Protecting Sensitive Information in the Volatile Memory from Disclosure Attacks

Abstract

The protection of the volatile memory data is an issue of crucial importance, since authentication credentials and cryptographic keys remain in the volatile memory. For this reason, the volatile memory has become a prime target for memory scrapers, which specifically target the volatile memory, in order to steal sensitive information, such as credit card numbers. This paper investigates security measures, to protect sensitive information in the volatile memory from disclosure attacks. Experimental analysis is performed to investigate whether the operating systems (Windows or Linux) perform data zeroization in the volatile memory. Results show that Windows kernel zeroize data after a process termination, while the Linux kernel does not. Next, we examine functions and software techniques in C/C++ programming language that can be used by developers to modify at process runtime the contents of the allocated blocks in the volatile memory. We have identified that only the Windows operating system provide a specific function named SecureZeroMemory that can reliably zeroize data. Finally, driven by the fact that malware scrapers primarily target web browsers, we examine whether it is feasible to extract authentication credentials from the volatile memory allocated by web browsers. The presented results show that in most cases we can successfully recover user authentication credentials from all the web browsers except when the user has closed the tab that used to access the website.