Re: Driver Model

Re: Driver Model

[Pascal Schmidt]

On Wed, 03 Sep 2003 20:00:24 +0200, you wrote in linux.kernel:

> The fact that GPL_ONLY horse sh*t exists means there is a restriction on
> usage.  So "GPL_ONLY" has in effect violated GPL, by imposing restrictions
> of usage.

Where is the restriction? You get the source code, you can roll your
own and remove the GPL_ONLY stuff. Apart from that I do not recall
to have seen anything about restrictions of usage in the GPL... the
only thing it tries to prevent is the source code becoming proprietary.

-- 
Ciao,
Pascal

Re: Driver Model

[Andre Hedrick]

On Wed, 3 Sep 2003, Pascal Schmidt wrote:

> On Wed, 03 Sep 2003 20:00:24 +0200, you wrote in linux.kernel:
> 
> > The fact that GPL_ONLY horse sh*t exists means there is a restriction on
> > usage.  So "GPL_ONLY" has in effect violated GPL, by imposing restrictions
> > of usage.
> 
> Where is the restriction? You get the source code, you can roll your
> own and remove the GPL_ONLY stuff. Apart from that I do not recall
> to have seen anything about restrictions of usage in the GPL... the
> only thing it tries to prevent is the source code becoming proprietary.

Pascal,

I agree with you and you make the point clear!
There are people who have issued letters to big linux distributors in the
US about changing a symbol types.  That is a restriction for usage, in my
opinion.

Cheers,

Andre

RE: Driver Model

[David Schwartz]

> On Wed, 03 Sep 2003 20:00:24 +0200, you wrote in linux.kernel:

> > The fact that GPL_ONLY horse sh*t exists means there is a restriction on
> > usage.  So "GPL_ONLY" has in effect violated GPL, by imposing
> > restrictions
> > of usage.

> Where is the restriction? You get the source code, you can roll your
> own and remove the GPL_ONLY stuff. Apart from that I do not recall
> to have seen anything about restrictions of usage in the GPL... the
> only thing it tries to prevent is the source code becoming proprietary.

	If the GPL_ONLY stuff is a license enforcement scheme, the DMCA prohibits
you from removing it. If the GPL_ONLY stuff is not a license enforcement
scheme, nothing prohibits you from stamping your module GPL when it's not.

	However, the GPL (section 2b) prohibits you from imposing any restrictions
other than those in the GPL itself. The GPL contains no restrictions that
apply to mere use and the GPL_ONLY stuff affects use, so it can't be a
license restriction, hence there is no restriction to enforce.

	I don't see anything preventing a GPL'd work from containing code that
imposes restrictions actually contained in the GPL and using the DMCA to
enforce them. But it would have to be a restriction contained in the GPL
itself, and there is no restriction about what code you can use with a GPL'd
work.

	DS

RE: Driver Model

[Pascal Schmidt]

On Wed, 3 Sep 2003, David Schwartz wrote:

> If the GPL_ONLY stuff is a license enforcement scheme, the DMCA 
> prohibits you from removing it.

-ENOTUSCITIZEN

> If the GPL_ONLY stuff is not a license enforcement scheme, nothing 
> prohibits you from stamping your module GPL when it's not.

I'd say its up to the lawyers and judges to find out whether having
MODULE_LICENSE("GPL") in a module means anything legally. It might
mean "I promise this module is made from GPL source", but it might
also mean nothing.

> However, the GPL (section 2b) prohibits you from imposing any
> restrictions other than those in the GPL itself.

Section 2b) in the file COPYING in the root dir of the kernel source
does not talk about restrictions. Are we talking about the same version
of the GPL?

> The GPL contains no restrictions that
> apply to mere use and the GPL_ONLY stuff affects use, so it can't be a
> license restriction, hence there is no restriction to enforce.

The GPL doesn't even cover use of the "product". It covers modification
and redistribution.

Well, it is still an open question whether kernel modules are derived
works or not, especially since we don't have a stable kernel ABI and
therefore modules have to use part of the kernel source (headers) and
module writers have to study kernel code to write their modules (since
there is no official complete documentation about functions in the 
kernel).

If modules are derived works, then legally, following the GPL, they
must be GPL too and GPL_ONLY is no problem but pointless.

Seems to me you could say GPL_ONLY is a way of the developer saying
"I consider your stuff to be a derived work if you use this symbol".
Ask a lawyer whether that's their decision to make. ;)

Apart from that, I fail to see how it is an addition restriction
when you still have the right to remove all the GPL_ONLY stuff. After
all, the kernel is GPLed work, so you have the right to remove
things and distribute the result. How is it a real restriction when
the license allows you to remove it?

-- 
Ciao,
Pascal

RE: Driver Model

[David Schwartz]

> On Wed, 3 Sep 2003, David Schwartz wrote:

> > If the GPL_ONLY stuff is a license enforcement scheme, the DMCA
> > prohibits you from removing it.

> -ENOTUSCITIZEN

	In that case, there is more than likely nothing that prevents you from
doing whatever you want.

> > If the GPL_ONLY stuff is not a license enforcement scheme, nothing
> > prohibits you from stamping your module GPL when it's not.

> I'd say its up to the lawyers and judges to find out whether having
> MODULE_LICENSE("GPL") in a module means anything legally. It might
> mean "I promise this module is made from GPL source", but it might
> also mean nothing.

	Probably so.

> > However, the GPL (section 2b) prohibits you from imposing any
> > restrictions other than those in the GPL itself.

> Section 2b) in the file COPYING in the root dir of the kernel source
> does not talk about restrictions. Are we talking about the same version
> of the GPL?

    b) You must cause any work that you distribute or publish, that in
    whole or in part contains or is derived from the Program or any
    part thereof, to be licensed as a whole at no charge to all third
    parties under the terms of this License.

	In other words, if you want to distribute the Linux kernel, you must
license it under the terms of the GPL. You may not impose additional
restrictions because if you do, you're not causing it to be distribute under
the terms of "this License".

	So if I download the Linux kernel from somewhere, someone distributed it to
me. Hence, if they complied with the GPL, I am under only the obligations
imposed by the GPL.

> > The GPL contains no restrictions that
> > apply to mere use and the GPL_ONLY stuff affects use, so it can't be a
> > license restriction, hence there is no restriction to enforce.

> The GPL doesn't even cover use of the "product". It covers modification
> and redistribution.

	It does cover use. Specifically, it permits unrestriced use. If you
received GPL'd code, you have the unrestricted right to use it. That's what
section 2b says.

> Well, it is still an open question whether kernel modules are derived
> works or not, especially since we don't have a stable kernel ABI and
> therefore modules have to use part of the kernel source (headers) and
> module writers have to study kernel code to write their modules (since
> there is no official complete documentation about functions in the
> kernel).

	Non-issue. I'm talking about your rights to *use* the kernel.

> If modules are derived works, then legally, following the GPL, they
> must be GPL too and GPL_ONLY is no problem but pointless.

	You must not be reading the same GPL I am. Can you please cite to me the
section that requires derived works to be placed under the GPL. I can't find
it.

> Seems to me you could say GPL_ONLY is a way of the developer saying
> "I consider your stuff to be a derived work if you use this symbol".
> Ask a lawyer whether that's their decision to make. ;)

	But that's not what it does. It prevents you from using the kernel in
certain ways. The GPL does not permit such usage restrictions. It also
restricts your ability to create and use derived works. The GPL similarly
does not permit such restrictions. The only restrictions the GPL allows a
distributed derived work to contain are those specifically imposed by the
GPL, and those restrictions only kick in at distribution.

	If there were distribution restrictions, you'd have an argument. But we are
talking about use restrictions.

> Apart from that, I fail to see how it is an addition restriction
> when you still have the right to remove all the GPL_ONLY stuff.

	You only have that right (in the United States) if the GPL_ONLY stuff is
*not* a copyright enforcement scheme.

> After
> all, the kernel is GPLed work, so you have the right to remove
> things and distribute the result. How is it a real restriction when
> the license allows you to remove it?

	Fine, so long as we all agree that the GPL_ONLY stuff is not a copyright or
license enforcement scheme and that evading or modifying it is not evading a
copyright/license enforcement scheme. In this case, you cannot argue that
the DMCA prohibits claiming a GPL license for purposes of compatability.

	DS

RE: Driver Model

[Pascal Schmidt]

On Wed, 3 Sep 2003, David Schwartz wrote:

> In other words, if you want to distribute the Linux kernel, you must
> license it under the terms of the GPL. You may not impose additional
> restrictions because if you do, you're not causing it to be distribute
> under the terms of "this License".

Correct.

> It does cover use.

In section 0:

Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope.

> Specifically, it permits unrestriced use. If you
> received GPL'd code, you have the unrestricted right to use it. That's
> what section 2b says.

No, section 2b gives you the the right to copy, distribute, and modify
the code (as the license only covers those rights, as per section 0) and 
no restrictions may be placed on those specific rights.

> Non-issue. I'm talking about your rights to *use* the kernel.

Well, I'm not buying the argument that the GPL has anything to say
about usage.

> You must not be reading the same GPL I am. Can you please cite to me the
> section that requires derived works to be placed under the GPL. I can't
> find it.

You quoted it yourself. 2b) 

    b) You must cause any work that you distribute or publish, that in
    whole or in part contains or is derived from the Program or any
    part thereof, to be licensed as a whole at no charge to all third
    parties under the terms of this License.

"work that ... is derived from the Program or any part thereof"

> But that's not what it does. It prevents you from using the kernel in
> certain ways. The GPL does not permit such usage restrictions.

Lots of GPL'ed programs refuse to be used in certain ways. For example,
fetchmail will refuse to run with a world-readable .fetchmailrc file.

> It also restricts your ability to create and use derived works. The GPL 
> similarly does not permit such restrictions.

That it does not, and such a restriction would violate the GPL, I'd
agree to that.

> You only have that right (in the United States) if the GPL_ONLY stuff is
> *not* a copyright enforcement scheme.

How can it be that? It does not restrict copying nor distribution
nor modification.

People here are saying that it's more of a hint to people that they
better think hard and ask a lawyer before implementing non-GPL'ed
kernel modules.

I think you bringing the DMCA into this shows an interesting aspect
of that law: who gets to say what is a copyright enforcement scheme
and what is not?

-- 
Ciao,
Pascal

RE: Driver Model

[David Schwartz]

> On Wed, 3 Sep 2003, David Schwartz wrote:
>
> > In other words, if you want to distribute the Linux kernel, you must
> > license it under the terms of the GPL. You may not impose additional
> > restrictions because if you do, you're not causing it to be distribute
> > under the terms of "this License".
>
> Correct.
>
> > It does cover use.
>
> In section 0:
>
> Activities other than copying, distribution and modification are not
> covered by this License; they are outside its scope.

	So are you arguing that I can distribute a derived work from the Linux
kernel and attach a 'you may not use this unless you pay me $100' clause and
it would be enforceable?

> > Specifically, it permits unrestriced use. If you
> > received GPL'd code, you have the unrestricted right to use it. That's
> > what section 2b says.
>
> No, section 2b gives you the the right to copy, distribute, and modify
> the code (as the license only covers those rights, as per section 0) and
> no restrictions may be placed on those specific rights.

	If you are right, you've discovered a serious fundamental flaw in the GPL.
I can distribute code under the GPL and prohibit anyone from using derived
works, hence effectively removing their freedom to modify.

> > Non-issue. I'm talking about your rights to *use* the kernel.
>
> Well, I'm not buying the argument that the GPL has anything to say
> about usage.

	Then you have no right to usage. The preamble contradicts this, but I doubt
it's binding.

> > You must not be reading the same GPL I am. Can you please cite to me the
> > section that requires derived works to be placed under the GPL. I can't
> > find it.
>
> You quoted it yourself. 2b)
>
>     b) You must cause any work that you distribute or publish, that in
>     whole or in part contains or is derived from the Program or any
>     part thereof, to be licensed as a whole at no charge to all third
>     parties under the terms of this License.
>
> "work that ... is derived from the Program or any part thereof"

	This is only about works that you distribute or publish. We're talking
about using modules.

> > But that's not what it does. It prevents you from using the kernel in
> > certain ways. The GPL does not permit such usage restrictions.
>
> Lots of GPL'ed programs refuse to be used in certain ways. For example,
> fetchmail will refuse to run with a world-readable .fetchmailrc file.

	Yes, but nobody's arguing that these are license enforcement schemes or
that this is a license restriction. Nobody would complain if you removed
those restrictions. The question here is whether the GPL_ONLY stuff is a
copyright enforcement mechanism or license restriction. If we agree it's not
and anyone is free to circumvent or remove it, then there's nothing to
dispute.

