Lightweight Shape Analysis Based on Physical Types

To understand and detect possible errors in programs manipulating memory, static analyses of various levels precision have been introduced, yet it remains hard to capture both information about the byte-level layout precise global structural invariants. Classical pointer struggle with latter, whereas advanced shape incur a higher computational cost. In this paper, we propose new memory analysis by abstract interpretation that summarizes heap means type invariant, using novel kind physical types, which express values memory. terms expressiveness, our abstraction aims at middle point between typical analyses, hence lightweight name. We pair summarizing retained staged points-to predicates refines regions are use, hereby allowing strong updates without introducing disjunctions. show combination abstractions suffices verify spatial safety non-trivial invariants presence low-level constructs such as arithmetic dynamic allocation, on C binary code.