Extracting the Secrets of OpenSSL with RAMBleed.

Area of Science:

• Computer Science
• Cybersecurity
• Hardware Security

Background:

• Increasing semiconductor density leads to new security threats for dynamic random-access memory (DRAM).
• The RAMBleed side-channel attack can exploit user-level accounts to access sensitive data, previously shown effective against OpenSSH keys.

Purpose of the Study:

• To develop and demonstrate a method for extracting Rivest-Shamir-Adleman (RSA) secret keys from OpenSSL using the RAMBleed attack.
• To assess the feasibility of this attack in realistic server environments.

Main Methods:

• Proposed a novel technique leveraging RAMBleed to target RSA private keys deployed in DRAM during Apache server operations.
• Combined RAMBleed data retrieval with RSA cryptanalysis techniques to correct for errors in the extracted information.

Main Results:

• Successfully retrieved OpenSSL RSA private keys from an Apache server environment.
• Demonstrated that error-free secret information is obtainable despite initial data corruption.
• Confirmed the attack's efficacy with high probability, even without administrator privileges.

Conclusions:

• RAMBleed poses a significant threat to OpenSSL security, enabling the extraction of sensitive RSA private keys.
• The attack is easily executable externally on vulnerable DRAM-equipped servers.
• Highlights the need for enhanced security measures against side-channel attacks in modern computing systems.