GPL-incompatible Module Error Message

GPL-incompatible Module Error Message

[Chris Bergeron]

Hello all,

Building the fglrx module against the current Linux kernel (2.6.20.7 as 
of this e-mail) I'm getting an error:

FATAL: modpost: GPL-incompatible module fglrx.ko uses GPL-only symbol 
'paravirt_ops'

which looks like someone (ATI?) might be doing something funny.  I'm 
just curious what the error means.. paravirt_ops is bit of the new 
virtualization code, correct?

Anyway, I'm curious if it's interesting or "nothing to see here, move 
along". I'm not looking for any kind of "fix" as the module itself not 
building isn't a kernel dev issue (currently having trouble with the 
Xorg Radeon drivers, but that's another story).

It just seemed like it might be interesting and I couldn't find anything 
to shed light on the error itself in the mailing list logs, and I'm 
curious at what's happening.

Thanks,
-- Chris

Re: GPL-incompatible Module Error Message

[Michael K. Edwards]

On 4/19/07, Chris Bergeron <chris@pcburn.com> wrote:
> It just seemed like it might be interesting and I couldn't find anything
> to shed light on the error itself in the mailing list logs, and I'm
> curious at what's happening.

What's happening is that some kernel developers don't like Linus's
stance on binary-only drivers and are trying to circumvent the norms
of software copyright law using EXPORT_SYMBOL_GPL.  (Why some people
think that the GPL is magically exempt from Lotus v. Borland, Lexmark
v. Static Control, and their analogues in other jurisdiction is beyond
me -- but then I gave up smoking the FSF's parallel-legal-universe
herb some time ago.)

Just s/EXPORT_SYMBOL_GPL/EXPORT_SYMBOL/ throughout the kernel and
you'll be fine -- at a technical level.  But be prepared, when later
changes break extra-volatile quasi-private in-kernel APIs, to keep
both pieces -- and to be shunned by EXPORT_SYMBOL_GPL partisans.

Cheers (IANAL, TINLA),
- Michael

Re: GPL-incompatible Module Error Message

[Alan Cox]

> What's happening is that some kernel developers don't like Linus's
> stance on binary-only drivers and are trying to circumvent the norms
> of software copyright law using EXPORT_SYMBOL_GPL.

The troll is back I see.

Why don't you give him some useful information instead

- Turn off the paravirt option - you don't need it, and its just bloat
and slows down the kernel. Then rebuild the kernel and other bits and it
should all work fine. 

The legality of the ati driver as a derivative work is another matter,
but I don't see what _GPL symbols have to do with its legality beyond
providing a hint.

Alan
 

Re: GPL-incompatible Module Error Message

[Michael K. Edwards]

On 4/19/07, Alan Cox <alan@lxorguk.ukuu.org.uk> wrote:
> The troll is back I see.

Troll, shmoll.  I call 'em like I see 'em.  As much as I like and
depend on Linux, and as much as I respect the contributions and the
ideals of the EXPORT_SYMBOL_GPL partisans, they're spreading needless
FUD by spraying "private-don't-touch-me" all over mechanisms that are
_explicitly_designed_ as interoperation boundaries.  They're also
aiding and abetting the FSF's hypocritical charlatanry about the
meaning of "derivative work".

> Why don't you give him some useful information instead

Alternate technical solutions are also useful.  You seem to know them;
I don't pretend to.  Thanks for providing them.

> - Turn off the paravirt option - you don't need it, and its just bloat
> and slows down the kernel. Then rebuild the kernel and other bits and it
> should all work fine.

Just out of curiosity -- it seems thoroughly unlikely that ATI has
intentionally touched paravirt_ops in fglrx.  Do you think that
redefining bog-standard Linux interfaces when CONFIG_PARAVIRT (or
whatever) is enabled suddenly makes fglrx a derivative work of
whatever code underlies paravirt_ops?

> The legality of the ati driver as a derivative work is another matter,
> but I don't see what _GPL symbols have to do with its legality beyond
> providing a hint.

Then surely you don't approve of spraying FATAL messages on people's
consoles under these circumstances.  Allowing code into one's kernel
whose integration problems can't or won't be diagnosed by mainline
developers may be foolish, but it's not FATAL.

Cheers,
- Michael

Re: GPL-incompatible Module Error Message

[Andi Kleen]

Chris Bergeron <chris@pcburn.com> writes:

> Hello all,
> 
> Building the fglrx module against the current Linux kernel (2.6.20.7
> as of this e-mail) I'm getting an error:
> 
> FATAL: modpost: GPL-incompatible module fglrx.ko uses GPL-only symbol
> 'paravirt_ops'
> 

It should probably be available for non GPL modules too. Otherwise
they can't even disable interrupts, which seems a bit extreme.
I will change that in my tree.

-Andi

Re: GPL-incompatible Module Error Message

[Adrian Bunk]

On Fri, Apr 20, 2007 at 03:47:27AM +0200, Andi Kleen wrote:
> Chris Bergeron <chris@pcburn.com> writes:
> 
> > Hello all,
> > 
> > Building the fglrx module against the current Linux kernel (2.6.20.7
> > as of this e-mail) I'm getting an error:
> > 
> > FATAL: modpost: GPL-incompatible module fglrx.ko uses GPL-only symbol
> > 'paravirt_ops'
> 
> It should probably be available for non GPL modules too. Otherwise
> they can't even disable interrupts, which seems a bit extreme.
> I will change that in my tree.

Don't do this, this would allow non-GPL modules write access to 
paravirt_ops.

Operations required by modules should be exported separately.

> -Andi

cu
Adrian

-- 

       "Is there not promise of rain?" Ling Tan asked suddenly out
        of the darkness. There had been need of rain for many days.
       "Only a promise," Lao Er said.
                                       Pearl S. Buck - Dragon Seed