On 10.10.23 14:11, Jarkko Sakkinen wrote:
Dave, since there was already sort of talk about detaching this code from kernel tree so that Jo could work on "pure C" runtime would it make sense to dual-license this first in the kernel tree? E.g. Jo could send a patch once this is merged with a new SPDX platter and we can then ack that? Just a proposal, with the emphasis on minimal amount of work required from each party. Also this would help with possible (and likely) bug fixes, i.e. minimal friction on fixing the same bug. Later on of course, we can consider adding "libsgx-dev" as depedency similarly as today there's a few dependencies like libelf-dev. I'm open for alternative proposals, just throwing something that came up mind.
Pitching in here: from my side, I'd also be fine to develop this bare-sgx "pure C" runtime under GPLv2 as is.
FWIW, I'd be mostly interested in and see most immediate use cases for such a runtime in research purposes (e.g., low-level benchmarking; rapid prototyping attacks/defenses; etc) and a copyleft license would be a good fit there IMHO.
This is not to say that I'm principally opposed to a more permissive (dual) license, especially if there would be a good use case for that.
But it seems to me that it may be non-trivial to build on the existing code base and re-license that, whereas GPLv2 would allow to fork immediately and also have any overlapping bug fixes seamlessly merged back upstream as you pointed out.
Also open to hearing alternative proposals of course! Best, Jo
