David Woodhouse wrote:
On Sun, 2008-06-15 at 20:12 -0300, jeff wrote:
> But tg3.o as distributed by RedHat/Fedora when it's compiled is *NOT* a GPL .o,
> it has the proprietary data in it. It isn't separate at all (like some
> firmware, say intel wireless, which is a completely separate file).
> I look at tg3.c and I can't tell where this "aggregation" begins and
> the *SAME FILE*. Can you clearly say which line numbers are GPL and which line
> numbers are not GPL?
It is quite clearly a violation of the GPL.
Feel free to submit a patch to shift the firmware out of that file and
fix the driver to use request_firmware() for it. I'm collecting such
patches in the git tree I referred to earlier.
Well, I did do a patched kernel with the GPL violations removed--but my goal
wasn't to make it so people could then re-add non-free software, so I don't
think the request_firmware() path is the one I'd take. Alexandre Oliva is now
maintaining this free software branch of the linux kernel (as you know) as he's
far more competent at this than me...
The "GPLing" of the driver (MODULE_LICENSE("GPL")) and the inclusion
of a chunk
of non-free code occurred in commit  2d8a9d3fd19147c808aa39ddc69a743d1c90f199.
The commit shows David S. Miller (davem(a)redhat.com) and Jeff Garzik
(jgarzik(a)mandrakesoft.com) as authors. If they have ever had it, couldn't they
be sued for the source code like other companies that are violating the GPL? I
suppose it would be RedHat itself that is liable, not them individually since
presumably David Miller was working on RH time. But he also has commits with
huge non-free chunks, such as b7fea8b8bd72166942e8b589e0e948af4aff3a37 coming
from "davemloft.net" (others from "ninka.net"), so perhaps he would be
individually responsible too.
I haven't thought through the implications of it yet, but lines in the firmware
itself get occassionally modified--it wasn't just a one time dump, so it
appears the source is there somewhere and someone is hacking away at it.
Perhaps Broadcom themselves (they added their copyright three years later, in
b8d263683391ab74f5794657a4b4edc6d7632937) have to give over the source code if
Miller and/or Garzik don't have it.
Has RedHat ever had the source code to the driver?