Section: New Results

Improving the Performance and Robustness of Systems Software in Multicore Architectures

Managed Runtime Environments

Participants : Bertil Folliot, Julia Lawall, Gilles Muller [correspondent] , Marc Shapiro, Julien Sopena, Gaël Thomas, Florian David, Lokesh Gidra, Jean-Pierre Lozi, Thomas Preud homme, Suman Saha, Harris Bakiras, Arie Middelkoop, Koutheir Attouchi.

Today, multicore architectures are becoming ubiquitous, found even in embedded systems, and thus it is essential that managed languages can scale on multicore processors. We have found that a major scalability bottleneck is the implementation of high contention locks, which can overload the bus, eliminating all performance benefits from adding more cores. To address this issue, as part of the PhD of Jean-Pierre Lozi, we have developed remote core locking (RCL), in which highly contended locks are implemented on a dedicated server, minimizing bus traffic and improving application scalability (USENIX ATC 2012 [24] ). This work initially targeted C code but is now being adapted to the needs of Java applications in the PhD of Florian David. Another bottleneck in the support for managed languages is the garbage collector. As part of the PhD of Lokesh Gidra, we have identified the main sources of overhead.

Systems software robustness

A new area of research for Regal, with the arrival of Gilles Muller in 2009 as Inria Senior Research Scientist and Julia Lawall in 2011 as Inria Senior Research Scientist, is on improving the reliability of operating systems code. Muller and Lawall previously developed Coccinelle, a scriptable program matching and transformation tool for C code that is now commonly used in the open-source development community, including by the developers of Linux, Wine and Dragonfly BSD. Based on Coccinelle, we have developed a new approach to inferring API function usage protocols from software, relying on knowledge of common code structures (Software – Practice and Experience [19] ).

We have also proposed a method for automatically identifying bug-fixing patches, with the goal of helping developers maintain stable versions of the software (ICSE 2012 [45] ) and have designed an approach to automatically generating a robust interface to the Linux kernel, to provide developers of new kernel-level code more feedback in the case of a misunderstanding of kernel API usage conventions (ASE 2012 [24] ).