Re: [Etherboot-developers] Q: GPL license clarification.

[Markus G. <ma...@gu...>]

Ken Yap wrote:
>>Would everyone be amenable to some additional language
>>in the GPL license clarifying that simple aggregation of
>>etherboot in the same rom as another BIOS.  Does not constitute
>>linkage and does not slurp the vendors ROM under the GPL.
>>
>>I don't think anyone has intended that but I have had at
>>least one nervous BIOS vendor say something about it.
> 
> Do you mean some clarification that putting Etherboot in the ROM does
> not suck in the BIOS under the GPL? I think this paragraph in COPYING is
> the relevant one:
> 
> In addition, mere aggregation of another work not based on the Program
> with the Program (or with a work based on the Program) on a volume of
> a storage or distribution medium does not bring the other work under
> the scope of this License.

[ Long legal discussion ahead; if it bores you, then the summary is:
   Go ahead! I see no problem with bundling etherboot with proprietary
   software in the same ROM. ]


I believe this is the same issue that Linus keeps mentioning on the 
kernel mailing list every so often.

If both pieces of code depend on each other and can not be used 
seperately, then one would be considered a derived work of the other. In 
this case, authors of both works could claim copyright on the combined 
work and it could only be licensed under conditions both of them agree to.

On the other hand, if each piece of code can be used completely 
independently, then each author only holds rights to the component they 
have written, and there is (probably) no way they could use a license 
for one piece of code to enforce restrictions on the other; e.g. it is 
probably not possible for a BIOS vendor to say that users cannot run 
etherboot on a machine that includes their BIOS; any attempt to enforce 
a restriction like this would likely be ruled null and void by the 
courts of most countries, as it would be considered in violation of 
fair-use rights.

While lawyers can argue this point forever, and while there does not 
seem to be much precedent for deciding this problem, I would think it is 
fair to assume that a good argument could be made for considering both 
etherboot and the BIOS to be seperate entities that are merely 
aggregated together.

There are two factors that lead me to this conclusion:

  a) there is a well-defined interface that both programs implement for 
interacting with each other. It is provably possible to combine each one 
of the two pieces of code with any one of multiple different 
implementations of the other component (i.e. etherboot can work with all 
  current PC BIOSs, and PC BIOS implementations can call ROM extensions 
other than just etherboot).

  b) it is possible to remove any one of the two components from the 
system ROM and to upgrade them independently without affecting the other 
one. In fact, if the BIOS vendor in question was really concerned about 
the legality of bundling etherboot with the BIOS, I would recommend they 
make a tool available that does exactly this: upgrade the etherboot 
implementation independently of the system BIOS.

If it makes anybody feel better, I would have no problem with adding a 
clause to the license saying that it is OK to aggregate etherboot 
together with proprietary software in the same ROM as long as the 
standard BIOS interface for communicating between etherboot and the 
other software is retained. In fact, if my legal analysis is correct 
(and please note that I am not a lawyer) this statement just reiterates 
what the law says anyway. On the other hand, if my analysis is wrong and 
both pieces of code are derived works of each other, than we could only 
grant this license extension if we managed to track down everybody who 
ever substantially contributed to etherboot -- now, that could be tricky.


Markus