Re:GPL Violations Related to Combining ZFS and Linux (2016)
by cyphar on Thursday July 27, 2017 @10:30PM
(#14871469)
The thing is, it's actually not clear whether the intention here might actually harm the GPL. Effectively, programmers aren't lawyers so is marking something as EXPORT_SYMBOL_GPL considered to be a legal restriction? And does that actually imply that EXPORT_SYMBOL is not considered a derived work? The problem is that "derived work" is something that needs to be handled on a case-by-case basis -- if you had a module that scanned the kernel's address space to violate the EXPORT_SYMBOL_GPL restriction, would that module have to be under the GPL (the answer is yes IMO, even though it didn't use any EXPORT_SYMBOL_GPL APIs).
This was brought up in a (very long) ML thread on LKML[1].
[1]: https://lists.linuxfoundation.org/pipermail/ksummit-discuss/...
In the kernel, functions which might be used by any modules are exported using EXPORT_SYMBOL. But if a module wants to use functions exported via EXPORT_SYMBOL_GPL, then the module has to use GPL.