Software Freedom Conservancy Funds GPL Suit Against VMWare

Jeremy Allison - Sam writes with this excerpt from a news release from the Software Freedom Conservancy: Software Freedom Conservancy announces today Christoph Hellwig's lawsuit against VMware in the district court of Hamburg in Hamburg, Germany. This is the regretful but necessary next step in both Hellwig and Conservancy's ongoing effort to convince VMware to comply properly with the terms of the GPLv2, the license of Linux and many other Open Source and Free Software included in VMware's ESXi products. Serge Wroclawski points out the SFC's technical FAQ about the suit. One nugget: This case is specifically regarding a combined work that VMware allegedly created by combining their own code (“vmkernel”) with portions of Linux's code, which was licensed only under GPLv2. As such, this, to our knowledge, marks the first time an enforcement case is exclusively focused on this type of legal question relating to GPL

Comments

How to help !

[Jeremy Allison - Sam]

To donate funds to Conservancy GPL compliance efforts see here:

http://sfconservancy.org/linux...

I'm dying of curiousity

[idontgno]

about VMWare's position on this. What on earth do they think trumps their obligations to the license they agreed to by using GPLv2 software in their product?

I wonder if they could possibly be as deluded and stubborn as SCO.

Re:I'm dying of curiousity

[MightyMartian]

We've seen enough corporate sociopathy to accurately answer that question.

Why didn't they just go with BSD?

Re:I'm dying of curiousity

[lolcutusofbong]

They probably figured they'd never get caught by anyone with the balls and money to take them to court. I'd love to see all of ESXi and vSphere get open sourced as the result of this lawsuit, but I suspect that's just a pipe dream.

Re:I'm dying of curiousity

[loosescrews]

They are not using the entire linux kernel. At this point most of vmkernel is their own code and they are (for the most part) only linking in the parts of the linux kernel that deal with hardware. Linux has way better hardware support than BSD.

Re:I'm dying of curiousity

[Tom]

They are taking a calculated risk knowing that very few GPL lawsuits actually went to court. They know it takes money to fight a legal battle and hope the opposing side doesn't have it, or will run out of it before reaching a final verdict. And finally, from the fact that they've been at this since 2012 - they probably think that it's a fairly cost-efficient way to buy more time and make business.

Re:I'm dying of curiousity

[pseudofrog]

I'll never get why the GPL is considered controversial with regard to its legality. Litigation should basically be:

It this your code?
No.
Why are you distributing this code?
We have a license.
What are the conditions of the license?
...

Re:I'm dying of curiousity

[Richard_at_work]

If, and thats a big if, VMWare have done anything in violation of the license at all - the "technical FAQ" is very light indeed on actual technical details, instead mostly talking about how Conservancy went to great lengths to do anything they could to open a dialog before suing. The "technical" part of the FAQ is a single diagram which explains very little in how they think VMWares approach is violating either the GPL or copyright.

This is going to be something to watch, as its going to be an interesting one.

Re:I'm dying of curiousity

[gstoddart]

They're a for-profit corporation, who see this as their bread and butter technology.

All you need is an idiot CEO to say "I want this" and an asshole lawyer to agree with him.

Are you honestly expecting principle, logic, and honesty from this?

It's a corporation looking out for its own interests. If that means fucking everybody else over or coming up with your own interpretation of the law? So be it.

I expect nothing less from this kind of situation.

Re:I'm dying of curiousity

[bill_mcgonigle]

it's a fairly cost-efficient way to buy more time and make business.

It sure is, and the people making such decisions face no consequences for violating the license. Yeah, maybe the corporation will get slapped with a tiny fine that reflects some small percentage of the money saved by incorporating the GPL'ed library, but how is that really any disincentive? It's more of an inconvenience, or simply a cost that gets processed through the EMC legal department, and then only maybe.

The money being spent on the prosecution won't actually change much behavior - there might be better causes to donate your money too (especially if you don't believe in imaginary property) than funding this expedition to behead a hydra.

Re:I'm dying of curiousity

[lgw]

I highly doubt there was any such forethought. Much more likely (at least, at companies I've worked at) that some junior dev checked in GPL code as his own work, and it somehow slipped past code review (as can happen at crunch time).

I worked for "shrinkwrap software" companies for `15 years, and all of them had ironclad rules against using GPL software in any way (without a multi-month lawyer-approval process anyhow). One place I worked ran open source detection tools (similar to the plagiarism detection tools, but seeded with all the big free projects) as part of the daily build, they were so paranoid. I'd be surprised if this was deliberate on VMware's part. But then, maybe they're just a shitty company now?

That will be the interesting part of this case, IMO: was this deliberate, against official policy that the dev teams ignored, or some junior guy cheating?

Re:I'm dying of curiousity

[phantomfive]

It's 'controversial' because the term 'derivative work' is controversial.

For example, imagine NVidia has a driver for their graphics card, and they decide to port the driver to Linux. Of necessity, they will need to use some Linux kernel code to do so. Would that be a derivative work, or not?

Re:I'm dying of curiousity

[zarthrag]

If that was the case VMware would (or should) have apologized, and removed the offending code to get into compliance. The fact that things are this far along signals at least some degree of maliciousness towards the terms of the GPL.

Hopefully, the penalty doesn't come out to be a meaningless fine. Instead, it should be a meaningless fine and forced compliance to the GPL - not through removal of the offending code (they have passed on that), but through open-sourcing of the entire product via GPLv2, effective immediately.

Re:I'm dying of curiousity

[LatePaul]

As someone said the technical FAQ is light on details, however they also link to this article from 2007 http://www.freerepublic.com/fo... which is how far back Hellwig's complaints about GPL violation go.

Based on that, it wasn't a case of whether VMware had included Linux code in their binaries, so much as they were distributing binaries which depended wholly on Linux to run but not GPLing them.

Re:I'm dying of curiousity

[darronb]

Yes, I'm quite curious about the lack of specifics.

It starts off with a very reasonable BusyBox violation that need to be corrected, and then veers off into claiming there's a much bigger problem without specifically stating it. It SOUNDS like they're saying VMWare's hypervisor is loaded by something that loads from the kernel and therefore it all must be GPL.

I'd like to be corrected if this is wrong.

Linus' own comment about a driver ported to Linux not falling under the GPL because the driver effort doesn't generally require Linux is simple from a general level, but not a technical one. Obviously the specific Linux driver requires Linux... you're not loading the same Linux driver on Windows. How much of the Linux-specific version of the driver can be custom just for Linux?

I'm sure the hypervisor was not from-scratch written for Linux, but existed before that. Anything that has a history before Linux should similarly be excluded.

If it's just the VMI layer that the discussion seems to indicate VMWare was close to open sourcing anyway at one point... it's hard to see what the big deal is.

Often lawyers have this awful tendency to want to 'score' as much as possible for their client to show their value... even if it's stuff that reasonably shouldn't be fought over.

Re:I'm dying of curiousity

[HornWumpus]

Only the headers (that define interfaces used) not a derivative work.

Uses actual implementations, derivative work.

This was covered three of four times in the commercial software world.

Re:I'm dying of curiousity

[jedidiah]

It's only "controversial" in software where people think that "what's mine is mine and what's yours is mine" is an acceptable way to think about other people's work.

