4 ms·

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 exploratio

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."

DyslexicAtheist 8y ago
> I was wondering what the fit with Qubes OS was. 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 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 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.
- yellowapple 8y ago
  Said attack surface is exactly why OpenBSD was so hesitant to buy into the virtualization trend up until they rolled their own hypervisor.
- heavenlyhash 8y ago
  I think the linked Rootkovska tweet (or more specifically, her own thread/replies to it) actually gracefully acknowledges that Xen is not magically free of attack surfaces. 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 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").
doener 8y ago
Actually no, I stumbled upon Genode OS on this list: https://en.wikipedia.org/wiki/Category:Microkernel-based_operating_systems https://en.wikipedia.org/wiki/Category:Microkernel-based_ope...

DyslexicAtheist 8y ago

> I was wondering what the fit with Qubes OS was. 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 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 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.

yellowapple 8y ago
Said attack surface is exactly why OpenBSD was so hesitant to buy into the virtualization trend up until they rolled their own hypervisor.
heavenlyhash 8y ago
I think the linked Rootkovska tweet (or more specifically, her own thread/replies to it) actually gracefully acknowledges that Xen is not magically free of attack surfaces. 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 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").

doener 8y ago

Actually no, I stumbled upon Genode OS on this list: https://en.wikipedia.org/wiki/Category:Microkernel-based_operating_systems https://en.wikipedia.org/wiki/Category:Microkernel-based_ope...