Shape analysis concerns the problem of determining "shape invariants" for programs that perform destructive updating on dynamically allocated storage. This paper presents a new algorithm that takes as input an abstract value (a 3-valued logical structure describing some set of concrete stores X) and a precondition p, and computes the most-precise abstract value for the stores in X that satisfy p. This algorithm solves several open problems in shape analysis: (ⅰ) computing the most-precise abstract value of a set of concrete stores specified by a logical formula; (ⅱ) computing best transformers for atomic program statements and conditions; (ⅲ) computing best transformers for loop-free code fragments (i.e., blocks of atomic program statements and conditions); (ⅳ) performing interprocedural shape analysis using procedure specifications and assume-guarantee reasoning; and (ⅴ) computing the most-precise over approximation of the meet of two abstract values. The algorithm employs a decision procedure for the logic used to express properties of data structures. A decidable logic for expressing such properties is described in [5]. The algorithm can also be used with an undecidable logic and a theorem prover; termination can be assured by using standard techniques (e.g., having the theorem prover return a safe answer if a time-out threshold is exceeded) at the cost of losing the ability to guarantee that a most-precise result is obtained. A prototype has been implemented in TVLA, using the SPASS theorem prover.
展开▼