The GPL is a clever hack of copyright law written by someone that knew what he was doing. I think people see RMS and think they can get away with sh*t because they think he's not smart enough to hire a real lawyer.

Re:I'm dying of curiousity

[Shinobi]

The entire thing is highly confusing, since ESXi tossed out the Linux kernel components from the hypervisor and runs on bare metal since 4.1(released in july 2010), and ESX, which is not the same thing as ESXi . And BusyBox is not a part of the kernel. Also, the source code for the Open Source components for ESXi 5 have been available for download since the start. So, judging by that technical FAQ, is their argument "You share kernel space with GPL'd stuff, now hand us the goodies or we'll sue you!"?. (Which wouldn't surprise me, given the religious zealotry behind GPL which can easily go into "What's yours is mine, and what's mine makes yours mine too")

Red Hat have made noises in that direction via support channels the last few years, and an argument has been made that their support techs should count as official users on the systems, for GPL reasons. Needless to say, some healthcare IT administrators, defense contractors, police IT etc weren't exactly thrilled by that argument, and quite a few of those have now either swapped to CentOS and do all support in-house, or dumped anything originating from Red Hat and either swap to another distro family or roll their own.

Re:I'm dying of curiousity

[dottrap]

Chances are VMWare will eventually just release the bare minimum code in question after they hit a small threshold in legal fees and make the lawsuit moot.

Waiting for somebody else bring up litigation didn't cost VMWare much. They can simply wait until somebody calls them on it, then do the bare minimum to make it go away.

And nobody will continue to press on them because seeking "damages" is really hard to demonstrate in this kind of case and continuing to pay lawyers to try to punish VMWare will be cost prohibitive.

There is also the case law risk that if VMWare does win, the GPL could be weakened/undermined, so there will be a lot of pressure on the plaintiff to not press their luck.

Re:I'm dying of curiousity

[phantomfive]

That's not true either. Anything in user-space uses the actual implementation, but is typically not considered a derivative work. Every driver is using more than the headers, it is using the actual implementation, because it is actually calling into the kernel.

Re:I'm dying of curiousity

[Shinobi]

But, ESX is not ESXi. ESXi is very different from ESX(and ESX was discontinued years ago). If they are really suing VMWare over ESXi but based on arguments about ESX, then it only looks like an ideology-based extortion-via-lawsuit attempt.

Re: I'm dying of curiousity

[Anonymous Coward]

Don't bet the farm on that, others have thought the same and list in court.

There is a solution, VMWare just needs to replace the existing GpL2 code with their own. Something VMWare should have done in the first place rather than not replace all GPL code when originally moving to a closed source kernel.

Re:I'm dying of curiousity

[dottrap]

That is, unless VMWare sincerely thinks they are in the right and have a defensible case. Then things get very interesting because then there is a chance the GPL could be undermined/weakened if they win and you will see a lot of groups start paying attention to make sure Software Freedom Conservancy doesn't screw up the case for GPL. (And you may see other parties interested in exploiting a weakened GPL.)

Re:I'm dying of curiousity

[omfglearntoplay]

Are they a shitty company now? My vote is, ever since EMC's CEO took over as VMware's CEO, they've been going downhill. Have you seen the travesty that is their web admin client that is now required? It's like go out of business bad interface for their paying customers? They are worried about lock in, money, cloud, and whatever is important to the money-loving decision makers at the top of their internal hierarchy right now. I'm worried and hope that they realize their problems. I also have some mid-level EMC products, and the amount of problems I've started having after the 1 year mark is terrible. Never will I buy another EMC branded product if I can help it. VMware started off great, and it's still better than other VM products, but for how long? I'm certainly not converting my non-VM environments over to VMware because of the amount they are pissing me off with their ridiculous web client/GUI.

Re:I'm dying of curiousity

[HornWumpus]

So binary only drivers violate the GPL?

Re:I'm dying of curiousity

[sjames]

That may have started it, but keep in mind, talks have been ongoing for several years now. They have had every opportunity to correct the problem. The suit didn't happen until VMware made it clear that they have no intention of remedying the situation.

Re:I'm dying of curiousity

[sjames]

The deciding factor is whether or not it is using the defined API. If so, it is mere use and not derivation.

If the kernel has to be modified to add an API, there would be a clear violation. If the driver is digging underneath the public API, it is on very shaky ground.

In the case of the Linux kernel, the line has been made very clear for modules. The module declares it's license and the kernel decides based on that what will or will not be linked.

Re:I'm dying of curiousity

[sjames]

The diagram was fairly clear that various large parts of the Linux kernel are modified and incorporated into the vmkernel. It doesn't say exactly what parts but at least one of them is the SCSI subsystem.

Re:I'm dying of curiousity

[tlambert]

So binary only drivers violate the GPL?

No, but they really piss off people who disagree with Linus' interpretation of the GPL when it comes to binary only drivers.

I imagine this case actually hinges on an EXPORT_SYMBOL_GPL symbol that VMWare uses, or an API underneath one of those as an API aggregator, to which the aggregator would like to force all sub-APIs to reverse-inherit the label. Which is more or less similar to const-poisoning all the way down.

My expectation is that (1) If there is not explicit user, or (2) it's possible to recompile the drivers, which is what's really in question, that if VMWare themselves don't do the compilation, it's going to be fine.

One of the reasons for bringing this action in the venue where it's being brought is that it's extremely unlikely to get a sympathetic hearing elsewhere, as people will quote Linux ad infinitum, and Helwig would almost certainly lose his case. He may lose anyway, as a matter of "standing", before they even make it to any other arguments.

Re:I'm dying of curiousity

[Anonymous Coward]

I definitely am no fan of their Web client. I can list my beefs about VMWare as well, from a user and IT person point of view:

1: The software is insanely expensive. So much that funding endeavors to use OpenStack and deal with its rough edges and incompatibilies is cheaper than licensing VMWare and its technologies.

2: There is an assumption you will have a SAN or NAS behind this, which of course... EMC will be happy to sell you. It would be nice if VMFS5 supported deduplication, or EMC would merge some of their Isilon technology into ESXi so you can plug in some infiniband cables, and add more compute nodes when more storage is needed.

However, even now, EMC/VMWare is pretty much the only game in town when it comes to serious (as in production critical) virtualization on large clusters. Yes, universities can use OpenStack, but they have a huge cheap labor pool to help support things. Hyper-V is constantly improving, but they still need to improve RAM management on compute nodes, have better failover, ditch CSV for a real clustered filesystem like VMFS, allow for easier migration of VMs across compute nodes and backing stores, and a number of other items.

Yes, there are other virtualization systems out there, but they tend to be nowhere as mature as VMWare's stuff, as VMWare can easily scale from an old 2008 MacBook to a large datacenter. As someone who has used Xen, KVM, and other items in large environments, there is definitely a big difference.

Of course, VMWare knows this, so their licensing prices are through the roof. I'm hoping Hyper-V gets more mature because if this becomes anywhere near as scalable and usable, it will provide fierce competition.

Re:I'm dying of curiousity

[Richard_at_work]

And yet we have had three versions of the GPL, several of the LGPL, and even more of the AGPL. Not to mention the rest of the derivatives...

It just goes to show that the GPL isn't even considered watertight by its creators, just like any other license.

