High performance software vulnerabilities detection system and methods

A method of detecting software vulnerabilities comprising the steps of:

• (a) accessing compiled code;
• (b) creating an instruction model for an instruction of said compiled code, said instruction model comprising an instruction location, a debug information, an instruction type and operands of said instruction;
• (c) creating a control flow graph associated with said instruction model, said control flow graph comprising all control flow paths of said instruction;
• (d) creating and populating a data flow model utilizing data flow merging;
• (e) obtaining a security finding for said instruction if it calls an unsafe function on unsafe data;
• (f) generating a security report comprising said security finding for said instruction if it calls an unsafe function on safe data;
• and
• (g) utilizing aliasing of data flows in said data flow merging, said aliasing including determining if a data location referenced by said instruction reached via a current control flow path has a data type that is the same as when said data location was referenced by said instruction when it was reached via a prior control flow path, said current control flow path and said prior control flow path chosen from amongst said all control flow paths.

A method of detecting software vulnerabilities comprising the steps of:

• (a) accessing compiled code;
• (b) creating an instruction model for an instruction of said compiled code, said instruction model comprising an instruction location, a debug information, an instruction type and operands of said instruction;
• (c) creating a control flow graph associated with said instruction model, said control flow graph comprising all control flow paths of said instruction;
• (d) creating and populating a data flow model utilizing data flow merging;
• (e) obtaining a security finding for said instruction if it calls an unsafe function on unsafe data;
• and
• (f) generating a security report comprising said security finding for said instruction if it calls an unsafe function on safe data;
• and
• (g) determining by said data flow merging if a data location referenced by said instruction reached via a current control flow path has a data type that is the same as when said data location was referenced by said instruction when it was reached via a prior control flow path, said current control flow path and said prior control flow path chosen from amongst said all control flow paths.

A method of detecting software vulnerabilities comprising the steps of:

• (a) accessing compiled code;
• (b) creating an instruction model for an instruction of said compiled code, said instruction model comprising an instruction location, a debug information, an instruction type and operands of said instruction;
• (c) creating a control flow graph associated with said instruction model, said control flow graph comprising all control flow paths of said instruction;
• (d) creating and populating a data flow model utilizing data flow merging;
• (e) obtaining a security finding for said instruction if it calls an unsafe function on unsafe data;
• and
• (f) generating a security report comprising said security finding for said instruction if it calls an unsafe function on safe data;
• and
• (g) if said instruction calls an unsafe function then marking said instruction location as containing unsafe data and adding said instruction to a worklist;
• (h) creating a set of concurrent worker threads, each thread selecting randomly a member instruction from said worklist, and processing it according to said control flow graph and said data flow model;
• and
• (i) concluding said creation of said worker threads if each said member instruction in said worklist has been processed, or if a predetermined time has elapsed.