Cerberus

C remains central to our computing infrastructure but still lacks a clear and complete semantics. Programmers lack tools to explore the range of behaviours they should expect; compiler development lacks test oracles; and formal verification and analysis must make (explicitly or implicitly) many choices about the specific C they target. The ISO standards for C have been developed over many years but they remain unclear in some important respects, and in some areas there are differences between the properties of C that are relied on by the corpus of systems code in production, the properties that compiler implementers aim to provide, and the properties of C as it is specified in ISO C11.

The Cerberus project is developing a semantic model for a substantial fragment of C. It has several distinctive features:

Where the ISO C11 standard is clear and corresponds with practice, Cerberus aims to follow that.
Where there are ambiguities or differences, chiefly in the memory layout model (the behaviour of pointers, pointer arithmetic, uninitialised values, etc.), we aim to clarify the de facto standards and understand how the ISO standard could be reconciled with them.
Cerberus precisely defines the range of allowed behaviour, not just that of some specific implementation.
It is executable, to explore either all behaviours or single paths of small test programs.
Its thread-local semantics is factored via an elaboration into a simpler Core language, to make it readable and conceptually and mathematically tractable; the dynamic semantics of Core can be linked with our sequential memory layout model or with that operational concurrency model.
Its front-end is written from scratch to closely follow the C11 standard, including a parser that follows the C11 standard grammar, and a typechecker.
It is designed to be integrated - though this is not currently maintained - with an operational model for C11 concurrency, with a mechanised proof of equivalence to the axiomatic C11 model of Batty et al.

This is a step towards a clear, consistent, and unambiguous semantics for C.

Cerberus Web Interface

The Cerberus web interface lets one interactively, randomly, or exhaustively explore the behaviour of small C test programs in the Cerberus semantics.

Papers

Into the Depths of C: Elaborating the De Facto Standards. Kayvan Memarian, Justus Matthiesen, James Lingard, Kyndylan Nienhuis, David Chisnall, Robert N. M. Watson, and Peter Sewell. In Proc. 37th ACM SIGPLAN Conference on Programming Language Design and Implementation (PLDI '16). Distinguished paper award.
An operational semantics for C/C++11 concurrency. Kyndylan Nienhuis, Kayvan Memarian, and Peter Sewell. In OOPSLA 2016. This gives an operational semantics for C11 concurrency, proved equivalent (in Isabelle/HOL) to the axiomatic model of Batty et al.

The C Memory Layout Model: ISO WG14 papers

Clarifying Pointer Provenance v4

WG14 Brno meeting working drafts 2018-04

WG14 Brno pre-meeting mailing, 2018-04

N2223: Clarifying the C Memory Object Model: Introduction to N2219 - N2222
N2219: Clarifying Pointer Provenance (Q1-Q20) v3
N2220: Clarifying Trap Representations (Q47) v3
N2221: Clarifying Unspecified Values (Q48-Q59) v3
N2222: Further Pointer Issues (Q21-Q46)
Structure and Union Padding (Q60-Q72): Section 3.3 of our revised N2013
Effective Types (Q73-Q81): Section 4 of our revised N2013

WG14 Pittsburgh meeting, 2016-10

WG14 London meeting, 2016-04

Here is a tarball of our test programs, including a tool to run them and log the compile- and run-time output:

defacto_tarball.tar

Survey

In April-September 2015 we distributed a survey of 15 questions about C (What is C in practice? (Cerberus survey v2)). We were asking what C is in current mainstream practice: the behaviour that programmers assume they can rely on, the behaviour provided by mainstream compilers, and the idioms used in existing code, especially systems code. We were explicitly not asking what the ISO C standard permits, which is often more restrictive, or about obsolete or obscure hardware or compilers. We focussed on the behaviour of memory and pointers. We had responses from 323 people, including many compiler and OS developers. Here is a summary of the results and the (anonymised) detailed comments:

What is C in practice? (Cerberus survey v2): Analysis of Responses, Kayvan Memarian and Peter Sewell. (WG14 N2014)
What is C in practice? (Cerberus survey v2): Analysis of Responses - with Comments, Kayvan Memarian and Peter Sewell. (WG14 N2015)

People

Contributors:

The current main developers are Kayvan Memarian and Victor Gomes. Kyndylan Nienhuis worked on the operational semantics for C11 concurrency. Cerberus originated with Justus Matthiesen's 2010-11 Part II project dissertation and his 2011-12 MPhil dissertation; James Lingard's 2013-14 MPhil dissertation developed a certifying translation validator for simple C programs for the Clang front-end, w.r.t. the Cerberus and Vellvm semantics.

Funding

This work is funded by REMS: Rigorous Engineering of Mainstream Systems, EPSRC Programme Grant EP/K008528/1, EPSRC grant EP/H005633 (Leadership Fellowship, Sewell), and a Gates Cambridge Scholarship (Nienhuis). This work is also part of the CTSRD projects sponsored by the Defense Advanced Research Projects Agency (DARPA) and the Air Force Research Laboratory (AFRL), under contract FA8750-10-C-0237. The views, opinions, and/or findings contained in this paper are those of the authors and should not be interpreted as representing the official views or policies, either expressed or implied, of the Department of Defense or the U.S. Government.