Re:I'm dying of curiousity

[Shinobi]

And GPL is very much "what's yours is mine and what's mine also makes yours mine"

Re:I'm dying of curiousity

[DarkOx]

There is no reason to think the court would require the entire stack / product to be opened. I don't anything any sensible read of the license would dictate that either.

At most vmkernel would have to be GPL licensed to use the code. The big question with the GPL (IANAL but have paid attention to this for a long time) is who can claim harm when its violated. Will the court agree the anyone who says they were harmed were harmed, is only the author harmed, are only users of the code harmed, is only the current copyright assignee harmed?

That all will have effects on standing. I can't succeed with a civil suit against you based on the claim your dog bit someone else kid and I read about it in the paper. I have not been harmed. This what the nature and magnitude of the harm is often affects the remedy in these cases too. If the court agreed for instance that everyone who used VMWare was harmed, the remedy might be they have to release the current version of vmkernel source and GPL it. If the court decides only the copyright holder(s) was harmed, who knows monetary valuation on code they don't sell won't be easy. The court might just say well you have to cease infringing their copyright if they ask you to. Which might not mean anymore than EMC has to remove the Linux code an issue a patch to existing customers.

Re:I'm dying of curiousity

[caseih]

Software licensing is always complicated, particularly when a product is a assembly of separate modules and pieces. If they have violated the license for the Linux kernel itself, I highly doubt this would touch their actual VMWare software at all, though a court could determine that. Conventional wisdom would indicate that the license does not cross the API boundary, so the copyright infringement ends at their kernel drivers. My understanding from the article is that it's the kernel drivers themselves (the vmkernel module anyway) that VMWare includes borrowed code from the kernel sources, which are then released with a proprietary license. Had VMWare wrote all the driver from scratch, there wouldn't even be a case, even if the drivers are proprietary, non-free code.

Re:I'm dying of curiousity

[Jokkey]

It's particularly interesting since VMware seems, in many respects, friendly to open source. They distribute a bunch of open source in their products and extensively document this, they partner with open source projects like Docker and OpenStack, and they're on GitHub. The sticking point here seems to be that they have kernel-level code that they think isn't covered by the GPL and Hellwig and the Conservancy do.

Re:I'm dying of curiousity

[petermgreen]

Where things get tricky is the "mere aggregation" clause in the GPL. AIUI a storage or distribution medium containing two unrelated programs is a derivative work of both those programs but thanks to the "mere aggregation" clause it's perfectly OK if one of those is GPL and one is propietary.

So the question becomes what level of interaction/interconnection/integration does there need to be between two works distributed together such that they no longer satisfy the "mere aggregation" clause and hence violate the GPL. This becomes especially tricky if the interface definition is under a GPL-compatible non-copyleft license such that both the GPL and propietary work could have been created without reference to the other. There is also the question of whether interfaces are copyrightable at all (see the recent oracle VS google lawsuits).

Re:I'm dying of curiousity

[HiThere]

IIUC, German law is different on this, and anyone is allowed to file suit to enforce anyone's copyright claim. I usually think this is a bad idea, but in this case I'm unsure.

Re:I'm dying of curiousity

[jellomizer]

I think it is mainly due to the issue, that enough hardware manufactures support linux officially that it makes it easier for them.
I expect if they lose the case, they may switch to BSD, or even work with a deal with Microsoft to use the NT Kernel, instead.

Sometimes if you force someones hand they just might leave the game.

Re:I'm dying of curiousity

[jellomizer]

Or it may be due to a general confusion of the GPL/Willing to push its meaning to the limits.

This is why if I do any coding professionally, that is meant to be distributed, I avoid the GPL like a plague, and go with MIT or BSD licenses. Just so I have the freedom to choose how I want to expand the code.

Re:I'm dying of curiousity

[phantomfive]

The deciding factor is whether or not it is using the defined API. If so, it is mere use and not derivation.

You are wrong; if you care to understand why, Linus explains in more depth.

Re:I'm dying of curiousity

[phantomfive]

It can. Here is a more in-depth discussion.

Re: I'm dying of curiousity

[phantomfive]

The userspacr process does not execute any kernel code, just request that the kernel do something.

Any function call can be characterized in the same way.

Re:I'm dying of curiousity

[Shinobi]

No, the diagram is very simplified, only showing 3 layers, and that the alleged Linux kernel and ESXi hypervisor kernel occupy kernel space. It is very disingenious, since ESXi is bare metal and stand-alone, unlike ESX, which was discontinued years ago. In fact, when reading the link to mailing lists someone in the thread posted, many of the complaints are about ESX, and invalid for ESXi.

In fact, the entire situation smells fishy.... I wonder if they are trying to appropriate VMware code somehow to improve Xen code.

Re:I'm dying of curiousity

[bug1]

What on earth do they think trumps their obligations to the license they agreed to by using GPLv2 software in their product?

Their profits i imagine.

A human brain will accept all sorts of compromised logic to get something it really wants.

Re:I'm dying of curiousity

[MightyMartian]

I guess the point has to be, what has VMWare bought to the game. They've essentially grabbed the Linux kernel, stacked their own kernel extensions on top of it and called it their own. I've never heard of them as major contributors to the Linux kernel itself.

Re:I'm dying of curiousity

[Cramer]

As best I can tell, from their limited "weasel lawyer speak", they are complaining about the linux driver compatibility layer of VMware's kernel. Something that has been depreciated for some time now -- it never was all that great for performance, but did allow many hardware vendors to offer ESXi drivers with very little effort.

Re:I'm dying of curiousity

[sjames]

Actually, he said the same thing I said, in different words.

For example, he points out that using the defined API, such as system calls is not a derived work, but digging into the internals is.

Re:I'm dying of curiousity

[sjames]

The diagram remains clear. It shows what I said it does. Your question is more about if the diagram shows reality or not.

I note though that the Linux kernel is at the bare metal level itself.

Re:I'm dying of curiousity

[HornWumpus]

Old thread. Linus apparently thought header files/interfaces could be copyrighted. He's been shown to be wrong in the USA. Thank dog.

Re:I'm dying of curiousity

[phantomfive]

Old thread. Linus apparently thought header files/interfaces could be copyrighted. He's been shown to be wrong in the USA.

A) If you're talking about Oracle VS Google, The appeals court overruled that

B) It's not just 'using the header files', it's literally using the kernel code, so even if interfaces couldn't be copyrighted, that wouldn't apply here.

Re:I'm dying of curiousity

[Shinobi]

But with ESXi, the linux kernel runs on top of the hypervisor.

Re:I'm dying of curiousity

[sjames]

And according to the diagram, parts of the linux kernel are incorporated INTO the hypervisor. That, I imagine, is what has lead to a legal dispute.

Re:I'm dying of curiousity

[gmack]

Why would the guy who wrote the driver in question not have standing?

Re:I'm dying of curiousity

[HornWumpus]

A. That sucks. Hopefully the supreme court returns sanity.

B. All interfaces 'literally use the code' If calling an interface makes my code a derived work that is very, very bad in a larger sense. SCO might have won with that argument.

Re:I'm dying of curiousity

[Shinobi]

Which is the bogus claim, as anyone who's worked with ESXi and developing drivers for it can testify(which I have done). ESX, otoh, ran sort of parallell with the Linux kernel, but ESX was discontinued years ago.

Re:I'm dying of curiousity

[sjames]

I have no opinion on the facts of the situation. I only know what the diagram clearly claims.

