Code pointer authentication for hardware flow control

Abstract

Techniques for enforcing flow control of a software program in a processor are provided. An example method according to these techniques includes analyzing program code of the software program to identify a code pointer in the program code, generating an authentication tag based on the code pointer, and modifying the code pointer in the program code with the authentication tag to generate a tagged code pointer.

Description

Background technique [0001] Software implementation errors, such as missing or incorrect input validation, can lead to out-of-bounds buffer accesses and memory corruption. These implementation errors can cause the software to become unstable or eventually fail under normal conditions. However, when the input is manipulated by a malicious party intent on attacking the computer system, the attacker can exploit these memory corruption and overflow errors to alter the expected behavior of the software code and execute code or functionality determined by the attacker. [0002] Attackers typically gain control of execution by overwriting data structures in memory such as return addresses, function pointers, or virtual table pointers. It is not always possible to find and fix all exploitable memory corruption bugs in large and legacy code bases. Accordingly, many computer systems contain a general defense feature known as an "exploitation mitigation mechanism" that is effective aga...

AI Technical Summary

Problems solved by technology

This approach makes exploitation difficult as the attacker needs to predict where the code / data / pointers they need to target / manipulate for a successful attack

Images