Twin Peaks v. Red Hat
Red Hat is no stranger to lawsuits, having grappled with the Firestar patent case in 2008 and dealt successfully with patent troll IP Innovations in 2010. But the company is now bringing a GPL compliance suit to court for the first time in its history. Red Hat filed the complaint in a countersuit against a patent infringement case that was launched in early 2012. If it goes to trial, it could bring several GPL-interpretation questions to the test.
The original litigant in the case is Twin Peaks Software (TPS), which makes proprietary network backup software. TPS sued Red Hat and Red Hat's recently-acquired subsidiary Gluster on February 23, 2012. TPS charges that the GlusterFS software violates TPS's US patent 7,418,439, which covers TPS's "Mirror File System" (MFS). GlusterFS is a network filesystem that aggregates multiple storage shares into a single volume. TPS's products include TPS Replication Plus, which automatically mirrors changes between two NFS filesystems over the network, and TPS Clustering Plus, which extends a similar feature set to larger clusters.
Red Hat initially responded to the patent infringement suit on August 2, both denying the infringement and asserting that the patent itself is invalid:
Had things stopped there, the case might have proceeded as a standard software patent infringement lawsuit. Red Hat's answer to the initial claim invoked numerous other counterarguments, such as denying that TPS has the right to ask for an injunction against the allegedly infringing Red Hat products, but it stuck to denying the claims of the initial suit. But Red Hat then followed up with a September 13 countersuit that charges TPS with a copyright violation claim, and asks for an injunction against the violating products. The products in question are TPS Replication Plus and TPS My Mirror, a freeware edition of Replication Plus. Red Hat claims that both products incorporate code from mount — specifically the 2.12a version from the util-linux package for which Red Hat is the registered copyright holder — and that TPS is in violation of the terms of the license by not providing or offering the corresponding source code.
At Groklaw, Mark Webbink argues that this action ups the stakes considerably, because even if TPS's suit against Red Hat were successful, Red Hat would experience only a small impact on its bottom line, due to the relatively minor role GlusterFS plays in Red Hat's core business. If Red Hat's countersuit were successful, however, TPS would lose the sales of 50% of its products — a hit few businesses could survive.
The countersuit is in most respects a standard GPL-violation charge, much like those brought against other proprietary software vendors by other enforcement entities. But it also brings to light some peculiarities of the free software licensing realm. Red Hat alleges that the mount code in question is under GPL version 2, specifically. Failure to comply with GPLv2's source code provisions automatically terminates the violator's rights to distribute the code (section 4). The most common interpretation of this section of GPLv2 was that only the copyright holder can reinstate the violator's right to distribute the copied software. In that case, if TPS is found to have copied mount code, Red Hat could effectively force TPS to rewrite its products by refusing to reinstate its rights under the GPLv2. But not everyone agrees with that interpretation; uncertainty over the meaning of that section was also one reason why GPLv3 added provisions for a violator to regain its right to distribute by coming into compliance with the license.
Another wrinkle to the copyright-violation issue is the possibility that there are portions of other GPL-licensed works inside TPS's products. The countersuit does not address this possibility, but it cannot rule it out, either. The difference between copying from one GPL-licensed work and copying from several could be great. In the event that there are multiple GPL violations of different copyrights, even if Red Hat agreed to reinstate TPS's right to distribute mount and all other Red Hat-copyrighted code, it cannot reinstate TPS's right to distribute software written by others. That problem is academic at the moment, but it may not remain so: Eben Moglen wrote on the Software Freedom Law Center (SFLC) blog that he is investigating whether TPS's products contain software that has been copied from SFLC clients.
Moglen also says that if a violation is proven in the TPS case, it
would be "a particularly severe offense
" because TPS has
chosen to sue a member of the free software community. Consequently,
it would be profiting from the work of free software developers while
simultaneously suing them. In contrast, most GPL violations are
reported to be unintentional; Bradley Kuhn estimated
in 2011 that 98% of the violation incidents he had worked on were
cases of negligence and not malice. "Malice" might be hard to pin
down, but the fact that TPS actively initiated this legal battle
certainly increases the chances that Red Hat will choose to fight it
out rather than settle.
If Red Hat does pursue the suit, this will also be the first GPL violation case brought by a commercial Linux distribution. Many of the high-profile GPL compliance cases in years past have been fought by independent projects like BusyBox or non-commercial groups like gpl-violations.org. Fighting out the GPL violation charge also has a different feel in this case because most other GPL enforcement actions are taken in order to bring the offending party into compliance. That is not the goal here: Red Hat is using the charge to wage an injunction-versus-injunction battle. The highest-grossing Linux distributor pursuing a GPL violation charge may not have the David-versus-Goliath feel of the other cases, but it could still be an important day in court — both for Red Hat and for anyone else who builds a business on free software.
Posted Sep 20, 2012 3:16 UTC (Thu)
by pabs (subscriber, #43278)
[Link]
Posted Sep 20, 2012 16:53 UTC (Thu)
by boog (subscriber, #30882)
[Link] (13 responses)
Posted Sep 20, 2012 17:08 UTC (Thu)
by k3ninho (subscriber, #50375)
[Link] (4 responses)
K3n.
Posted Sep 20, 2012 18:11 UTC (Thu)
by boog (subscriber, #30882)
[Link]
I'm assuming they would buy their copy before the judgement. In any case, it would be an additional insurance policy.
Posted Sep 20, 2012 18:16 UTC (Thu)
by gidoca (subscriber, #62438)
[Link] (2 responses)
Posted Sep 22, 2012 10:36 UTC (Sat)
by Max.Hyre (subscriber, #1054)
[Link] (1 responses)
Thus, unless mount uses the patent in question, buying a copy would afford no protection. Even if it did, the implicit license for mount wouldn't extend to any other random use of that patent.
