Unifying Compilers for SNARKs, SMT, and More

The programming languages community, the cryptography community, and others rely on translating programs in highlevel source languages (e.g., C) to logical constraint representations. Unfortunately, building compilers for this task is difficult and time consuming. In this work, we show that all of these communities can build upon a shared compiler infrastructure, because they all share a common abstraction: stateless, non-deterministic computations that we call existentially quantified circuits, or EQCs. To make our approach concrete we create CirC, an infrastructure for building compilers to EQCs. CirC makes it easy to add support for new EQCs: we build support for two, one used by the PL community and one used by the cryptography community, in ≈2000 LOC. It’s also easy to extend CirC to support new source languages: we build a featurecomplete compiler for a cryptographic language in one week and ≈700 LOC, whereas the reference compiler for the same language took years to write, comprises ≈24000 LOC, and produces worse-performing output than our compiler. Finally, CirC enables novel applications that combine multiple EQCs. For example, we build the first pipeline that (1) automatically identifies bugs in programs, then (2) automatically constructs cryptographic proofs of the bugs’ existence.