Re:I'm dying of curiousity

[HornWumpus]

Specifically he says if the symbol is exported with _GPL at it's end he thinks it's protected.

Apparently he invented an 'intent of author standard'. Which would suck in the bigger sense.

I sure hope Oracle loses the header case.

Re:I'm dying of curiousity

[mysidia]

You are mistaken in thinking they use the Linux kernel in ESXi. There is no Linux kernel anywhere in ESXi.

They have written their own operating system from scratch, and they did a complete rewrite of the kernel in the update from ESXi 3.5 to 4.0.

What they have done is copied a subset the interface API from the Linux kernel. Much how like the Wine Project has copied API details from Win32 without permission from Microsoft.

This allows existing driver source code that already works in Linux to be compiled using the VMware driver development kit into a binary that can be loaded as a driver in ESXi.

This means that hardware vendors can write the driver once, and then it could be built for either Linux or ESXi, so that seems beneficial for Linux users to have more drivers still being written for Linux.

This is considered a legacy framework, and VMware is already phasing this out... see details on the new native driver framework

This will be sad, as the native driver framework is proprietary, and it will likely no longer be possible to write your own drivers for ESXi, once vmklinux is gone, without purchasing the driver development tools at high $$$.

Also, major enterprises are running ESXi on much of their hardware, so the incentive may go away for many manufacturers to release information or develop Linux drivers; they can just produce their binary ESXi drivers and be done with it.

Re:I'm dying of curiousity

[phantomfive]

All interfaces 'literally use the code' If calling an interface makes my code a derived work that is very, very bad in a larger sense.

Do you understand that the case of the linux kernel is different than the oracle vs google case? Google literally rewrote all the code from scratch, merely maintaining the APIs. In the case of the Linux kernel, drivers use the actual code written by Linus et al, it's not just using the APIs.

Re:I'm dying of curiousity

[Shinobi]

But you still made factual claims, based merely on the diagram

Re:I'm dying of curiousity

[cas2000]

if someone sees the terms of the GPL and then chooses not to use GPL-ed code, then that is the GPL working as intended. You're only supposed to use GPL code if you are willing to pay the licensing "fee", which is that your derivative work can only be distributed under the same terms.

If you choose not to use GPL code, that's perfectly OK. That's not a bug in the GPL, it's an intended feature.

vmware's fault is that they chose to use GPL-ed code while refusing to pay the license "fee". This directly harms not only their users (whose GPL rights to the derivative work they've bought from vmware have been infringed) and Christoph Hellwig but also everyone who has ever contributed in any way to the linux kernel, as those contributions were made with an expectation that any derivative work would only be distributed under the terms of the GPL.

Re:I'm dying of curiousity

[Tom]

You may have noticed I don't care how it got there, only why they are acting now the way they are.

Many companies have this immune system response that if something happens that shouldn't have, they will at the same time punish someone internally, and defend themselves externally claiming everything is proper.

Re:I'm dying of curiousity

[cas2000]

no, it's not the mistake of a junior dev - it's quite clear that vmkernel is entirely based around the linux kernel, most of the code is linux with some changes and additions.

either:

1. they're recklessly ignorant about software licenses

2. they're unable/unwilling to see the difference between BSD licensed code and GPL licensed code

3. they thought they'd get away with it, either because they thought nobody would notice or because the legal costs of license enforcement would be prohibitive.

Re:I'm dying of curiousity

[sjames]

Yes, I made factual claims ABOUT THE DIAGRAM.

Re:I'm dying of curiousity

[sumdumass]

I guess the question is does he have copyrights for the parts infringed upon? If he wrote the scheduler and the infringement is with the driver API which he may not have any copyright, its the same as claiming your dog bit someone else child. Neither the GPL or the copyright law grant you rights to the entire copyrighted entity for partial contribution ( unless it's the same file carrying the copyright). And that is limited further to the extend that the contributed copyright remains (I can rewrite your contributions in a non derivative way and end up with a file that you no longer have a copyright interest in)

I cannot find exactly where they suppose the infringement is other than they used the linux kernel somewhere in the process and the outcomes appear similar. The articles are scarce on details.

Comparing it to a dog bite is a little difficult though. The harm is two parts, first a legal right to control distribution of a copyrighted material and second losses or harm caused by the violation of those rights. If it is the first, they can stop VMware from using the code but they will either need to show that their use deprived him from something concerning his copyright or that he is entitled to royalty for the illegal uses in order to get monetary rewards. Seeing how linux is distributed freely, I suspect he has a good faith belief his copyright is being violated and wants to stop that from happening as outside of recovering court costs, it will be difficult to get much more.

Re:I'm dying of curiousity

[Darinbob]

Because that's what a lot of customers want and it's easy to market a Linux product to IT departments. I also think there may be some synergy between a VM running Linux and the underlying vmklinux host.

Re:I'm dying of curiousity

[Darinbob]

There's another step in here: is the code copyrighted? Without that part the rest does not matter. That's why GPL has a copyright even though Stallman seems to be personally opposed to the concept with software.

Re:I'm dying of curiousity

[sumdumass]

Actually, I believe the claim is that it was which is why Nvidia ended up using a two part driver module in the early days with one being licensed under the LGPL.

I'm not sure how they do it now but I don't think they use the two part modules any more. I believe Travolds has made statements that binary only blobs in the kernel violate the GPL too as they are derivative. Actually, here is an interesting discussion concerning the linux kernel and derived works the confusion possibilities.

Re:I'm dying of curiousity

[sumdumass]

IS that what happened? As far as we can tell from the provided links is that the guy in question has copyright in the kernel and somehow the VMware software uses parts of the kernel and can be graphed to look similar to the operations of the linux kernel. I have yet to find anything detailing the exact claim of infringement involved as in what files where.

Re: I'm dying of curiousity

[buchanmilne]

Have you tried Red Hat Enterprise Virtualisation ( http://www.redhat.com/promo/rh... )? It adds the management features like vCenter Server adds the vSphere to ESXi, and costs less for a suppprted environment than the cheapest VMWare option, and has feature parity and performs better (kvm).

Re:I'm dying of curiousity

[Paul Jakma]

Which means their own code depends heavily on the Linux code they've borrowed. Which, to my layman's understanding, likely means the resulting combined work is a derived work of the Linux code concerned, at least under US norms. Which means they would need to follow the GPL for *all* the code concerned. Alternatively, if they don't like that, they have the choice of not using the Linux code they don't own.

My understanding is based on advice from US corporate counsel that I have dealt with. Corporate counsel gave us in engineering a rule of thumb that adding such dependencies on others' code likely introduced licensing issues (i.e. created a derived work).

Re:I'm dying of curiousity

[Paul Jakma]

My understanding is that your description is inaccurate.

Yes, they have implemented a number of Linux APIs in their own code. Additionally, they have sucked in bits of GPL Linux code that implemented bits of those APIs (i.e. NOT reimplemented, as WINE does). This is to allow ESXi to be able to re-use drivers from Linux, as you say. However, they didn't stop there, from what I understand. They also have ESXi use Linux GPL drivers.

