Michel Gien

A Practical Look at Micro-Kernels and Virtual Machine Monitors

In this paper, we look at two different approaches used to provide embedded system support for virtualization and virtual machine monitors for consumer electronics and mobile devices. We compare the micro-kernel approach, which has been a popular choice for building embedded operating systems with the Virtual Machine Monitor (VMM) or hypervisor approach widely deployed in general purpose computing environments such as desktops and data center servers. Comparison criteria are based on virtualization use cases that are typical of Consumer Electronics (CE) systems such as mobile devices and IPTV. These approaches are further evaluated based on performance and on their ability to allow re-use of existing (often real-time) software as well as modern open operating systems such as Linux while remaining as transparently as possible. Such transparency can come through different paths, including: leveraging of hardware virtualization support, minimal modifications to the original operating system internals (kernel, device drivers, etc.), and the ability to use existing operating system applications as-is and without the need to port them to a new environment. An analysis of the fundamental principles behind each approach is presented with a discussion of their impact on existing operating environments, together with practical performance results based on existing micro-kernels and real-time hypervisor benchmarks. We conclude that mapping the VMM (hypervisor) approach used in data centers to the needs of embedded systems is a better option for the support of complete operating systems (as guests) than extending micro-kernels for such functionality.

Architectural issues in microkernel-based operating system: the CHORUS experience

Abstract An important trend in operating system development is the restructuring of the traditional monolithic operating system kernel into independent servers running on top of a minimal nucleus or ‘microkernel’. This approach arises out of the need for modularity and flexibility in managing the ever-growing complexity caused by the introduction of new functions and new architectures. In particular, it provides a solid architectural basis for distribution, faulttolerance and security. Microkernel-based operating systems have been a focus of research for a number of years, and are now beginning to play a role in commercial UNIX ∗ systems. The ultimate feasibility of this attractive approach is not yet widely recognized, however. A primary concern is efficiency: can a microkernel-based modular operating system provide performance comparable to that of a monolithic kernel when running on comparable architectures? The elegance and flexibility of the client-server model may exact a cost in message-handling and context-switching overhead. If this penalty is too great, commercial acceptance will be limited. Another pragmatic concern is compatibility: in an industry relying increasingly on portability and standardization, compatible interfaces are needed not only at the level of application programs, but also for device drivers, streams modules and other components. In many cases, binary as well as source compatibility is required. These concerns affect the structure and organization of the operating system. The Chorus team has spent the past six years studying and experimenting with UNIX ‘kernelization’ as an aspect of its work in distributed and real-time modular systems. Aspects of the current Chorus † system are examined here in terms of its evolution from the previous version. The focus is on pragmatic issues such as performance and compatibility, as well as considerations of modularity and software engineering.

Shared device driver model for virtualized mobile handsets

One of the major expected benefits of hardware virtualization in mobile handsets is the ability to use a peripheral device, for which an original device driver has been written for a particular operating system, from applications running on other operating systems. The savings and new opportunities generated by avoiding writing multiple device drivers for the same peripheral because for different operating systems (Symbian, Windows, Linux, and proprietary real-time operating systems) are tremendous, and relevant to the whole eco-system (chipset and hardware peripheral vendors, device manufacturers, Independent Software Vendors (ISVs), service providers and even end-users. This paper quantifies the scope of the problem in terms of investments, time and effort, and lost opportunities. The authors then propose an architecture based on a virtualization layer, allowing sharing and control of physical peripheral device drivers among multiple execution environments running/hosting different operating systems. The paper concludes with a discussion on research topics generated by such distributed device driver architecture within a single handset, in the areas of performance optimization, sharing policies to guarantee quality of service, access control policies, dependability (stop and restart of a device driver), power management optimization, and partial or incremental system reconfiguration.

Virtual Memory Management in Chorus

The Chorus technology has been designed for building “new generations” of open, distributed, scalable operating systems. It is based on a small kernel onto which operating systems are built as sets of distributed cooperating servers. This paper presents the Virtual Memory Management service provided by the Chorus kernel. Its abstractions, interfaces and some implementation issues are discussed. Some examples of the use of this interface by our distributed Unix implementation are given.

Resource-level autonomy in CHORUS

The Chorus project, at INRIA, designed and developed two operational versions of the CHORUS I Distributed Operating System, between 1980 and 1986 ([9,10]). CHORUS V2 has a UNIX ~ System V interface [I], and is currently used as a basis for supporting half a dozen of research distributed applications. A new version, CHORUS V3, has been developed by Chorus syst~mes [4], building on the experience of state-of-the art research systems such as CHORUS V2 [7], the V-System [3], Amoeba [5] and Mach [2], while taking into account constraints of the industrial environment. CHORUS V3 emphasizes portability, compatibility with standards and scalability [8], implying the ability to configure a system on a wide range of operational environments.