Any thoughts about adding an implementation of KVM to the redox kernel?
I do know it would require changes to the paging subsystem in the kernel, possibly the threading logic in the kernel.
The initial idea that I had was implementing a libvirt compatible interface, but it would likely need a new scheme. Also, the paging system would be updated to include hardware-assisted nested paging first then software implemented nested paging next.
The goal is to see if we can run QEMU with Redox and Linux guests with Redox as the host.
Personally, I feel that Redox is more likely to be running within a virtual machine on some more established system rather than hosting other stuff as it is unlikely to challenge Linux in the server room for quite some time.
Also bear in mind that KVM is just one type of virtual machine that could be chosen, albeit possibly the best known, and is written in C. Would a clean sheet design be better than simply porting an existing feature to rust?
I know I might be comparing apples to oranges here, but suggest a rock solid ‘web technology’ application framework such as Electron or Tauri offers desktop users a lot of the same type of advantages that hypervisor would offer the server admin and thus would benefit most users in the foreseeable future. Whilst of no technical consequence, adopting such blatantly web based technologies riff quite nicely off Redox’s trademark ‘everything as an url’.