My understanding, from what I've read, is that ESXi didn't just re-implement Linux APIs. ESXi also heavily depends on GPL licensed Linux code, both in the partial-reimplementation of Linux APIs, and in sucking in Linux GPL drivers. The issue is this direct re-use of GPL code, and ESXi's heavy dependence on that GPL code. That dependence likely makes ESXi a derived work of the Linux GPL code, and as such it - in its *entirety* - must be distributed in accordance with the GPL.

Alternatively, VMWare are quite free to not use code they didn't write and don't own, if they don't like the licence conditions.

This is very different from what WINE does, and to characterise the situation as like WINE seems to be quite inaccurate.

Re:I'm dying of curiousity

[sjames]

Headers and kernel linking are separate issues. The headers are definitely part of the documented public API.

Meanwhile, when the module system was built for the kernel, all of the modules were GPL and the question wasn't even considered. It's a bit of a special case, when it's Free software, what constitutes the public API? Linux clarifies that marking the function as _GPL makes it the public API in kernel space.

In non-Free software, non-public API is that which is not in the headers or that which no proper way to access it is provided. Effectively, "intent of author" has always been the standard, Linus just documents it a lot more clearly than others.

Oracle should lose. I simply can't see bare documentation of an API as a creative work and simply offering the same API is more akin to writing a distinct work in the same genre than it is to copyright violation. Otherwise, there could only ever be one detective novel (if you've seen one McGuffin, you've seen them all) and one space opera, etc. That is, nearly every environment has a read function that takes some sort of handle (generally returned by an open function), a size, and a destination. There's not many ways to express that in a header.

Re:I'm dying of curiousity

[Paul Jakma]

Uh, no, not at all.

The GPL says "You're free to go build on what's mine, but that makes yours ours".

If you don't like that aspect of the GPL, then you're perfectly free to not use that code and go build your own instead.

Re:I'm dying of curiousity

[mysidia]

However, they didn't stop there, from what I understand. They also have ESXi use Linux GPL drivers.

GPL drivers, such as?

Note.... ESXi does not ship with a large collection of drivers. They have a rather exclusive hardware compatibility list. If it's not in the list, then there is probably not a driver included.

To say this might be a GPL violation, you would need to find a driver that is GPL'd and also only available under GPL. Many of the GPL'd drivers are developed by the hardware vendors who can provide the binary code for use with VMware under other times, since it's their driver, they can license their work for other uses freely, as they will.

VMware could also get around this problem by distributing just the source code for the GPL'd driver, since these are discrete software modules that you compile, then copy the binary into ESXi, then load; VMware can distribute the driver modules which could even be the same source code as in the Linux kernel, without needing to GPL vmklinux or vmkernel itself.

For the small number of hardware drivers that they do have included, they are distributed as loadable files called .VIB files. As far as I know, these are supplied by the hardware manufacturers, for example:

~ # esxcli software vib list | grep ixg
net-ixgbe 3.21.4-1OEM.550.0.0.1331820 Intel VMwareCertified 2014-09-30
~ #

Re:I'm dying of curiousity

[Paul Jakma]

The Software Conservancy FAQ has a diagramme giving an abstract of what they allege has been copied:

http://sfconservancy.org/linux...

With respect to the drivers, it seems they've copied SCSI, USB and network drivers. Christoph Hellwig holding copyright to at least some of the SCSI drivers concerned, in addition to core code VMWare are alleged to have copied to implement required APIs for the drivers.

It seems you could give us a full list from an ESXi installation, if you wished, rather than just a selected driver (selected why?).

Re:I'm dying of curiousity

[Paul Jakma]

BTW, a vendor that wrote a Linux driver could give a different, non-GPL licence to that driver code, of course. However, that still leaves the issue that Linux drivers are written within a framework of core Linux code (driver sub-system specific frameworks and further more generic services and data-structures). The driver vendor can not give a non-GPL licence for that core code they didn't write.

VMWare are alleged to have copied such core code too. Further, they are alleged to also have used GPL driver code (e.g. Hellwig's SCSI). So VMWare, according to the allegations, have borrowed GPL code on /both/ sides of the "line" between drivers and their supporting code. Without fulfilling the conditions required by the GPL for legal, licensed use of GPL code...

Re:I'm dying of curiousity

[countach]

I'm not sure that they do have the choice of "not using the Linux code they don't own" to avoid following the GPL, because they *already* distributed it. That was an option before they went and distributed it, but now they've done it, I think they've burnt that bridge.

Re:I'm dying of curiousity

[Paul Jakma]

Good point. I should have used "had" instead of "have": "they had the choice of not using the Linux code they don't own".

Ta!

Re:I'm dying of curiousity

[Richard_at_work]

No, your opinion is being formed from what the diagram is insinuating - that and the actual truth of the matter can be two entirely different things. The diagram is certainly far from clear, changing a few colours on a chart is not a technical explanation.

Re:I'm dying of curiousity

[mysidia]

Driver sub-system specific frameworks and further more generic services and data-structures

VMware can implement all the data structures, generic service definitions, and frameworks they want, such as having the same function names, and struct files, the "Headers", without necessarily having to copy the executable code that is defined in those frameworks.

The software program is subject to copyright. The use or re-implementation of interfaces to a software program are considered fair use, and it's an important concept behind open source, that the interfaces when different packages talk to each other allow drop-in replacement of one program for another, such as a new open source software program to control your proprietary piece of hardware.

These function prototypes, struct formats, and interface definitions, "the headers"; are either not subject to copyright, or Linux itself and many GNU projects would be in deep trouble, because in the Linux world copying structs and headers from proprietary software is a common practice, that actually is part of what allows GNU to even exist.

So if they made that argument in court that the frameworks are subject to copyright, it would also be a major setback for many open source software projects such as Wine, or even various GCC projects.

And the Linux kernel as well.... if the Linux kernel struct formats and headers were subject to copyrights, then struct formats used to control hardware would be as well, and Microsoft could sue Linux developers over making a driver for hardware device X that used a struct for control based on a Microsoft copyrighted standard.

Also, SCO probably should have won if this was the case, and since POSIX itself is copyrighted, and header files in many BSD-related OSes include fundamental interface definitions that are copyrighted with no license to use.....

Re:I'm dying of curiousity

[sjames]

No, it isn't. I have formed an opinion of what the diagram is trying to communicate. Understanding a communication is not the same as believing it.

But just in case, you must send me all your money!

Re:I'm dying of curiousity

[HornWumpus]

I believe the words you are looking for is 'undocumented API'.

An undocumented API is almost guaranteed to change. But is less protected than a normal API as you don't even have .h files to point at and claim.

Undocumented APIs in commercial software are not protected, they are just a bad idea to use.

Re:I'm dying of curiousity

[cas2000]

Pragmatism is only a secondary concern for free software (as opposed to open source).

People who license their software under copyleft terms don't want people who wont license their software as copyleft to benefit from their work. In that sense, them wasting their time is also the GPL working as intended. If they're not willing to pay the license "fee", then why the hell should they benefit? In other words, if you won't share, then write your own code or pay a commercial licence for proprietary code.

The GPL is more than just a software license, it's a political statement and political action about sharing and freedom for ALL users, not just for a handful of developers who want to use other people's work for free. If you want to benefit from sharing, then you also have to share.

if you don't like it or don't want to accept the terms, that's fine - but write your own code.

