An embedded variable granularity control flow verification method and system based on probabilistic prediction

Abstract

The invention relates to an embedded variable granularity control flow verification method and system based on probability prediction, which is divided into a verification end and a proof end. Build a trusted path feature set, predict the vulnerability probability of each function in the target source program, and then perform coarse-grained instrumentation preprocessing on the function, and then use different inputs to obtain all possible dynamic controls of the target program on the verification side Flow graphs, these control flow graphs are replaced by hash values ​​and stored in the database; the collection of dynamic paths, the proof end runs the processed target program after receiving the request sent by the verification end, and calculates this control in the secure world The hash value of the flow graph is signed; the verification of the dynamic path, the proof end sends the signature to the verification end, and the verification end matches the hash value obtained by decrypting the signature in the database. If it can match, it means that the control flow of this operation is complete, otherwise the control flow integrity is violated.

Description

technical field [0001] The invention relates to a method system and method for verifying the integrity of software control flow on an embedded platform, and belongs to the field of dynamic measurement for key applications of embedded terminals. Background technique [0002] With the rapid development of embedded devices, especially the emergence of the Internet of Things (IoT), the integrity protection of key applications of embedded terminals has received increasing attention, and various integrity protection models and implementation methods have been proposed accordingly. . During the running of systems and applications, attackers can use various means to attack the systems or applications and destroy the integrity of these software. For example, Control-Flow Attack is one of the attack means, which is very common and can cause a relatively large threat. A control flow attack affects the runtime behavior of a program by hijacking the control flow of the program to an ar...

AI Technical Summary

Problems solved by technology

The former verification scheme will check all control flow transfer instructions in the program, including branch instructions, function calls and return events. This kind of granular verification can guarantee safety but will cause a large overhead

The latter has less running overhead, it only checks the call relationship between functions (that is, function call and return events), but it is not safe enough

Images