Bounded Model Checking for C programs in an enterprise environment
Research Scientist at Amazon Web Services and Lecturer in Theoretical Computer Science at Queen Mary, University of London
Software model checking tools promise to deliver genuine traces to errors, and sometimes even proofs of their absence. As static analysers, they do not require concrete execution of programs, which may be even more beneficial when targeting new platforms. Academic research focusses on improving scalability, yet largely disregards practical technical challenges to make tools cope with real-world code. At Amazon, both scalability requirements as well as real-world constraints apply. Our prior work analysing more than 25,000 software packages in the Debian/GNU Linux distribution containing more than 400 million lines of C code not only led to more than 700 public bug reports, but also provided a solid preparation for the challenges at Amazon.