Re:I'm dying of curiousity

[Paul Jakma]

Except they didn't re-implement the sub-system framework and data-structure APIs themselves. The lawsuit alleges that they took some code from Linux (e.g. radix tree, timer API stuff). Even if VMWare /had/ reimplemented those APIs from scratch, then there is still the issue that they /also/ have appropriated the code GPL-only drivers (as alleged by the lawsuit at least) for ESXi.

Also, will you provide that list of drivers? If you won't, I have to wonder if you're favourably predisposed to VMWare in some way.

More than curious,

[piRSqrd]

I need to know if I should stop using VMWare.

Re:More than curious,

[pak9rabid]

The answer to this should have been obvious 8 years ago when they:

1. Made their management tools run on only Windows, shitting on the Linux community
2. Deprecated VMWare Server 1.x (free and very functional) for VMWare Server 2 (free and barely functional)

There are far better free alternatives out there nowadays if you're not managing a full-blown cloud infrastructure (see: LXC and KVM). And if you are, there's OpenStack.

Re:More than curious,

[GerryGilmore]

Of course not. Why would you even think that? If VMWare loses, which they should, IMO, all that will happen is that they'll either pay $$ to Christoph or just rewrite the offending sections. VMWare isn't going anywhere...

Re:More than curious,

[alen]

we are wiping it tomorrow and rerolling all of our vm's with hyper-v or xen

Re:More than curious,

[ledow]

Judging by the diagram, the "offending sections" are huge swathes of virtualisation-specific code that ESXi runs on, written or modified by VMWare.

Being forced to open-source their largest software project is a quite conceivable (even if unlikely) outcome. Not everyone who has a part in the Linux kernel is going to accept "Here's some cash, ssshh, don't tell anyone we cocked up" compared to forcing them to open their code.

VMWare accepted a (believed to be) legally binding agreement to open any code that modified or become part of the kernel under GPL licences to anyone who asked. You can't always buy your way out of such obligations with money, and the courts may well end up showing that.

And that... that's gotta hurt the bottom line to have vast portions of ESXi available as open-source software.

Re:More than curious,

[DarkFencer]

Their management tools are transitioning, slowly to web based. There are tasks that ONLY work in the web client now but there are also tasks that ONLY exist in the Windows only desktop client.

If you're only running guests, then you can get away with the slow web client. If you're managing the hosts you need to use both, for different things now.

Re:More than curious,

[bheading]

Being forced to open-source their largest software project is a quite conceivable (even if unlikely) outcome.

It's neither conceivable nor likely. VMware includes a lot of very, very clever technology. (I note in VMware ESXi 6 that they can now do fault tolerance with 4-way SMP. FT, for those who aren't aware, means that two physical servers between them can keep a VM continuously available even if one server fails.

It sounds to me as if VMware are prepared to contest the matter, which comes down to whether or not a court will accept that VMware have complied with the terms of the license.

The most likely outcome is that they will settle out of court. That will, most likely, involve an agreement to pay the litigant's legal costs, and some sort of time-bounded agreement to modify their software in such a way that it no longer breaches the GPL.

Re:More than curious,

[mikeroySoft]

vSphere 6 web client has full feature parity with the Windows-only client, and sucks 100% less in terms of snappy performance.
Sorry it took so long, but it was a resourcing decision at the time. Spend Engineer's time building new features, or porting the client to Linux.... seems like an easy decision for me in 2005. I am certain lots of homework went into this decision.
As for OP's comment #2... really? You're still upset over a decision made ~10 years ago?
I find it interesting how people are quick to attribute malice to things without understanding the context or rationale behind a given decision.. "They did it because they're evil!" sure sounds simple enough to say.
Life's never that simple, friends.

Disclaimer: I am sitting at my desk in VMware HQ, been with the company since 2009.

Re:More than curious,

[mikeroySoft]

Seems like a brash decision, assuming it's based on this report.
Not even going to wait until the courts reach a verdict?
Is the disruption to your organization worth it?

Re:More than curious,

[Cramer]

When have their management tools (viclient) run on anything other than windows?

They dropped Server v1 because it's a thousand f'ing years old.

Re:More than curious,

[phantomfive]

So you prefer ones that shit on the Windows community?

Yes.

Re:More than curious,

[Cramer]

FLASH based, through a web page. And it's a horrible mash of crap compared to the native client.

Re:More than curious,

[Cramer]

Name one (common) thing that can only be done via the native client? The flash client is a pain in the ass, but I've yet to run into a task it couldn't complete. (it might not be the fastest, or most straight forward method, 'tho)

Re:More than curious,

[rubycodez]

and not able to be used by browsers in Linux because of flash version needed

it can be used by browser on OS X though

Re:More than curious,

[pak9rabid]

Those aren't the reasons why I currently shy away from VMWare (it sounded like he was asking more from a moral point of view, hence the answers I gave), but those reasons, plus others, prompted me to start looking for alternatives.

The biggest problems I had with VMWare back then was 1.) when they made the switch to the web-based managed interface for VMWare Server 2 (it was sluggish and very unresponsive compared to the VMWare Server 1 management console) and 2.) the lack of a mature CLI toolset for managing VMs (this was big for me because I wanted to be able to SSH into a VM server and do trivial things, like start a VM without having to get a GUI involved).

I first made the switch to VirtualBox because it was free, had a good GUI management utility (for managing local testing or development VMs), and a nice CLI utility (VBoxHeadless) for managing remote VMs on a headless server. Eventually I started running into weird performance issues with VirtualBox which prompted me to give KVM a second hard look.

I had first looked into using KVM a few years prior due to the hypervisor being built-into the standard Linux kernel, which sounded really appealing to me, but it wasn't quite ready for production use. Once KVM/libvirt became stable enough to use, I made the switch to that and I haven't looked back. It has everything I need: a rich CLI environment (virsh), a rich API for managing it at a lower level (libvirt + bindings to all the popular languages), a decent GUI that works surprisingly well over X11-forwarded sessions (virt-manager) for when it make more sense just to do things in a GUI environment, and a nice range of storage options for VMs (raw disk image files, qcow2 images, LVM logical volumes (my personal favorite), block devices, etc).

Just recently I've started playing with LXC containers. For instances where I don't need the benfits of full paravirtualization and don't mind sharing the host machine's kernel it's awesome. It performs better than paravirtualization because there's less overhead (thanks to kernel and process namespaces), and if I need to change something in the guests' filesystem I can access the files directly from the host machine, or chroot into it if I need to do something fancy.

I'll give VMWare the credit they deserve; they paved the way for paravirtualization on the x86 front in the early 2000s and for many years I was a faithful user (even fanboy) of theirs. When they had their IPO back in 2007 I bought as many shares as I could afford and advised my friends and family to do the same because I knew they were the industry leader in an industry that was exploding.

I also realize that I am not VMWare's target audience and have since moved on to tools that are more suitable for my needs.

Re:More than curious,

[harryjohnston]

One possibility is that suit is based on the theory that simply sharing an address space between the GPL code and the proprietary code, e.g., because they are both running in kernel mode, is enough to create a derivative work in violation of the GPL.

As far as I know, this theory has not yet been tested in court - am I mistaken?

Interpreting these conditions

[tepples]

The controversial part, as I understand it, is the difference in interpretation of a license's conditions. For example, the difference between an "aggregation" and a "combined work" in the GPLv2 confused at least one Slashdot user.

