Xilinx Legal

Ed McGettigan wrote:
> 
> The previous link that I cited only discussed a narrow issue that
> was raised to an appellate court.  Try this one instead which has
> notes on the on the  Software License Agreement.  Altera was not
> claiming that they "owned" the bitstream only that use of the
> bitstream was restricted to Altera only devices by the license of
> the software that created it.
> 
> http://www.iplawobserver.com/2005/09/using-softwares-output-to-copy-chips.html 
> 

One more link that which is the actual ruling made by the US Court of Appeals
for the 9th Circuit:

http://www.ca9.uscourts.gov/ca9/newopinions.nsf/01512427B6AF4BF88825707C0076B4A3/$file/0317323.pdf?openelement

The first half discusses the mask claims, the latter half discusses the SW License
claims on pages 17-23.  Specifically on the bottom of page 23 the court found:

     Altera customers cannot use the software, and
     therefore create the bitstreams, without agreeing to the licensing
     agreement, including the permitted use restriction. In
     essence, a valid contract is a prerequisite to the creation of a
     bitstream from Altera software, and the jury could logically
     conclude that valid contracts were formed via the Altera
     licensing agreements before customers sent bitstreams to
     Clear Logic. We therefore affirm the district court�s denial of
     judgment (sic) as a matter of law on the final claim.

Maybe we could now move any further legal oriented threads to comp.arch.fpga.legal :-)
and get back to technical issues and discussions.  I know at least that this will
be my last post on this thread and the XDL/Open Source License threads.

Ed
--
Xilinx Inc.

Ed McGettigan schrieb:
> The previous link that I cited only discussed a narrow issue that
> was raised to an appellate court.  Try this one instead which has
> notes on the on the  Software License Agreement.  Altera was not
> claiming that they "owned" the bitstream only that use of the
> bitstream was restricted to Altera only devices by the license of
> the software that created it.
> 
> http://www.iplawobserver.com/2005/09/using-softwares-output-to-copy-chips.html

Interesting.
They invoke the Semiconductor Chip Protection Act by stating that the
bitstream format contains information on the structure of alteras FPGA
circuits and that it therefore something like a copy of alteras circuit
layout was created.

Clear Logic seems to believe that the jury was confused about what
exactly was copied. I think that is very likely.

You are allowed to reverse engineer if you incorporate the results in an
original work. As I understand it clear logics mask is very different
from alteras in that there is no configuration logic, sram cells, etc.
There should be less then 1/6 of the transistors. Of course the
structure will be similar, but what else should "incorporate the
results" mean, clearly similarities are allowed?


Still, that ruling does not apply to using an altera bitstream in a
Xilinx FPGA oder implementing an altera bitstream in an ASIC (that is
not similar to alteras FPGA structure)

Kolja Sulimma

Ed McGettigan wrote:
>      Altera customers cannot use the software, and
>      therefore create the bitstreams, without agreeing to the licensing
>      agreement, including the permitted use restriction. In
>      essence, a valid contract is a prerequisite to the creation of a
>      bitstream from Altera software, and the jury could logically
>      conclude that valid contracts were formed via the Altera
>      licensing agreements before customers sent bitstreams to
>      Clear Logic. We therefore affirm the district court's denial of
>      judgment (sic) as a matter of law on the final claim.
>
> Maybe we could now move any further legal oriented threads to comp.arch.fpga.legal :-)
> and get back to technical issues and discussions.  I know at least that this will
> be my last post on this thread and the XDL/Open Source License threads.

Thanks for the clearification Ed, and the small dose of contract law is
good
for everyone to keep them out of the defendants chair and having to
learn
a whole lot more. This statement about the enforcability of an EULA
contract
terms should be a wake up call to those that haven't thought these
issues
thru.

In summary, the EULA contract NDA supercedes all other rights you might
have to similar information in other settings. The confusion this week
over the
concepts of copyright fair use while under NDA, over contridictory
statements
about the open nature of VDL, and other rights issues have all been
useful
to learn and think about so we can protect ourselves and the IP we
agree to
use.

Hopefully it's also been a learning exercise for Xilinx too. The real
economic
value for Xilinx is the sale of it's chips, and at some point open
source access
to the tool chain will greatly benefit the companies sales by expanding
uses
into new markets that the existing tool chain doesn't support.
Certainly easy
access to Xilinx product for personal research by students and
hobbiests is
a market builder that yields long term benefits as those individuals
influence
purchasing decisions for their employers and the companies they own.

Surely this is purely academic nitpicking going on here.  Xilinx's
ownership of your bitstream (which you agreed to when you installed
their software) simply provides them a legal recourse should you try to
retarget it to a different vendor's technology.

They're in the business of selling chips, and with development systems
like WebPack they make available free tools (which they invest massive
amounts of money in) to reduce the cost to you to develop.  They're not
going to try to assert ownership of your Whizbang5000 IP - get real.


