Eben Moglen from Software Freedom Law Center thinks its OK as long as the source is distributed and no proprietary changes is made.
https://www.softwarefreedom.org/resources/2016/linux-kernel-...
>In this specific sense, then, the conduct which falls outside the words of GPLv2 falls within the "equity of the license," or its "spirit." As all Western legal systems have known since Aristotle, literal interpretation of any legal material will sometimes produce unintended unjust results, which can and should be corrected by the invocation of "equity." This present issue is evidently an example in which the tension between literal and equitable interpretation is raised, and it is the consensus of the kernel copyright holders' intention which determines which mode of interpretation is to be employed.
Linus is one central figure among kernel copyright holders. When he has stated his interpretation and it seems that most other maintainers are not objecting either, kernel copyright holders' intention seems to be closer to the "equity of the license".
Clearly ZFS is not a derivative work of Linux and Linux is not a derivative work of ZFS. Only the glue code is in violation (because it derives from both).
To determine whether there is glue here that violates the GPL would require looking at the particulars. Said glue might be licensed under the GPL, and might export the sorts of interfaces that Illumos does, thus allowing the ZFS module to link without reference to either the glue or the kernel! Then how could such glue be said to violate the GPL? It couldn't! Nor could ZFS be said to be violating the GPL in that case.
I don't think you can avoid an escape hatch for the GPL in the kernel without disallowing dynamic loading or modifying the GPL to have harsher terms. But recall that the GPL, and/or much software distributed under the GPL, is designed to have escape hatches: so you can run proprietary software using standard interfaces, for example (i.e., user-land code is not required to be GPLed just because it runs on a GPLed kernel, though a kernel could require it, it's just that _Linux_ doesn't).
It seems here you can't have your cake and eat it too.
I'm not sure where the legal line is going to get drawn, but clearly it's still fuzzy.
We each write something that passes as 'English' the language. Yet each of our brains has a different implementation of that interface. We do not copy each other even though we happen to be able to interact.
I'd like to see more legal argument, with citation to applicable precedent, on this conclusion. It necessarily, for functional reasons, makes some use of copyrighted kernel interfaces, but whether this use of functional elements makes it a legal derivative work is a question I've seen many people present cobflicting conclusions on, but never with much argument supporting the conclusion.
This then requires looking for a way in which the GPL prevents creating such adaptor modules. I don't see how it does. Perhaps it could if modified, but as it is I don't think it does. IANAL though.
Is it possible to glue a kernel module shim to something else under License A, glued to another shim under License B, etc until you have license compatibility with ZFS?
At the very least, this would seem to violate the very spirit of what the GPL intends.