Re:Interpreting these conditions

[Marginal Coward]

To take that a step further, the GPL is largely untested in court. Per TFS: "this, to our knowledge, marks the first time an enforcement case is exclusively focused on this type of legal question relating to GPL." IANAL, but any element of it that is reasonably subject to interpretation can be interpreted any way you like - until a case comes along to codify some other interpretation.

Therefore, from a strictly business point of view, you can interpret it in whatever way favors your business. However, if you push the boundaries of that, you run the risk of some future court case - against either you or someone else - interpreting it differently.

To me, the real purpose of most open source software licenses is to give honest people a set of rules to play by. It's likely that there are many violations of all but the most permissive licenses that we don't know about - and never will. There have only been a few situations that have been publicized of anybody allegedly violating the GPL, so it's likely that most users are complying, and those who aren't have found it easy to go unnoticed and/or unchallenged.

Re:Interpreting these conditions

[Anonymous Coward]

To take that a step further, the GPL is largely untested in court.

Mostly because every violator's lawyers have looked at it and said "Give up. There's no way we can win this in a court case. This thing is ironclad."

Re:Interpreting these conditions

[hawkinspeter]

This could be interesting as VMWare are most likely going to have to rely on the GPL being valid as otherwise they are not allowed to distribute that code, only their own code. So, presumably they need to show that the GPL is valid and that they are complying with it (which will be subject to interpretation).

Users don't have to worry about the GPL as anyone is allowed to use GPL code however they like, but it's only when you distribute it that you come up against its restrictions.

Re:Interpreting these conditions

[bmo]

the GPL is largely untested in court.

No it isn't. It's been tested at the federal level.

Daniel Wallace tried to get the GPL declared invalid through stretching of legal concepts, and was thusly shown how stupid /that/ is.

Wallace v. International Business Machines Corp.
From Wikipedia, the free encyclopedia
Wallace v. International Business Machines Corp. et al., 467 F.3d 1104 (7th Cir. 2006), was a significant case in the development of free software. The case decided, at the Court of Appeals for the Seventh Circuit, that in United States law the GNU General Public License (GPL) did not contravene federal antitrust laws. Daniel Wallace, a United States citizen, sued the Free Software Foundation (FSF) for price fixing. In a later lawsuit, he unsuccessfully sued IBM, Novell, and Red Hat. Wallace claimed that free Linux prevented him from making a profit from selling his own operating system.[1]

And this quote from the decision shows that the courts completely understand the values behind the GPL and copyleft.

From the 7'th Circuit decision of the Wallace vs. IBM appeal:

http://www.internetlibrary.com...
  People may make and distribute derivative works if and only if they come under the same license terms as the original work. Thus the GPL propagates from user to user and revision to revision: neither the original author, nor any creator of a revised or improved version, may charge for the software or allow any successor to charge. Copyright law, usually the basis of limiting reproduction in order to collect a fee, ensures that open-source software remains free: any attempt to sell a derivative work will violate the copyright laws, even if the improver has not accepted the GPL. The Free Software Foundation calls the result âoecopyleft.â

And notice the subsequent utter silence from Darl and the lawyers at SCO, who were jumping up and down about the so-called unconstitutionality of the GPL. Among other things.

The validity of the GPL is now settled law.

but any element of it that is reasonably subject to interpretation can be interpreted any way you like

This is why you aren't a lawyer.

--
BMO - not a lawyer, but someone who doesn't agree with people who think that lawyers perform magic. They don't.

Re:Interpreting these conditions

[Kjella]

The controversial part, as I understand it, is the difference in interpretation of a license's conditions. For example, the difference between an "aggregation" and a "combined work" in the GPLv2 confused at least one Slashdot user.

Actually the ugliest part of the GPL which is clear as ink in law is what - if anything - makes inter-module communication derivative. The theory of derivative works mainly involve sections or elements reappearing in the derivative, like a composite made from a photo. It doesn't cover interfaces where independently developed code calls each other at all. If I wrap a GPL library into a web service, is calling it derivative? If the answer is yes, the GPL is extremely viral. If the answer is no, the GPL is in big trouble. Which is why you never get a straight answer.

This directly links in with the "mere aggregation" clause, if you can for example distribute a distro that has an application that sends mail and a mail server without those being derivative, can you also distribute proprietary software and this web service? Your software needs it, this software happens to provide it but it could in theory be provided by a different implementation. I'm sure Stallman says no, but it's entirely unclear to me if a judge would agree.

Re:Interpreting these conditions

[Anonymous Coward]

For sure it will be valid, but the question is how to interprete some things. For example: When is something a derived work, when is it mere aggregation ('when you call the GPL thing as a binary? If you just put them on the same disk? Maybe even if you link against it?)

It uses many terms that have never really been tested in court except for really obvious cases.

Re:Interpreting these conditions

[Anonymous Coward]

If I wrap a GPL library into a web service, is calling it derivative? If the answer is yes, the GPL is extremely viral. If the answer is no, the GPL is in big trouble. Which is why you never get a straight answer.

You obviously do no understand the GPL. What you say here has specifically been addressed by the Affero GPL (https://gnu.org/licenses/agpl.html) and is a major problem for GPL code. There is an uncountable number of websites out there running GPL code which do not comply with the "spirit" of copyleft simply because they attached a web interface to it, but are not actually distributing their modified work.

Re:Interpreting these conditions

[Kjella]

You obviously do no understand the GPL. What you say here has specifically been addressed by the Affero GPL

That's not what I'm talking about, because it lacks the "distribution" part. What I'm talking about is what level of detachment is necessary to say that these bits of software depend on each other, but they're not derivative of each other. And thus the GPL wouldn't apply, even if you distribute them together.

Re:Interpreting these conditions

[mysidia]

Well... a software license is a type of contract. There's a principle in contract law; that if there are multiple ways which a condition can be interpreted, then it will be interpreted for purposes at time of adjuication in the manner that most favors the parties who did not produce the contract term / present the offer.

The same contract or license text can be interpreted in different ways for different cases.

Re:Interpreting these conditions

[harryjohnston]

Not true. The FSF insists that dynamically linking proprietary code and GPL code (for example, by installing a proprietary extension in a GPL product) creates an infringing derivative work. Linus has said that creating a proprietary Linux device driver violates the GPL, even if it uses no GPL code, IIRC on the basis that the Linux driver API is itself a copyright work subject to the GPL.

(The GPLv2, at least, has language that seems to explicitly rule this interpretation out, but that doesn't appear to bother the FSF.)

I'm actually wondering if the question of dynamic linking is involved in this case.

Re:Interpreting these conditions

[harryjohnston]

The FSF disagrees with you:

http://www.gnu.org/licenses/gp...

Re:Silly people

[Mikawo]

After skimming the GPLv2, found under section 3:

For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable.

Seems like it includes building and installation to me.

Here's a suggestion for a verdict

[Opportunist]

Anyone found in willful and deliberate violation of the GPL showed that they have no interest in copyright or its protection. Hence they implicitly and irrevocably agree that they will not pursue anyone violating their copyright.

That should take care of this pretty quickly. You don't even have to look for GPL violations in products anymore, corporations will do that for you in the products of their competitor, hoping to kick them out of the market that way.

Re:Here's a suggestion for a verdict

[Anonymous Coward]

That assumes they are guilty. I'm very skeptical of Christoph Hellwig's claims. I used to be a developer on vmkernel and ESXi, but that was many years ago. But back then we were pretty firm that people could not cut and paste code from GPL projects into the platform.