> > It also restricts your ability to create and use derived works. The GPL
> > similarly does not permit such restrictions.
>
> That it does not, and such a restriction would violate the GPL, I'd
> agree to that.

	Okay.

> > You only have that right (in the United States) if the GPL_ONLY stuff is
> > *not* a copyright enforcement scheme.

> How can it be that? It does not restrict copying nor distribution
> nor modification.

	Copyright enforcement schemes can also restrict usage. Access cards for
satellite TV are purely usage restriction devices.

> People here are saying that it's more of a hint to people that they
> better think hard and ask a lawyer before implementing non-GPL'ed
> kernel modules.

	I'd agree with that.

> I think you bringing the DMCA into this shows an interesting aspect
> of that law: who gets to say what is a copyright enforcement scheme
> and what is not?

	The law has a somewhat incomprehensible definition of what consitutes such
a scheme. I don't think anybody really knows what things would be considered
copyright enforcement schemes and what would not.

	Your argument that the GPL does not grant usage rights is troubling. If
it's correct, then I have no right to use the Linux kernel, since the
copyright holders never granted it to me!

	DS

RE: Driver Model

[Pascal Schmidt]

On Wed, 3 Sep 2003, David Schwartz wrote:

> So are you arguing that I can distribute a derived work from the Linux
> kernel and attach a 'you may not use this unless you pay me $100' clause
> and it would be enforceable?

Well, the GPL does not allow you to impose retrictions on copying and
redistribution, so you cannot disallow others redistributing your work
without your usage clause attached, and it would be legal for them to
do that under the GPL.

Your restriction could only effect the part of the kernel you
actually modified, because that would be the only part you have
copyright on. How are you going to prove people used that part? They
have the right to modify the copy, unrestricted, so they could've
taken your piece of the code out before using the kernel.

> If you are right, you've discovered a serious fundamental flaw in the 
> GPL. I can distribute code under the GPL and prohibit anyone from using 
> derived works, hence effectively removing their freedom to modify.

No, I don't see how you have influence over derived works. Nothing in
the GPL allows you to bind redistributors to your usage clause, since
copying and redistribution is covered by the GPL.

> Then you have no right to usage. The preamble contradicts this, but I
> doubt it's binding.

Well, under German law, if I have the right to legally obtain something,
I automatically have the right to usage. Law only comes into play again if
I redistribute it or use it in public in some illegal way.

Under German law, you need a real contract in place for usage
restrictions to be effective or you need to make certain kinds of
usage technically impossible (because if you don't, courts will say
you did not even attempt to protect your interests). And no, MS-type 
"open this bag and you're bound to our license" is not a legal contract 
over here. Violation of the usage restriction would then fall under 
contract law, not copyright law.

I don't see how GPL'ed code can have technical restrictions since I
can easily compile and change it and I don't have to sign a contract to 
get it since the license grants unrestricted redistribution right.

This may all be different in the US, of course.

> This is only about works that you distribute or publish. We're talking
> about using modules.

Well, somebody must have distributed the module, and if that act was
illegal because of a GPL violation, I don't know whether you have any
right to usage anyway.

> Copyright enforcement schemes can also restrict usage. Access cards for
> satellite TV are purely usage restriction devices.

These come with a contract here in Germany, and once you have a contract,
you can of course, within some limits, put usage restrictions in it as
much as you like. You just have to make sure you only give the card to
someone who has indeed signed the contract.

-- 
Ciao,
Pascal

People, not GPL [was: Re: Driver Model]

[Timothy Miller]

I'm still 1600 messages behind in reading the list, but I have spent 
enough time reading the discussion about GPL and drivers that I feel 
compelled to comment.  I don't intend to comment further because I don't 
want to contribute to a flamewar any more than this already will.  But I 
feel the need to defend those who contribute to Free Software against 
those who don't.

The argument I have been reading has been centered around the idea of 
working around the GPL to support binary-only driver and various other 
things which are counter to the spirit of the GPL and Linux.  But 
someone who is trying to find a legal GPL loophole is not considering 
the root of the situation and that the GPL is an effect, not a cause.

A point someone else made that I feel compelled to reiterate is that it 
is the nature of the Linux development model and the attitude of the 
developers which has made Linux what it is and has made you want to use it.

But I have another point.  You are not dealing with a license here.  The 
license is there to satisfy lawyers and make clear the INTENT of the 
authors.  The keyword here is INTENT in that someone who has developed 
something is telling you how they feel about the use of their work 
which, under many circumstances, they could have chosen not to share 
with you.  What you are dealing with is real people who have put an 
incredible amount of time and effort into developing Linux.  Those 
people, to whom you owe much respect for sharing their contributions, 
have decided that their software should be used with certain 
restrictions, that being the GPL.  If you abuse Linux, it is not the GPL 
that you are insulting, but the people who developed Linux.

The GPL_ONLY restriction for driver modules may seem unfair, but it is 
far from it.  There are both valid technical and philosophical reasons 
for working that way.  No one forces you to use Linux, and when you made 
the choice to use it, you are entering into a community with a specific 
philosophy.  You know that philosophy in advance, so when you discover 
that you have a restruction you don't like, you have no room to complain.

As someone said, if you want to write drivers for a UNIX which does not 
have these restrictions, there are plenty of commercial UNIXes out there 
that you have to choose from.  The fact that they are perhaps less 
popular is one reason why Linux developers do not want to imitate them.

So, the discussions about finding ways to make a non-GPL driver look 
like a GPL driver and get away with it legally are all moot.  The reason 
you should not violate this is because the architects of Linux do not 
want you to.  If you choose to violate that, you are being unethical, 
pure and simple.  Or more to the point, you're being an asshole to a lot 
of hard-working people who have chosen to freely share their work with 
you.  Since they are the authors and you are not, their feelings about 
their softare are more important than yours.  You may be able to screw 
them over and get away with it -- people do that sort of thing all the 
time -- but the fact that you may find a legal loophole doesn't make you 
any less of an abject asshole.

In short:  Be honorable.

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> But I have another point.  You are not dealing with a license here.  The
> license is there to satisfy lawyers and make clear the INTENT of the
> authors.  The keyword here is INTENT in that someone who has developed
> something is telling you how they feel about the use of their work
> which, under many circumstances, they could have chosen not to share
> with you.  What you are dealing with is real people who have put an
> incredible amount of time and effort into developing Linux.  Those
> people, to whom you owe much respect for sharing their contributions,
> have decided that their software should be used with certain
> restrictions, that being the GPL.  If you abuse Linux, it is not the GPL
> that you are insulting, but the people who developed Linux.

	In other words, information does not want to be free. You shouldn't use
code the way you want to use it but the way the authors want you to use it.
After all, they didn't have to give it to you if they didn't want to.

	However, Richard Stallman does not agree with this view. It's his view that
if the authors chose to give you the code, you can use it any way you want
to, regardless of how the authors feel about that type of usage. This is why
he created the GPL.

> So, the discussions about finding ways to make a non-GPL driver look
> like a GPL driver and get away with it legally are all moot.  The reason
> you should not violate this is because the architects of Linux do not
> want you to.

	If you really believe that the Linux authors wished to continue to control
how their code was used, you have to think that they were stupid to release
the code under the GPL. After all, the whole point of the GPL is to prohibit
such restrictions. The reason Linux is under the GPL is so that developers
*can't* put restrictions on how the package can be used. That's the "open"
in open source.

> If you choose to violate that, you are being unethical,
> pure and simple.  Or more to the point, you're being an asshole to a lot
> of hard-working people who have chosen to freely share their work with
> you.

	The person who tries to put other people's GPL'd works under his license
restrictions is the asshole. I have contributed code to the Linux kernel
under the GPL license (bonus points to anyone who can find my 25 lines of
code or so). It is nobody else's right to add code to my code and add usage
restrictions to it. The GPL expressly forbids this.

> Since they are the authors and you are not, their feelings about
> their softare are more important than yours.

	I'm just baffled. You don't seem to understand at all why the Linux kernel
is organized under the GPL. It's precisely so that some developers can't
hijack the project and encumber the growing source base with usage and
distribution restrictions.

> You may be able to screw
> them over and get away with it -- people do that sort of thing all the
> time -- but the fact that you may find a legal loophole doesn't make you
> any less of an abject asshole.

	The asshole is the person who thinks that they have the right to change the
express wishes of all the other contributors to the kernel who chose to
contribute to a project that operates under the GPL license. The GPL license
is about there being no restrictions on usage.

> In short:  Be honorable.

	I am. The hijackers are not.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Timothy Miller]

David Schwartz wrote:
>>But I have another point.  You are not dealing with a license here.  The
>>license is there to satisfy lawyers and make clear the INTENT of the
>>authors.  The keyword here is INTENT in that someone who has developed
>>something is telling you how they feel about the use of their work
>>which, under many circumstances, they could have chosen not to share
>>with you.  What you are dealing with is real people who have put an
>>incredible amount of time and effort into developing Linux.  Those
>>people, to whom you owe much respect for sharing their contributions,
>>have decided that their software should be used with certain
>>restrictions, that being the GPL.  If you abuse Linux, it is not the GPL
>>that you are insulting, but the people who developed Linux.
> 
> 
> 	In other words, information does not want to be free. You shouldn't use
> code the way you want to use it but the way the authors want you to use it.
> After all, they didn't have to give it to you if they didn't want to.

EXACTLY.  Fortunately in this case, the authors place few restrictions 
on your usage.  Indeed, the "restrictions" are more a matter of being 
nice to the people who made the stuff you're using.

You're stuck thinking about "law" and "rules".  I'm thinking instead 
about "honor".

One of the (unresolved) discussions I had with RMS involved the role of 
the author.  He believes that the user is more important than the 
author.  I believe the author is more important, because had the author 
not written what the user has, then the user would not have it! 
Actually, he partially agrees with me but does not feel that the author 
should be able to apply restrictions (IF the work is PUBLISHED).


> 	However, Richard Stallman does not agree with this view. It's his view that
> if the authors chose to give you the code, you can use it any way you want
> to, regardless of how the authors feel about that type of usage. This is why
> he created the GPL.

But it seems clear to me that the GPL places some very strong 
restrictions on your usage.  Those restrictions are that although your 
specific usage of the code doesn't matter, anything (published) that you 
derive from GPL work MUST also be published under GPL.

So my point is that when someone publishes something under GPL that you 
find useful, give respect to the person who wrote it by obeying the 
spirit of the license.  But it is not the license that is important as 
that the author chose to release his work under those terms.  People, 
not rules; honor, not law.

There are other licenses besides the GPL.  When authors release under 
those terms, you should respect those terms as well.  The thing that 
sets the GPL apart from, say, closed-source licensing is that the GPL is 
based on a system of honor, while proprietary systems can (and often do) 
become unfairly restrictive to the users.

> 
> 
>>So, the discussions about finding ways to make a non-GPL driver look
>>like a GPL driver and get away with it legally are all moot.  The reason
>>you should not violate this is because the architects of Linux do not
>>want you to.
> 
> 
> 	If you really believe that the Linux authors wished to continue to control
> how their code was used, you have to think that they were stupid to release
> the code under the GPL. After all, the whole point of the GPL is to prohibit
> such restrictions. The reason Linux is under the GPL is so that developers
> *can't* put restrictions on how the package can be used. That's the "open"
> in open source.

The restrictions they are applying are the only ones in the GPL.  It was 
smart for them to use the GPL because it is compatible with their 
wishes.  Follow the GPL not because the GPL is a copyright license, but 
because you are grateful for the efforts of those who published under GPL.

Furthermore, if someone publishes under GPL, and your insult them, they 
may become less willing to release under GPL, thereby limiting your 
ability to rip them off.

Also, just to be pedantic, because you mentioned RMS, "open source" and 
"free software" are not the same thing.

> 
>>If you choose to violate that, you are being unethical,
>>pure and simple.  Or more to the point, you're being an asshole to a lot
>>of hard-working people who have chosen to freely share their work with
>>you.
> 
> 
> 	The person who tries to put other people's GPL'd works under his license
> restrictions is the asshole. I have contributed code to the Linux kernel
> under the GPL license (bonus points to anyone who can find my 25 lines of
> code or so). It is nobody else's right to add code to my code and add usage
> restrictions to it. The GPL expressly forbids this.

And I completely agree with all of this.  I'm getting the feeling that 
you and I don't actually disagree on any of this.  :)

The point of divergence is with regard to modules and the symbol 
restrictions for non-GPL drivers.  Here is a gray area where the GPL may 
not apply really.  These gray areas are where ethics and honor must come 
into play.

