You think Microsoft, who has put tens of thousands of lines into Linux, has a right to change the license of Linux away from GPL? Is that really what you're positing?
So far, open source projects act like copyright either requires unanimous agreement among the copyright owners to make changes, or each contributor has copyright only in their specific contribution (so, in practice, you need unanimous agreement). That's simply wrong for joint works, and joint works are defined as works meant to be combined together to make one work (the alternative being a collective work, where the parts are simply combined into a collection, does follow the "each contributor has copyright in just their contribution" model).
Copyright law is based on certain assumptions, like "if the co-authors get into an argument, it will be about how to divvy up royalties, and there will be a contract involved." Another is "co-authors work together at the same time, and know each other." Those assumptions are wrong for open source projects.
A party simply can't change the license for code that they did not write. They can only change the license for their own work.
The FSF avoids this problem by requiring copyright assignment. To be honest, I'm not sure that's enough, but I think they're hoping for a sympathetic judge. Most projects have never looked at what the law actually says regarding joint work, and think it works the way they would write the law. I know what the outcome of that will be.
It's like claiming that someone shouting outside of a recording studio has the right to relicense the music created inside the recording studio. While it might be true, I'd want to see an actual legal case before claiming things that are that outlandish.
Not to mention that the GPL has been tested in court and has held. So there's that.
I know what the arguments would be, but I'm not sure how the court would rule. The hypothetical depends on legally making a fork, and the GPL (as an example) only gives permission to fork if you agree that any distribution will be under the terms of the GPL. If you violate those terms, you've committed copyright infringement. But can a joint copyright owner be guilty of infringing his own work?
I remember Eric Raymond discussing a commercial copyright infringement case where a contractor was hired to update a program, and started distributing the updated program without permission. This was his argument.
However, he hadn't made significant changes to the program. Using your words, the program was several orders of magnitude larger than what the contractor added. The court got creative, and declared the contractor only had a collective copyright -- like somebody who has copyright on a single article in an encyclopedia -- and not a joint copyright -- even though his changes were meant to be combined with the original work to create a single work and not a collection of works. A collective copyright owner only has copyright in his contribution.
You are right that courts would be reluctant to allow the tail to wag the dog, but there is some threshold where the contributor has made enough changes that he has joint copyright, and at that point he can do what the law says a joint copyright holder can do.
Every individual contributor could also potentially sue Microsoft in that case. The FSF wouldn't even have to lift a finger, and the contributors could get a nice class-action lawsuit going.
Usually, when a project changes its license, it either tries to get permission from each contributor; or it makes an announcement that if any contributor objects, they will remove that person's contributions on request. But US law doesn't actually require that. In the US ( http://copyright.universityofcalifornia.edu/ownership/joint-... ):
"Each joint author has the right to exercise any or all of the exclusive rights inherent in the joint work. Each author may:
"• Grant third parties permission to use the work on a nonexclusive basis without the consent of other joint authors
"• Transfer his or her entire ownership interest to another person without the other joint authors' consent".
In other words, unless you have some kind of enforceable agreement between all contributors, any contributor can relicense the work, without the consent of the other contributors. I'm not sure whether the GPL qualifies as the kind of agreement that could prevent this; the FSF has spent years saying the GPL doesn't meet all the standard requirements of a contract because it doesn't need to be a contract. The FSF famously requires contributors to transfer their copyright interest to the FSF because Stallman says an attorney told him that makes it easier to enforce the GPL. Or, as Eben Moglen says ( https://www.gnu.org/licenses/why-assign.en.html ):
"If there are multiple authors of a copyrighted work, successful enforcement depends on having the cooperation of all authors.
"... [I]n order to be able to enforce the GPL most effectively, FSF requires that each author of code incorporated in FSF projects provide a copyright assignment, and, where appropriate, a disclaimer of any work-for-hire ownership claims by the programmer's employer."
Many open source licenses other than the GPL don't have any chance of qualifying as the kind of agreement that could prevent a single contributor from relicensing the project on a whim.
Also note that "contributor" doesn't necessarily mean to "contributor to the parent project." I haven't been able to find any legal reason somebody can't fork a project (following the license), make enough changes to become a joint copyright owner as far as their fork is concerned, and then relicense the fork. I guess the original contributors would also be joint copyright holders as far as the fork is concerned, and could relicense the fork as well. I'm not sure where that would end; and I'm glad that so far nobody's been creative enough to start such a battle.
So, back to LLVM: Apple is, without question, a joint copyright holder and can relicense the code as desired. De Raadt's idea is that a European could say "well, the relicensing was OK under US law, but not under European law." But I don't see how moral rights would prevent relicensing. The Berne Convention establishes moral rights as ( https://en.wikipedia.org/wiki/Moral_rights ):
"Independent of the author's economic rights, and even after the transfer of the said rights, the author shall have the right to claim authorship of the work and to object to any distortion, modification of, or other derogatory action in relation to the said work, which would be prejudicial to the author's honor or reputation."
Note, (1) that is "independent of an author's economic rights" and (2) the contributor has the ability to object to any modification, mutilation, distortion, etc., of the work; not of the terms the work is distributed under. Countries can establish their own idea of moral rights, but I'm pretty sure they will follow the Berne Convention's definition pretty closely.
So I would predict the outcome would be a court opinion that any contributor has the right to change the license, and that moral rights aren't the right tool to stop a relicensing. And if anybody can fork, become a contributor in the fork, and relicense the fork, I don't see how anyone can enforce any license restrictions.