Formal Verification of Pentium ® 4 Components with Symbolic Simulation and Inductive Invariants

Abstract

We describe a practical methodology for large-scale formal verification of control-intensive industrial circuits. It combines symbolic simulation with human-generated inductive invariants, and a proof tool for verifying implications between constraint lists. The approach has emerged from extensive experiences in the formal verification of key parts of the Intel IA-32 Pentium 4 microprocessor designs. We discuss it the context of two case studies: Pentium 4 register renaming mechanism and BUS recycle logic.

Industrial strength refinement checking

“…5 Finally, our work advocates manually writing a RM, rather than using more automated techniques such as flushing [5]. Like other work on RMs for real industrial RTL designs [9], [15], we have found manually-written maps more appropriate.…”

Compositional model checking of software product lines using variation point obligations

“…It is a powerful technique for enhancing the quality of software systems, e.g., by identifying flaws that would not have been caught otherwise (Havelund et al 2001;Kaivola 2005). Especially for software product lines, since parts of the systems are reused in multiple products, it is important to detect flaws in those reused parts.…”

Safety analysis of software product lines using state-based modeling and compositional model checking

“…Model checking is a powerful technique for enhancing the quality of software systems [Clarke et al (2000)], e.g., by identifying flaws that would not have been caught otherwise [Havelund et al (2001); Kaivola (2005)]. However, there is currently insufficient support for model checking in product lines, most specifically, for property specification and management.…”

“…Model checking [Clarke et al (2000); Huth and Ryan (2004)] takes a model of a given system's design, and checks if it satisfies certain properties of the system, interpreted in terms of logic formulas. It is a powerful technique for enhancing the quality of software systems, e.g., by identifying flaws that would not have been caught otherwise [Havelund et al (2001); Kaivola (2005)]. As such, model checking can play a vital role in verifying key properties of products in high-integrity product lines such as pacemakers, medical imaging systems, and avionics control systems.…”