DDRop attack bypasses Intel and AMD confidential computing defenses

As reported by The Hacker News, researchers from KU Leuven, ETH Zurich, Durham University, and Google have disclosed a novel hardware attack named DDRop that undermines the memory protection mechanisms of Intel and AMD confidential computing technologies. This attack operates by silently dropping writes to a server's memory, causing the processor to continue reading outdated encrypted data as if it were current.

The DDRop attack requires an adversary with existing software control of a server and brief physical access to install a custom interposer circuit board between the processor and memory modules. This interposer, costing under $200, exploits a weakness in confidential computing implementations like Intel TDX, Intel Scalable SGX, and AMD SEV-SNP, which encrypt server memory to protect data privacy. The attack capitalizes on the absence of a memory freshness guarantee in these systems, allowing old encrypted data to be read as if it were the latest value.

Researchers demonstrated that DDRop can achieve full control of protected virtual machines on Intel TDX by manipulating page table writes, enabling unauthorized memory access and attestation forgery. While the impact on AMD SEV-SNP is more limited, all three technologies are affected due to their lack of a freshness check. The researchers have released the attack's design and code, noting that a simple software patch is not feasible due to the hardware-level vulnerability. Both Intel and AMD acknowledge the findings, stating the attack falls outside their defined threat models due to the physical access requirement.

Source: The Hacker News