My annual plea for a thing: I want a type 1 hypervisor that just has a small isolated VM and then passes through the rest of the hardware to the main VM which runs Linux. The small VM is intended to be used to run small pieces of code that the main OS should not be able to interfere with. Does such a thing exist? (Think Xen, but with a Dom0 that can't see into DomUs)
@rzeta0 A hypervisor doesn't have to let a privileged VM see into other VMs - Xen allowing that for Dom0 is an artifact of their design rather than anything inherent. The primary Windows VM can't see into the Credential Guard VM, since the hypervisor has drawn a hardware-enforced barrier in between them.
if you have the patience to argue with me I would value your reply..
for me, a security barrier is cryptographic
software and software-controlled-hardware barriers are just inconveniences
so the hypervisor you mention happens not to make it easy to see into a vm, but with the right combination of software failures or subversion it is possible
the openssl and CPU vulnerabilities (heartbleed, spectre) are illustrative examples
am i being too pendantic about a security boundary?
@rzeta0 Cryptography doesn't remove side channels - if you keep the secrets in a TPM but it doesn't use constant time operations, or if I'm able to monitor the power rails, that's not an absolute barrier. Very little is absolute - the level of security appropriate for a given problem will vary depending on what your threat model is, and I'm broadly ok with having my WebAuthn secrets in a separate VM running on the same CPU
@agraf My recollection is that Jailhouse does static partitioning and no scheduling, ie you need to give it a CPU? It also starts from Linux which makes it harder to sequester secrets that Linux can't get at.
@agraf I'm pretty sure the lack of scheduling is a design choice that would need to be retrofitted. Launching from Linux is more about how it's managed, so that's probably an easier thing to fix.
@hyc I'm fine with the hypervisor being able to see what's happening in arbitrary guests, but there needs to be isolation between the primary VM and the security VM (Hyper-V manages this fine in Windows land)
@fl0_id@hyc it's a hypervisor, it simply imposes a barrier between the resources? This isn't a conceptually complicated situation, modern CPUs support it just fine
@mjg59@hyc sure, but I just meant if the hv can technically see into all guests, who enforces the rules for security vm? The cpu or the hv or both? If the hv, this is likely more easily overridden.
Add comment