On Tue, Oct 13, 2026 at 09:05:37PM +0800, GONG Ruiqi wrote: > I'd rather not base our work on Jiri's patch at this stage: it is > a pre-RFC draft of the new attestation subsystem, whose design is > itself still in progress. Building upon it instead of current > mainline would be quite risky — we would be rebasing on a moving > target, and merging would be blocked until the whole framework > lands, which is realistically far off.
I'm going to be strongly against any work to try and put a ROT multiplexor into IMA. This is architecturally the wrong direction and is clearly very a short term project that will expose uABI that will be difficult to unwind later. If you want this to go faster please work with Jiri. > I'd suggest we develop the proposal on current mainline, and > switch to the new attestation subsystem once it lands. As I > understand it, for us that is mostly just a matter of switching > to a different set of APIs. It isn't, one activity is moving the policy and ROT multiplexer into a shared subsystem, the other is building a ROT multiplexer into IMA for only IMA to use. Jaon
