Building efficient and highly run-time adaptable virtual machines

Abstract

Programming language virtual machines (VMs) realize language semantics, enforce security properties, and execute applications efficiently. Fully Reflective Execution Environments (EEs) are VMs that additionally expose their whole structure and behavior to applications. This enables developers to observe and adapt VMs at run time. However, there is a belief that reflective EEs are not viable for practical usages because such flexibility would incur a high performance overhead.To refute this belief, we built a reflective EE on top of a highly optimizing dynamic compiler. We introduced a new optimization model that, based on the conjecture that variability of low-level (EE-level) reflective behavior is low in many scenarios, mitigates the most significant sources of the performance overheads related to the reflective capabilities in the EE. Our experiments indicate that reflective EEs can reach peak performance in the order of standard VMs. Concretely, that a) if reflective mechanisms are not used the execution overhead is negligible compared to standard VMs, b) VM operations can be redefined at language-level without incurring in significant overheads, c) for several software adaptation tasks, applying the reflection at the VM level is not only lightweight in terms of engineering effort, but also competitive in terms of performance in comparison to other ad-hoc solutions.

A Metaobject Protocol for Optimizing Application-Specific Run-Time Variability

“…What follows is an informal characterization of di erent kinds of variability we have observed in our experiments using Truf-eMate as an adaptive platform [3]. While we assume they have been observed by many VM implementers before, to the best of our knowledge, they have not been widely documented yet.…”

“…Tru eMate uses a self-optimizing AST interpreter featuring dynamic objects and dispatch chains for speculative optimizations. Preliminary evidence [3] suggests that, in combination with the Graal [11] JIT compiler, Tru eMate can run e ciently.…”

Fully Reflective Execution Environments: Virtual Machines for More Flexible Software

“…When designing a reflective system, this tension must be resolved. Fortunately, just-in-time compilers can significantly reduce the overhead of intercession handlers whenever the usage of the reflective capabilities are moderate or exhibit stable usage patterns [43], [44].…”

“…To complement our qualitative evaluation and show indications that fully reflective VMs have the potential to be competitive, we report here on some of the performance results obtained in [44]. Specifically, we use the latest version of TruffleMATE to rerun two of the experiments presented in that work: i) a performance analysis of the VM when no metaobject is activated and ii) the performance of the immutable references scenario.…”

“…To assess the performance impact in applications using reflective capabilities, we took the read-only protection experiment from [44] which resembles the scenario presented in section 7.2.2. The benchmark essentially traverses a linked list, attempting to write some of its elements.…”