But there was also a pretty poor understanding of the lack of a dynamic linking clause in GPL (as opposed to LGPL), many would argue with me that GPL software was fine to use (like busybox) as long as 1. the source was available somewhere (wrong) 2. we didn't compile and link it together in the build system and let dynamic linking do the dirty business (also wrong)

Re:Here's a suggestion for a verdict

[Danathar]

That's actually a VERY good point.

We could see MICROSOFT help fund this GPLv2 lawsuit against VMWare since they have a competing product.

Wouldn't that be bizarre.

Re:Here's a suggestion for a verdict

[preflex]

You [wouldn't] even have to look for GPL violations in products anymore, corporations will do that for you in the products of their competitor, hoping to kick them out of the market that way.

Great idea!

I'm concerned my competition is sucking up all of my users with their superior product and marketing. So, naturally, I was wondering: How can I discourage users from using my competitors product?

That's it! I should force my competitor to make his product free! The users will come flocking straight to me. Brilliant!

Re:Here's a suggestion for a verdict

[Opportunist]

If your competitor fraudulently used OSS to compete with you on unfair terms (because you're playing by the rules and develop your software yourself while he is cutting corners pirating OSS) then yes, you should.

If your competitor doesn't, all you accomplish is running up legal fees.

Re:Here's a suggestion for a verdict

[Opportunist]

No, I think there's nobody within the OSS movement that would bother to cry out when copyright was dropped and source code of every and all projects was to be disclosed and used as needed.

Re:Here's a suggestion for a verdict

[Opportunist]

I dimly remember the copyright trolls were pushing hard to get copyright into the criminal code. Maybe it's time to relent and give them a chance to hang themselves...

Installation on what machine?

[tepples]

But does it include "compilation and installation" on the end user's machines, or only on developer hardware available only to a select few? The latter interpretation leads to the Tivoization loophole in the GPLv2. GPLv3 tightened this by defining "Installation Information", its counterpart to GPLv2's "scripts used to control [...] installation", to require that execution be possible "in that User Product" if the work is designed for a consumer platform.

Re:Installation on what machine?

[TemporalBeing]

But does it include "compilation and installation" on the end user's machines, or only on developer hardware available only to a select few? The latter interpretation leads to the Tivoization loophole in the GPLv2. GPLv3 tightened this by defining "Installation Information", its counterpart to GPLv2's "scripts used to control [...] installation", to require that execution be possible "in that User Product" if the work is designed for a consumer platform.

Well, having used VMware Workstation 8 and 9, I can was able to download and modify the Linux drivers provided by VMware, necessary to fix some kernel related bugs (http://clocksmind.blogspot.com/2013/04/vmware-workstation-8-and-linux-kernel-38.html) as kernels changed over time but VMware simply didn't keep up with compatibility.

So frankly, I'm a little surprised but then, may this is for a different edition or something.

Re:Installation on what machine?

[mysidia]

Well, having used VMware Workstation 8 and 9, I can was able to download and modify the Linux drivers provided by VMware, necessary to fix some kernel related bugs

I don't think the lawsuit is over vmware tools. VMware provides source code to most of the VMware tools components; often they are installed by building the source code.

The VMware hypervisor includes a special management Virtual Machine run by the vmkernel which uses Busybox.

They do not include the source code for Busybox. However, there is a written offer for customers to request source code on CD for the product you purchased, valid for 3 years after you purchased the software product from VMware, by sending a request to an address given of VMware General Counsel, Attn: Open Source Files Request.

Versions of ESXi prior to 5.5 supported an architecture for drivers called vmklinux; essentially, the VMware kernel supported a framework compatible with Linux drivers ---- you could compile Linux drivers from source code and load them into the vmklinux system. ESXi5.5 introduced a new thing called Native Drivers, but they still support the Linux kernel driver SDK. There is no Linux kernel code, other than drivers themselves, however, they have only copied the Kernel driver API interfaces.

Re:Installation on what machine?

[Shinobi]

Wrong.

BusyBox is part of the source code ISO's on their website. Don't even need a login to grab them.

Re:Installation on what machine?

[mysidia]

BusyBox is part of the source code ISO's on their website. Don't even need a login to grab them.

Have you inspected these? If I recall correctly, the complaint made in the past was that the version on their website contained versions of open source components that had not been updated with new releases of ESXi, and some of them may have contained the original component, and not the source to VMware's highly customized versions.

Re:Installation on what machine?

[Shinobi]

Yes, I have. The isos contain everything needed to build all the open source components used for that version of the suite

Location, location, location.

[Anonymous Coward]

Those who whined about the Apple - Samsung patent cases being heard in Northern California will no doubt remain mute about this venue.

The kernel is GPL, you can change it ...

[perpenso]

The deciding factor is whether or not it is using the defined API. If so, it is mere use and not derivation. If the kernel has to be modified to add an API, there would be a clear violation.

That is simply a fork of the kernel. If the forked kernel source is published there would seem to be no violation.

Re:The kernel is GPL, you can change it ...

[sjames]

But then they MUST offer the modified source to anyone they distribute a binary to. They also get to field all of the support issues since the kernel maintainers won't touch a bug report if the kernel is tainted by a non-free module (for good reason).

Re:The kernel is GPL, you can change it ...

[perpenso]

But then they MUST offer the modified source to anyone they distribute a binary to. They also get to field all of the support issues since the kernel maintainers won't touch a bug report if the kernel is tainted by a non-free module (for good reason).

Seems a non-issue. If the bug is not in their code it could be replicated in the pristine kernel and reported to maintainers. Since the forked kernel would be used at the core of one of their virtual machines they would be the natural contact point. From the customer's perspective VMware failed, not some embedded kernel that they don't know or care about.

Re:The kernel is GPL, you can change it ...

[sjames]

If it's a non issue, they should release the modified source and be done with it. They've had 3 years to do so and flatly refuse, so the law suit.

Re:Summary a little misleading...

[Jokkey]

Reading the FAQ and TFA, this is more about BusyBox than the Linux Kernel.

No, it started with BusyBox, but it's now mostly about the Linux kernel. The BusyBox issues are largely or completely resolved. As explained in the FAQ: "VMware made substantial and good efforts toward compliance on BusyBox. However, VMware still refused to fix a few minor and one major compliance problem that we discovered during the process. Namely, there was a major violation regarding Linux itself that ultimately became Christoph's key complaint in this lawsuit."

Re:GPL is about control, not freedom

[CAPSLOCK2000]

It is about freedom, just not the freedom of the programmer. It's about the users. Freedom is not an absolute condition, it's always a balance the rights of various stakeholders. The cliche is that my right to swing my fist ends where your nose begins. Most software licenses restrict the rights of the users in favour of the programmer. The BSD licenses are vary liberal but they only focus on programmers that want to use the code. As a user you don't know anything about your rights if code is based on BSD code. Usually its 'free', but there is no guarantee. The programmer has no obligation to the user. The GPL is about giving assurances to users. If software is based on GPL code the user knows for sure that he will be able to get the code and use it.

Re:Mere aggregation is well understood.

[harryjohnston]

The FSF insists that dynamic linking - in fact anything that puts two pieces of code into the same address space - creates a derivative work.