On Wed, May 18, 2016 at 4:00 PM, Frank Rowand <frowand.list@xxxxxxxxx> wrote: > Warner Losh asked a question about the use of cpp #include in dts > source files, and while the topic of license of existing devicetree > source files in the linux source tree has a bearing on the question, > he adeptly tried to avoid stirring up a hornet's nest. Instead of > sidetracking his thread with license issues, this is a new thread > with my license question. > > I'm sure I'll regret asking this, but what the heck. > > The result of compiling a devicetree source file (.dts and .dtsi) > generates a .dtb file aka devicetree blob aka flattened device > tree (FDT). The FDT is a binary and ascii data structure that > contains no code. If the devicetree source file(s) license is > gpl, then it seems to me that the license of the blob is also > gpl. > > If an operating system kernel "reads" an FDT, would that somehow > be a license, copyright, or other issue if the kernel was not also > gpl licensed? > > My naive, "I'm not a lawyer", view is that there would not be > an issue, but my third hand impression is that the BSDs are > concerned that there may be an issue. > > Are there any good authoritative sources that have addressed > this issue or a similar issue? My belief, based on my layman's understanding of the law, is that the DTB would be a separate work. Using the DTB by the kernel would not infect the kernel any more than running a GPL'd binary. It's my belief that the extent to which the dts files are a creative work, then the resulting DTB would be covered under the licenses for each file. For the overwhelming majority of users, this is an acceptable situation since the exposure for releasing the DTB source is low. And most of the source can be produced by dtc outputting dts format anyway. As this question hasn't been litigated to my knowledge, we have to follow similar precidents like arguing an analogy to an executable. One could also argue that it's similar to ACPI in that it loads a blob from elsewhere, then interprets the blob based on a fixed set of rules. Since dts is purely descriptive of the hardware, an argument could be made that it's not creative enough to enjoy copyright protection. The bar is usually pretty low to get copyright protection, but many cases that had novel aggregations of facts failed to gain copyright protection because you can't copyright facts. So that makes it an open question, since it hasn't been litigated. On the other hand, there are several examples of works that are barely creative at all that have copyright protection. For certainty, thought, I'd prefer a definitive answer in the form of a legal opinion. It would be better from the BSD community's perspective if there were an explicit license that was more like BSDL than GPL, but I understand the sensitivities in other communities may not allow that. Such a license, though, would render this point moot. I'd be interested if anybody has an expert legal opinion that's been prepared on the topic as well. Warner -- To unsubscribe from this list: send the line "unsubscribe devicetree-spec" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