> 
>>Since they are the authors and you are not, their feelings about
>>their softare are more important than yours.
> 
> 
> 	I'm just baffled. You don't seem to understand at all why the Linux kernel
> is organized under the GPL. It's precisely so that some developers can't
> hijack the project and encumber the growing source base with usage and
> distribution restrictions.

Which is precisely why those authors placed Linux under the GPL.  If 
they didn't like that, they would be working on proprietary OS's.  I'm 
saying that people should respect the authors by honoring the GPL.

> 
>>You may be able to screw
>>them over and get away with it -- people do that sort of thing all the
>>time -- but the fact that you may find a legal loophole doesn't make you
>>any less of an abject asshole.
> 
> 
> 	The asshole is the person who thinks that they have the right to change the
> express wishes of all the other contributors to the kernel who chose to
> contribute to a project that operates under the GPL license. The GPL license
> is about there being no restrictions on usage.

Well, aside from the "it must remain free clause" (which I interpret as 
an important restriction), I agree with you.  That is to say, the 
"contributor" who hijacks the work of other contributors is being 
dishonorable, because he is not respecting the other contributors.

> 
>>In short:  Be honorable.
> 
> 
> 	I am. The hijackers are not.

Agreed.

Re: People, not GPL [was: Re: Driver Model]

[Geert Uytterhoeven]

On Fri, 12 Sep 2003, Timothy Miller wrote:
> David Schwartz wrote:
> You're stuck thinking about "law" and "rules".  I'm thinking instead 
> about "honor".

Indeed. The GPL may not be perfect, but IMHO the GPL is the only way to make
sure everyone plays the game according to the rules. I.e. I give you something,
you can use it, but please play according to the rules of fair play.

BSD may look OK, but it doesn't protect against people who don't (want to) play
according to the rules.

> The point of divergence is with regard to modules and the symbol 
> restrictions for non-GPL drivers.  Here is a gray area where the GPL may 
> not apply really.  These gray areas are where ethics and honor must come 
> into play.

I consider binary-only modules some kind of loophole to avoid playing according
to the rules of fair play.

Unfortunately it seems to be almost impossible to design a license that forces
you to play according to the rules of fair play, and doesn't have any
loopholes or grey areas.

Gr{oetje,eeting}s,

						Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Sad, 2003-09-13 at 15:18, Geert Uytterhoeven wrote:
> Unfortunately it seems to be almost impossible to design a license that forces
> you to play according to the rules of fair play, and doesn't have any
> loopholes or grey areas.

Fair play is awfully hard to define. Fair use likewise. Currently almost
all countries legal systems have a clear notion of "derived work", and
copyright (unlike patents) extends no further. That limits how far the
GPL can extend, but its the same line in the sand (well fuzzy patch in
the sand in truth) that stops a lot of other things you wouldnt like.
Which and what modules count as derivative works is a lawyer question
and not it seems a trivial one.

Patents do extend beyond just the derived work and since Linux contains
patented material with rights granted for GPL use as per the GPL(but not
for non GPL use) there is a murky area around modules and patents - one
example of the issues that raises being RTLinux.

Folks using binary modules may also find third party software licenses
invalid (eg the OpenMotif one)

Re: People, not GPL [was: Re: Driver Model]

[James Clark]

In many ways the original intention of the debate was lost in the battle. 
There are a lot of zealots on both sides of the GPL debate - personally, my 
undertanding is that the GPL forbids usage restrictions, thus once you 
release code under the GPL you cannot control it. This seems especially 
important in a project, such as Linux, that has been collaboratively 
developmed by a whole community.

The original question was would a binary driver interface allow easier usage 
by 'normal' users with any compatible kernel (rather than specific versions) 
and perhaps simplify module development cycle? 

Would the performance hit involved be worth the potential 
compatibility/simplicity? 

A lot of people have, assumed that this is a call to arms for binary-only 
modules. This is not true although this type of change would make such things 
more common.

James


On Wednesday 10 Sep 2003 3:59 pm, Timothy Miller wrote:
> I'm still 1600 messages behind in reading the list, but I have spent
> enough time reading the discussion about GPL and drivers that I feel
> compelled to comment.  I don't intend to comment further because I don't
> want to contribute to a flamewar any more than this already will.  But I
> feel the need to defend those who contribute to Free Software against
> those who don't.
>
> The argument I have been reading has been centered around the idea of
> working around the GPL to support binary-only driver and various other
> things which are counter to the spirit of the GPL and Linux.  But
> someone who is trying to find a legal GPL loophole is not considering
> the root of the situation and that the GPL is an effect, not a cause.
>
> A point someone else made that I feel compelled to reiterate is that it
> is the nature of the Linux development model and the attitude of the
> developers which has made Linux what it is and has made you want to use it.
>
> But I have another point.  You are not dealing with a license here.  The
> license is there to satisfy lawyers and make clear the INTENT of the
> authors.  The keyword here is INTENT in that someone who has developed
> something is telling you how they feel about the use of their work
> which, under many circumstances, they could have chosen not to share
> with you.  What you are dealing with is real people who have put an
> incredible amount of time and effort into developing Linux.  Those
> people, to whom you owe much respect for sharing their contributions,
> have decided that their software should be used with certain
> restrictions, that being the GPL.  If you abuse Linux, it is not the GPL
> that you are insulting, but the people who developed Linux.
>
> The GPL_ONLY restriction for driver modules may seem unfair, but it is
> far from it.  There are both valid technical and philosophical reasons
> for working that way.  No one forces you to use Linux, and when you made
> the choice to use it, you are entering into a community with a specific
> philosophy.  You know that philosophy in advance, so when you discover
> that you have a restruction you don't like, you have no room to complain.
>
> As someone said, if you want to write drivers for a UNIX which does not
> have these restrictions, there are plenty of commercial UNIXes out there
> that you have to choose from.  The fact that they are perhaps less
> popular is one reason why Linux developers do not want to imitate them.
>
> So, the discussions about finding ways to make a non-GPL driver look
> like a GPL driver and get away with it legally are all moot.  The reason
> you should not violate this is because the architects of Linux do not
> want you to.  If you choose to violate that, you are being unethical,
> pure and simple.  Or more to the point, you're being an asshole to a lot
> of hard-working people who have chosen to freely share their work with
> you.  Since they are the authors and you are not, their feelings about
> their softare are more important than yours.  You may be able to screw
> them over and get away with it -- people do that sort of thing all the
> time -- but the fact that you may find a legal loophole doesn't make you
> any less of an abject asshole.
>
> In short:  Be honorable.
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/

RE: Driver Model

[Andre Hedrick]

Pascal,

SUPER HIGH FIVE!

You have made the obvious clear, and most will not even follow or listen.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Thu, 4 Sep 2003, Pascal Schmidt wrote:

> On Wed, 3 Sep 2003, David Schwartz wrote:
> 
> > If the GPL_ONLY stuff is a license enforcement scheme, the DMCA 
> > prohibits you from removing it.
> 
> -ENOTUSCITIZEN
> 
> > If the GPL_ONLY stuff is not a license enforcement scheme, nothing 
> > prohibits you from stamping your module GPL when it's not.
> 
> I'd say its up to the lawyers and judges to find out whether having
> MODULE_LICENSE("GPL") in a module means anything legally. It might
> mean "I promise this module is made from GPL source", but it might
> also mean nothing.
> 
> > However, the GPL (section 2b) prohibits you from imposing any
> > restrictions other than those in the GPL itself.
> 
> Section 2b) in the file COPYING in the root dir of the kernel source
> does not talk about restrictions. Are we talking about the same version
> of the GPL?
> 
> > The GPL contains no restrictions that
> > apply to mere use and the GPL_ONLY stuff affects use, so it can't be a
> > license restriction, hence there is no restriction to enforce.
> 
> The GPL doesn't even cover use of the "product". It covers modification
> and redistribution.
> 
> Well, it is still an open question whether kernel modules are derived
> works or not, especially since we don't have a stable kernel ABI and
> therefore modules have to use part of the kernel source (headers) and
> module writers have to study kernel code to write their modules (since
> there is no official complete documentation about functions in the 
> kernel).
> 
> If modules are derived works, then legally, following the GPL, they
> must be GPL too and GPL_ONLY is no problem but pointless.
> 
> Seems to me you could say GPL_ONLY is a way of the developer saying
> "I consider your stuff to be a derived work if you use this symbol".
> Ask a lawyer whether that's their decision to make. ;)
> 
> Apart from that, I fail to see how it is an addition restriction
> when you still have the right to remove all the GPL_ONLY stuff. After
> all, the kernel is GPLed work, so you have the right to remove
> things and distribute the result. How is it a real restriction when
> the license allows you to remove it?
> 
> -- 
> Ciao,
> Pascal
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: People, not GPL [was: Re: Driver Model]

[Pascal Schmidt]

On Wed, 10 Sep 2003 22:40:14 +0200, you wrote in linux.kernel:

> 	However, Richard Stallman does not agree with this view. It's his
> view that if the authors chose to give you the code, you can use it any
> way you want to, regardless of how the authors feel about that type of
> usage. This is why he created the GPL.

Use in any way you want to is the BSD license, not the GPL. The GPL
does restrict what you're allowed to do in order to keep the source
free...

-- 
Ciao,
Pascal

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> On Wed, 10 Sep 2003 22:40:14 +0200, you wrote in linux.kernel:

> > However, Richard Stallman does not agree with this view. It's his
> > view that if the authors chose to give you the code, you can use it any
> > way you want to, regardless of how the authors feel about that type of
> > usage. This is why he created the GPL.

> Use in any way you want to is the BSD license, not the GPL.

	Please show me one restriction on *use* in the GPL.

"Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope.  The act of
running the Program is not restricted, and the output from the Program
is covered only if its contents constitute a work based on the
Program (independent of having been made by running the Program).
Whether that is true depends on what the Program does."

	Licenses that place restrictions on usage are *not* open source licenses.

> The GPL
> does restrict what you're allowed to do in order to keep the source
> free...

	Yes, it restricts your ability to distribute and your ability to create
derived works if and only if you distribute those derived works. It places
no restrictions whatsoever on use. And since it requires distributors to
place no restrictions not in the GPL, distributors cannot place *any*
restrictions on usage either.

	DS

RE: People, not GPL [was: Re: Driver Model]

[Pascal Schmidt]

On Wed, 10 Sep 2003, David Schwartz wrote:

> Please show me one restriction on *use* in the GPL.

Well, you may not *use* GPL'd code to produce a derived work and
distribute it in binary form only. Use of the code, not use of
the product, sure.

-- 
Ciao,
Pascal

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> On Wed, 10 Sep 2003, David Schwartz wrote:

> > Please show me one restriction on *use* in the GPL.

> Well, you may not *use* GPL'd code to produce a derived work and
> distribute it in binary form only.

	That is a restriction on distribution, not use. You may use GPL'd code to
produce a derived work and you may use that derived work. The only
restrictions kick in when and if you distribute that derived work.

> Use of the code, not use of
> the product, sure.

	The GPL puts no restrictions on use. The GPL_ONLY stuff does.

	It's really this simple: The GPL says you can use the code for whatever you
want. It also says that if you want to distribute the GPL'd work or works
based on the GPL'd work, you may not impose any restrictions other than
those imposed by the GPL. It very specifically prohibits restrictions on
mere use or upon the mere creation of derived works. All of its restrictions
kick in on distribution.

	The GPL_ONLY stuff is an attempt to restrict use. There is nothing
inherently wrong with attempts to restrict use. One could argue that the
root permission check on 'umount' is a restriction on use. Surely the GPL
doesn't mean you can't have any usage restrictions at all.

	What it does mean is that such usage restrictions *cannot* be licensing
restrictions. In other words, it cannot be a license violation to remove
them or circumvent them. So long as nobody tries to claim the GPL_ONLY is a
license enforcement technique, there is no dispute.

	However, some people seem to be arguing that the GPL_ONLY symbols are in
fact a license enforcement technique. If that's true, then when they
distribute their code, they are putting additional restrictions not in the
GPL on it. That is a GPL violation.

	Other people who contributed to the Linux kernel relied upon the GPL
license to ensure that their code, and works derived from it, would be
available *without* use restrictions. Nobody has the right to turn around
and impose usage restrictions on the Linux kernel source code.

	The GPL prohibits the imposition of any licensing requirements other than
those contained in the GPL itself, all of which kick in only upon
distribution. It's really that simple. Anyone who distributes the Linux
kernel with a licensing restriction that kicks in on mere use is violating
the GPL.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Eric W. Biederman]

"David Schwartz" <davids@webmaster.com> writes:

> 	The GPL_ONLY stuff is an attempt to restrict use. There is nothing
> inherently wrong with attempts to restrict use. One could argue that the
> root permission check on 'umount' is a restriction on use. Surely the GPL
> doesn't mean you can't have any usage restrictions at all.

