I support Victor Lopez choosing whatever terms and conditions he
wants, for the things he produces.


...
... I'm not a lawyer, but I have heard of a few cases involving copying chip hardware: * "Apple Computer, Inc. v. Franklin Computer Corp." (1983) case * "Intel Corp. v. Advanced Micro Devices, Inc. 1990". In 1990, Intel brought a copyright infringement action alleging illegal use of its 287 microcode. * "The first chip-layout copying case", IEEE Micro, Aug 1991, which describes "Brooktree Corp. v. Advanced Micro Devices, Inc. 1990" in detail, has illustrations of the SRAM memory cell in question. I've seen other things that seem to imply even more cases involving copying chip hardware and copyright (or something similar to copyright): * chip art. Have you seen the Silicon Zoo? Supposedly, some of that art was intended to prove that a copied chip was a direct copyright violation, and not innocent convergent design. * I've also been told that "The exclusive rights in a mask work are somewhat like those of copyright ... rights in semiconductor mask works last 10 years", from the Semiconductor Chip Protection Act of 1984 .. _Apple Computer, Inc. v. Franklin Computer Corp. http://en.wikipedia.org/wiki/Apple_Computer,_Inc._v._Franklin_Computer_Corp. .. _Franklin Computer Corp. http://en.wikipedia.org/wiki/Franklin_Computer_Corporation .. _Intel v. AMD http://en.wikipedia.org/wiki/Advanced_Micro_Devices#Litigation_with_Intel .. _first chip-layout copying case http://docs.law.gwu.edu/facweb/claw/BrooktAMD.pdf .. _chip art http://en.wikipedia.org/wiki/Chip_art .. _Silicon Zoo http://micro.magnet.fsu.edu/creatures/index.html .. _"Steal The Best" http://micro.magnet.fsu.edu/creatures/pages/russians.html .. _mask work http://en.wikipedia.org/wiki/Mask_work .. _Semiconductor Chip Protection Act of 1984 http://en.wikipedia.org/wiki/Semiconductor_Chip_Protection_Act_of_1984 -- David Cary 918.81.2279 http://carybros.com/ http://en.wikibooks.org/wiki/Microprocessor_Design

Its always been my interpretation that FPGA configuration files are essentially object code. As such, including a (L)GPL core in the same FPGA configuration as a closed module would be a forbidden. That being said, if you make one FPGA open source only cores, can you run signals out of it and interact with a closed module on a separate FPGA or custom hardware? I would argue that if the license is LGPL it would be permissable, but not if the license is GPL. I think this would be analogous to dynamic linking. However, it is permissible to run GPL programs (gcc) from non-gpl programs (like WindRiver Workbench), so is this a better analogy in this case? In any case, the GPL as written is just not equipped with the correct language to address its applicability to HDL code. ----- Original Message ----- From: Jeremy Bennettjeremy.bennett at e...> To: Date: Wed Jan 28 13:22:19 CET 2009 Subject: [oc] GPL License - again

At 10:24 AM 1/29/2009, you wrote:


This case only relates to protection of software embodied in a
computer ROM, not IP describing the hardware itself.



I am familiar with this case. As you say, it involved copyright as
applied to microcode, although I didn't think it was limited to the
287 code, but rather applied to the 486 code. Maybe there was more
than one case between these two. In the 486 case, it was ruled that
AMD could not literally copy the 486 microcode, but rather had to use
a "'cleanroom" approach to reproduce the code in as much detail as
could be specified in a document by a team of engineers reverse
engineering the code and a team of engineers creating the new
code. This lead to AMD producing a functionally equivalent, but not
copied, version of the 486 microcode. Since they never had any of
the Intel IP, this case is not relevant to the question of copyright.



This does not appear to be a copyright case, "Brooktree Corporation
brought suit against Advanced Micro Devices, Inc. (herein AMD) for
patent infringement"




Yes, if a chip is copied as in photographically copied, that would be
a copyright violation. But we are talking about using the source to
produce an implementation embodied in a chip. Or in more specific
terms, will the GPL applied to HDL prevent its use in a proprietary product?

The GPL covers only specific acts. "Activities other than copying,
distribution and modification are not
covered by this License" (translation is specifically included in
modification). The question is whether producing a chip from an HDL
program is the same as compiling software to machine code; is
creating chip masks or FPGA bitstreams from IP considered
"modification" or "translation" of the original IP?

There are a lot of interesting analogies that could be made to this
situation. I wonder how much difference in application would come
from the use of a program, such as a synthesis tool, to produce the
end result. If I obtain plans to a copyrighted house design and
construct my own house from that, I would not expect that to be a
violation of that copyright. I would not consider that to be
"copying" "distributing" or "translating" the plans. IF there was a
standard that told you exactly how to produce a doorway from the
symbol on the plan and how to make the walls and floors and roof from
the dimensions on that page, maybe that could be considered
"translating". To me, the difference is how much this "construction"
requires intelligent thought be used to produce the result vs. how
much of it is obvious from the plans.

In the case of expression of an HDL program, I can see where much of
the HDL clearly specifies *what* is being produced. Normally the
functionality of logic is clear even though the actual implementation
in LUTs or gates is not specified. The functionality of registers is
clear even if it does not specify whether those registers are single
FFs in several blocks or a single LUT used as a shift
register. Often even the intent of using a block RAM vs an array of
FFs is clear and so a block RAM could be said to be "specified" in
the original work. So I would have to say that synthesizing a design
from an HDL description would be considered "translating" and so
covered under the GPL even if it is decided to *not* be considered a
violation of copyright.

However, the other side of the coin is the amount of additional
effort is embodied in producing the resulting work. Any synthesis is
followed by the layout process. The objects must be positioned on
the device and they much be interconnected by routing. I believe it
has already been ruled that a "circuit" may be copyrighted in the
form of a schematic, but the use of the inherent circuit in physical
form is can not be protected by copyright.

But the GPL does not depend on copyright. It is a license. So in
the end, I believe the question comes down to whether the act of
placement and routing of the objects specified in the original HDL IP
is considered "modification" or "translation" of the original IP or
if it would be outside of that definition. Since it is a level of
specification that goes beyond the original IP, I believe it is *not*
"modification" as specified (which specifically includes translation)
in the GPL and so hardware that is produced from GPL'd HDL IP is not
covered in the same way that machine code in a ROM would be.

Rick