I know people like to rant and rave about stuff like this.  But for all
practical purposes it's a no-op.

Paul.

Austin Lesea <austin@xilinx.com> writes:
> There is nothing "open source" about any of Xilinx's software.

Then why did the installer made me agree to the GPL and LGPL?  I thought
it was because Xilinx's software (ISE Foundation) included some open
source software.

Actually.. yes.. if you look at the output files .. you will see it
detecting and replacing counters, comparators, state machines, ram etc.

Simon

"DJ Delorie" <dj@delorie.com> wrote in message
news:xn3bj4icdg.fsf@delorie.com...
>
> [disclaimer: I'm a GCC developer and former Cygwin developer]
>
> One key difference between Cygwin and Xilinx, is that apps built with
> Cygwin also *include* part of cygwin (almost verbatim) in the
> resulting binary.  Do bitstreams built by Xilinx tools *include*
> portions of the Xilinx tools in the resulting bitstream?  Can Xilinx
> point to a bitstream and say "these 1000 bits are copied from our
> library" ?
>
> A better comparison is comparing Xilinx to GCC.  The GCC license
> explicitly states that binaries built *with* GCC are not affected in
> any way by GCC's license.
>
> Note that binaries built *from* GCC (derived works) are a different
> story.  GCC's runtime libraries have a specific clause that covers
> linking; if you build with GCC, linking doesn't incur the GPL.  If you
> build with something else, linking does incur the GPL.

DJ Delorie wrote:
> [disclaimer: I'm a GCC developer and former Cygwin developer]
> 
> One key difference between Cygwin and Xilinx, is that apps built with
> Cygwin also *include* part of cygwin (almost verbatim) in the
> resulting binary.  Do bitstreams built by Xilinx tools *include*
> portions of the Xilinx tools in the resulting bitstream?  Can Xilinx
> point to a bitstream and say "these 1000 bits are copied from our
> library" ?
> 
> A better comparison is comparing Xilinx to GCC.  The GCC license
> explicitly states that binaries built *with* GCC are not affected in
> any way by GCC's license.

That is not quite entirely true - binaries build with gcc *are* affected 
by gcc's licenses.  In particular, patterns of assembly code generated 
by gcc are generated verbatim from gcc's source code (or in some cases, 
gcc's low-level libraries' source code), and these sections are 
therefore directly affected by gcc's licenses and copyrights.  There 
main license for gcc's source code is GPL, but there are explicitly 
stated exceptions to remove all restrictions and copyright assignments 
from the generated code, precisely so that you can do as you will with 
gcc-generated binaries.

If Xilinx' tools also have such verbatim copying through to the 
generated bitstreams, and they do not have such stated exceptions, then 
they are in a position (in my interpretation - IANAL) to claim joint 
copyright ownership of the bitstream.

> 
> Note that binaries built *from* GCC (derived works) are a different
> story.  GCC's runtime libraries have a specific clause that covers
> linking; if you build with GCC, linking doesn't incur the GPL.  If you
> build with something else, linking does incur the GPL.

David Brown schrieb:

> If Xilinx' tools also have such verbatim copying through to the
> generated bitstreams, and they do not have such stated exceptions, then
> they are in a position (in my interpretation - IANAL) to claim joint
> copyright ownership of the bitstream.

In germany falsely claiming copyright ownership is a crime that can be
punished by 5 years in prison, so Xilinx should make sure it only claims
joint ownership if there are such parts in the bitstream.

Think about connecting input to output in FPGA editor with a wire.

Kolja Sulimma

Kolja Sulimma wrote:

> In germany falsely claiming copyright ownership is a crime that can be
> punished by 5 years in prison, so Xilinx should make sure it only claims
> joint ownership if there are such parts in the bitstream.

Hmm, and what about falsely accusing someone of a crime?

> Think about connecting input to output in FPGA editor with a wire.

May very well instantiate a non-trivial "straight through" bit pattern,
or more likely several necessary patterns - really you have to
configure two fairly complex IOB's, and a bunch of routing resources.
And I wouldn't be suprised if there's some pattern of bits you need to
put in unused areas of the chip to optimally idle them.

Also, even if there are no verbatim bits, think about this.  Say
instead of running the software on your computer, you had to send your
HDL to Xilinx and they sent you back a bitstream.  Clearly they can
claim some ownership rights to that, unless they signed away the rights
under a work-for-hire type of contract.  It would appear that their
tools license is basically implying that this is what is going on, only
as a convenience for fast turnaround they are letting you borrow the
software to run on your computer.  From the software industry
perpsective, you can't purchase software the way you can purchase
something like a drillpress - though it's unclear if all legal systems
will uphold that view in all circumstances, especially things like
software bundled with boxed retail eval kits.