Skip to contentSkip to conversation

Linux GPL and binary module exception clause?

DISCUSSION Kendall Bennett 25 participants 59 messages Dec 3, 2003 – Dec 13, 2003 Licenses and spdx stuff

lore
Share & tools
lore
#0 lore raw

Hi All,

I have heard many people reference the fact that the although the Linux Kernel is under the GNU GPL license, that the code is licensed with an exception clause that says binary loadable modules do not have to be under the GPL. Obviously today there are vendors delivering binary modules (not supported by the kernel maintainers of course), so clearly people believe this to be true. However I was curious about the wording of this exception clause so I went looking for it, but I cannot seem to find it. I downloaded the 2.6-test1 kernel source code and looked at the COPYING file, but found nothing relating to this (just the note at the top from Linus saying user programs are not covered by the GPL). I also looked in the README file and nothing was mentioned there either, at least from what I could see from a quick read.

So does this exception clause exist or not? If not, how can the binary modules be valid for use under Linux if the source is not made available under the terms of the GNU GPL?

Lastly I noticed that the few source code modules I looked at to see if the exception clause was mentioned there, did not contain the usual GNU GPL preable section at the top of each file. IMHO all files need to have such a notice attached, or they are not under the GNU GPL (just being in a ZIP/tar achive with a COPYING file does not place a file under the GNU GPL). Given all the current legal stuff going on with SCO, I figured every file would have such a header. In fact some of the files I looked at didn't even contain a basic copyright notice!!

Regards,

---
Kendall Bennett
Chief Executive Officer
SciTech Software, Inc.
Phone: (530) 894 8400
http://www.scitechsoft.com

~ SciTech SNAP - The future of device driver technology! ~

#2 lore raw
Quoted 11 lines · [email protected] wrote:
Quoted 6 lines · Kendall Bennett

Hi All,

I have heard many people reference the fact that the although the Linux Kernel is under the GNU GPL license, that the code is licensed with an exception clause that says binary loadable modules do not have to be under the GPL.

there is no such exception, for example see:

http://www.kernelnewbies.org/kernels/rh9/SOURCES/COPYING.modules

Thanks for the link, that pretty much explains what I needed to know!

Regards,

---
Kendall Bennett
Chief Executive Officer
SciTech Software, Inc.
Phone: (530) 894 8400
http://www.scitechsoft.com

~ SciTech SNAP - The future of device driver technology! ~

#3 lore raw
In reply to Kendall Bennett
Quoted 15 lines · Kendall Bennett

Hi All,

I have heard many people reference the fact that the although the Linux Kernel is under the GNU GPL license, that the code is licensed with an exception clause that says binary loadable modules do not have to be under the GPL. Obviously today there are vendors delivering binary modules (not supported by the kernel maintainers of course), so clearly people believe this to be true. However I was curious about the wording of this exception clause so I went looking for it, but I cannot seem to find it. I downloaded the 2.6-test1 kernel source code and looked at the COPYING file, but found nothing relating to this (just the note at the top from Linus saying user programs are not covered by the GPL). I also looked in the README file and nothing was mentioned there either, at least from what I could see from a quick read.

Quoted 4 lines

So does this exception clause exist or not? If not, how can the binary modules be valid for use under Linux if the source is not made available under the terms of the GNU GPL?

I'll jump into this fray first stating that it is really great that the CEO of a company that is producing high-performance graphics cards and acceleration software is interested in finding out this information. It seems that some other companies just hack together some general-purpose source-code under GPL and then link it with a secret object file. This, of course, defeats the purpose of the GPL (which is or was to PUBLISH software in human readable form).

It is certainly time for a definitive answer.

Maybe Linus knows the answer.

Quoted 9 lines

Lastly I noticed that the few source code modules I looked at to see if the exception clause was mentioned there, did not contain the usual GNU GPL preable section at the top of each file. IMHO all files need to have such a notice attached, or they are not under the GNU GPL (just being in a ZIP/tar achive with a COPYING file does not place a file under the GNU GPL). Given all the current legal stuff going on with SCO, I figured every file would have such a header. In fact some of the files I looked at didn't even contain a basic copyright notice!!

I have been told by lawyers who do intellectual property law for a living that under US Copyright law, the INSTANT that something is written anywhere in a manner that allows it to be read back, it is protected by the writer's default copyright protection. The writer may alter that protection or even assign ownership to something or somebody else, but nobody needs to put a copyright notice anywhere in text. Now, if you intend to sue, before that suit starts, the text must be registered with the United States Copyright Office. In that case, it still doesn't need a copyright notice or the famous (c) specified by the act. It just needs to be identified by the writer, like:

File: TANGO.FOR Created 12-DEC-1988 John R. Doe

Grin... from my VAX/VMS days.

Cheers,
Dick Johnson
Penguin : Linux version 2.4.22 on an i686 machine (797.90 BogoMips).
Note 96.31% of all statistics are fiction.

#4 lore raw
Quoted 10 lines · Richard B. Johnson
Quoted 3 lines

So does this exception clause exist or not? If not, how can the binary modules be valid for use under Linux if the source is not made available under the terms of the GNU GPL?

I'll jump into this fray first stating that it is really great that the CEO of a company that is producing high-performance graphics cards and acceleration software is interested in finding out this information. It seems that some other companies just hack together some general-purpose source-code under GPL and then link it with a secret object file.

Well one of the primary reasons we are interested in researching this is because we are in the process of preparing to release the SciTech SNAP DDK under the GNU GPL license (dual licensed for proprietry developers), and want to understand the ramifications of this approach. I must say I am surprised to see that such a clause does not exist, since binary only modules seem to be pretty abundant in the community.

Then again the whole 'derived works' clause in the GPL is a huge gray area IMHO, and even just sitting there arguing with myself I can come up with pretty good arguments in both directions for why a binary only module is legal or not. Especially when the only reference to this in the GNU GPL FAQ is about plugins, and they use a really silly distinction about whether a plugin uses fork/exec or not. I mean if you use fork/exec to start the plugin, yet the plugin uses RPC and shared memory to communicate with the main process, that is the same as local procedure calls and local memory IMHO.

Quoted 9 lines
Quoted 8 lines

Lastly I noticed that the few source code modules I looked at to see if the exception clause was mentioned there, did not contain the usual GNU GPL preable section at the top of each file. IMHO all files need to have such a notice attached, or they are not under the GNU GPL (just being in a ZIP/tar achive with a COPYING file does not place a file under the GNU GPL). Given all the current legal stuff going on with SCO, I figured every file would have such a header. In fact some of the files I looked at didn't even contain a basic copyright notice!!

Quoted 4 lines

I have been told by lawyers who do intellectual property law for a living that under US Copyright law, the INSTANT that something is written anywhere in a manner that allows it to be read back, it is protected by the writer's default copyright protection.

Yes of course, but if there is no notice in the file, who is the copyright holder? Only the original author knows for sure, and for someone to *use* that file without permission from the original author would be inviting a lawsuit.

Quoted 11 lines

The writer may alter that protection or even assign ownership to something or somebody else, but nobody needs to put a copyright notice anywhere in text. Now, if you intend to sue, before that suit starts, the text must be registered with the United States Copyright Office. In that case, it still doesn't need a copyright notice or the famous (c) specified by the act. It just needs to be identified by the writer, like:

File: TANGO.FOR Created 12-DEC-1988 John R. Doe

Grin... from my VAX/VMS days.

Right, but there are files in the source code with *nothing* at the top. No copyright header, not notice of who wrote the file, nothing.

In my opinion, unless a specific source file has a copyright notice attached to it and specifically says 'you are licensed to use this file under the terms of the GNU GPL - see the COPYING file' etc, then that file is *not* under the GNU GPL. What that means is that I would have zero rights to use that file without direct permission from the author, and realistically should not use it.

If there are lots of files like that in the Linux kernel, then that means that none of us really have the right to use Linux on our machines! Yikes. Considering that all GNU projects contain this type of header, I have always just assumed the Linux kernel source code would too. Especially now with this SCO debacle going on.

Regards,

---
Kendall Bennett
Chief Executive Officer
SciTech Software, Inc.
Phone: (530) 894 8400
http://www.scitechsoft.com

~ SciTech SNAP - The future of device driver technology! ~

#5 lore raw
In reply to Richard B. Johnson

In article <Pine.LNX.4.53.0312031648390.3725@chaos>, Richard B. Johnson <[email protected]> wrote:

| On Wed, 3 Dec 2003, Kendall Bennett wrote:
| 
| > Hi All,
| >
| > I have heard many people reference the fact that the although the Linux
| > Kernel is under the GNU GPL license, that the code is licensed with an
| > exception clause that says binary loadable modules do not have to be
| > under the GPL. Obviously today there are vendors delivering binary
| > modules (not supported by the kernel maintainers of course), so clearly
| > people believe this to be true. However I was curious about the wording
| > of this exception clause so I went looking for it, but I cannot seem to
| > find it. I downloaded the 2.6-test1 kernel source code and looked at the
| > COPYING file, but found nothing relating to this (just the note at the
| > top from Linus saying user programs are not covered by the GPL). I also
| > looked in the README file and nothing was mentioned there either, at
| > least from what I could see from a quick read.
| >
| > So does this exception clause exist or not? If not, how can the binary
| > modules be valid for use under Linux if the source is not made available
| > under the terms of the GNU GPL?
| >
| 
| I'll jump into this fray first stating that it is really great
| that the CEO of a company that is producing high-performance graphics
| cards and acceleration software is interested in finding out this
| information. 

Really? I guess I'm just suspicious, but when someone who might have an interest in only providing a binary driver asks about the legality of doing that, "great" is not my first thought.

| information. It seems that some other companies just hack together some
| general-purpose source-code under GPL and then link it with a secret
| object file. This, of course, defeats the purpose of the GPL (which is
| or was to PUBLISH software in human readable form).

