I was wondering what the fit with Qubes OS was. Found this
entry on Genode challenges page:
"Genode as virtualization layer for Qubes OS - ...This exploration project pursues the goal of replacing Xen by Genode as virtualization layer for Qubes."
"Genode as virtualization layer for Qubes OS - ...This exploration project pursues the goal of replacing Xen by Genode as virtualization layer for Qubes."
Rootkovska has been (imvho rightfully) criticized in the past for selling isolation but dismissing the attack surface in Xen. I think Genode can help here:
https://twitter.com/rootkovska/status/949297922998489088
Though I believe thegrugq / ioerror have a point when they say that hardware compartmentalization is superior than layers of SW virtualization:
https://twitter.com/thegrugq/status/515085244198703105
The threat model is relevant though: In that aspect QubesOS has some problem finding an audience. I personally like it (I'm an active user) ... but mostly because I have a fetish for privacy+security and I like to play with new stuff. However if you push a journalist or activist or anyone facing real risk to learn QubesOS (the learning curve is huge for the general user), it makes more sense to just invest in HW based compartmentalization instead. HW compartmentalization can easier be trained and there is less likely a chance of shooting yourself in the foot. It's not a tech that anybody whose life is under threat should be using. First rule here not to use tech in the first place, and for cases where the risk is marginal use HW based isolation ... then there is nothing for a long time followed by things like QubesOS, Signal etc. These things are just good enough to hide an affair but it's reckless selling them as serious solution to tech-illiterate folk who have something to loose.
She's also written some fairly long-form articles about Xen security and the (sometimes significant) room for improvement, including, for example, this Black Hat talk (from 2008!) about breaking Xen: https://invisiblethingslab.com/resources/bh08/part3.pdf
I don't disagree with the rest of your comment :) I just have the impression that any discussion about Qubes+Xen is less "dismissal" and more that supporting a variety of hypervisors is "a small matter of programming" (where "small" is used to mean "absolutely not small").