[uClinux-dev] insmod fails with Unknown symbol

[Gregg Levine]

On 10/5/07, Crane, Matthew <mcrane03 at harris.com> wrote:
>
> Hi,
>
> Linksys was sued for GPL compliance: http://lwn.net/Articles/73848/
>
> Linksys conformed the router in question; on wikipedia:
> http://en.wikipedia.org/wiki/WRT54G
>
> "The WRT54G is notable for being the first consumer-level network device
> that had its firmware source code released to satisfy the obligations of
> the GNU GPL."
>
> I've downloaded the package that conforms.  It contains about 4mb, that
> I noticed from a quick inspection, of binaries, some that are built into
> the kernel.
>
> Linksys, wether they respect the original intent of the GPL, has
> contributed significantly over the past few years to linux.  They do not
> release their driver source still.
>
> Are they illegal?  Is there any real chance that linksys will ever have
> to release their propietary firmware??  Isn't the benefits to the open
> source comm. of their releasing mass quanities fo flashable embedded
> linux devices considerable?
>
> Matt
>
>
> -----Original Message-----
> From: uclinux-dev-bounces at uclinux.org
> [mailto:uclinux-dev-bounces at uclinux.org] On Behalf Of Robin Getz
> Sent: Friday, October 05, 2007 10:49 AM
> To: Gavin Lambert
> Cc: uclinux-dev at uclinux.org
> Subject: Re: [uClinux-dev] insmod fails with Unknown symbol
>
> On Wed 3 Oct 2007 17:57, Gavin Lambert pondered:
> > Quoth Robin Getz:
> > > I think that nearly everyone would agree that the patch is
> > > bad/illegal/immoral/not acceptable - so why is a gasket that does
> the
> > > same thing OK?
> >
> > Well, the wrapper module would be GPL licensed so other people would
> be able
> > to use it too.  (And I wasn't thinking of a simple redirection
> wrapper, I
> > was thinking of a proper split where as much code as possible is in
> the
> > wrapper module and only a few things in the proprietary module.)  But
> really
> > it depends on what the thing is and what it's used for, as Jamie said.
>
> >
> > Especially in the embedded world, sometimes policy-style code needs to
> be
> > implemented at the kernel level rather than in userspace, and
> sometimes that
> > policy will be proprietary.
> >
> > But some of the GPL-only exports do seem misapplied to me.  For
> example, why
> > should every SPI-bus device driver have to be GPL?  Ok, maybe it
> encourages
> > people to write general-purpose drivers rather than task-specific
> drivers,
> > but sometimes you really need to write that task-specific driver :)
>
> I forwarded this on the Greg KH - at OLS he told me he (personally) was
> suing
> someone over this, and I wanted to see how things had progressed...
>
> His response verbatim (since he didn't want to subscribe to the list)
> =======================================
>
> On Thu, Oct 04, 2007 at 11:26:58AM -0400, Robin Getz wrote:
> > I thought based on previous discussion you may be interested in this
> (since
> > you might not read the uClinux-dev mailing list)
> >
> >
> http://mailman.uclinux.org/pipermail/uclinux-dev/2007-October/044210.htm
> l
>
> Ugh, that sucks.
>
> > Someone asked about why when their module was marked Proprietary, they
> > couldn't use GPL symbols, a suggestion was to:
> >
> > > 3. create a GPL-licensed module that uses the GPL-only symbols and
> make
> > > your proprietary driver talk through that module.
> >
> > Although technically correct - I don't think it would meet the spirit
> or the
> > letter of the law/GPL.
>
> It's flat out illegal, and Linus has said so many times in the past.
>
> However, some companies are trying to do just that, and I am personally
> taking legal action right now againt one that does it.  It is slowly
> moving its way through the legal process, as all legal things seem to
> take forever...
>
> >
> http://mailman.uclinux.org/pipermail/uclinux-dev/2007-October/044215.htm
> l
> >
> > > The GPL-only symbols are hint that, in the opinion of kernel
> > > developers, any code which uses them must be so intimately dependent
> > > on the kernel code as to be a derived work of it.  If you produce a
> > > wrapper module, then depending on what the module does, that may
> apply
> > > to the module which uses the wrapper module, or it may not.  There's
> > > no technical way to discern this, unfortunately; it takes a legal
> > > opinion.
> >
> > feel free to follow up on the uClinux-dev mailing list. Unfortunately,
> it is
> > subscriber only.
>
> Sorry, I don't feel like signing up for another list.  Feel free to
> forward my comment if you feel it will help anything out.
>
> The fact that some companies still seem to think that closed source
> Linux kernel modules are legal to ship these days is bloody amazing.
> Especially when so many very large companies with copyright on the
> kernel itself have come out and said so in public (Novell, IBM, Red Hat,
> HP, etc.)  Does a company feel that it is somehow ok to violate IBM's
> copyright on the kernel?  Are they crazy?
>
> thanks,
>
> greg k-h
> =======================================

Hello!
Actually twice, but they settled out of court for the NSLU2. Since all
of their boards are built for them by outsourced processes they didn't
know until someone familiar with the GPL pointed out that by running
Linux in them they were obligated to release the code for each device.

They claim that some of the items used in building the gizmos actually
need binary only drivers/modules. An example is the hardware used by
the router in question. Same goes for the NSLU2. All balderdash.
Eventually they will realize that they are shooting themselves with a
shotgun, and in all of their feet.

-- 
Gregg C Levine gregg.drwho8 at gmail.com
"This signature was once found posting rude
 messages in English in the Moscow subway."