Google, AMD Release Security Audit of Epyc Processors Used in Google Cloud’s Confidential Computing


An unusual association between Google and AMD can offer a plan on how the technology industry can better address processor security risks before they get out of hand. The only problem? Configuration requires an equally rare level of trust, which can be difficult for other companies to replicate.

On Tuesday, Google Cloud released a detailed audit of AMD’s confidential computer technology produced in a collaboration between Google’s Project Zero bug search group, two Google Cloud Security teams, and AMD’s firmware group. The audit follows Google Cloud’s years of increasing emphasis on its offerings for confidential computing, a set of capabilities that keep customer data encrypted at all times, even during processing. There is a lot at stake, as customers increasingly depend on the privacy and security protections offered by these services and the underlying physical infrastructure, which is based on AMD’s special, secure processors. An exploitable vulnerability in confidential computing could be disastrous.

Defects in the way processors are designed and implemented carry enormous risks, making widely used chips a single point of failure in the computers, servers, and other devices on which they are installed. Vulnerabilities in specialized security chips have especially dire potential ramifications because these processors are designed to be immutable and provide a “trusted root” that all other components of a system can rely on. If hackers can exploit a security chip failure, they can poison a system at that root and potentially gain undetectable control. Thus, AMD and Google Cloud have developed an unusually close partnership for over five years to collaborate on auditing Epyc processors used in the sensitive infrastructure of Google Cloud and try to plug as many holes as possible.

“When we find something and know that security is improving, that’s the best thing to do,” says Nelly Porter, Google Cloud Group Product Manager. “It’s not a point and a finger, it’s a combined effort to fix things. Opponents have incredible ability and their innovation is growing, so we need to not only catch up, but move forward. “

Porter points out that the partnership with AMD is unusual because the two companies have been able to build enough trust for the chip maker to be willing to allow Google teams to analyze the source code closely. Brent Hollingsworth, director of AMD’s Epyc software ecosystem, notes that the relationship also creates room to expand the boundaries on what types of attacks researchers can test. For example, in this audit, Google security researchers used specialized hardware to mount physical attacks against AMD technology, an important and valuable exercise in which more and more chip makers are also focusing, but which goes beyond of the traditional security guarantees offered by chip makers.

Pentesting of PCIe hardware using an IO breederPhotography: Google



Source link

Leave a Reply