Yes, I am a devout fundamentalist paranoid, but I've based my life on the assumptions that I should treat others fairly and expect them to screw me if they could, and both have served me well.

I do not mean to cast aspersions on the original poster, about whom I know nothing. There are many companies who have provided full source drivers, and I have rewarded them with my business. I have chosen less performance video over binary module hardware, and would be very happy if there were some guilt-free hardwaree to use. I'm just starting to do video processing, I'd be *really* happy, ecstatic even.

Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#6 lore raw
In reply to Kendall Bennett
Quoted 5 lines · Kendall Bennett

I have heard many people reference the fact that the although the Linux Kernel is under the GNU GPL license, that the code is licensed with an exception clause that says binary loadable modules do not have to be under the GPL.

Nope. No such exception exists.

There's a clarification that user-space programs that use the standard system call interfaces aren't considered derived works, but even that isn't an "exception" - it's just a statement of a border of what is clearly considered a "derived work". User programs are _clearly_ not derived works of the kernel, and as such whatever the kernel license is just doesn't matter.

And in fact, when it comes to modules, the GPL issue is exactly the same. The kernel _is_ GPL. No ifs, buts and maybe's about it. As a result, anything that is a derived work has to be GPL'd. It's that simple.

Now, the "derived work" issue in copyright law is the only thing that leads to any gray areas. There are areas that are not gray at all: user space is clearly not a derived work, while kernel patches clearly _are_ derived works.

But one gray area in particular is something like a driver that was originally written for another operating system (ie clearly not a derived work of Linux in origin). At exactly what point does it become a derived work of the kernel (and thus fall under the GPL)?

THAT is a gray area, and _that_ is the area where I personally believe that some modules may be considered to not be derived works simply because they weren't designed for Linux and don't depend on any special Linux behaviour.

Basically:
- anything that was written with Linux in mind (whether it then _also_ works on other operating systems or not) is clearly partially a derived work.
- anything that has knowledge of and plays with fundamental internal Linux behaviour is clearly a derived work. If you need to muck around with core code, you're derived, no question about it.

Historically, there's been things like the original Andrew filesystem
module: a standard filesystem that really wasn't written for Linux in the first place, and just implements a UNIX filesystem. Is that derived just because it got ported to Linux that had a reasonably similar VFS interface to what other UNIXes did? Personally, I didn't feel that I could make that judgment call. Maybe it was, maybe it wasn't, but it clearly is a gray area.

Personally, I think that case wasn't a derived work, and I was willing to tell the AFS guys so.

Does that mean that any kernel module is automatically not a derived work? HELL NO! It has nothing to do with modules per se, except that non-modules clearly are derived works (if they are so central to the kenrel that you can't load them as a module, they are clearly derived works just by virtue of being very intimate - and because the GPL expressly mentions linking).

So being a module is not a sign of not being a derived work. It's just one sign that _maybe_ it might have other arguments for why it isn't derived.

Linus

#7 lore raw
Quoted 4 lines · Linus Torvalds

So being a module is not a sign of not being a derived work. It's just one sign that _maybe_ it might have other arguments for why it isn't derived.

Side note: historically, the Linux kernel module interfaces were really quite weak, and only exported a few tens of entry-points, and really mostly effectively only allowed character and block device drivers with standard interfaces, and loadable filesystems.

So historically, the fact that you could load a module using nothing but these standard interfaces tended to be a much stronger argument for not being very tightly coupled with the kernel.

That has changed, and the kernel module interfaces we have today are MUCH more extensive than they were back in '95 or so. These days modules are used for pretty much everything, including stuff that is very much "internal kernel" stuff and as a result the kind of historic "implied barrier" part of modules really has weakened, and as a result there is not avery strong argument for being an independent work from just the fact that you're a module.

Similarly, historically there was a much stronger argument for things like AFS and some of the binary drivers (long forgotten now) for having been developed totally independently of Linux: they literally were developed before Linux even existed, by people who had zero knowledge of Linux. That tends to strengthen the argument that they clearly aren't derived.

In contrast, these days it would be hard to argue that a new driver or filesystem was developed without any thought of Linux. I think the NVidia people can probably reasonably honestly say that the code they ported had _no_ Linux origin. But quite frankly, I'd be less inclined to believe that for some other projects out there..

Linus

#8 lore raw

Hello Linus,

Quoted 11 lines · Linus Torvalds

Similarly, historically there was a much stronger argument for things like AFS and some of the binary drivers (long forgotten now) for having been developed totally independently of Linux: they literally were developed before Linux even existed, by people who had zero knowledge of Linux. That tends to strengthen the argument that they clearly aren't derived.

In contrast, these days it would be hard to argue that a new driver or filesystem was developed without any thought of Linux. I think the NVidia people can probably reasonably honestly say that the code they ported had _no_ Linux origin. But quite frankly, I'd be less inclined to believe that for some other projects out there..

Since the last time this was mentioned, I have been thinking that this argument can really be read as an invitation to do just what's being
described: first implement a driver/module in a non-Linux OS (this may even imply requiring that whoever works on the driver/module have NO Linux experience whatsoever; yes there will always be candidates for this) and then have this driver/module ported to Linux by Linux-aware developers.

Sure, one could argue about "intent", but that's going to be really difficult, especially if the right-hand doesn't know what the left-hand is doing. IOW, can't this position be abused as much as, if not more than, publishing an "approved" set of characteristics for non-GPL modules?

Just thinking aloud,

Karim

Signature · 3 lines
Author, Speaker, Developer, Consultant
Pushing Embedded and Real-Time Linux Systems Beyond the Limits
http://www.opersys.com || [email protected] || 514-812-4145
#9 lore raw
Quoted 6 lines · Karim Yaghmour

Since the last time this was mentioned, I have been thinking that this argument can really be read as an invitation to do just what's being
described: first implement a driver/module in a non-Linux OS (this may even imply requiring that whoever works on the driver/module have NO Linux experience whatsoever; yes there will always be candidates for this) and then have this driver/module ported to Linux by Linux-aware developers.

So you have a loadable module made of two sections; a GPL'd wrapper layer clearly based on the kernel, and your original driver. The latter is clearly an identifiable section of that compound work which is _not_ derived from Linux and which can reasonably be considered an independent and separate work in itself.

The GPL and its terms do not apply to that section when you distribute it as a separate work.

But when you distribute the same section as part of a _whole_ which is a work based on the Linux kernel, the distribution of the whole must be on the terms of the licence, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.

For the precise wording which I've paraphrased above, see §2 of the GPL.

Note that 'is this a derived work' is only part of the question you should be asking yourself. The GPL makes requirements about the licensing even of works which are _not_ purely derived.

Some claim that copyright law does not allow the GPL to do such a thing. That is incorrect. I can write a work and license it to you under _any_ terms I see fit. I can, for example, license it to you _only_ on condition that you agree to release _all_ your future copyrightable work, including works of fiction and other completely unrelated things, under terms I decree.

You either do that or you don't have permission to use my work. Whether your own work is derived or not is completely irrelevant; if you don't agree to the terms of _my_ licence, you don't get to use _my_ code.

Signature · 1 line
dwmw2
#10 lore raw
Quoted 3 lines · David Woodhouse

You either do that or you don't have permission to use my work. Whether your own work is derived or not is completely irrelevant; if you don't agree to the terms of _my_ licence, you don't get to use _my_ code.

If and only if, as has been pointed out here many times, your license is enforceable.

Signature · 2 lines
---
Larry McVoy              lm at bitmover.com          http://www.bitmover.com/lm
#11 lore raw
Quoted 7 lines · Larry McVoy
Quoted 3 lines · David Woodhouse

You either do that or you don't have permission to use my work. Whether your own work is derived or not is completely irrelevant; if you don't agree to the terms of _my_ licence, you don't get to use _my_ code.

If and only if, as has been pointed out here many times, your license is enforceable.

This is true. The above applies only if the licence is enforceable, as Larry says.

As has also been pointed out many times; if my licence is not enforceable then it is void, and you have no right to use my work whether you obey its conditions or not.

Signature · 1 line
dwmw2
#12 lore raw
Quoted 9 lines · David Woodhouse
Quoted 7 lines · Larry McVoy
Quoted 3 lines · David Woodhouse

You either do that or you don't have permission to use my work. Whether your own work is derived or not is completely irrelevant; if you don't agree to the terms of _my_ licence, you don't get to use _my_ code.

If and only if, as has been pointed out here many times, your license is enforceable.

Quoted 6 lines

This is true. The above applies only if the licence is enforceable, as Larry says.

As has also been pointed out many times; if my licence is not enforceable then it is void, and you have no right to use my work whether you obey its conditions or not.

Interesting... So if it turns out that the use license of <insert any (commercial) software product here> is not enforceable, no one is allowed to use that product anymore, even if he paid for it?

Gr{oetje,eeting}s,

Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But when I'm talking to journalists I just say "programmer" or something like that.
-- Linus Torvalds

#13 lore raw
In reply to David Woodhouse
Quoted 5 lines · David Woodhouse

So you have a loadable module made of two sections; a GPL'd wrapper layer clearly based on the kernel, and your original driver. The latter is clearly an identifiable section of that compound work which is _not_ derived from Linux and which can reasonably be considered an independent and separate work in itself.

I didn't exactly specify how the interfacing would be done because that's besides the point I'm trying to make (in fact, it's the later part of my email which was most important). But here's two other ways to do it just for the sake of discussion:
a) Hard-wired assembly in the driver that calls on the appropriate address with the proper structure offsets etc. No headers used here.
b) User-space interrupt callbacks. Start app -> mlockall -> open GPL driver -> use ioctl to pass callback address -> open /dev/mem -> ... I've skipped a few things, but the essentials are there. Basically, you get the interrupts in user-space and can access whatever you want through /dev/mem.

