Zapscape (CVE-2026-64561): Guest-to-Host Escape in KVM/x86

Posted by john_strinlai 13 hours ago

Counter69Comment11OpenOriginal

Comments

Comment by eqvinox 1 hour ago

Anyone know if "-cpu ${CPU},vmx=off,svm=off" in QEMU is a safe workaround for this?

(To disable nested virtualization on a per-VM basis. Only against exploitation from within that specific VM, obviously does nothing against users with access to /dev/kvm on the host.)

[That did work around Januscape: https://news.ycombinator.com/item?id=48815819]

Comment by minimaltom 12 hours ago

Oh yay another one lol. This one seems much more general than the prior one that needed nested page tables.

Patch Thursday for cloud VM ppl lol

Comment by esjeon 4 hours ago

This one exploit "shadow MMU" in the nested virtualization path of KVM, so this one is more-limited than EPT/NPT vul'n (KVM defaults to EPT/NPT, nested virt'n is disabled by default).

Nested virtualization is rather a niche feature, and, tbh, considering that shadow MMU is highly complicated legacy code (i.e. outside of the main happy path) and has been source of critical vul'n, I would avoid nested virtualization on KVM.

Comment by metadat 12 hours ago

Do most cloud providers have live-migration or what is the approach to make this seamless? What kind of interruption might tenants notice?

Comment by cudder 12 hours ago

It depends. If you have a plain old VM with no GPU and networked storage it's easy to migrate. If you're using low latency HW on the host it gets a lot trickier. Kuberenetes or some other fleet management helps there, you spin up new nodes and drain the old ones as capacity allows. If you have your PodDisruptionBudgets set right you won't notice much.

You can also use ksplice to live patch the vulnerable kernels as a first aid. I'm not an expert but I think there are some limits to what you can accomplish with it. Also guessing that the embargo period allows the big names to roll the patches out gradually enough that you won't see any panic shutdowns.

Comment by minimaltom 10 hours ago

Yeah most cloud providers have live migration and a reasonably-well-tested playbook for updating the node kernel + hypervisor.

Comment by tryauuum 10 hours ago

Doesn't this one need it as well? I see the shadow mmu

Comment by minimaltom 10 hours ago

Right, but what x86 KVM setup is there in practice that doesnt present a mmu to its guest and hence keep its own track (shadow) of memory mappings?

Comment by Veserv 8 hours ago

Shadow memory mappings are only needed when you do not have hardware virtualization or when doing nested virtualization.

From the page: "it can threaten the guest-host isolation of KVM/x86 hosts that accept untrusted guests and expose nested virtualization"

That is not to say that it is not a serious vulnerability though.

Comment by rvz 13 hours ago

Another serious critical vulnerability that almost no-one cares about, when they should.

Comment by inigyou 11 hours ago

Every LLM bug in the Linux kernel (warehouse rave edit).

(This comment is a reference to https://youtu.be/v1Mfirg2-Z8