright is that free/open source software has more or less won the argument, and nearly everyone is using it these days. As such, it's grown well beyond the FSF.

        When copyright in a work is widely distributed and the license is GPL, it's not possible for a single party to take it proprietary. Thus companies are more willing to participate in large projects because they know they won't be screwed over by someone else.

        When copyright is held by a single party or the license is not GPL, as an outside participant you have to "trust" whoever is controlling the project to not screw over the other parties whenever there is a change of management or control. While the FSF may be fairly trustworthy, how far do you trust Oracle with their ownership of MySQL (which they own all the copyrights to)?

        With Android, how far would you trust Google to not take it completely proprietary if Linux used a BSD license instead of GPL? They're gradually tightening the screws on the Apache/BSD licensed bits of it. If I was a major phone manufacturer, I would much prefer GPL licensed code as then I would know that I always had an escape hatch if Google went totally "evil".

        Generally, a GPL license with widely distributed copyright is "friendlier" to end users, small developers, and businesses who have ongoing support obligations. If you want to create a "community" around a software project it's still the best bet because it doesn't allow anyone to get into a position of privilege with respect to anyone else.

POST COMMENT House rules

Not a member of The Register? Create a new account here.

  • Enter your comment

  • Add an icon

Anonymous cowards cannot choose their icon

Other stories you might like