Sure the above isn't as powerful as a properly coded driver for Linux, but it should work for a few things.

Quoted 2 lines

The GPL and its terms do not apply to that section when you distribute it as a separate work.

Right, but my argument has little to do with the GPL. In fact, as I said before (http://www.embeddedtux.org/pipermail/etux/2003-October/000415.html),
I don't personally think the GPL has all the answers to this issue. Not just that, but I don't really see myself debating with any kernel developer what the law says he has the right to do or not do with his code. It just seems to me that the copyright holder has the upper hand here. What I'm really looking for is to understand how to apply the GPL to binary-only modules given the copyright holders' interpretation. So far, I have heard the following (this is a summary, so please correct me if you think I've miss-characterized your take on this):

Linus Torvalds: modules API not a GPL barrier, must prove your work is not a derived work - preexisted Linux. Ex.: NVidia is clearly not a derived work. Alan Cox: Unclear if Linus can dictate terms for binary-only modules since he's not the only copyright owner. Talk to your lawyer.
Russell King: Linus isn't the only copyright owner and hence can't change the terms outright.
Theodore Ts'o: Modules API constitutes license barrier.
David Woodhouse: There's no such barrier and applications tightly packaged with the kernel (i.e. embedded systems) _may_ be subject to derived work clauses of the GPL.
etc.

Frankly, I don't know what to make of all this. I wish I could collect the input of all kernel developers to come up with a table of X owns N% of kernel copyright and his statement about binary-only modules is best characterized as A, B, C, or D, or ... where each of these is an already stated position about such modules. We could then come up with a weighed percentage of validity for each of A, B, C, ... The exercise may not have any legal weight, especially since as Linus stated a judge may just give him more credence than any other kernel developer, but it would at least tell kernel developers where they all stand on this issue.

As it stands now, however, it seems to me that any kernel developer attempting to enforce the GPL across the modules API would have quite a few problems. Mainly because:
1) There is no clear consensus among the copyright holders as to how the GPL is to be interpreted across the modules API.
2) Established technical practice has been that hardware manufacturers do indeed ship drivers with different licenses than those of the host OS.

I had mentioned #2 elsewhere before
(http://www.embeddedtux.org/pipermail/etux/2003-October/000415.html) and it has been discussed in this thread by other folks in slightly different words. However, it may be that #1 could end up causing the most damage in the case where a kernel developer, or a few, try to prosecute a real case of binary-only module infringement on the GPL.

In sum, I agree with Jonathan Corbet's assessment that it's about time that kernel developers agree where the axe falls. Not just for outsiders, but also for themselves.

Cheers,

Karim

Signature · 3 lines
Author, Speaker, Developer, Consultant
Pushing Embedded and Real-Time Linux Systems Beyond the Limits
http://www.opersys.com || [email protected] || 514-812-4145
#14 lore raw

In article <[email protected]>,
Karim Yaghmour <[email protected]> wrote:

| 
| David Woodhouse wrote:
| > So you have a loadable module made of two sections; a GPL'd wrapper
| > layer clearly based on the kernel, and your original driver. The latter
| > is clearly an identifiable section of that compound work which is _not_
| > derived from Linux and which can reasonably be considered an independent
| > and separate work in itself.
| 
| I didn't exactly specify how the interfacing would be done because that's
| besides the point I'm trying to make (in fact, it's the later part of my
| email which was most important). But here's two other ways to do it just
| for the sake of discussion:
| a) Hard-wired assembly in the driver that calls on the appropriate address
| with the proper structure offsets etc. No headers used here.

Well, the addresses and offset specs came from *somewhere*, and I would love to hear someone argue that they "just seemed like good values," or that reading the header file and then using absolute numbers isn't derivative.

| b) User-space interrupt callbacks. Start app -> mlockall -> open GPL
| driver -> use ioctl to pass callback address -> open /dev/mem -> ...
| I've skipped a few things, but the essentials are there. Basically, you
| get the interrupts in user-space and can access whatever you want through
| /dev/mem.
| 
| Sure the above isn't as powerful as a properly coded driver for Linux,
| but it should work for a few things.

And people who would do this so they can violate the spirit of the GPL without violating the language would sue if someone reverse engineered their secret code...

| 
| > The GPL and its terms do not apply to that section when you distribute
| > it as a separate work.
| 
| Right, but my argument has little to do with the GPL. In fact, as I
| said before (http://www.embeddedtux.org/pipermail/etux/2003-October/000415.html),
| I don't personally think the GPL has all the answers to this issue. Not
| just that, but I don't really see myself debating with any kernel developer
| what the law says he has the right to do or not do with his code. It just
| seems to me that the copyright holder has the upper hand here. What I'm
| really looking for is to understand how to apply the GPL to binary-only
| modules given the copyright holders' interpretation. So far, I have heard
| the following (this is a summary, so please correct me if you think I've
| miss-characterized your take on this):
| 
| Linus Torvalds: modules API not a GPL barrier, must prove your work is not a
| derived work - preexisted Linux. Ex.: NVidia is clearly not a derived work.
| Alan Cox: Unclear if Linus can dictate terms for binary-only modules since
| he's not the only copyright owner. Talk to your lawyer.
| Russell King: Linus isn't the only copyright owner and hence can't change
| the terms outright.
| Theodore Ts'o: Modules API constitutes license barrier.
| David Woodhouse: There's no such barrier and applications tightly packaged
| with the kernel (i.e. embedded systems) _may_ be subject to derived work
| clauses of the GPL.
| etc.
| 
| Frankly, I don't know what to make of all this. I wish I could collect the
| input of all kernel developers to come up with a table of X owns N% of
| kernel copyright and his statement about binary-only modules is best
| characterized as A, B, C, or D, or ... where each of these is an already
| stated position about such modules. We could then come up with a weighed
| percentage of validity for each of A, B, C, ... The exercise may not have
| any legal weight, especially since as Linus stated a judge may just give
| him more credence than any other kernel developer, but it would at least
| tell kernel developers where they all stand on this issue.
| 
| As it stands now, however, it seems to me that any kernel developer
| attempting to enforce the GPL across the modules API would have quite a
| few problems. Mainly because:
| 1) There is no clear consensus among the copyright holders as to how
| the GPL is to be interpreted across the modules API.
| 2) Established technical practice has been that hardware manufacturers
| do indeed ship drivers with different licenses than those of the host
| OS.
| 
| I had mentioned #2 elsewhere before
| (http://www.embeddedtux.org/pipermail/etux/2003-October/000415.html) and
| it has been discussed in this thread by other folks in slightly different
| words. However, it may be that #1 could end up causing the most damage in
| the case where a kernel developer, or a few, try to prosecute a real case
| of binary-only module infringement on the GPL.
| 
| In sum, I agree with Jonathan Corbet's assessment that it's about time
| that kernel developers agree where the axe falls. Not just for outsiders,
| but also for themselves.

I don't think the opinion of the copyright holders counts a bit, the text of the license and the opinion of a court count. And based on asking a total of one lawyer, any Linux user has standing to sue because (if) the copyright infringement interferes with the user's right to use the software. I report that opinion without defending it, the lawyer was unwilling to be named.

| Cheers,
| 
| Karim
| -- 
| Author, Speaker, Developer, Consultant
| Pushing Embedded and Real-Time Linux Systems Beyond the Limits
| http://www.opersys.com || [email protected] || 514-812-4145
Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#15 lore raw
Quoted 9 lines · bill davidsen

In article <[email protected]>,
Karim Yaghmour <[email protected]> wrote:

| I didn't exactly specify how the interfacing would be done because that's
| besides the point I'm trying to make (in fact, it's the later part of my
| email which was most important). But here's two other ways to do it just
| for the sake of discussion:
| a) Hard-wired assembly in the driver that calls on the appropriate address
| with the proper structure offsets etc. No headers used here.
Quoted 4 lines

Well, the addresses and offset specs came from *somewhere*, and I would love to hear someone argue that they "just seemed like good values," or that reading the header file and then using absolute numbers isn't derivative.

INAL. Observable facts (such as absolute numbers) aren't
derivative (in the U.S.) because there's no "creativity"***. See the famous court decision (... web search ...) "Feist Publications v. Rural Telephone Serv. Co.", for example. Of course, the DCMA (or other fell beasts) may have superseded that legal doctrine.

Craig Milo Rogers

*** This raises the possibility that structured numbers might be
copyrightable.

#16 lore raw
Quoted 10 lines · Craig Milo Rogers
Quoted 9 lines · bill davidsen

In article <[email protected]>,
Karim Yaghmour <[email protected]> wrote:

| I didn't exactly specify how the interfacing would be done because that's
| besides the point I'm trying to make (in fact, it's the later part of my
| email which was most important). But here's two other ways to do it just
| for the sake of discussion:
| a) Hard-wired assembly in the driver that calls on the appropriate address
| with the proper structure offsets etc. No headers used here.
Quoted 4 lines
Quoted 4 lines

Well, the addresses and offset specs came from *somewhere*, and I would love to hear someone argue that they "just seemed like good values," or that reading the header file and then using absolute numbers isn't derivative.

Using the absolute numbers and/or structures (i.e. the interface) needed to progam a device seems like a fine example of fair use, even when the header file containing them is copyrighted.

Fair use trumps derivation.

Quoted 5 lines

INAL. Observable facts (such as absolute numbers) aren't
derivative (in the U.S.) because there's no "creativity"***. See the famous court decision (... web search ...) "Feist Publications v. Rural Telephone Serv. Co.", for example. Of course, the DCMA (or other fell beasts) may have superseded that legal doctrine.

I guess a hardware manufacturer who feels their programming information is worth keeping secret may well argue that they _chose_ the numbers that needed to be fed to their device, and thus they're creative.

But that's silly, and we're in deep shit if ever that argument is taken seriously because _lots_ of Linux and BSD open source drivers were developed by studying the behaviour or code of another driver.

