Beyond decompilation: multi-level lifting for automatic software understanding
Last year we celebrated 30 years of decompilation. The idea that lifting a binary into a well-structured source-level form could be done algorithmically was so hugely powerful that it arguably obscured other worthwhile questions: into what other useful representations could a binary be lifted? Could these lifted forms enable some kinds of automated software understanding better than source-like representations? Could such representations be mathematically defined and "baked to order" for a particular kind of automated program analyses?
Various research efforts over the years provided partial answers to these questions. This year's panel will revisit these questions and these answers, and will try to map out promising practical directions of lifting for program analysis.