No the GPL_ONLY stuff is an attempt to document that there is no conceivable
way that using a given symbol does not create a derived work.  

If you use an unmodified kernel it is only a one liner to ensure it does
not complain about your code.  So this only shows up as a real
impediment when code that uses the symbol is distributed.

Beyond which copying code into the kernel is when this is checked so this is
a valid place to check things.  There is a strong tying between using
programs and copying them into memory.  And that copying is the
justification for most usage restrictions even in commercial software.

The code is also quite a small nit that really should not affect
anything.

Eric

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

	Combined responses.

> No the GPL_ONLY stuff is an attempt to document that there is no
> conceivable
> way that using a given symbol does not create a derived work.

	and:

> > 	The GPL puts no restrictions on use. The GPL_ONLY stuff does.

> No it doesn't. It allows people to track whether you get support or not.
> It prevents nothing.

	So long as it's not an attempt to use the GPL license as a way to restrict
use, there's no problem. There is no license restriction or enforcement
scheme here that one could possibly circumvent because the GPL_ONLY stuff is
not a license enforcement scheme. It cannot be a license enforcement scheme
because there are no license requirements that kick in upon any act other
than distribution.

	The GPL_ONLY stuff does in fact restrict use. However, this is a
restriction anyone is free to remove or circumvent if it bothers them. This
isn't a dishonorable act because nobody intended this mechanism to restrict
use -- I hope.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Mike Fedyk]

On Thu, Sep 11, 2003 at 11:39:33AM -0700, David Schwartz wrote:
> 	The GPL_ONLY stuff does in fact restrict use. However, this is a
> restriction anyone is free to remove or circumvent if it bothers them. This
> isn't a dishonorable act because nobody intended this mechanism to restrict
> use -- I hope.

If effect it does affect use because of lack of support.  But personally I
agree with that...

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On Thu, 11 Sep 2003, Mike Fedyk wrote:

> On Thu, Sep 11, 2003 at 11:39:33AM -0700, David Schwartz wrote:
> > 	The GPL_ONLY stuff does in fact restrict use. However, this is a
> > restriction anyone is free to remove or circumvent if it bothers them. This
> > isn't a dishonorable act because nobody intended this mechanism to restrict
> > use -- I hope.
> 
> If effect it does affect use because of lack of support.  But personally I
> agree with that...

I disagree that use == support.  Whatever happen to RTFM ?
Wait I forgot TFM became GPL_ONLY so you can not read it freely anymore.

Try getting support from RedHat without a support contract, LOL.

Try getting support from XMission for anything without a support contract.

Cheers,

Andre

Re: People, not GPL [was: Re: Driver Model]

[Erik Andersen]

On Sat Sep 13, 2003 at 10:52:20AM -0700, Andre Hedrick wrote:
> Try getting support from XMission for anything without a support contract.

News flash, This just in!  XMission is an ISP and Eric Biederman
does not work for them
    http://www.xmission.com/

Eric Biederman, myself, and pretty much anybody around here that
wants a decent always-up internet connection run by folks with a
clue uses them.  They do not sell "support contracts".  As with
any company, they support their customers and nobody else.  And
their support is excellent.  Want to learn to install Linux?
They'll even show you how http://www.xmission.com/lif3/index.html

If you expect companies to support non-customers I have some fine
ocean-front property in New Mexico to sell you.

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On 11 Sep 2003, Eric W. Biederman wrote:

> "David Schwartz" <davids@webmaster.com> writes:
> 
> > 	The GPL_ONLY stuff is an attempt to restrict use. There is nothing
> > inherently wrong with attempts to restrict use. One could argue that the
> > root permission check on 'umount' is a restriction on use. Surely the GPL
> > doesn't mean you can't have any usage restrictions at all.
> 
> No the GPL_ONLY stuff is an attempt to document that there is no conceivable
> way that using a given symbol does not create a derived work.  

Bzzit ... GPL_ONLY stuff is an attempt to retrict usage by removing access
to the unprotectable API.  And for anyone claiming there is not API to
protect, the kernel source is the manual to the API.  The foolish intent
and design to hide the API has caused the kernel itself to become the
manual.

This is even obvious to people, like myself, who are not lawyers.

Andre

Re: People, not GPL [was: Re: Driver Model]

[Sean Estabrooks]

On Sat, 13 Sep 2003 10:34:12 -0700 (PDT)
Andre Hedrick <andre@linux-ide.org> wrote:
> 
> On 11 Sep 2003, Eric W. Biederman wrote:
> 
> > "David Schwartz" <davids@webmaster.com> writes:
> > 
> > > 	The GPL_ONLY stuff is an attempt to restrict use. There is nothing
> > > inherently wrong with attempts to restrict use. One could argue that the
> > > root permission check on 'umount' is a restriction on use. Surely the GPL
> > > doesn't mean you can't have any usage restrictions at all.
> > 
> > No the GPL_ONLY stuff is an attempt to document that there is no conceivable
> > way that using a given symbol does not create a derived work.  
> 
> Bzzit ... GPL_ONLY stuff is an attempt to retrict usage by removing access
> to the unprotectable API.  And for anyone claiming there is not API to
> protect, the kernel source is the manual to the API.  The foolish intent
> and design to hide the API has caused the kernel itself to become the
> manual.
> 
> This is even obvious to people, like myself, who are not lawyers.

Andre,

You seem to be mixing two forms of "use"; the right to use GPL'd source
and the right to use an operational Linux kernel.    

You are free to take the source and do whatever you want with it, restricted
only by the terms of the GPL.  Other people have the right to do the same.
There is nothing in the GPL that says how a program must behave when 
_executing_.  Therefore, if you don't like how the resulting executable 
operates, _tough_.   Your option, and the _freedom_ provided by the GPL, 
is to fork the source. 

No different than a kernel provided by the Church of Holy Computation 
which refuses to operate on Sunday.   You may disagree with the 
restriction, but surely they are free to add such a restriction to their
kernel source.

As it stands, you are complaining about a _runtime_ restriction;  ignoring 
your ability to change the source.   By insisting that other people remove 
checks for license type, you are trying to restrict  what others do with 
_their_ source.  By design, the kernel is already full of runtime restrictions. 
There is nothing special about a runtime limitation imposed on modules 
lacking the GPL symbol.   Just to drive the point home,  please consider that there is nothing wrong with a kernel that refuses 
to load modules that _DO_ contain the GPL symbol.

In short, fork off...
Sean

RE: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Iau, 2003-09-11 at 02:35, David Schwartz wrote:
> 	The GPL puts no restrictions on use. The GPL_ONLY stuff does.

No it doesn't. It allows people to track whether you get support or not.
It prevents nothing.

RE: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On Thu, 11 Sep 2003, Alan Cox wrote:

> On Iau, 2003-09-11 at 02:35, David Schwartz wrote:
> > 	The GPL puts no restrictions on use. The GPL_ONLY stuff does.
> 
> No it doesn't. It allows people to track whether you get support or not.
> It prevents nothing.

Then allow the usage and functionality of the SYMBOLS and not "steal" them
from usage.  If you say thay can be used without support.

PS it is "The Troll" again.

Just execute the reject of the License and have the means to reject
support.  You are so full of it to say it is for rejecting support when
anyone knows it removes usage of the SYMBOL == API.

Cheers,

Andre

Re: People, not GPL [was: Re: Driver Model]

[Timothy Miller]

David Schwartz wrote:

> 	However, some people seem to be arguing that the GPL_ONLY symbols are in
> fact a license enforcement technique. If that's true, then when they
> distribute their code, they are putting additional restrictions not in the
> GPL on it. That is a GPL violation.

Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.

Binary-only modules are inherently untrustworthy (no open code review) 
and undebuggable.  It is therefore of technical merit to restrict both 
what they can access in the kernel (GPL_ONLY) and limit how much kernel 
developers should have to tolerate when they're involved.

But beyond this, there are some social issues.  If someone finds a way 
to work around this mechanism, they are breaking things to everyone 
else's detriment.  For a commercial entity to violate the GPL_ONLY 
barrier is an insult to kernel developers AND to their customers who 
will have trouble getting problems solved.

So, if a company works around GPL_ONLY, are they violating the GPL 
license?  Probably not.  Does that make it OKAY?  Probably not.

This is like finding a way to give a user space program access to kernel 
resources.  There are barriers put in place for a REASON because people 
make mistakes when they write software.  If no one did, we wouldn't have 
any need for memory protection, would we.

Re: People, not GPL [was: Re: Driver Model]

[Matt D. Robinson]

On Fri, 12 Sep 2003, Timothy Miller wrote:
|>David Schwartz wrote:
|>
|>> 	However, some people seem to be arguing that the GPL_ONLY symbols are in
|>> fact a license enforcement technique. If that's true, then when they
|>> distribute their code, they are putting additional restrictions not in the
|>> GPL on it. That is a GPL violation.
|>
|>Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.
|>
|>Binary-only modules are inherently untrustworthy (no open code review) 
|>and undebuggable.  It is therefore of technical merit to restrict both 
|>what they can access in the kernel (GPL_ONLY) and limit how much kernel 
|>developers should have to tolerate when they're involved.
|>
|>But beyond this, there are some social issues.  If someone finds a way 
|>to work around this mechanism, they are breaking things to everyone 
|>else's detriment.  For a commercial entity to violate the GPL_ONLY 
|>barrier is an insult to kernel developers AND to their customers who 
|>will have trouble getting problems solved.

I think you are all missing the point.  This isn't a Linux kernel
problem.  This is a customer and distributor problem.  If Red Hat,
Mandrake, SuSE, etc., choose to remove these GPL_ONLY() barriers
and release the new "free" code under GPL, they're entitled.  Heck,
they can even add in new ones in their own kernels.

I'm surprised you are even discussing this issue on this list.

If third party developers are restricted based on GPL_ONLY(),
they can either (A) release their product under other OSes or Linux
distributions without the GPL_ONLY() restrictions, (B) modify their
product to be more GPL-friendly, or (C) avoid Linux support entirely.

It ultimately means either more support for third party products with
one distribution over another, which may or may not be financially
beneficial to that distribution, or it means that some Linux
distributors continue to supress supporting new third party devices
that don't believe in the GPL.  Either way, it's a distribution
decision based on open-source beliefs and how that balances with
financial benefit (both of which would matter to me as a stockholder).

This really has nothing to do with the Linux kernel code itself.
Very few customers that want third party device support will go to
vger to roll their own kernel -- they'll go to a Linux distributor,
get them to include device support and pay for it (or start with a
Linux distribution and roll their own).

So include GPL_ONLY(), don't include GPL_ONLY(), whatever.  If you
don't like it, Mr. Customer, find a Linux distributor that will
fix the problem for you.

If you want to create a new driver model that supports third party
devices without regard to their GPL status, make a patch.  Who
knows, maybe some distribution will actually start using it.  But
don't even think about bugging this list to get it included by
default (too many GPL purists out there).

--Matt

Re: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Gwe, 2003-09-12 at 22:47, Matt D. Robinson wrote:
> So include GPL_ONLY(), don't include GPL_ONLY(), whatever.  If you
> don't like it, Mr. Customer, find a Linux distributor that will
> fix the problem for you.

Linux vendors have already recieved, and decided to act on cease and
desist letters involving adding hooks (ie EXPORT_SYMBOL stuff) for non
free modules that were not in the base distro. I think that speaks for 
part of the legal view.

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> On Gwe, 2003-09-12 at 22:47, Matt D. Robinson wrote:

> > So include GPL_ONLY(), don't include GPL_ONLY(), whatever.  If you
> > don't like it, Mr. Customer, find a Linux distributor that will
> > fix the problem for you.

> Linux vendors have already recieved, and decided to act on cease and
> desist letters involving adding hooks (ie EXPORT_SYMBOL stuff) for non
> free modules that were not in the base distro. I think that speaks for
> part of the legal view.

	Who is sending these letters? Who has no respect for the GPL and seeks to
add additional restrictions? IMO, these letters are almost as bad as the SCO
letters. Nobody has any business putting additional licensing restrictions
on code was placed under the GPL.