-- Jamie

#17 lore raw
Quoted 20 lines · Jamie Lokier
Quoted 6 lines · Craig Milo Rogers
Quoted 4 lines · bill davidsen

Well, the addresses and offset specs came from *somewhere*, and I would love to hear someone argue that they "just seemed like good values," or that reading the header file and then using absolute numbers isn't derivative.

Using the absolute numbers and/or structures (i.e. the interface) needed to progam a device seems like a fine example of fair use, even when the header file containing them is copyrighted.

Fair use trumps derivation.

Quoted 5 lines

INAL. Observable facts (such as absolute numbers) aren't
derivative (in the U.S.) because there's no "creativity"***. See the famous court decision (... web search ...) "Feist Publications v. Rural Telephone Serv. Co.", for example. Of course, the DCMA (or other fell beasts) may have superseded that legal doctrine.

Quoted 7 lines

I guess a hardware manufacturer who feels their programming information is worth keeping secret may well argue that they _chose_ the numbers that needed to be fed to their device, and thus they're creative.

But that's silly, and we're in deep shit if ever that argument is taken seriously because _lots_ of Linux and BSD open source drivers were developed by studying the behaviour or code of another driver.

After thinking about this, and why this would be allowed if disassembly or reverse engineering Windows drivers is (in some cases) prohibited, I conclude that the license on the drivers prohibits such reverse engineering.

Taking that one step farther, could the license on the moduling ABI be changed, not requiring consent of everyone with code in the kernel just the designer of the ABI, to prohibit using it for binary-only modules? Did Rusty redo this when he attacked the module code? Is there a lawyer in the house?

Signature · 3 lines
bill davidsen <[email protected]>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.
#18 lore raw
In reply to David Woodhouse

David,

How about this ...

If you impose EXPORT_SYMBOL_GPL or impose restrictions on binary only drivers you are forbidden to use my work?

Last check 96% of all linux boxes have IDE/ATAPI/SATA in them. Oh yeah that ATAPI stuff is shared with others, but the setup code for the majority of the hba's I own copyright.

So the sword cuts two ways.

Can we stop with the stupidity?

Andre Hedrick
LAD Storage Consulting Group

Quoted 50 lines · David Woodhouse
Quoted 6 lines · Karim Yaghmour

Since the last time this was mentioned, I have been thinking that this argument can really be read as an invitation to do just what's being
described: first implement a driver/module in a non-Linux OS (this may even imply requiring that whoever works on the driver/module have NO Linux experience whatsoever; yes there will always be candidates for this) and then have this driver/module ported to Linux by Linux-aware developers.

So you have a loadable module made of two sections; a GPL'd wrapper layer clearly based on the kernel, and your original driver. The latter is clearly an identifiable section of that compound work which is _not_ derived from Linux and which can reasonably be considered an independent and separate work in itself.

The GPL and its terms do not apply to that section when you distribute it as a separate work.

But when you distribute the same section as part of a _whole_ which is a work based on the Linux kernel, the distribution of the whole must be on the terms of the licence, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.

For the precise wording which I've paraphrased above, see §2 of the GPL.

Note that 'is this a derived work' is only part of the question you should be asking yourself. The GPL makes requirements about the licensing even of works which are _not_ purely derived.

Some claim that copyright law does not allow the GPL to do such a thing. That is incorrect. I can write a work and license it to you under _any_ terms I see fit. I can, for example, license it to you _only_ on condition that you agree to release _all_ your future copyrightable work, including works of fiction and other completely unrelated things, under terms I decree.

You either do that or you don't have permission to use my work. Whether your own work is derived or not is completely irrelevant; if you don't agree to the terms of _my_ licence, you don't get to use _my_ code.

Signature · 8 lines
dwmw2


-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
#19 lore raw
Quoted 1 line · Andre Hedrick

Can we stop with the stupidity?

I'm not sure which part you think is stupid. Do you:

1. Disagree with my paraphrasing of the GPL, in the first two paragraphs?

2. Disagree with the obviously correct statement that the GPL makes requirements about the licensing even of non-derived works?

3. Disagree with my example, given in the final two paragraphs, which makes it clear that a copyright licence _may_ make restrictions on the licensing of even non-derived works. Note that although my example is a licence where the licensee must release _all_ future work under the same licence, I'm not claiming that the GPL does this; I only claim that such a licence is _possible_.

Signature · 1 line
dwmw2
#20 lore raw
Quoted 7 lines · David Woodhouse

3. Disagree with my example, given in the final two paragraphs, which makes it clear that a copyright licence _may_ make restrictions on the licensing of even non-derived works. Note that although my example is a licence where the licensee must release _all_ future work under the same licence, I'm not claiming that the GPL does this; I only claim that such a licence is _possible_.

Unless I need more coffee (which is certainly possible, it's early), yeah, I disagree with this. A contract could do this but a copyright based license doesn't seem like it can.

Signature · 2 lines
---
Larry McVoy              lm at bitmover.com          http://www.bitmover.com/lm
#21 lore raw
Quoted 3 lines · Larry McVoy

Unless I need more coffee (which is certainly possible, it's early), yeah, I disagree with this. A contract could do this but a copyright based license doesn't seem like it can.

Why so? I can license my work under whatever terms I please.

I certainly can't force you to _accept_ the terms of my licence -- you always have the option to decline -- but in that case you may not use my work.

Signature · 1 line
dwmw2
#22 lore raw
Quoted 10 lines · David Woodhouse
Quoted 3 lines · Larry McVoy

Unless I need more coffee (which is certainly possible, it's early), yeah, I disagree with this. A contract could do this but a copyright based license doesn't seem like it can.

Why so? I can license my work under whatever terms I please.

I certainly can't force you to _accept_ the terms of my licence -- you always have the option to decline -- but in that case you may not use my work.

You may license *your* work under whatever terms you want. Those terms can't extend to things that aren't your work in a copyright license. You need a contract to do that and even then there are limits to what you can do.

Signature · 2 lines
---
Larry McVoy              lm at bitmover.com          http://www.bitmover.com/lm
#23 lore raw
Quoted 15 lines · Larry McVoy
Quoted 11 lines · David Woodhouse
Quoted 3 lines · Larry McVoy

Unless I need more coffee (which is certainly possible, it's early), yeah, I disagree with this. A contract could do this but a copyright based license doesn't seem like it can.

Why so? I can license my work under whatever terms I please.

I certainly can't force you to _accept_ the terms of my licence -- you always have the option to decline -- but in that case you may not use my work.

Quoted 4 lines

You may license *your* work under whatever terms you want. Those terms can't extend to things that aren't your work in a copyright license. You need a contract to do that and even then there are limits to what you can do.

At the horrible risk of getting involved in something here, I would like to point out that the process of creating a "derived work" would be a use of the original. Therefore, restrictions placed on the original about usage would not apply directly to the derived work, but they would apply to the process of creating the derived work.

Aron

Signature · 7 lines
ssh [email protected] cat /dev/brain | grep ^work:

Aron Rubin                       Member, Engineering Staff
Lockheed Martin                  E-Mail: [email protected]
Advanced Technology Laboratories Phone:  856.792.9865
3 Executive Campus               Fax:    856.792.9930
Cherry Hill, NJ USA 08002        Web:    http://www.atl.lmco.com
#24 lore raw
In reply to Larry McVoy
Quoted 2 lines · Larry McVoy

You may license *your* work under whatever terms you want. Those terms can't extend to things that aren't your work in a copyright license.

Please elaborate. We're talking about a situation in which I grant a licence for you to copy and use my work, in exchange for something I desire from you. In the absence of a contract, this is limited to me effectively saying "You may copy and use my work if and only if you do XXX".

If you do not do 'XXX' then you may not use my work.

Are you saying that there are things which I may not ask for? Could you elaborate?

Certainly criminal law prohibits me from actually inciting you to murder by really asking for the sacrifice of your first-born; I'll grant you that one.

But what's wrong with postcardware, where I ask you to send me a postcard in exchange for permission to use my work?

What's wrong with shareware, where you have permission to copy my work and use it for evaluation purposes, but you're asked to pay me if you continue to use it?

What's wrong with a licence which requires you to bathe daily in creosote in order to receive my permission?

What's wrong with a licence on my software which requires you to relinquish copyright on any musical score you ever write?

What's wrong with a licence which requires you to release any unrelated software you ever write under the same licence?

It may be increasingly unlikely that you'll _accept_ some of the above licences, but there's no fundamental problem with them.

Quoted 2 lines

You need a contract to do that and even then there are limits to what you can do.

I need a contract to force you to do anything; this is true.

Without a contract, all I can do if you don't abide by my conditions is observe that you don't have my permission to copy my work, and hence that if you do so you are committing a criminal offence.

Signature · 1 line
dwmw2
#25 lore raw
Quoted 10 lines · David Woodhouse
Quoted 2 lines · Larry McVoy

You may license *your* work under whatever terms you want. Those terms can't extend to things that aren't your work in a copyright license.

Please elaborate. We're talking about a situation in which I grant a licence for you to copy and use my work, in exchange for something I desire from you. In the absence of a contract, this is limited to me effectively saying "You may copy and use my work if and only if you do XXX".

Quoted 4 lines

If you do not do 'XXX' then you may not use my work.

Are you saying that there are things which I may not ask for? Could you elaborate?

I think at this point you need to talk to a lawyer. I can guess where this conversation is going and I'm not going to be effective.

Quoted 5 lines

I need a contract to force you to do anything; this is true.

Without a contract, all I can do if you don't abide by my conditions is observe that you don't have my permission to copy my work, and hence that if you do so you are committing a criminal offence.

Huh? I don't think that copyright violation makes you a felon.

