Advertisement
Front, Kernel, Security, Distributions, Development. See your byline here on LWN.net.
| |||||||||||||
|
| |||||||||||||
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch)Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch)Posted Apr 6, 2007 19:00 UTC (Fri) by khim (subscriber, #9252)In reply to: Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) by NAR Parent article: Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch)
Not really. ATI and NVidia binary drivers are legal because binary part is clearly not derived work of Linux kernel (it contains code ported from Windows) and glue code is free. So it's clear that AMD and NVidia are not violating anything. It's not clear if what the Ubuntu is doing is allowed by GPL - but that other question. Now with PeerFS it's not really known if this FS is totally separable from Linux (like AFS) or is it deeply ingrated (like Ext3) - and that's quite a question because it determines if it's GPL violation or not. This question is not discussed in article at all...
(Log in to post comments)
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 6, 2007 23:26 UTC (Fri) by drag (subscriber, #31333) [Link] I wouldn't say that it's clear that they are not violating the GPL. They certainly can be.
I think it's much more accurate to say that it's reasonable to assume that they are not violating it.
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 7, 2007 17:55 UTC (Sat) by mikov (subscriber, #33179) [Link] Are you talking about NVIDIA and ATI or PeerFS ?
These are the things that I am curious about in regards to PeerFS:
- Are there any other closed-source kernel file systems for Linux ?
- Does PeerFS taint the kernel ? Even if we assume that it doesn't technically violate the GPL, a tainted kernel is not something I'd like to run.
This is the second instance in the last couple of months that I have came across products utilizing closed source Linux kernel code. The other instance is Virtuozzo which apparently has proprietary kernel modules providing functionality beyond OpenVZ.
Is this a growing trend ?
In both these cases it is assumed that the GPL is not violated, because the code in question cannot be examined. It seems to me that the logical assumption is exactly the opposite - that the GPL _is_ violated. Both companies should at least post detailed explanations on their websites of why it isn't.
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 7, 2007 19:28 UTC (Sat) by NAR (subscriber, #1313) [Link] Are there any other closed-source kernel file systems for Linux ?There was (or is?) a binary module for the mvfs used by ClearCase.
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 8, 2007 2:02 UTC (Sun) by madscientist (subscriber, #16861) [Link] MVFS existed for other UNIX systems including SunOS, Solaris, HP-UX, and AIX for years before it was ported to Linux, so although it was not GPL'd originally it's reasonable to assume it is not a derivative work for the same reasons the video drivers discussed above are not (if they are not). The glue bits were released as source code (to be honest I can't remember the license) just as with other binary modules.
Although I switched jobs recently and am no longer using ClearCase, my understanding is that IBM (which bought Rational Software, which bought PureAtria, which was the combined company of Purify and Atria, which was the original vendor providing ClearCase) released the MVFS kernel module code under the GPL as of ClearCase 6.0, released last year or so.
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 8, 2007 6:06 UTC (Sun) by drag (subscriber, #31333) [Link] You have a similar issue with OpenAFS.
AFS is ancient, as old as X Windows or Kerberos. Part of that Athena project.
IBM released code for it and it has gone through tremendous improvements in terms of performance, code quality, and stability. Especially with newer releases.
But it's not under a GPL compatable license. Nobody can do anything about it since the original copyright owners are far gone and numerous. Still open source, though, and not a derivative of the kernel.
Radiant Data launches 64-bit Linux HA replicating file system (Linux-Watch) Posted Apr 7, 2007 22:38 UTC (Sat) by drag (subscriber, #31333) [Link] I was talking about the Nvidia/ATI driver. Mostly the Nvidia driver.
They use software that was developed for a different operating system first, then adopted to be used in a Linux module. Any code that they have that is 'linux derived' is in GPL form.
So given my limited understanding about how the copyright works and how what is and what is not 'derived' is up to a judge to decide, and not the Linux devs, the GPL license, or FSF.. I'd have to guess that they are not violating the GPL.
But they could be. I am sure that if a Linux developer wanted to he could easily get a lawyer to take them to court over it. It's definately not cut and dry.
Now with this FS, I have no freaking clue. I know that I wouldn't touch a closed source FS with a ten foot pole if I could help it.
> Is this a growing trend ?
It's a cultural problem that I don't think that you'd realy be able solve with a license.
I'd bet good money that GPL violations are failing per thousand users or per hundred companies involved in Linux development. But I'd also bet that since Linux is growing in popularity, maturity, and usefullness and the overal competitive operating systems are dwindling that issues with people trying to squeeze around GPL is _increasing_ in number.
So the critical thing is to avoid having a bad reputation placed on open soruce drivers for being inadiquete.
The major examples of this is wireless drivers for Linux. Linux developers dropped the ball early on with wireless drivers by treating them as just your normal everyday ethernet device with 'wireless extensions'.
This was a mistake and lead to having drivers that had mismatched and limited functionality. Made it much more difficult to develop drivers and drivers were generally not mature or stable.
So now you have people who realy beleive that it's impossible to have a Linux operating system running a modern laptop and not use any closed source software.
They beleive that unless you use closed source software your not going to have 3d support or only have very poor wireless support if any at all.
This is not true. But what is needed is a campaign of education on the part of major Linux vendors and distributions on _why_ users should avoid closed source drivers.
With examples and everything. How people had bugs that corrupted files, but they went away with removing a closed source drivers. How performance and stability was bad and there was nothing anybody can do about.
How you may have instability and issues with open source drivers, but they can be fixed and people are allowed to find out why things went bad.
With Redhat coming out and stating why they won't support closed source drivers. That not only it is a license violation most of the time, but that users you are stuck depending on closed source drivers have a worse experiance then those that don't.
That it's for the end user's own _self_interest_ to specificly seek out vendors and hardware that has proper Linux support and support manufacturers with healthy appreciation for end users and for Linux developers.
Stuff like that.
It's like that driver development kit for Linux. It was a success, but it was ment as sort of a joke. But people used it and liked it.
Or like how Linux developers advertising no-cost driver development for vendor's hardware like it was a special deal down at the hardware store. That is something that for you and I is obvious, but for other people they didn't realise. And that helped people get together with the Linux developers very well.
People have been using closed source software for years. They have assumptions and expectations. If they had good success with using closed source drivers for Windows then why would they expect any difference from Linux? They probably figure as long as the vendor is smart and pays attention to their issues that they have nothing to fear, right?
So education is a nice way to show that 'there is a better way'.
|
|
||||||||||||