	DS

RE: People, not GPL [was: Re: Driver Model]

[Matt D. Robinson]

On Fri, 12 Sep 2003, David Schwartz wrote:
|>> On Gwe, 2003-09-12 at 22:47, Matt D. Robinson wrote:
|>> > So include GPL_ONLY(), don't include GPL_ONLY(), whatever.  If you
|>> > don't like it, Mr. Customer, find a Linux distributor that will
|>> > fix the problem for you.
|>
|>> Linux vendors have already recieved, and decided to act on cease and
|>> desist letters involving adding hooks (ie EXPORT_SYMBOL stuff) for non
|>> free modules that were not in the base distro. I think that speaks for
|>> part of the legal view.
|>
|>	Who is sending these letters? Who has no respect for the GPL and seeks to
|>add additional restrictions? IMO, these letters are almost as bad as the SCO
|>letters. Nobody has any business putting additional licensing restrictions
|>on code was placed under the GPL.
|>
|>	DS

FUD, Inc.  If these really did exist, they would be more commonly
known and I'm certain would affect customer choice.  IMHO, I think
this is the vendor's (and I specifically think of Red Hat) way of
excusing themselves from having to work with third party vendors
in order to cover themselves legally with groups like the FSF.
Whether that's true or not, or if there is some other developer
contingent at these distribution houses that is preventing the
EXPORT_SYMBOL{,GPL}() changes due to personal bias, I don't know.
And I doubt we'll ever know.

These letters are like WMD -- they exist, you know they do, but you
just can't find them. :)

--Matt

RE: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Sad, 2003-09-13 at 01:29, Matt D. Robinson wrote:
> |>> Linux vendors have already recieved, and decided to act on cease and
> |>> desist letters involving adding hooks (ie EXPORT_SYMBOL stuff) for non
> |>> free modules that were not in the base distro. I think that speaks for
> |>> part of the legal view.

> FUD, Inc.  If these really did exist, they would be more commonly

It has happened. If you want to call me a liar feel free. Its your right
in a free society. It doesn't change the facts however.

Adding hooks for a proprietary module is a dubious practice. Doing it as
part of a two way arrangement with a proprietary vendor producing a non
free module is extremely unwise indeed.

Re: People, not GPL [was: Re: Driver Model]

[jw schultz]

On Fri, Sep 12, 2003 at 04:26:19PM -0700, David Schwartz wrote:
> 
> > On Gwe, 2003-09-12 at 22:47, Matt D. Robinson wrote:
> 
> > > So include GPL_ONLY(), don't include GPL_ONLY(), whatever.  If you
> > > don't like it, Mr. Customer, find a Linux distributor that will
> > > fix the problem for you.
> 
> > Linux vendors have already recieved, and decided to act on cease and
> > desist letters involving adding hooks (ie EXPORT_SYMBOL stuff) for non
> > free modules that were not in the base distro. I think that speaks for
> > part of the legal view.
> 
> 	Who is sending these letters? Who has no respect for the GPL and seeks to
> add additional restrictions? IMO, these letters are almost as bad as the SCO
> letters. Nobody has any business putting additional licensing restrictions
> on code was placed under the GPL.

I don't know who but i can say that Linus or anyone he
assigns can do so based not on GPL but on trademark.  That is
to say that while it wouldn't be a violation of the
copyright license to distribute such modified code but it
would be within the rights of the trademark holder (Linus)
to refuse them the _privilege_ of calling the modified code
Linux.  Not only would it be within his rights but it is
necessary for him to define and enforce restrictions on what
may be called Linux or he would lose that authority.

Just like you can make a copy of Red Hat's distribution but
without Red Hat Inc's explicit permission you cannot call it Red
Hat or Maroon Chapeau, etc.  For that matter the Open group
could, in theory, deny SCO the privilege of of calling their
own products UNIX.

-- 
________________________________________________________________
	J.W. Schultz            Pegasystems Technologies
	email address:		jw@pegasys.ws

		Remember Cernan and Schmitt

RE: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Sad, 2003-09-13 at 00:26, David Schwartz wrote:
> > On Gwe, 2003-09-12 at 22:47, Matt D. Robinson wrote:
> 	Who is sending these letters? Who has no respect for the GPL and seeks to
> add additional restrictions? IMO, these letters are almost as bad as the SCO
> letters. Nobody has any business putting additional licensing restrictions
> on code was placed under the GPL.

Nobody I've talked to (legal folks included) seems to think they are 
additional restrictions, merely people enforcing their copyright rights
against folks working as a group to conspire to create a derivative work
prohibited by the GPL.

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On Fri, 12 Sep 2003, Timothy Miller wrote:

> 
> 
> David Schwartz wrote:
> 
> > 	However, some people seem to be arguing that the GPL_ONLY symbols are in
> > fact a license enforcement technique. If that's true, then when they
> > distribute their code, they are putting additional restrictions not in the
> > GPL on it. That is a GPL violation.
> 
> Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.

Technical if and only if it did not remove the functionality of the
symbol when called.

Since it remove the ability to call and it work, creates a restriction of
usaged.  Usage being, one calls the function and it works.

I have not decided yet to expose one of the grossest example of API thieft
yet, but will do so in time.

> Binary-only modules are inherently untrustworthy (no open code review) 
> and undebuggable.  It is therefore of technical merit to restrict both 
> what they can access in the kernel (GPL_ONLY) and limit how much kernel 
> developers should have to tolerate when they're involved.

Who cares, it reports a tainted and the community says, thank you have a
nice day.

> But beyond this, there are some social issues.  If someone finds a way 
> to work around this mechanism, they are breaking things to everyone 
> else's detriment.  For a commercial entity to violate the GPL_ONLY 
> barrier is an insult to kernel developers AND to their customers who 
> will have trouble getting problems solved.

So when is the last time a business sent it problems to LKML to be
solved?  If I were a customer of such a company, I would be gone.

> So, if a company works around GPL_ONLY, are they violating the GPL 
> license?  Probably not.  Does that make it OKAY?  Probably not.

GPL_ONLY violate usage, thus it violate GPL.
Not a hard concept.
Also what if a company produces a "private gpl" product?
Open Source to customers but not to the world?

Nan, nobody would do that silly idea.

> This is like finding a way to give a user space program access to kernel 
> resources.  There are barriers put in place for a REASON because people 
> make mistakes when they write software.  If no one did, we wouldn't have 
> any need for memory protection, would we.

Stop, the laughing hurts "memory protection" "vm" ...

Cheers,

Andre

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

SYMBOL in question "dequeue_signal"

*************************************

pwd :: /usr/src/linux-2.4.18-27/kernel/signal.c

/*
 * Dequeue a signal and return the element to the caller, which is
 * expected to free it.
 *
 * All callers must be holding current->sigmask_lock.
 */

int
dequeue_signal(sigset_t *mask, siginfo_t *info)
{
        int sig = 0;

#if DEBUG_SIG
printk("SIG dequeue (%s:%d): %d ", current->comm, current->pid,
        signal_pending(current));
#endif

        sig = next_signal(current, mask);
        if (sig) {
                if (current->notifier) {
                        if (sigismember(current->notifier_mask, sig)) {
                                if (!(current->notifier)(current->notifier_data)) {
                                        current->sigpending = 0;
                                        return 0;
                                }
                        }
                }

                if (!collect_signal(sig, &current->pending, info))
                        sig = 0;

                /* XXX: Once POSIX.1b timers are in, if si_code == SI_TIMER,
                   we need to xchg out the timer overrun values.  */
        }
        recalc_sigpending(current);

#if DEBUG_SIG
printk(" %d -> %d\n", signal_pending(current), sig);
#endif

        return sig;
}

EXPORT_SYMBOL(dequeue_signal);
EXPORT_SYMBOL(flush_signals);

*************************************

pwd :: /usr/src/linux-2.4.20-18.9/kernel/signal.c

static int __dequeue_signal(struct sigpending *pending, sigset_t *mask,
                        siginfo_t *info)
{
        int sig = 0;

        sig = next_signal(pending, mask);
        if (sig) {
                if (current->notifier) {
                        if (sigismember(current->notifier_mask, sig)) {
                                if (!(current->notifier)(current->notifier_data)) {
                                        current->sigpending = 0;
                                        return 0;
                                }
                        }
                }

                if (!collect_signal(sig, pending, info))
                        sig = 0;

                /* XXX: Once POSIX.1b timers are in, if si_code == SI_TIMER,
                   we need to xchg out the timer overrun values.  */
        }
        recalc_sigpending();

        return sig;
}

/*
 * Dequeue a signal and return the element to the caller, which is
 * expected to free it.
 *
 * All callers have to hold the siglock.
 */
int dequeue_signal(sigset_t *mask, siginfo_t *info)
{
        int signr = __dequeue_signal(&current->pending, mask, info);
        if (!signr)
                signr = __dequeue_signal(&current->signal->shared_pending,
                                         mask, info);
        return signr;
}


EXPORT_SYMBOL(recalc_sigpending);
EXPORT_SYMBOL_GPL(dequeue_signal);
EXPORT_SYMBOL(flush_signals);

*************************************


Now it is totally clear this is an attempt to strip and remove
functionality of the API, period.

But this is to obvious to miss.

Regards,

Andre Hedrick
LAD Storage Consulting Group

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

more examples of SYMBOL rip offs.

2.6.0-test5

./kernel/exit.c

static inline void __exit_fs(struct task_struct *tsk)
{
        struct fs_struct * fs = tsk->fs;

        if (fs) {
                task_lock(tsk);
                tsk->fs = NULL;
                task_unlock(tsk);
                __put_fs_struct(fs);
        }
}

void exit_fs(struct task_struct *tsk)
{
        __exit_fs(tsk);
}

./kernel/fork.c

static inline struct fs_struct *__copy_fs_struct(struct fs_struct *old)
{
        struct fs_struct *fs = kmem_cache_alloc(fs_cachep, GFP_KERNEL);
        /* We don't need to lock fs - think why ;-) */
        if (fs) {
                atomic_set(&fs->count, 1);
                fs->lock = RW_LOCK_UNLOCKED;
                fs->umask = old->umask;
                read_lock(&old->lock);
                fs->rootmnt = mntget(old->rootmnt);
                fs->root = dget(old->root);
                fs->pwdmnt = mntget(old->pwdmnt);
                fs->pwd = dget(old->pwd);
                if (old->altroot) {
                        fs->altrootmnt = mntget(old->altrootmnt);
                        fs->altroot = dget(old->altroot);
                } else {
                        fs->altrootmnt = NULL;
                        fs->altroot = NULL;
                }
                read_unlock(&old->lock);
        }
        return fs;
}

struct fs_struct *copy_fs_struct(struct fs_struct *old)
{
        return __copy_fs_struct(old);
}

EXPORT_SYMBOL_GPL(exit_fs);
EXPORT_SYMBOL_GPL(copy_fs_struct);

This is what the issue is about!
Taking away the functionality of the API.

So much for choice anymore.



Andre Hedrick
LAD Storage Consulting Group

Re: People, not GPL [was: Re: Driver Model]

[Erik Andersen]

On Fri Sep 12, 2003 at 04:58:03PM -0400, Timothy Miller wrote:
> 
> 
> David Schwartz wrote:
> 
> >	However, some people seem to be arguing that the GPL_ONLY symbols 
> >	are in
> >fact a license enforcement technique. If that's true, then when they
> >distribute their code, they are putting additional restrictions not in the
> >GPL on it. That is a GPL violation.
> 
> Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.

I'll go even farther, and say that one might call the GPL_ONLY
symbols an "effective technological measure" that "effectively
controls access to a work" and "effectively protects a right of a
copyright owner ... in the ordinary course of its operation....".
Bypassing GPL_ONLY symbols, as recently advocated by David
Schwartz and Andre Hedrick, could be considered circumvention of
an effective technological measure.  Remember Dmitry Sklyarov,

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

Please Erik,

Go have your "DADDY" write another legal letter for you and send it my
way.  I will be happy to shove it down your pie hole.

There is a difference in blanket theift of the kernel API and any court
will point out that "fair usage" is proper.

PS how is your IRS media forensics gig going now that you do not have me
on your side to help you do the dirty work of sneaking around people's
hard drives.  I was smart enough to never teach you what I knew you could
never learn.

Later!

Andre Hedrick
LAD Storage Consulting Group

On Sat, 13 Sep 2003, Erik Andersen wrote:

> On Fri Sep 12, 2003 at 04:58:03PM -0400, Timothy Miller wrote:
> > 
> > 
> > David Schwartz wrote:
> > 
> > >	However, some people seem to be arguing that the GPL_ONLY symbols 
> > >	are in
> > >fact a license enforcement technique. If that's true, then when they
> > >distribute their code, they are putting additional restrictions not in the
> > >GPL on it. That is a GPL violation.
> > 
> > Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.
> 
> I'll go even farther, and say that one might call the GPL_ONLY
> symbols an "effective technological measure" that "effectively
> controls access to a work" and "effectively protects a right of a
> copyright owner ... in the ordinary course of its operation....".
> Bypassing GPL_ONLY symbols, as recently advocated by David
> Schwartz and Andre Hedrick, could be considered circumvention of
> an effective technological measure.  Remember Dmitry Sklyarov,
> 
>  -Erik
> 
> --
> Erik B. Andersen             http://codepoet-consulting.com/
> --This message was written using 73% post-consumer electrons--
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: People, not GPL [was: Re: Driver Model]

[Erik Andersen]

On Sat Sep 13, 2003 at 07:40:52PM -0700, Andre Hedrick wrote:
> 
> Please Erik,
> 
> Go have your "DADDY" write another legal letter for you and send it my
> way.  I will be happy to shove it down your pie hole.

