On 2/6/25 18:17, Sasha Finkelstein wrote: > On Wed, 29 Jan 2025 at 15:40, Dmitry Osipenko > <dmitry.osipenko@xxxxxxxxxxxxx> wrote: >> Otherwise, the proper solution would be to pass info about host's page >> size to guest using extended virtio protocol. > > It is not fully clear to me, as to what exactly is meant by that. > IIUC, virtio is a spec managed externally, so does that mean > that i should write to the virtio people to add the relevant bits > to the specification? Virtio people will only approve the new bits. You who will have to research how to add new virtio feature, propose and discuss the virtio spec change with virtio people till the change is accepted, implement new feature in the code. Anyone can propose spec changes, the process is documented and you may look up ML archives to find what other people did for upstreaming new features. Virtio spec is managed externally, but in reality is tightly coupled with Linux, all virtio people are also Linux devs. For the first step you may write a quick *kernel* RFC patches to spark the discussion. Cc all relevant people, see how much interest will be in the new feature, ask how to properly add the new property if unsure. Could be that virtio crowd will reject the idea of having a common generic property and then we may try to make it private to the virtio-gpu device. -- Best regards, Dmitry