Signature · 2 lines
---
Larry McVoy              lm at bitmover.com          http://www.bitmover.com/lm
#26 lore raw
Quoted 5 lines · Larry McVoy
Quoted 2 lines

Are you saying that there are things which I may not ask for? Could you elaborate?

I think at this point you need to talk to a lawyer. I can guess where this conversation is going and I'm not going to be effective.

I was asking you to elaborate on what you were talking about in your previous mail -- in particular why _you_ think I can't make arbitrary conditions when granting a licence to use my work.

A lawyer could not answer that question for me.

You seemed to be saying that I may not place arbitrary conditions on my permission for you to use my work.

That is not the case. I _may_ place arbitrary conditions on my licence.

So if I say you may only cross my land when walking on your hands, or that you may only copy my software if you release all your own unrelated software under the GPL, that's fine.

You of course retain the right to _not_ cross my land, and to _not_ use my software.

Quoted 1 line

Huh? I don't think that copyright violation makes you a felon.

It's definitely a criminal offence in the UK:
http://www.hmso.gov.uk/acts/acts1988/Ukpga_19880048_en_7.htm#mdiv107

I can't bring myself to believe it's a criminal offence here but not in a country where the legislature is owned by Disney Corp., but I could be wrong.

Signature · 1 line
dwmw2
#27 lore raw
In reply to David Woodhouse

David,

For the simple folks like myself, please show me and everyone else where the "contract" terms are specifically? I am having a hard time seeing them, and they exist and real, it should be pointed out clearly.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

Quoted 56 lines · David Woodhouse
Quoted 2 lines · Larry McVoy

You may license *your* work under whatever terms you want. Those terms can't extend to things that aren't your work in a copyright license.

Please elaborate. We're talking about a situation in which I grant a licence for you to copy and use my work, in exchange for something I desire from you. In the absence of a contract, this is limited to me effectively saying "You may copy and use my work if and only if you do XXX".

If you do not do 'XXX' then you may not use my work.

Are you saying that there are things which I may not ask for? Could you elaborate?

Certainly criminal law prohibits me from actually inciting you to murder by really asking for the sacrifice of your first-born; I'll grant you that one.

But what's wrong with postcardware, where I ask you to send me a postcard in exchange for permission to use my work?

What's wrong with shareware, where you have permission to copy my work and use it for evaluation purposes, but you're asked to pay me if you continue to use it?

What's wrong with a licence which requires you to bathe daily in creosote in order to receive my permission?

What's wrong with a licence on my software which requires you to relinquish copyright on any musical score you ever write?

What's wrong with a licence which requires you to release any unrelated software you ever write under the same licence?

It may be increasingly unlikely that you'll _accept_ some of the above licences, but there's no fundamental problem with them.

Quoted 2 lines

You need a contract to do that and even then there are limits to what you can do.

I need a contract to force you to do anything; this is true.

Without a contract, all I can do if you don't abide by my conditions is observe that you don't have my permission to copy my work, and hence that if you do so you are committing a criminal offence.

Signature · 7 lines
dwmw2

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to [email protected]
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/
#28 lore raw
In reply to Linus Torvalds
Quoted 9 lines · Linus Torvalds

Side note: historically, the Linux kernel module interfaces were really quite weak, and only exported a few tens of entry-points, and really mostly effectively only allowed character and block device drivers with standard interfaces, and loadable filesystems.

So historically, the fact that you could load a module using nothing but these standard interfaces tended to be a much stronger argument for not being very tightly coupled with the kernel.

Quoted 7 lines

That has changed, and the kernel module interfaces we have today are MUCH more extensive than they were back in '95 or so. These days modules are used for pretty much everything, including stuff that is very much "internal kernel" stuff and as a result the kind of historic "implied barrier" part of modules really has weakened, and as a result there is not avery strong argument for being an independent work from just the fact that you're a module.

FWIW, it would be very nice if somebody did hard and messy work and
produced lists of in-tree modules using given symbols. Ideally - automated that, but that won't be easy to do (quite a few are used only via inlined wrappers and in some cases - under an ifdef; many arch-specific exports are of that sort).

Aside of "hey, nothing uses that at all" and "only core uses it"
we'd get a bunch of "hmm, we really should've exported higher-level function instead" and "WTF does that lone driver use this function?". I'd played with that for fs/* exports and so far results look interesting. I'm using grep, but that's pretty much hopeless - we have literally thousands of exported symbols and any manual approach will break on that.

Some approximation might be obtained by building all modules and
doing nm on them, with manual work for non-obvious cases. I've done that on x86 (allmodconf + enabling whatever could be enabled, even if broken). Statistics is interesting, to put it mildly.

First of all, there are ~3600 symbols used by some in-tree drivers.
~600 of them are have 10 users or more. ~2000 have only one or two users. And we have ~7500 EXPORT_... in the tree. Now, that number is inflated by duplicates between architectures (and other stats are deflated by incomplete coverage). And yes, there are things that have every reason to be exported, but only a few modules care to use them.

However, it certainly looks like a large fraction of export list
should go away. Moreover, we probably should introduce EXPORT_FOR(symbol, module list)
and use it for stuff like jbd poking very deep in buffer.c guts - deeper than anybody else. Ditto for ipv6 / ipv4 interaction - they really have a special relationship and it makes no sense whatsoever to treat everything in TCPv4 guts that happens to be shared with TCPv6 as public export. There's a lot of cases like that and I suspect that they cover ~50-60% of the in-tree imports.

Real interface is somewhere around 400-500 symbols and it can be
split into several more or less compact parts. Having more than an order of magnitude more than that, and having it as a big pile... Not a good thing(tm), IMO.

#29 lore raw

[catching up with vast backlog]

On Thu, 4 Dec 2003, [email protected] mused:

Quoted 5 lines

FWIW, it would be very nice if somebody did hard and messy work and
produced lists of in-tree modules using given symbols. Ideally - automated that, but that won't be easy to do (quite a few are used only via inlined wrappers and in some cases - under an ifdef; many arch-specific exports are of that sort).

ISTM that this could be done quite easily with a hacked-up libcpp that does token pasting and #include processing, but does *not* process #if statements themselves.

Preprocess every translation unit in the kernel with that, and grep each of them for every exported symbol in turn, and bingo. :)

I'll have a hack sometime this weekend if nobody else gets around to it.

(Of course, this will say that stuff in #if 0's and other #ifdef branches that can't be reached is in fact used, but such code isn't common.)

Quoted 2 lines

Some approximation might be obtained by building all modules and
doing nm on them, with manual work for non-obvious cases.

Hang on, surely you can tell which symbols in modules are exported just by looking for expansions of EXPORT_SYMBOL{_GPL}? Why is this bit hard?

I think the problem is you're trying to compile rather than doing a hacked -E :) we don't need to compile for this, just do most of the preprocessing phase.

Signature · 2 lines
`...some suburbanite DSL customer who thinks kernel patches are some
 form of military insignia.' --- Bob Apthorpe
#30 lore raw
Quoted 5 lines · Nix
Quoted 2 lines

Some approximation might be obtained by building all modules and
doing nm on them, with manual work for non-obvious cases.

Hang on, surely you can tell which symbols in modules are exported just by looking for expansions of EXPORT_SYMBOL{_GPL}? Why is this bit hard?

It's not a question of which symbols are exported by module, it's what symbols are _imported_.

IOW, the hard question is "what modules use foo()", not "where do we define foo()". And while it's easy for a single symbol, we want it for _all_ exported symbols in the tree at once.

#31 lore raw

On Sat, 13 Dec 2003, [email protected] yowled:

Quoted 9 lines · Nix
Quoted 5 lines
Quoted 2 lines

Some approximation might be obtained by building all modules and
doing nm on them, with manual work for non-obvious cases.

Hang on, surely you can tell which symbols in modules are exported just by looking for expansions of EXPORT_SYMBOL{_GPL}? Why is this bit hard?

It's not a question of which symbols are exported by module, it's what symbols are _imported_.

Yes, of course. That's what you use the hacked-up non-macro-expanding-or- #if-processing preprocessor for. :)

Quoted 3 lines

IOW, the hard question is "what modules use foo()", not "where do we define foo()". And while it's easy for a single symbol, we want it for _all_ exported symbols in the tree at once.

Hm, OK, so a hacked cpp isn't good enough because it still requires more than grep over preprocessed-save-for-#if-and-macro-expansion sources to determine if you'll get an external symbol reference.

(Hm, or does it? We'll get FPs doing it that way, from text in literal strings, but how many? How many modules contain the names of exported symbols from other modules in literal strings? Not many, I guess, but there's no way to tell but to try... I'll have a whack at it.)

Signature · 2 lines
`...some suburbanite DSL customer who thinks kernel patches are some
 form of military insignia.' --- Bob Apthorpe
#32 lore raw
In reply to Linus Torvalds
Quoted 8 lines · Linus Torvalds
Quoted 5 lines · Kendall Bennett

I have heard many people reference the fact that the although the Linux Kernel is under the GNU GPL license, that the code is licensed with an exception clause that says binary loadable modules do not have to be under the GPL.

Nope. No such exception exists.

Right, after following a few of the links and reading the previous discussions, that is clear to me now ;-)

Quoted 9 lines

But one gray area in particular is something like a driver that was originally written for another operating system (ie clearly not a derived work of Linux in origin). At exactly what point does it become a derived work of the kernel (and thus fall under the GPL)?

THAT is a gray area, and _that_ is the area where I personally believe that some modules may be considered to not be derived works simply because they weren't designed for Linux and don't depend on any special Linux behaviour.

Or what if a device driver is written such that it is completely binary portable between different operating systems, such that the same binary can work on Linux, OS/2, Windows, QNX, BeOS and multiple embedded systems? Clearly I am talking about our SciTech SNAP drivers here, as that is what they are. Would writing glue code to allow those modules to be loaded inside the Linux kernel make the binary driver modules derived works? Clearly the glue code would be Linux specific and hence need to be GPL, but would it extend to the binary module? I can also come up with some arguments both for and against this scenario, and I agree 100% that it is a huge gray area.