Your intellect is truly dizzying.  I am left stunned by your
carefully reasoned arguments.

> There is a difference in blanket theift of the kernel API and any court
> will point out that "fair usage" is proper.

"fair usage" != "derivitive works" as any court will point out.

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On Sat, 13 Sep 2003, Erik Andersen wrote:

> On Sat Sep 13, 2003 at 07:40:52PM -0700, Andre Hedrick wrote:
> > 
> > Please Erik,
> > 
> > Go have your "DADDY" write another legal letter for you and send it my
> > way.  I will be happy to shove it down your pie hole.
> 
> Your intellect is truly dizzying.  I am left stunned by your
> carefully reasoned arguments.
> 
> > There is a difference in blanket theift of the kernel API and any court
> > will point out that "fair usage" is proper.
> 
> "fair usage" != "derivitive works" as any court will point out.
> 
>  -Erik

Wow ... Does "Original Work" have meaning?

Does an "Original Work" using only the standard kernel API headers to
interface mean it is a derived work?  You better go find a new lawyer.

"fair usage" of .h files as the API is standard.

Using any .c or kernel C code is a NO NO.

Later.

Andre Hedrick
LAD Storage Consulting Group

Re: People, not GPL [was: Re: Driver Model]

[Erik Andersen]

On Sat Sep 13, 2003 at 08:36:36PM -0700, Andre Hedrick wrote:
> Wow ... Does "Original Work" have meaning?
> 
> Does an "Original Work" using only the standard kernel API headers to
> interface mean it is a derived work?  You better go find a new lawyer.

You seem to be somewhat confused as to who needs a lawyer.  I'm
not the one asking this question.   I am also not the one trying
to make a closed source binary only product that runs within the
context of the Linux kernel, and then complaining that the GPL
wackos are ruining my business...  It seems to be that doing such
a thing would be a really stupid business model.

As I recall it is the One True(tm) iSCSI stack you are working
on, right?

> "fair usage" of .h files as the API is standard.
> 
> Using any .c or kernel C code is a NO NO.

I invite you to read the COPYING file included in each and every
kernel tarball.  There is exactly ONE exception granted in the
linux kernel copyright:

    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".

All the noise in the world about other exceptions is precisely
that, since the license granting use of the Linux kernel does
not contain any additional provisions.

Anything that can be identified as a "user program" that "use[s]
kernel services by normal system calls" is, by virtue of the above
license grant, doing so with permission and is therefore within
its rights.  So you can make all the closed source user space
only One True(tm) iSCSI stacks you want.

Anything that is not a "user program" (and I think everyone can
agree a kernel module is not a "user program") is therefore a
derivitive work.

Anything that is linked into the kernel (and I think everyone can
agree a kernel module is linked into the kernel) and is therefore
interfacing with kernel internals, rather than using "kernel
services by normal system calls" is therefore a derivitive work.

Laugh at people, mock people, rant, rave, wantever you want.
When you are done making noise, please have your laywer explain
how a closed source binary only product that runs within the
context of the Linux kernel is not a derivitive work, per the
very definition given in the kernel COPYING file that grants you
your limited rights for copying, distribution and modification,

 -Erik

--
Erik B. Andersen             http://codepoet-consulting.com/
--This message was written using 73% post-consumer electrons--

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> I invite you to read the COPYING file included in each and every
> kernel tarball.  There is exactly ONE exception granted in the
> linux kernel copyright:
>
>     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".

	You are mistaken about what this is. This is not an exception being
granted. Nobody has the authority to grant GPL exceptions unless they are
the sole author and the Linux kernel does not have a sole author.

> All the noise in the world about other exceptions is precisely
> that, since the license granting use of the Linux kernel does
> not contain any additional provisions.

	This is not a license provision, it's a license interpretation. If you can
show me why the interpretation doesn't apply to kernel modules, please do.

> Anything that can be identified as a "user program" that "use[s]
> kernel services by normal system calls" is, by virtue of the above
> license grant, doing so with permission and is therefore within
> its rights.  So you can make all the closed source user space
> only One True(tm) iSCSI stacks you want.

	Except it's not a license grant. If it is a license grant, please tell me
who had the authority to grant it.

> Anything that is not a "user program" (and I think everyone can
> agree a kernel module is not a "user program") is therefore a
> derivitive work.

	Hahahahhahaha! Oh, that's too funny. "All dogs are mammals" means things
that aren't dogs aren't mammals? My pencil isn't a "user program". Is it a
derivative work of the Linux kernel in your opinion?

> Anything that is linked into the kernel (and I think everyone can
> agree a kernel module is linked into the kernel) and is therefore
> interfacing with kernel internals, rather than using "kernel
> services by normal system calls" is therefore a derivitive work.
>
> Laugh at people, mock people, rant, rave, wantever you want.
> When you are done making noise, please have your laywer explain
> how a closed source binary only product that runs within the
> context of the Linux kernel is not a derivitive work, per the
> very definition given in the kernel COPYING file that grants you
> your limited rights for copying, distribution and modification,

	The same way a user program that runs within the context of the Linux
operating system is not a derivative work. A defined interface creates a
license boundary. If you can draw a line between the two works, neither is
derivative of the other.

	And you are still ignoring the entire point, which is that it doesn't
matter whether or not such modules are derivative works. The GPL puts no
restrictions whatsoever on the creation or usage of derivative works that
are not distributed. The GPL does not say that you cannot make derivative
works unless you GPL them.

	So the restriction GPL_ONLY places is *not* a restriction that exists in
the GPL which is the *only* license. Hence it does not enforce a license
restriction. It really is that simple.

	And when it comes to respecting the author's wishes, I respect the wishes
of the many people who contributed to the Linux kernel expecting the GPL to
ensure that usage remained unrestricted. I don't not respect the wishes of
the people who try to impose usage restrictions on other people's GPL'd code
and I thoroughly disrespect those who would wield the DMCA to restrict
modifications to code placed under the GPL.

	DS

Re: People, not GPL [was: Re: Driver Model]

[David S. Miller]

On Sat, 13 Sep 2003 19:40:52 -0700 (PDT)
Andre Hedrick <andre@linux-ide.org> wrote:

> Go have your "DADDY" write another legal letter for you and send it my
> way.  I will be happy to shove it down your pie hole.

Umm, that's enough of this thread ok?

Take it to gnu.misc.discuss or whipme.beatme.i.like.it

And this is a one time warning, open defiance and continuing this
thread will cost anyone their access to vger.  This kind of
"let's abuse people" threads cannot continue unchecked, and Andre
you are on terribly thin ice ok?

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

Erik:

One more thing you forgot!

Before you run around ranting about DMCA and CPRM and making suttle legal
threats.  Everyone here knows who put it on the line to gain control of
technology to prevent Content Protection Recordable Media from covering
the storage industry wide.  That is correct it was me.  I was the one and
only person ever to stand up and have a position to help everyone here to
take control of technology.  While all of my hard work was destroyed by
the brain-dead of the EFF by causing it to go underground.

So lets be real clear about your wild swings of DMCA, because the irony
will be back to haunt.  Threaten to strike me down with DMCA, what a child
you have become.

So this is where it goes ... Mighty Linux kills usage for business with
lawsuits like the MPAA and RIAA.

GPL == DMCA, this is the brush you swing paint well dude.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Sat, 13 Sep 2003, Erik Andersen wrote:

> On Fri Sep 12, 2003 at 04:58:03PM -0400, Timothy Miller wrote:
> > 
> > 
> > David Schwartz wrote:
> > 
> > >	However, some people seem to be arguing that the GPL_ONLY symbols 
> > >	are in
> > >fact a license enforcement technique. If that's true, then when they
> > >distribute their code, they are putting additional restrictions not in the
> > >GPL on it. That is a GPL violation.
> > 
> > Agreed.  GPL_ONLY is not a license restriction.  It is a technical issue.
> 
> I'll go even farther, and say that one might call the GPL_ONLY
> symbols an "effective technological measure" that "effectively
> controls access to a work" and "effectively protects a right of a
> copyright owner ... in the ordinary course of its operation....".
> Bypassing GPL_ONLY symbols, as recently advocated by David
> Schwartz and Andre Hedrick, could be considered circumvention of
> an effective technological measure.  Remember Dmitry Sklyarov,
> 
>  -Erik
> 
> --
> Erik B. Andersen             http://codepoet-consulting.com/
> --This message was written using 73% post-consumer electrons--
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> David Schwartz wrote:

> But beyond this, there are some social issues.  If someone finds a way
> to work around this mechanism, they are breaking things to everyone
> else's detriment.  For a commercial entity to violate the GPL_ONLY
> barrier is an insult to kernel developers AND to their customers who
> will have trouble getting problems solved.

	While I understand this point of view, I do not share it. People
contributed to the Linux kernel project largely because it *was* an open
process. Nobody has the right to take offense when it's used for a different
reason than they had intended. If you want control over how your code is
used and modified, *don't* GPL it. If you're going to take offense when
people remove restrictions you impose, regardless of how much you like the
restrictions, *don't*' GPL it. It's that simple.

> So, if a company works around GPL_ONLY, are they violating the GPL
> license?  Probably not.  Does that make it OKAY?  Probably not.

	What's not okay is trying to inject your own rules on how GPL'd software
can be used. That's perhaps tolerable if you're the sole author. It's
condemnable when you're one among many.

> This is like finding a way to give a user space program access to kernel
> resources.  There are barriers put in place for a REASON because people
> make mistakes when they write software.  If no one did, we wouldn't have
> any need for memory protection, would we.

	Yet there is a project that removes all of these boundaries in the name of
improved performance for trusted applications. These are engineering
trade-offs and one of the good things about the GPL is that I'm not stuck,
by law or custom, when your engineering trade-offs if I don't think they
apply to me.

	DS

Re: People, not GPL [was: Re: Driver Model]

[David S. Miller]

Do not even _THINK_ about bringing this thread back to life.

I said weeks ago for this thread to end, and I really meant it.

Any more postings in this thread and the people who do it are going to
be removed from the lists.  This is the one and only warning.

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

	This message went through a time warp. Please don't reply to it.

	DS

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> On Wed, 10 Sep 2003, David Schwartz wrote:

> > Please show me one restriction on *use* in the GPL.

> Well, you may not *use* GPL'd code to produce a derived work and
> distribute it in binary form only. Use of the code, not use of
> the product, sure.

	Sure, and you may not *use* GPL'd code to produce a derived work and bring
a firearm into a court of law. I asked for a restriction on *use*, not a
restriction on use and something else.

	DS

RE: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

On Thu, 11 Sep 2003, Pascal Schmidt wrote:

> On Wed, 10 Sep 2003, David Schwartz wrote:
> 
> > Please show me one restriction on *use* in the GPL.
> 
> Well, you may not *use* GPL'd code to produce a derived work and
> distribute it in binary form only. Use of the code, not use of
> the product, sure.

Prove an original work which uses the proper "unprotectable API" to
operate in the Linux kernel is a "derived work".

You can't and never will.  It is an "original work".

"Original works" are exempt, what is not clear?

Oh I get it, because it happens to work in "Linux" it is not an original
work?  What a lame cast of mental giants.


Andre

RE: People, not GPL [was: Re: Driver Model]

[Pascal Schmidt]

On Sat, 13 Sep 2003, Andre Hedrick wrote:

> Prove an original work which uses the proper "unprotectable API" to
> operate in the Linux kernel is a "derived work".

I never claimed such a thing. Whether an API is unprotectable or not
is a question for the lawyers.

If people put GPL_ONLY symbol exports in their code, that's their call
to make, is it not? It's their code and they're free to say "well, this
is my code, and if you use this symbol, I consider your stuff to be a
derived work". Once again it's up to the lawyers to decide whether
this has legal value or not.

-- 
Ciao,
Pascal

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

 
> If people put GPL_ONLY symbol exports in their code, that's their call
> to make, is it not? It's their code and they're free to say "well, this
> is my code, and if you use this symbol, I consider your stuff to be a
> derived work". Once again it's up to the lawyers to decide whether
> this has legal value or not.

	If it has legal value, then it's an additional restriction.