Posted Sep 24, 2012 9:15 UTC (Mon)
by gidoca (subscriber, #62438)
[Link]
As to the discussion below whether you have to explicitly agree to a contract, at least in Swiss law (which is the only one I have the faintest knowledge about) there are three ways to enter a contract:
Posted Sep 20, 2012 21:44 UTC (Thu)
by iabervon (subscriber, #722)
[Link] (6 responses)
If you put a $1M price tag on the television in your living room, and someone breaks in and steals it, you can't go to court and say they bought your TV for $1M but now refuse to pay you. Even if you provide a standing offer to make their actions legal, they don't have to accept it.
Posted Sep 20, 2012 23:22 UTC (Thu)
by boog (subscriber, #30882)
[Link] (5 responses)
Posted Sep 20, 2012 23:50 UTC (Thu)
by iabervon (subscriber, #722)
[Link] (4 responses)
Posted Sep 21, 2012 0:10 UTC (Fri)
by boog (subscriber, #30882)
[Link] (3 responses)
Seems to me that it's rather like "click through" (which is hard to wriggle out of). By distributing they have accepted the licence and through that licence they (implicitly) license their patents to recipents of the distributed GPL programs that practise them. A court will decide this for us one day...
Posted Sep 21, 2012 0:42 UTC (Fri)
by dlang (guest, #313)
[Link]
As such, it's very possible for a source tree to have a copy of the GPl that includes such language, but where the person using that source tree does not have to follow the GPL because they have some other license.
This is why the GPL is not 'viral' in the sense that the FUD being spread around implies. Putting a GPL file into your proprietary code does not automatically make your entire codebase GPL, it just means that you have no valid way to distribute that codebase. When you get called on it by a GPL license holder, you have three choices.
1. continue to distribute the codebase, but comply with the GPL
2. negotiate some other license to distribute the codebase.
3. remove the GPL code from your codebase
granting an additional license actually happens quite frequently in the FOSS world. Not in cases where someone has started legal action against a GPL infringer, but it's very common to see someone ask if a particular piece of code can be dual licensed (frequently under BSD) so that some other project with a non-GPL license can make use of it. If the function is fairly small, the reaction is commonly "sure"
If you take option #3, the version you are distributing now is still infringing, and the license holder can still get an injunction and go after you for damages for the version you are distributing, but if you release a new version without the GPL code (and without anything "derived" from that code, which is a slippery definition), then the GPL license holder has no claim against your new version.
In this case, since Twin Peaks was sloppy enough to use one piece of GPL code like this, the odds are really good that they have used more. Every copyright holder of such code can now go after Twin Peaks, and they can each do so individually.
Doing so is a nuclear option that could scare a lot of people away from FOSS code. The incident with Cisco that Rob Landley has been so upset about is a mild example of this. If you were to have 500 individuals sue one company for copyright infringement for the Linux kernel as separate lawsuits (nothing says they need to band together and jointly hire a lawyer), the mere cost of lawyers to handle so many separate cases would be devastating.
But if companies want to go nuclear against FOSS on patents because they don't think FOSS can defend themselves, they need to be prepared to have FOSS fight back with the tools that are available, namely copyrights.
Posted Sep 21, 2012 0:46 UTC (Fri)
by iabervon (subscriber, #722)
[Link]
Posted Sep 24, 2012 6:35 UTC (Mon)
by branden (guest, #7029)
[Link]
I'm not convinced.
Posted Sep 20, 2012 22:24 UTC (Thu)
by jhhaller (guest, #56103)
[Link]
The mount command is a standalone piece of software. It's unlikely that Twin Peaks' patent is practiced in the mount command, but in some kernel module. Hence, providing the mount command and their patented software is mere aggregation, and wouldn't provide a GPL license to the rest of the software even without some of the other arguments posted by others. Twin Peaks could also likely write their own mount command which does not infringe on Red Hat's copyright, which would eliminate the infringement, and make an injunction moot. Then, one gets into what the maximum statutory damages would be for distributing mount. It's unlikely to be per per item, but for the work being infringed. Twin Peaks would have to spend some effort to rewrite the mount command, but that would cost less than the amount of damages they expect to collect. Alternatively, Twin Peaks' sales might be poor enough that they could stop selling their product without a significant change in revenue, as SCO did. Finally, if Twin Peaks downloads another copy of util-linux and releases their mount sources with future distribution of their product, US courts have never weighed in on if that is a new license, stopping the infringement.
The more interesting question is whether there are any GPL-only interfaces used by Twin Peaks's filesystem code, whether wrapped or not. Since Twin Peaks doesn't release this code, it's harder to determine this. Plus, one then gets into who holds the copyright on that code, and if it's not Red Hat, they can't use it in their counter claims. I'm betting that someone would be willing to transfer their copyright to Red Hat in that case. This would make it infinitely more difficult for Twin Peaks to avoid an injunction, and make a settlement more likely. SFLC is on this aspect of the case, and hopefully others.
Twin Peaks v. Red Hat
Patent licence
Patent licence
Patent licence
Patent licence
The comments above give the impression (at least to me) that folks are thinking of Twin Peaks' offerings as a monolith. In fact, RH is suing over the mount program. mount being in violation of the GPL says nothing about the rest of their suite (FUD to the contrary notwithstanding). (Of course, if you're selling a filesystem, not having mount could put a crimp in your operations. :-)
Keep in mind the different programs involved
Keep in mind the different programs involved
- written, by signature
- orally
- by taking an action that implies your intention to enter the contract (this is called "stillschweigend", which literally means tacitly, in German)
Patent licence
Patent licence
Patent licence
Patent licence
Patent licence
Patent licence
Patent licence
Patent licence