This person can also quit IBM, therefore being subject to no contractual obligation, and then change the "IBM Power SRIOV Virtual NIC Device Driver" entirely free of constraint as long as their code pass review. IBM has no control over the code, only the people it pays to change it.

That's not quite right. If IBM is listed as maintainers for this driver, then any patches would necessarily go through them to end up in the mainline tree. And they could decide to reject patches from the person in question.

Although if this happened and legitimate code was rejected by IBM, what could happen is a community backlash against IBM as a maintainer and the code could enter that way.

People higher up the directory tree can effectively override subdirectory maintainers, all the way up to Linus at the top.

> This person can also quit IBM, therefore being subject to no contractual obligation, and then change the "IBM Power SRIOV Virtual NIC Device Driver" entirely free

I'd assume the person has received confidential information about the IBM technology in question during his employment. Of course the Linux driver is open source so it cannot be a business secret. But typically there is much more knowledge than just the Linux source code. Using business secrets after your employment is most likely enforcable in many legislations. Damages or criminal charges or both.

That's a fair point about clean room implementation. However, I can see IBM shooting themselves in the foot suing a kernel developer over an email address change, if said developer quits beforehand.

Such an action would only decrease developer goodwill towards IBM. Needless to say, IBM is already not one of the "cool" companies to work for.

not really.

there is this stipulation of using a clean room in which to introduce your idea in order not to be sued for “stealing” intellectual property.

And often times, the same person cannot be used in that clean room.