	DS

RE: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

David,

Agreed the total intent by the author is to impose a restriction.
Thus the author by default lost his/her right to use, period.
Ironic, how imposing a usage restriction will terminate ones own write to
use.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Sat, 13 Sep 2003, David Schwartz wrote:

> 
>  
> > If people put GPL_ONLY symbol exports in their code, that's their call
> > to make, is it not? It's their code and they're free to say "well, this
> > is my code, and if you use this symbol, I consider your stuff to be a
> > derived work". Once again it's up to the lawyers to decide whether
> > this has legal value or not.
> 
> 	If it has legal value, then it's an additional restriction.
> 
> 	DS
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

RE: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Sad, 2003-09-13 at 22:19, David Schwartz wrote:
>  
> > If people put GPL_ONLY symbol exports in their code, that's their call
> > to make, is it not? It's their code and they're free to say "well, this
> > is my code, and if you use this symbol, I consider your stuff to be a
> > derived work". Once again it's up to the lawyers to decide whether
> > this has legal value or not.
> 
> 	If it has legal value, then it's an additional restriction.

If it has legal value in showing the work is derivative thats not an
additional restriction. Its merely showing the intent of the author. If 
someone creates a work and its found to be derivative and they didnt
make it GPL compatible they get sued, thats also not an additional
restriction its what the GPL says anyway.

That is the whole point of EXPORT_SYMBOL_GPL, it doesn't enforce
anything and Linus was absolutely specific it should not do the
enforcing. Its a hint and a support filter.

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> On Sad, 2003-09-13 at 22:19, David Schwartz wrote:

> > > If people put GPL_ONLY symbol exports in their code, that's their call
> > > to make, is it not? It's their code and they're free to say
> > > "well, this
> > > is my code, and if you use this symbol, I consider your stuff to be a
> > > derived work". Once again it's up to the lawyers to decide whether
> > > this has legal value or not.

> > 	If it has legal value, then it's an additional restriction.

> If it has legal value in showing the work is derivative thats not an
> additional restriction.

	If the work would not have been restricted without it and is restricted
with it and you can't remove it, it's an additional restriction. If not,
what would an additional restriction be?

> Its merely showing the intent of the author.

	The intent of the author has no bearing on whether or not a work is
derived.

> If
> someone creates a work and its found to be derivative and they didnt
> make it GPL compatible they get sued, thats also not an additional
> restriction its what the GPL says anyway.

	Show me where the GPL says you have to GPL derived works that you don't
distribute. That restriction is found nowhere in the GPL and if you
attempted to impose such a restriction, it would be an additional one.

> That is the whole point of EXPORT_SYMBOL_GPL, it doesn't enforce
> anything and Linus was absolutely specific it should not do the
> enforcing. Its a hint and a support filter.

	If it doesn't enforce anything and isn't a license restriction, then it's
perfectly legal and kosher to remove it.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Andrew Pimlott]

On Sat, Sep 13, 2003 at 03:30:35PM -0700, David Schwartz wrote:
> 	If the work would not have been restricted without it and is restricted
> with it and you can't remove it, it's an additional restriction. If not,
> what would an additional restriction be?

You can remove it.  But if you remove for with the obvious purpose
of abetting the distribution of non-GPL derived work, don't be
surprised to get smacked by courts who don't care for your technical
sophistry.

> > Its merely showing the intent of the author.
> 
> 	The intent of the author has no bearing on whether or not a work is
> derived.

I've noticed it's become common to say this, but (NAL) I doubt it's
true.  I would expect a court to respect the author's intent within
some narrow range that would otherwise be ambiguous.  Intent and
community standards play a large role in law.  If enough people wear
a path across private property, it can become an easement.

Andrew

RE: People, not GPL [was: Re: Driver Model]

[Martin Schlemmer]

On Sun, 2003-09-14 at 00:30, David Schwartz wrote:

This is starting to get one of those threads that really
should have ended a _long_ time ago.

> 
> > > 	If it has legal value, then it's an additional restriction.
> 
> > If it has legal value in showing the work is derivative thats not an
> > additional restriction.
> 
> 	If the work would not have been restricted without it and is restricted
> with it and you can't remove it, it's an additional restriction. If not,
> what would an additional restriction be?
> 

Read further on - it in itself is not a legal issue as such.

> > Its merely showing the intent of the author.
> 
> 	The intent of the author has no bearing on whether or not a work is
> derived.
> 

If you read the ending part of Alan's message, you will see where
this fits into.  Removing it, once again do not have any real
legal value while still within GPL 'usage boundaries'.

If the party that is accused of breaching GPL however did also
remove the GPL_ONLY symbols, it does once again have no legal
implication, except maybe enforcing a case of 'look, his intent
was from day one to breach GPL'.  It is however not binding
legally, and its worth will depend on the country, legal system
and so on.

> > If
> > someone creates a work and its found to be derivative and they didnt
> > make it GPL compatible they get sued, thats also not an additional
> > restriction its what the GPL says anyway.
> 
> 	Show me where the GPL says you have to GPL derived works that you don't
> distribute. That restriction is found nowhere in the GPL and if you
> attempted to impose such a restriction, it would be an additional one.
> 

I do not see why point you want to make.  If you do _NOT_ distribute
modified versions of the kernel, or works based on it, wtf worry ?
If you however do (hello, driver derived from kernel source/examples,
etc), you are bound by GPL (section 2a) to also distribute it under
GPL, and thus also the sources, which is exactly what manufactures
doing binary only drivers do not do.

> > That is the whole point of EXPORT_SYMBOL_GPL, it doesn't enforce
> > anything and Linus was absolutely specific it should not do the
> > enforcing. Its a hint and a support filter.
> 
> 	If it doesn't enforce anything and isn't a license restriction, then it's
> perfectly legal and kosher to remove it.
> 

Yes, didn't Alan just say it does not enforce anything?

What it does though do, what the whole idea behind it is, is if
party A, say do a binary driver, and they changed EXPORT_SYMBOL_GPL
to EXPORT_SYMBOL, then the kernel devs will _not_ support party A,
as Alan said with 'Its a hint and a support filter.'.  Sure, you
are free to remove it, but if you do, do not expect any support.

The company I work for is a supplier of PC components.  We have
the same type of support 'safety system' build in - we do not have
the infrastructure to support the public masses, thus we sell at
low prices to vendors that have a much higher markup, and good
support.  If then somebody come to us, that is not a dealer, or
do not have an Invoice, sorry sir, go to who you bought it from.
Its basically the same thing if you think about it.

Can we now stop this, or could you continue this in private if
you still choose not to understand?


Thanks,

-- 
Martin Schlemmer

RE: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

Alan,

Taking you at face value, as always in the past, how does one parse
between "usage"/"enforcement"?

Can a non-gpl object call, use, and it work "EXPORT_SYMBOL_GPL"?

If EXPORT_SYMBOL_GPL allows functionality, then it does not "enforce" any
license issues.  If it is called and then allowed to quietly fail or not
function unless the caller's associated LICENSE is "GPL", then it does
enforce, restrict and prevents 'functional' usage.

If this is a "NEW SYMBOL" one may have an arguement to hold.

If it is an "OLD SYMBOL", or the old one is removed and replaced with an
random new one, yet the core functionality of the operations are the same,
the intent is to break or terminate usage.

If EXPORT_SYMBOL_GPL is to be a notifier to let end-user be aware that a
binary vendor may be operating in a grey region of "derived" work, fine.
Allow it to operate but make noise.

Pretend it is doing the NOVELL thing of exceeding license count, and make
noise.

It is all to simple to bypass, and if the TAINTING issues for detection
is truly only to ignore issues from non-gpl sourced drivers, then great.

I know you have killfiled me by now, but how goes business school for the
MBA?  Does this mean you are going to quit Linux?

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Sat, 13 Sep 2003, Alan Cox wrote:

> On Sad, 2003-09-13 at 22:19, David Schwartz wrote:
> >  
> > > If people put GPL_ONLY symbol exports in their code, that's their call
> > > to make, is it not? It's their code and they're free to say "well, this
> > > is my code, and if you use this symbol, I consider your stuff to be a
> > > derived work". Once again it's up to the lawyers to decide whether
> > > this has legal value or not.
> > 
> > 	If it has legal value, then it's an additional restriction.
> 
> If it has legal value in showing the work is derivative thats not an
> additional restriction. Its merely showing the intent of the author. If 
> someone creates a work and its found to be derivative and they didnt
> make it GPL compatible they get sued, thats also not an additional
> restriction its what the GPL says anyway.
> 
> That is the whole point of EXPORT_SYMBOL_GPL, it doesn't enforce
> anything and Linus was absolutely specific it should not do the
> enforcing. Its a hint and a support filter.
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: People, not GPL [was: Re: Driver Model]

[Timothy Miller]

David Schwartz wrote:
>>On Wed, 10 Sep 2003 22:40:14 +0200, you wrote in linux.kernel:
> 
> 
>>>However, Richard Stallman does not agree with this view. It's his
>>>view that if the authors chose to give you the code, you can use it any
>>>way you want to, regardless of how the authors feel about that type of
>>>usage. This is why he created the GPL.
>>
> 
>>Use in any way you want to is the BSD license, not the GPL.
> 
> 
> 	Please show me one restriction on *use* in the GPL.
> 
> "Activities other than copying, distribution and modification are not
> covered by this License; they are outside its scope.  The act of
> running the Program is not restricted, and the output from the Program
> is covered only if its contents constitute a work based on the
> Program (independent of having been made by running the Program).
> Whether that is true depends on what the Program does."
> 
> 	Licenses that place restrictions on usage are *not* open source licenses.


What about "usage" of source code?

GPL says you are not allowed to "use" GPL source in a non-free program 
that you publish.

> 
>>The GPL
>>does restrict what you're allowed to do in order to keep the source
>>free...
> 
> 
> 	Yes, it restricts your ability to distribute and your ability to create
> derived works if and only if you distribute those derived works. It places
> no restrictions whatsoever on use. And since it requires distributors to
> place no restrictions not in the GPL, distributors cannot place *any*
> restrictions on usage either.

I don't think anyone was talking about use of applications, but rather 
use of source code.

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> > 	Licenses that place restrictions on usage are *not* open
> > source licenses.

> What about "usage" of source code?

	Same thing.

> GPL says you are not allowed to "use" GPL source in a non-free program
> that you publish.

	This is a publication restriction, not a usage restriction. Your phrasing
above is like saying, "you can't put bullets into a gun that you use to
shoot a police officer". The restriction is on the shooting, not the
loading.

	Again, quoting the GPL:

Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope.  The act of
running the Program is not restricted, and the output from the Program
is covered only if its contents constitute a work based on the
Program (independent of having been made by running the Program).
Whether that is true depends on what the Program does.

> I don't think anyone was talking about use of applications, but rather
> use of source code.

	I'm not sure why you think this is an important distinction.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Timothy Miller]

David Schwartz wrote:

> 
> 	This is a publication restriction, not a usage restriction. Your phrasing
> above is like saying, "you can't put bullets into a gun that you use to
> shoot a police officer". The restriction is on the shooting, not the
> loading.

Are you saying that publishing a program does not constitute a "usage" 
of the program and/or its source code?

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> David Schwartz wrote:

> > 	This is a publication restriction, not a usage restriction.
> > Your phrasing
> > above is like saying, "you can't put bullets into a gun that you use to
> > shoot a police officer". The restriction is on the shooting, not the
> > loading.

> Are you saying that publishing a program does not constitute a "usage"
> of the program and/or its source code?

	No, I'm saying that publishing a program is not *just* a usage of the
program. You can have full usage rights to a program and not be able to
publish it because publication requires granting rights to other people,
which is not use.

	Similarly, I can give you full rights to use my house but you still can't
sell my house. That's not because "selling" isn't a way of "using", it's
because selling is "using and then some". The additional thing that
distribution has that usage doesn't is the ability to grant rights to
others.

	The GPL is quite clear on this point:

"Activities other than copying, distribution and modification are not
covered by this License; they are outside its scope.  The act of
running the Program is not restricted, and the output from the Program
is covered only if its contents constitute a work based on the
Program (independent of having been made by running the Program).
Whether that is true depends on what the Program does."

	When you distribute a copyrighted work, you are not just distributing the
physical work but you are also conveying at last some rights of usage to
other people. Conveying rights of usage is legally distinct from having
rights of usage.

	Unrestricted use does not mean you can do whatever you please so long as it
somehow involves usage, it just means that the restrictions aren't upon the
usage itself. So anything that is *just* usage is okay. Anything that is
usage and something else is okay so long as the something else is okay. You
can have full unrestricted use of my bat and you still can't use it to beat
a stranger, but that's because you couldn't use anything to beat a stranger,
not because you don't have full use of the bat.

	Distribution is usage and something else. The something else being the
conveying of usage rights to others. If you have unrestricted use, you can
still only distribute if you have the rights to the something else.