To date we have not even attempted to load our SNAP drivers inside the Linux kernel, even though it would be quite easy to do so, primarily because of this gray area. We may need to write some kernel modules shortly to support a few things that the SNAP drivers may need, but since the user space drivers would interface to this kernel module via system calls, only the kernel module would need to the GPL and the user space driver that uses it would not. So we may never try to load the SNAP drivers inside the kernel, but it is an interesting exercise to consider the legality of whether it is possible.

Anyway loading SNAP drivers inside the Linux kernel, while interesting, wasn't the reason I brought up this discussion. The primary reason is that we are going to be releasing the SciTech SNAP DDK under the GNU GPL soon, and I wanted to understand how this expection clause works. Since it doesn't exist, it does open up a few questions ;-)

It is our position that drivers that would be built from our GNU GPL DDK are in themselves separate works as they follow a clearly defined, binary compatible interface and can be used with software building using a non- GPL version of our SDK (the SDK is actually LGPL). The driver binary itself on the other hand must be GPL, as it links against generic code provided to the developer as part of the DDK, so full source to any driver built with the GPL DDK must be provided. Even your own writings on this issue would indicate that you would agree with me on that front.

The lowest level binary modules have a 'less public' interface, in that we do not directly document it (well, at least not yet except for the header files but some may consider that valid documentation). It does evolve more quickly over time than the public SDK API, but even so the drivers ABI is strictly maintained for backwards compatibility, even though it does evolve. We have a clearly defined interface for evolving the API while maintaining backwards compatibility, and even doing hot bug fixes for existing binary drivers as necessary. But above the low level driver is the public SciTech SNAP API, which evolves much, much more slowly and is completely documented (both at the header file level and external documentation level).

Does my position make sense?

Clearly the intention here is to make sure that the GPL version of our DDK is available for free use by the Free Software community, but if someone like Intel or NVIDIA wants to write a SNAP driver that they don't provide source code for, they need to license a proprietry licensed version of our DDK.

Regards,

---
Kendall Bennett
Chief Executive Officer
SciTech Software, Inc.
Phone: (530) 894 8400
http://www.scitechsoft.com

~ SciTech SNAP - The future of device driver technology! ~

#33 lore raw
In reply to Linus Torvalds
Quoted 8 lines

So being a module is not a sign of not being a derived work. It's just one sign that _maybe_ it might have other arguments for why it isn't derived.

Linus

My question is a natural extension of this point and subsequent posts, When is a kernel so tainted it can no longer be considered GPL, or can no longer be considered free software? I write software for astronomical applications where some vendors give binary only drivers, or give you restricted access to the source code so I will some times load 5 or 6 devices that are binary only or at least non GPL. So what taint is allowable?

Thanks,
Aaron Smith
Virginia Astronomical Instrumentation Laboratory
Programmer

PS sorry if this is a stupid question.

#34 lore raw

there is no way to taint something so much that it's no longer covered by the GPL (if it was that would mean that you have no legal way to have a copy of it).

Remember that the GPL doesn't put any restrictions on what you do with the code as long as you are not distributing it.

becouse of this you could take the kernel and include any propriatary code in it that you want and run it. You don't even need to use modules, just paste in th code and compile (make sure you have a legal right to the code you are pasting in though :-)

the GPL comes in when you distribute it (and even here there are grey areas, does it count as distributing code to put it on a companies desktops for example?, but if you sell it or incude it in a product that you sell you definantly do have to release the source)

if you are talking about the 'tainted' flag in the kernel, that is just a flag that you should not expect to get support for this configuration from the opensource developers.

does this answer your question?

David Lang

Quoted 34 lines · Aaron Smith

Date: Wed, 03 Dec 2003 19:52:26 -0500
From: Aaron Smith <[email protected]>
To: [email protected]
Subject: Re: Linux GPL and binary module exception clause?

Quoted 7 lines

So being a module is not a sign of not being a derived work. It's just one sign that _maybe_ it might have other arguments for why it isn't derived.

Linus

My question is a natural extension of this point and subsequent posts, When is a kernel so tainted it can no longer be considered GPL, or can no longer be considered free software? I write software for astronomical applications where some vendors give binary only drivers, or give you restricted access to the source code so I will some times load 5 or 6 devices that are binary only or at least non GPL. So what taint is allowable?

Thanks,
Aaron Smith
Virginia Astronomical Instrumentation Laboratory
Programmer

PS sorry if this is a stupid question.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 3 lines
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." - Brian W. Kernighan
#35 lore raw

Thank you for clarifiing that. Now I wont feel so guilty for modifying my init scripts :-)

-Aaron Smith

#36 lore raw
In reply to David Lang
Quoted 4 lines · David Lang

becouse of this you could take the kernel and include any propriatary code in it that you want and run it. You don't even need to use modules, just paste in th code and compile (make sure you have a legal right to the code you are pasting in though :-)

To elaborate on this: most patches which apply to the Linux kernel are clearly derived works of the kernel.[*]

You can write and apply such a patch , but if you distribute the patch it must be licensed under the GPL.

Thus while _you_ may take a Linux kernel and include any proprietary code you want in it, a vendor _may not_ send you a patch which applies to the Linux kernel if the patch is not licensed under the GPL or a compatible license. A vendor is not allowed to restrict your rights in this way.

[*] Exceptions would be patches which are very small (<10 lines is sometimes suggested, but it's not a hard boundary), and patches which simply add code which is clearly not derived form the kernel, as Linus described.

-- Jamie

#37 lore raw
In reply to Linus Torvalds
Quoted 7 lines · Linus Torvalds

And in fact, when it comes to modules, the GPL issue is exactly the same. The kernel _is_ GPL. No ifs, buts and maybe's about it. As a result, anything that is a derived work has to be GPL'd. It's that simple.
...
- anything that has knowledge of and plays with fundamental internal Linux behaviour is clearly a derived work. If you need to muck around with core code, you're derived, no question about it.

If that is the case, why the introduction of EXPORT_SYMBOL_GPL and MODULE_LICENSE()?

Specifying explicit boundaries for the module interface has legitimised binary-only modules.
This was the signal to developers of proprietary code that binary-only modules are tolerable.

Note that I said tolerable, not acceptable. Ref also the 'tainted' flag ("man 8 insmod")
My personal view is that Linux should mandate GPL for all modules in 2.6 and beyond.

The Kevin Dankwardt article gives an alternative perspective for Linux embedded use:
http://www.linuxdevices.com/articles/AT5041108431.html

#38 lore raw
Quoted 7 lines · Jason Kingsland
Quoted 3 lines

- anything that has knowledge of and plays with fundamental internal Linux behaviour is clearly a derived work. If you need to muck around with core code, you're derived, no question about it.

If that is the case, why the introduction of EXPORT_SYMBOL_GPL and MODULE_LICENSE()?

It is really just documentation.

This is exactly so that it is more clear which cases are black-and-white, and where people shouldn't even have to think about it for a single second. It still doesn't make the gray area go away, but it limits it a bit ("if you need this export, you're clearly doing something that requires the GPL").

Note: since the kernel itself is under the GPL, clearly anybody can modify the EXPORT_SYMBOL_GPL() line, and remove the _GPL part. That wouldn't be against the license per se. But it doesn't make a module that needs that symbol any less needful of the GPL - exactly because the thing is just a big cluehint rather than anything else.

Linus

#39 lore raw
In reply to Jason Kingsland
Quoted 4 lines

My personal view is that Linux should mandate GPL for all modules in 2.6 and
beyond.

I'm baffled how you think this is a choice that can be made. The license is
the GPL itself and even the Linux kernel developers have no power to change it.

DS

#40 lore raw
Quoted 2 lines

My personal view is that Linux should mandate GPL for all modules in 2.6

and beyond.

Quoted 1 line · David Schwartz

I'm baffled how you think this is a choice that can be made. The license

is

Quoted 1 line

the GPL itself and even the Linux kernel developers have no power to

change it.

Modules are essentially dynamically linked extensions to the GPL kernel. In some cases they can be shown to be independent, prior works where GPL can reasonably be argued not to apply - which as Linus stated earlier on this thread was the original intention of allowing binary-only modules.

But in most of the more recent cases the driver/module code is written specifically for Linux, so it seems more appropriate that they would be considered as derived works of the kernel. But those various comments from Linus are being taken out of context to somehow justify permission for the non-release of source code for binary loadable modules.

Linux is not pure GPL, it also has the Linus "user program" preamble in copying.txt - that preamble plus other LKML posts from Linus are commonly used as justifications for non-disclosure of source code to some classes of modules.

But with all due respect, Linus is not the only author of Linux and his words to tend to convey an artificial sense of authority or justification for such attitudes. Here is a typical example:
http://www.linuxdevices.com/articles/AT9161119242.html

All I am suggesting is that the preamble could be extended to clearly state the position for kernel binary-only modules, and that the upcoming 2.6 release might be an opportunity for a quorum of the Linux authors to agree to revised wording.

#41 lore raw
Quoted 10 lines · On Thu, 4 Dec 2003 21:43:01 -0500 "Jason Kingsland"…

But in most of the more recent cases the driver/module code is written specifically for Linux, so it seems more appropriate that they would be considered as derived works of the kernel. But those various comments from Linus are being taken out of context to somehow justify permission for the non-release of source code for binary loadable modules.

Linux is not pure GPL, it also has the Linus "user program" preamble in copying.txt - that preamble plus other LKML posts from Linus are commonly used as justifications for non-disclosure of source code to some classes of modules.

what I forsee happening is, if the mandation of modules is enforced, people will start writing modules that just export the module interfaces to userspace and start writing userspace drivers, which will just be slower and uglier than the binary only drivers being sent right now. I know about these drivers, I use an nforce2 motherboard. Fortunately, people have reverse engineered the network driver for this (and I do love forcedeth) and people have reverse engineered other drivers, or brute force hacked them. And I'm guessing people will have to continue to do this until linux becomes common on the desktop.

