In this talk I will demonstrate some features of my experimental
software model checker Hector, which combines different abstraction
domains to prove safety properties of heap-manipulating imperative
programs. I will concentrate in particular on the cooperative use of
predicate abstraction (as in SLAM) alongside a heap shape analysis (as
in the TVLA system).

“Modular” combination of domains is achieved, during the construction of
a transition-system-like model of the program, by allowing the various
abstraction domains to exchange constraints on program state, using a
single intermediate logic. Models of programs can then be checked with a
fragment of LTL, which can express many properties, including the usual
memory safety and assertion non-violation. (Joint work with Michael Huth)

Registration required:

No

Free:

Yes