	DS

Re: People, not GPL [was: Re: Driver Model]

[Andre Hedrick]

Timothy,

Who is talking about using "GPL source code", that is a given NO NO!
Calling symbols in the header files used as the generic functionality of
the kernel is all one could use in "Driver Model".  Beyond the simple
usage of header files, one is no longer in the land of API but usage of
raw GPL source.

Cheers,

Andre Hedrick
LAD Storage Consulting Group

On Fri, 12 Sep 2003, Timothy Miller wrote:

> 
> 
> David Schwartz wrote:
> >>On Wed, 10 Sep 2003 22:40:14 +0200, you wrote in linux.kernel:
> > 
> > 
> >>>However, Richard Stallman does not agree with this view. It's his
> >>>view that if the authors chose to give you the code, you can use it any
> >>>way you want to, regardless of how the authors feel about that type of
> >>>usage. This is why he created the GPL.
> >>
> > 
> >>Use in any way you want to is the BSD license, not the GPL.
> > 
> > 
> > 	Please show me one restriction on *use* in the GPL.
> > 
> > "Activities other than copying, distribution and modification are not
> > covered by this License; they are outside its scope.  The act of
> > running the Program is not restricted, and the output from the Program
> > is covered only if its contents constitute a work based on the
> > Program (independent of having been made by running the Program).
> > Whether that is true depends on what the Program does."
> > 
> > 	Licenses that place restrictions on usage are *not* open source licenses.
> 
> 
> What about "usage" of source code?
> 
> GPL says you are not allowed to "use" GPL source in a non-free program 
> that you publish.
> 
> > 
> >>The GPL
> >>does restrict what you're allowed to do in order to keep the source
> >>free...
> > 
> > 
> > 	Yes, it restricts your ability to distribute and your ability to create
> > derived works if and only if you distribute those derived works. It places
> > no restrictions whatsoever on use. And since it requires distributors to
> > place no restrictions not in the GPL, distributors cannot place *any*
> > restrictions on usage either.
> 
> I don't think anyone was talking about use of applications, but rather 
> use of source code.
> 
> 
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@vger.kernel.org
> More majordomo info at  http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at  http://www.tux.org/lkml/
> 

Re: People, not GPL [was: Re: Driver Model]

[Nicolas Mailhot]

[-- Attachment #1: Type: text/plain, Size: 1755 bytes --]

"David Schwartz" wrote :

[ sorry to interupt your flamewar but the amount of nonsence produced
here  starts to irritate me ]

| Who is sending these letters? Who has no respect for the GPL and seeks
| to add additional restrictions?

This is no additional restriction.
Check your history. The linux kernel was always under the GPL, not the
LGPL ie distributing stuff that links with the kernel means this stuff
must be distributed under the gpl.

At some point Linus decreeted linking closed modules was ok with him
(note this was done without consulting anyone, so others contributors
could have objected - they did choose to release stuff under the gpl
after all - but this being Linus they let it pass)

At a later point however the scope of closed linking had grown so big
people started saying enough is enough and GPL-ONLY was born with
Linus's approval.

It is not a licensing change. It's an hint the associated kernel symbols
are not covered by Linus' previous informal exemption and full GPL rules
apply. To avoid rewriting history symbols that could be used in non-free
stuff previously are not GPL-ONLY. People that ignore the hint can and
will be sued (people that link to symbols not GPL-ONLY could be sued too
but everyone seems to have agreed to let it pass). Removing the software
GPL-ONLY checks or working around them has nothing to do with it - it
does not change the basic kernel license nor the stated intentions of
its authors to enforce it. Hiding a do-not-trespass sign does not give
you the right to do it (if you think so do a reality check).

So please stop making horrified noises the GPL is being enforced in a
GPL project. Don't you realise how ridiculous it is ?

-- 
Nicolas Mailhot

[-- Attachment #2: Ceci est une partie de message numériquement signée --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

RE: People, not GPL [was: Re: Driver Model]

[David Schwartz]

> "David Schwartz" wrote :

> [ sorry to interupt your flamewar but the amount of nonsence produced
> here  starts to irritate me ]

> | Who is sending these letters? Who has no respect for the GPL and seeks
> | to add additional restrictions?

> This is no additional restriction.
> Check your history. The linux kernel was always under the GPL, not the
> LGPL ie distributing stuff that links with the kernel means this stuff
> must be distributed under the gpl.

	Yes, *distributing* stuff that links with the kernel means this stuff must
be distributed under the GPL. Note that this is a restriction that only
kicks in when you distribute something. It places no restrictions on how you
can use derived works you don't distribute.

> At some point Linus decreeted linking closed modules was ok with him
> (note this was done without consulting anyone, so others contributors
> could have objected - they did choose to release stuff under the gpl
> after all - but this being Linus they let it pass)

	Right.

> At a later point however the scope of closed linking had grown so big
> people started saying enough is enough and GPL-ONLY was born with
> Linus's approval.

	Okay.

> It is not a licensing change. It's an hint the associated kernel symbols
> are not covered by Linus' previous informal exemption and full GPL rules
> apply.

	Fine, so long as it's not a license enforcement mechanism.

> To avoid rewriting history symbols that could be used in non-free
> stuff previously are not GPL-ONLY. People that ignore the hint can and
> will be sued (people that link to symbols not GPL-ONLY could be sued too
> but everyone seems to have agreed to let it pass).

	Sued for what? Violating a restriction that isn't part of the license?
That's no more illegal than removing the security checks on 'mount'.

> Removing the software
> GPL-ONLY checks or working around them has nothing to do with it - it
> does not change the basic kernel license nor the stated intentions of
> its authors to enforce it. Hiding a do-not-trespass sign does not give
> you the right to do it (if you think so do a reality check).

	Except that the GPL does not permit any usage restrictions.

> So please stop making horrified noises the GPL is being enforced in a
> GPL project. Don't you realise how ridiculous it is ?

	All of the GPL's restrictions kick in upon distribution. The GPL_ONLY
restrictions affect use even in the absence of distribution. Thus, the
GPL_ONLY stuff *cannot* be a license enforcement because what it enforces is
*not* part of the license. Anyone who distributed Linux claiming that it had
such a license restriction would be in violation of the GPL's prohibition
against distribution with additional restrictions. Can you please reply to
that specific argument?

	And this is not some whacko obsure legalistic argument. This is a
fundamental point. Many people who contributed to the Linux kernel
contributed *because* they believed in the GPL and felt assured that nobody,
not even Linus, could close the evolving works by putting usage restrictions
on it. You GPL a work because you want to keep not only the current code
open to unrestricted use but all future distributed derived works as well.
Nobody has the right to add new license restrictions beyond those present in
the GPL.

	DS

RE: People, not GPL [was: Re: Driver Model]

[Nicolas Mailhot]

[-- Attachment #1: Type: text/plain, Size: 2571 bytes --]

Le sam 13/09/2003 à 11:49, David Schwartz a écrit :

[...]

> > To avoid rewriting history symbols that could be used in non-free
> > stuff previously are not GPL-ONLY. People that ignore the hint can and
> > will be sued (people that link to symbols not GPL-ONLY could be sued too
> > but everyone seems to have agreed to let it pass).
> 
> 	Sued for what? Violating a restriction that isn't part of the license?
> That's no more illegal than removing the security checks on 'mount'.

Sued for not infringing the software license, in this case the GPL. They
won't be sued for removing GPL_ONLY checks ; they won't even be sued for
using GPL_ONLY symbols. They will be sued for distributing a non-GPLed
derived work when the license forbids it.

A symbol is marked GPL-only :
1. when there is no bloody way it could be used in another work without
it being a derived work
2. when the author of the associated kernel code publicly stated he will
go after you if you do not respect the license.

All the software "workarounds" proposed are only ways to infringe
without getting a big red warning in the face. They do not change the
reality of the infraction per see.

> > Removing the software
> > GPL-ONLY checks or working around them has nothing to do with it - it
> > does not change the basic kernel license nor the stated intentions of
> > its authors to enforce it. Hiding a do-not-trespass sign does not give
> > you the right to do it (if you think so do a reality check).
> 
> 	Except that the GPL does not permit any usage restrictions.

There is no usage restriction - you have the code, you can remove the
checks if you want to. Unlike a license enforcement restriction you
won't be sued just for this.

Now the next step is usually to link a big fat closed module to your
modified kernel. And there you *will* be sued if you distribute the
resulting work (which pisses of some unethical corporations because the
sole way to do it legally is to have the end-user perform the operation,
and the end-user does not want to bother with it. ie its closes them
some markets unless they open up their stuff. Guess what ? You won't
find any people on this list cry for them).

I even suspect if you submit a patch to have GPL-ONLY a kernel build
option it will be accepted (provided it taints the whole kernel
probably). Distributions of course will never ship binaries build
without it, and appliance manufacturers that use it will be submitted to
heavy review, but that's already the case.

-- 
Nicolas Mailhot

[-- Attachment #2: Ceci est une partie de message numériquement signée --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: People, not GPL [was: Re: Driver Model]

[Alan Cox]

On Sad, 2003-09-13 at 10:08, Nicolas Mailhot wrote:
> At some point Linus decreeted linking closed modules was ok with him
> (note this was done without consulting anyone, so others contributors
> could have objected - they did choose to release stuff under the gpl
> after all - but this being Linus they let it pass)

Other contributors objected at the time. Furthermore there is merging of
third party GPL code from bodies like the FSF who also object and didnt
submit their code themselves. 

> will be sued (people that link to symbols not GPL-ONLY could be sued too
> but everyone seems to have agreed to let it pass). Removing the software

No. Sorry its neccessary to keep correcting people here but I have no
intention of allowing anyone creating a derivative work to claim
estoppel when they get their arses kicked some years on. Their sole
defence is that the work is not derivative. In some cases I can believe
that could be a valid claim.

As to _GPL. Its primarily a way of ensuring people who mix binary code
in with their kernel in unsupportable ways are made aware of it, and so
we can all use filter rules to discard their bugs.

Some vendors do support certain third party binary modules in some
specific enterprise configurations, others don't. In the community case
the general rule is we don't.

Alan

Re: People, not GPL [was: Re: Driver Model]

[Pascal Schmidt]

On Sat, 13 Sep 2003 23:30:12 +0200, you wrote in linux.kernel:

>> If people put GPL_ONLY symbol exports in their code, that's their call
>> to make, is it not? It's their code and they're free to say "well, this
>> is my code, and if you use this symbol, I consider your stuff to be a
>> derived work". Once again it's up to the lawyers to decide whether
>> this has legal value or not.
> 	If it has legal value, then it's an additional restriction.

Derived works are already restricted (when it comes to distributing them,
otherwise none of this is relevant since you can easily locally make
all EXPORT_SYMBOL_GPL be plain EXPORT_SYMBOL). It would be a restriction
on who gets to say what constitutes a derived work and what not - but
that is not governed by the GPL anyway but by the relevant local laws.

The GPL does not and cannot outrank the law. If the law were to say the
authors of a piece of code get to say what is a derived work and what
not, then so it is and nothing the GPL says can stand against it.

-- 
Ciao,
Pascal

Re: People, not GPL [was: Re: Driver Model]

[walt]

Andre Hedrick wrote:
> Erik:
> 
> One more thing you forgot!
> 
> Before you run around ranting about DMCA and CPRM and making suttle legal
> threats.  Everyone here knows who put it on the line to gain control of
> technology to prevent Content Protection Recordable Media from covering
> the storage industry wide.  That is correct it was me...

Hi Andre,

I'm not a technical person but I've been following the linux-kernel list
for enough years to know who you are and approximately what you have
contributed to the open-source community.  Let me start by saying THANK
YOU!  I very much appreciate what you do for all of us.

I would like to help you in return, if I can.

I have tried my best for several years to understand your posts and I
simply cannot.  I want to understand you because I know you know you
have a great deal to offer -- but I just don't know what the fuck you
are talking about most of the time, no matter how many times I read
your posts.

Please don't be insulted, I beg you.  I don't mean to criticize in any
way, but I see that you need help with communication skills.  Many of
your posts in this mailing list probably are ignored simply because
very few people understand what you are trying to say.

I'm a married man so I understand how frustrating it is to have every
word misinterpreted.  But such frustration should not be necessary in
a technical forum like this one.  Anyone who can talk to a disk should
be able to talk to this crowd, but I think you need a wrapper for the
PMLI (Programmer Mailing List Interface).

I can't help feeling that a lot of your frustration is caused by
miscommunciation rather than real issues.

I hate to see you so frustrated and upset when your coronaries are at
risk!  Take your heart medication and hire a translator....please!
We are depending on you to be around for many more years.