Pat Erley

#42 lore raw
In reply to Jason Kingsland
Quoted 6 lines · Jason Kingsland

Modules are essentially dynamically linked extensions to the GPL kernel. In some cases they can be shown to be independent, prior works where GPL can reasonably be argued not to apply - which as Linus stated earlier on this thread was the original intention of allowing binary-only modules.

Quoted 6 lines

But in most of the more recent cases the driver/module code is written specifically for Linux, so it seems more appropriate that they would be considered as derived works of the kernel. But those various comments from Linus are being taken out of context to somehow justify permission for the non-release of source code for binary loadable modules.

I would agree that a module/driver written specifically for Linux would fall under the GPL. By extension however, any user program written specifically for Linux that does *not* run anywhere else would also fall under the GPL. Clearly there are not many programs that fit into this category, as most are portable to other platforms. However there are clear instances of code that is Linux specific such as the installers used by distro vendors to install their version of Linux. By extension if Linux specific modules must be GPL, so too must Linux specific installation programs.

Which means all the proprietry installers done by many distro vendors that are not GPL would be in violation.

Quoted 4 lines

Linux is not pure GPL, it also has the Linus "user program" preamble in copying.txt - that preamble plus other LKML posts from Linus are commonly used as justifications for non-disclosure of source code to some classes of modules.

No, Linux is pure GPL. I always thought the same but Linus and others have cleared this up in the last few days within this discussion. The pre- amble at the top of the COPYING file is not legalese and not legally binding. It is just Linus' interpretation of the GPL as it applies to user programs, and as such is certainly not legally binding.

Regards,

---
Kendall Bennett
Chief Executive Officer
SciTech Software, Inc.
Phone: (530) 894 8400
http://www.scitechsoft.com

~ SciTech SNAP - The future of device driver technology! ~

#43 lore raw
In reply to Jason Kingsland

Sorry as the principle author of the taskfile transport for storage. And all points of distribution under stupid anal EC law apply.

I will vote NO to enforce such brain dead concepts.

See since it is not easy to get all the authors/contributors/surviors to agree to switch to a new/better/license, it takes only one to say NO to stop the process of change.

Well here is the one NO, and imposing all modules are GPL is a no can do.

Bye have a nice day!

Andre Hedrick
LAD Storage Consulting Group

Quoted 42 lines · Jason Kingsland
Quoted 2 lines

My personal view is that Linux should mandate GPL for all modules in 2.6

and beyond.

Quoted 1 line · David Schwartz

I'm baffled how you think this is a choice that can be made. The license

is

Quoted 1 line

the GPL itself and even the Linux kernel developers have no power to

change it.

Modules are essentially dynamically linked extensions to the GPL kernel. In some cases they can be shown to be independent, prior works where GPL can reasonably be argued not to apply - which as Linus stated earlier on this thread was the original intention of allowing binary-only modules.

But in most of the more recent cases the driver/module code is written specifically for Linux, so it seems more appropriate that they would be considered as derived works of the kernel. But those various comments from Linus are being taken out of context to somehow justify permission for the non-release of source code for binary loadable modules.

Linux is not pure GPL, it also has the Linus "user program" preamble in copying.txt - that preamble plus other LKML posts from Linus are commonly used as justifications for non-disclosure of source code to some classes of modules.

But with all due respect, Linus is not the only author of Linux and his words to tend to convey an artificial sense of authority or justification for such attitudes. Here is a typical example:
http://www.linuxdevices.com/articles/AT9161119242.html

All I am suggesting is that the preamble could be extended to clearly state the position for kernel binary-only modules, and that the upcoming 2.6 release might be an opportunity for a quorum of the Linux authors to agree to revised wording.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#44 lore raw
In reply to David Schwartz
Quoted 7 lines
Quoted 3 lines

My personal view is that Linux should mandate GPL for all modules in 2.6 and
beyond.

I'm baffled how you think this is a choice that can be made. The license is
the GPL itself and even the Linux kernel developers have no power to change it.

I'm not so sure about that. If Linus and a few core developers decide to explicitly state that you can't use the Linux kernel with binary only modules than you can't. They have the right under copyright to restrict the usage of their contributions - if this means that they say "you can't use our contributions to the kernel with any binary only ring 0 code" - then you can't. And it is totally irrelevant whether your work is derived or not. They can't force you to release your work as GPL - but they can forbid you to utilise the linux kernel with non-GPL'ed work - which in the end is quite close.

Or am I totally wrong here?

Cheers,
MaZe.

#45 lore raw

Lets have some fun now and play this game.

As principle author of the "taskfile transport", any an all operations using, storing, execution, transfering, copying, opening ... anything may not operate with non-source-published binary modules.

So everyone one with/sells a PVR, NAS, SAN, Laptop, Workstation, Server which uses IDE/ATA/SATA is forbidden to operate unless written terms of use are set forward.

We can kill Linux in minutes, shall we?

Andre Hedrick
LAD Storage Consulting Group

Quoted 29 lines · Maciej Zenczykowski
Quoted 7 lines
Quoted 3 lines

My personal view is that Linux should mandate GPL for all modules in 2.6 and
beyond.

I'm baffled how you think this is a choice that can be made. The license is
the GPL itself and even the Linux kernel developers have no power to change it.

I'm not so sure about that. If Linus and a few core developers decide to explicitly state that you can't use the Linux kernel with binary only modules than you can't. They have the right under copyright to restrict the usage of their contributions - if this means that they say "you can't use our contributions to the kernel with any binary only ring 0 code" - then you can't. And it is totally irrelevant whether your work is derived or not. They can't force you to release your work as GPL - but they can forbid you to utilise the linux kernel with non-GPL'ed work - which in the end is quite close.

Or am I totally wrong here?

Cheers,
MaZe.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#46 lore raw
Quoted 2 lines · Andre Hedrick

Lets have some fun now and play this game.

Sorry, you need to learn the rules before you can play.

Quoted 3 lines

As principle author of the "taskfile transport", any an all operations using, storing, execution, transfering, copying, opening ... anything may not operate with non-source-published binary modules.

That's against the GPL, and you can't modify the terms of the license. At most, you personally can say that you will not sue even when the license isn't followed - you can tell people that as far as _you_ are concerned, you can losen the license further, and that actually puts a legal onus on _you_ but nobody else.

But while you have the right to say "I will not sue over this" and the GPL doesn't care one whit, you can _not_ say "I have my own list of additional requirements that would trigger copyright infringement".

Quoted 3 lines

So everyone one with/sells a PVR, NAS, SAN, Laptop, Workstation, Server which uses IDE/ATA/SATA is forbidden to operate unless written terms of use are set forward.

"The act of running the Program is not restricted" according to the GPL, and "You may not impose any further restrictions on the recipients' exercise of the rights granted herein."

So basically you _cannot_ take rights away outside the ones the GPL requires (which boil down to the requirement of having source available).

Quoted 1 line

We can kill Linux in minutes, shall we?

Trust me, when you said that the GPL is badly written, you have no clue what you're talking about. It's a very solid license, and your rants about it have no basis in fact. I personally actually like the OSL slightly better in the way it was written (see opensource.org), but your arguments against the GPL are just fundamentally wrong.

Linus

#47 lore raw

Okay,

If I am wrong I can step up and freely admit it. :-)

If nothing else I got one point for getting you to point out you like OSL which I do also.

So given RMS and company state OSL and GPL are not compatable, how does the two exist in the current kernel? Earlier, iirc, there were comments about dual license conflicts.

I was actually trying to show how silly David's arguement was about imposing bogus rules and taking it to the logical ends of insanity. It is lonely out here, I need more people to go over the edge on the bobsled too.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

PS: I am not attached to the the flame war any more, I just enjoy the warmth of the heat :-0

Quoted 43 lines · Linus Torvalds
Quoted 2 lines · Andre Hedrick

Lets have some fun now and play this game.

Sorry, you need to learn the rules before you can play.

Quoted 3 lines

As principle author of the "taskfile transport", any an all operations using, storing, execution, transfering, copying, opening ... anything may not operate with non-source-published binary modules.

That's against the GPL, and you can't modify the terms of the license. At most, you personally can say that you will not sue even when the license isn't followed - you can tell people that as far as _you_ are concerned, you can losen the license further, and that actually puts a legal onus on _you_ but nobody else.

But while you have the right to say "I will not sue over this" and the GPL doesn't care one whit, you can _not_ say "I have my own list of additional requirements that would trigger copyright infringement".

Quoted 3 lines

So everyone one with/sells a PVR, NAS, SAN, Laptop, Workstation, Server which uses IDE/ATA/SATA is forbidden to operate unless written terms of use are set forward.

"The act of running the Program is not restricted" according to the GPL, and "You may not impose any further restrictions on the recipients' exercise of the rights granted herein."

So basically you _cannot_ take rights away outside the ones the GPL requires (which boil down to the requirement of having source available).

Quoted 1 line

We can kill Linux in minutes, shall we?

Trust me, when you said that the GPL is badly written, you have no clue what you're talking about. It's a very solid license, and your rants about it have no basis in fact. I personally actually like the OSL slightly better in the way it was written (see opensource.org), but your arguments against the GPL are just fundamentally wrong.

Linus

#48 lore raw
Quoted 4 lines · Andre Hedrick

So given RMS and company state OSL and GPL are not compatable, how does the two exist in the current kernel? Earlier, iirc, there were comments about dual license conflicts.

They don't "co-exist".

Some parts of the kernel are dual-licensed, which basically means that the author says "you can use this code under _either_ the GPL or the OSL".

