AMD's EPYC SEV encryption reportedly "SEVered" by German researchers

[WYP]

AMD's EPYC SEV encryption has been reportedly "SEVered" by German researchers. Why does every exploit need a clever name these days?



Read more about AMD's EPYC SEV encryption being "SEVered" by security researchers.

[BFBC2033]

Unfortunately a key detail is missing. It needs full admin access on the host, which truns this into a Bug not a Security Flaw. Sad to see the same kind of fake news as with the amdflaws. But it seems like the SEV feature does not truely isolate the hyper-visor, maybe we will get there soon.

[WYP]

Originally Posted by BFBC2033

Unfortunately a key detail is missing. It needs full admin access on the host, which truns this into a Bug not a Security Flaw. Sad to see the same kind of fake news as with the amdflaws. But it seems like the SEV feature does not truely isolate the hyper-visor, maybe we will get there soon.

Thanks for the correction, the article has been updated. A note has also been added at the base of the article to credit you for your help.

[RobM]

Seems these security firms are quite happy to trash AMD

[Kleptobot]

Originally Posted by RobM

Seems these security firms are quite happy to trash AMD

I think AMD has a little bit of a target on their backs because they were less susceptible to specter and meltdown

[Emu]

This attack apparently requires a modified hypervisor to work along with a service on the virtual machine that will provide access to resources. Basically the hypervisor fiddles with the physical location of logical memory locations and then accesses those via a resource providing service on the virtual machine.

I don't think that this would be an issue if you are running your own virtual machine setup as you could use secure boot to protect the hypervisor from tampering. As for third party VM hosting, how well do you trust the third party?