Formal Change Impact Analyses of Extended Finite State Machines Using a Theorem Prover

Abstract

This paper describes a formal change impact analysis approach for systematic evolution of communicating systems. Systems are modeled using a network of communicating extended finite state machines (CEFSMs) with variables ranging over commonly used data types including numbers, booleans, arrays, and object fields. Parameterized messages exchanged over queues and shared variables are used for communication. Changes to the system are performed at the transition level by adding/deleting transitions. Given a changed transition, the impacted system transitions are automatically computed using a bounded, selective, state exploration based on the inductive assertion approach. A theorem prover extended with queue axioms is used to discharge the verification conditions. Multiple symbolic values for each variable present in a system state are represented as a set of rewrite rules to minimize state space overheads. Rewrite-rule based procedures are described for reducing the number of symbolic values in system states. We also describe heuristics to identify simultaneously enabled and disabling transitions and describe a procedure to reduce the number of verification conditions generated during the impact analysis. The effectiveness of the proposed approach is illustrated on several applications including web services and cache coherence protocols.

Code-Change Impact Analysis Using Counterfactuals: Theory and Implementation

“…Guo et al [11] undertake the task of performing change-impact analysis in the context of distributed systems by using model checking. The authors chose a model that consists of a set of networked state machines (i.e.…”

“…Adding and/or deleting transitions from the composite state machine represents a change in the system. The work presented in [11] di®ers from ours in two ways: (1) the authors Code-Change Impact Analysis Using Counterfactuals: Theory and Implementation 1461 apply change-impact analysis to distributed systems via model checking; our approach performs change impact analysis in the context of sequential programs and it is based on proof-theory. (2) The authors of [11] manage change by applying partial order reduction to the global state space.…”

Using Change Impact Analysis to Select Tests for Extended Finite State Machines

“…This work builds on our earlier work on formal change impact analysis [4]. There, we developed an approach to identify transitions affected by a maintenance change whereas this work identifies tests affected by a change.…”

“…The post-image of t is automatically generated using t; I(O)Q 0 and I(O)Q 1 stand for the queues before and after executing t respectively; hd, deq, and enq denote the head of queue and dequeue and enqueue operations respectively; − → ne denotes the parameter expressions in the output message using the latest instances of variables; nA t is similarly obtained from the action statements A t after translating them into equalities [4], [12].…”

“…The pre-image of a transition t from a state g, P re(t) denotes the set of global states in which t can be executed to reach state g. It can be similarly computed from t and g [4], [12].…”