When used in the kernel, the GPL is the one that matters. But being dual-licensed means that the same thing may be used somewhere else under another license (so you could use that particular instance of code under the OSL in some _other_ project where the OSL would be ok).

This is pretty common. We have several drivers that are dual-GPL/BSD, and there are some parts that are dual GPL/proprietary (which is just another way of saying that the author is licensing it somewhere else under a proprietary model - common for hardware manufacturers that write their own driver and _also_ use it somewhere else: when in Linux, they license it under the GPL, when somewhere else, they have some other license).

This isn't Linux-specific - you'll find the same thing in other projects. Most well-known perhaps perl - which is dual Artistic/GPL (I think. That's from memory).

And ghostscript was (is?) dual-licensed too (proprietary/GPL).

Linus

#49 lore raw

Linus:

Okay, can you add to the approved license list "GPL/proprietary" and "proprietary/GPL" ?

This would allow a legal shim for the hardcare folks to choke down, and provide the comfort level for others.

Then the trick is when does the license flip modes? Compile time?
Execution time?

This starts to become more fuzzy than I care to look at right now.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

Quoted 33 lines · Linus Torvalds
Quoted 4 lines · Andre Hedrick

So given RMS and company state OSL and GPL are not compatable, how does the two exist in the current kernel? Earlier, iirc, there were comments about dual license conflicts.

They don't "co-exist".

Some parts of the kernel are dual-licensed, which basically means that the author says "you can use this code under _either_ the GPL or the OSL".

When used in the kernel, the GPL is the one that matters. But being dual-licensed means that the same thing may be used somewhere else under another license (so you could use that particular instance of code under the OSL in some _other_ project where the OSL would be ok).

This is pretty common. We have several drivers that are dual-GPL/BSD, and there are some parts that are dual GPL/proprietary (which is just another way of saying that the author is licensing it somewhere else under a proprietary model - common for hardware manufacturers that write their own driver and _also_ use it somewhere else: when in Linux, they license it under the GPL, when somewhere else, they have some other license).

This isn't Linux-specific - you'll find the same thing in other projects. Most well-known perhaps perl - which is dual Artistic/GPL (I think. That's from memory).

And ghostscript was (is?) dual-licensed too (proprietary/GPL).

Linus

#50 lore raw
Quoted 3 lines · Andre Hedrick

Then the trick is when does the license flip modes? Compile time?
Execution time?

a license does not 'trigger' or 'flip'. Either the full source code is licensed under the GPL (by the copyright holders) or not.

a given piece of code might be licensed under an infinite number of other licenses as well, but this doesnt matter a bit, as long as the GPL is one of them.

Quoted 1 line

This starts to become more fuzzy than I care to look at right now.

then ask a lawyer.

Ingo

#51 lore raw

Hi Ingo,

I have and the lawyers tell me that it is one or the other and can not be both. So explain to me how a GPL/BSD or BSD/GPL works again?

Also if one does an md5sum on the "COPYING" file from FSF and compares it from the one in the kernel source they differ.

Since the original version from FSF protects the content inside because it protects the shell of the file, bridge point to stating the kernel is w/o a license is relativily easy. I will have to re-read the "original COPYING" file from FSF for version 1 and version 2.

Not sure but could the veil of GPL be now pierced because of the simple additions to the top of "COPYING" ?

Cheers,

Andre Hedrick
LAD Storage Consulting Group

Quoted 20 lines · Ingo Molnar
Quoted 3 lines · Andre Hedrick

Then the trick is when does the license flip modes? Compile time?
Execution time?

a license does not 'trigger' or 'flip'. Either the full source code is licensed under the GPL (by the copyright holders) or not.

a given piece of code might be licensed under an infinite number of other licenses as well, but this doesnt matter a bit, as long as the GPL is one of them.

Quoted 1 line

This starts to become more fuzzy than I care to look at right now.

then ask a lawyer.

Ingo

#52 lore raw
Quoted 2 lines · Andre Hedrick

I have and the lawyers tell me that it is one or the other and can not be both. So explain to me how a GPL/BSD or BSD/GPL works again?

ugh. Are your lawyers saying that the tons of dual-licensed code is not a valid license? Seems like your lawyers disagree with lots of other lawyers.

Quoted 2 lines

Also if one does an md5sum on the "COPYING" file from FSF and compares it from the one in the kernel source they differ.

here's the (trivial) diff. Draw your own conclusions.

COPYING +16 -0
@@ -1,3 +1,19 @@
+
+   NOTE! This copyright does *not* cover user programs that use kernel
+ services by normal system calls - this is merely considered normal use
+ of the kernel, and does *not* fall under the heading of "derived work".
+ Also note that the GPL below is copyrighted by the Free Software
+ Foundation, but the instance of code that it refers to (the Linux
+ kernel) is copyrighted by me and others who actually wrote it.
+
+ Also note that the only valid version of the GPL as far as the kernel
+ is concerned is _this_ particular version of the license (ie v2, not
+ v2.2 or v3.x or whatever), unless explicitly otherwise stated.
+
+			Linus Torvalds
+
+----------------------------------------
+
 		    GNU GENERAL PUBLIC LICENSE
 		       Version 2, June 1991
 
#53 lore raw

Ingo,

I suggest asking FSF how they play with GPL+another license. They will tell you GPL can not co-exist, period.

The reply is a double bounce back of ideas.

Jason,

What you are really saying is the Linux does not have a GPL license. It has a GPL-like or LKL.

Linux Kernel License.

Hmmm, new thought ...

I will be kind to it because it is in a strange place :-)

Cheers,

Andre Hedrick
LAD Storage Consulting Group

Quoted 43 lines · Ingo Molnar
Quoted 2 lines · Andre Hedrick

I have and the lawyers tell me that it is one or the other and can not be both. So explain to me how a GPL/BSD or BSD/GPL works again?

ugh. Are your lawyers saying that the tons of dual-licensed code is not a valid license? Seems like your lawyers disagree with lots of other lawyers.

Quoted 2 lines

Also if one does an md5sum on the "COPYING" file from FSF and compares it from the one in the kernel source they differ.

here's the (trivial) diff. Draw your own conclusions.

COPYING +16 -1
@@ -1,3 +1,19 @@
+
+   NOTE! This copyright does *not* cover user programs that use kernel
+ services by normal system calls - this is merely considered normal use
+ of the kernel, and does *not* fall under the heading of "derived work".
+ Also note that the GPL below is copyrighted by the Free Software
+ Foundation, but the instance of code that it refers to (the Linux
+ kernel) is copyrighted by me and others who actually wrote it.
+
+ Also note that the only valid version of the GPL as far as the kernel
+ is concerned is _this_ particular version of the license (ie v2, not
+ v2.2 or v3.x or whatever), unless explicitly otherwise stated.
+
+			Linus Torvalds
+
+----------------------------------------
+
 		    GNU GENERAL PUBLIC LICENSE
 		       Version 2, June 1991
 
-

To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

#54 lore raw
Quoted 2 lines · Andre Hedrick

I suggest asking FSF how they play with GPL+another license. They will tell you GPL can not co-exist, period.

They cannot coexist _at the same time in the same agreement_, but a single work can be licensed out hundreds of times in different ways.

If I hold the copyright on something, I can provide a different license to each licensee of the product if I so choose. That is pretty common, in fact.

The GPL and some-other-license are incompatible in the sense that you cannot put another license on _top_ of the GPL in the _same_ licensing agreement. But you are free to license something in the GPL _or_ BSD _or_ some-evil-EULA, and that is what Linus et al are talking about, and _not_ what the FSF means when they say that some license is incompatible with the GPL.

Robert Love

#55 lore raw
In reply to Andre Hedrick
Quoted 4 lines · Andre Hedrick

Ingo,

I suggest asking FSF how they play with GPL+another license. They will tell you GPL can not co-exist, period.

Think of it as a fork.
I own a piece of code I just made. So far I can license it the way I want, OK ?
So I fork it, license one branch under the GPL and license the other branch under a proprietary license. Voilà, it's dual-licensed ! When integrated to the linux kernel, it's technically fully GPLed, even if called "proprietary/GPL" in the header file.

Of course, contributors to my work are free to tell me "I want my patches to your code only for the GPL version", but generally they won't, so I can integrate them back to the proprietary code. Pending good legal advice, of course.

It it clearer ?

Xav

#56 lore raw
In reply to Linus Torvalds
Quoted 5 lines · Linus Torvalds

Trust me, when you said that the GPL is badly written, you have no clue what you're talking about. It's a very solid license, and your rants about it have no basis in fact. I personally actually like the OSL slightly better in the way it was written (see opensource.org), but your arguments against the GPL are just fundamentally wrong.

I believe OSL's patent termination clause would have prevented IBM from using patents in its counter-suit against SCO.

This license stuff isn't quite as easy as it looks. :)

Quoted 1 line

Linus

Rob

#57 lore raw
In reply to Linus Torvalds
Quoted 2 lines · Linus

There's a clarification that user-space programs that use the standard system call interfaces aren't considered derived works

If it said "user-space" or "non-kernel address space" in the Linux license then I would agree.

But the exact wording is much more vague:

"user programs that use kernel services by normal system calls"

Any binary loadable kernel module can be considered a "user program" Any interface defined in the kernel header files can be considered a "normal system call"

This is why I think further clarification is warranted in future versions of copying.txt - because we are needlessly giving away the source-code freedom that GPL is intended to protect.

The proponents of binary-only kernel modules currently use the above as a defense to argue their case for GPL non-compliance.

#58 lore raw
Quoted 4 lines · Jason Kingsland

Any binary loadable kernel module can be considered a "user program" Any interface defined in the kernel header files can be considered a "normal system call"

I disagree. No definition of "system call" implies "any random function". That's a technical term with a technical definition, and that technical definition is in no way ambiguous when we are talking about a kernel.

Linus