M. Seltzer, Y. Endo, C. Small, and K. Smith. An introduction to the architecture of the VINO kernel. Technical Report TR-34-94, EECS Department, Harvard University, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....many groups have applied online reconfiguration to systems and achieved a variety of benefits. Many different adaptive techniques have been implemented to improve system performance [2, 25, 35] Extensible operating systems have shown performance benefits for a number of interesting applications [7, 19, 49]. Technologies such as compiler directed I O prefetching [10] and storage latency estimation descriptors [39] improve application performance using detailed knowledge about the state of system structures. Incremental and optimistic specialization [43] can remove unnecessary logic for common case ....

M. Seltzer, Y. Endo, C. Small, and K. A. Smith. An introduction to the architecture of the VINO kernel. Technical report. 1994.

.... terme de temps au niveau de leur utilisation h la compilation (approche statique) au chargement ou h l ex6cution (approche dynamique) Une 6tude comparative de ces techniques revient h confronter 1 amortissement Fonctionnalit extension sdcurisde Scurit PCC[3d Virtual Virtual Machines VINO [40] TAL[3 VVM [14] Exokernel [12] SFI [43] Camille [16] micro noyau Vrifieur Java [27] SNAP[32] temps rdel ouvert disponibilitd algorithme d ordonnancement micro noyau temps rel EMERALDS [44] SPRING [35] Qualit de Service Figure 1 Classification possible des techniques et systmes ....

....cornroe les noyaux, les exo noyaux, les systames extensibles, peu de systames ont essay de mettre la scurit comme point central. Parmi ceux qui se sont penchs sur le problame, on peut citer: SPIN [2] qui intagre un systame de capacits [19] audessus d un langage s0r: Modula 3 [5] VINO [40] qui compose un systame transactionnel [39] au dessus de techniques SFI intgres dans un compila teur [41 ] Les exokernels [12] qui utilisent un systame de capacitts hitrarchiques [30] au plus pras du mattriel, m6me si d autres techniques peuvent leur 6tre appliqutes [ 13] Camille [16] ....

SELTZER, M., ENDO, Y., SMALL, C., AND SMITH, K. A. An Introduction to the Architecture of the VINO Kernel. Tech. Rep. 34-94, VINO: The 1994.

....implemented as a user process, the errors due to the module are not propagated to the rest of the operating system. However, the performance is sacrificed because the overheads of interprocess communication and context switches are large. The extensible operating systems such as SPIN [2] and VINO [8] alleviate the di#culty of debugging distributed file systems by allowing the users to install an extension module into the kernel on demand. These systems provide a certain fixed level of fail safety using type safe language Modula 3 [4] and software fault isolation [9] respectively, but it is ....

M. I. Seltzer, Y. Endo, C. Small, and K. A. Smith. An Introduction to the Architecture of the VINO Kernel. Technical Report TR--34--94, Harvard University Computer Science, 1994.

....modules and the ability to recover from the errors. However, there is the trade o# between fail safety and performance, and it is di#cult to achieve both su#cient fail safety and good performance. Many 1 microkernel operating systems like Mach [1] and many extensible operating systems like VINO [43] sacrifice the performance for the fail safety more or less. On the other hand, an approach using loadable kernel modules gives up the fail safety for the performance. To avoid this dilemma, we propose a new fail safe mechanism called multi level protection [17, 18, 20, 21, 19] It enables users ....

....A LKM is implemented as a part of the kernel and runs very e#ciently after linked with the kernel. However, the errors due to the module can make the whole operating system crash because fail safety is not considered at all. To solve the problem of LKM, some extensible operating systems like VINO [43, 44] and DECADE [29] protect the extension modules downloaded into the kernel from the rest of the operating system. VINO use software fault isolation [49, 45] to protect the kernel from illegal memory accesses due to the downloaded extension modules. It also limits the maximum amount of resources ....

Seltzer, M. I., Y. Endo, C. Small, and K. A. Smith, "An Introduction to the Architecture of the VINO Kernel," Tech. Rep. TR--34--94, Harvard University Computer Science, 1994.

....virtualization can result in a complex and inefficient OS. In contrast, our approach exports hardware resources rather than emulates them, allowing an efficient and fast implementation. OS extensibility has a long history [15, 20] Current attempts include SPIN [4] Bridge [17] and Vino [22]. Some of the techniques used in these systems, such as type safe languages and software faultisolation [25] are also applicable to exokernels. The commercial world has long looked at this issue in the form of unsafe dynamically loaded device drivers. 6 Conclusions Two decades ago, Lampson ....

Margo Seltzer et al. An introduction to the architecture of the VINO kernel, November 1994.

....or two of these features. Recent operating system research [4] 5] shows the advantage of providing reusable system components as basic building blocks. Other operating systems focus on the flexibility of the system that allows to adapt itself to changes in the application set it must support [13][12]. System software on tomorrow s embedded systems (such as personal digital assistants (PDA s) cellular phones or cars) must be intelligent enough to adapt the internal structure to new, even non anticipated services and to integrate the necessary support for it (such as a new communication ....

M. I. Seltzer, e.a.. An Introduction to the architecture of the VINO Kernel. Harvard University Center for Research in Computing Technology Technical Report TR-34-94, 1994.

....clients. Recently, various proposals have been made to allow application specific extension code to be downloaded dynamically into a server software, so as to customize its access policies to meet the specific need of a client software. Typical examples include extensible operating systems [11, 102, 37], active network [111, 20] active disk [4] and many others. 4. Avoiding distribution of state: In traditional client server applications, the state of computation is distributed among the servers and the clients. As a consequence, it is difficult to maintain consistency of the distributed ....

....this, one needs to be able to hierarchically construct a child protection domain for executing untrusted code within the invoker s process. Layered protection is a characteristic feature in single address space operating systems like OPAL [19] and extensible operating systems like SPIN [11] VINO [102], and Exokernel [37] In such operating systems, untrusted code may be (dynamically) introduced into a privileged protection domain (e.g. the kernel) One wants to avoid the untrusted code units from exploiting the resources in that domain. In the following discussion, we will refer to some of ....

[Article contains additional citation context not shown here]

Margo Seltzer, Yasuhiro Endo, Christopher Small, and Keith A. Smith. An introduction to the architecture of the vino kernel. Technical Report 34-94, Harvard Computer Center for Research in Computing Technology, 1994. Also available at ftp://das-ftp.harvard.edu/ techreports/ tr-34-94.ps.gz.

....application programmer explicitly, which is contrary to the required transparency. Kernel based systems currently devote a lot of attention to satisfy the requirement of adaptability and extensibility of operating system services. Approaches like SPIN[BSP 95] Paramecium [vDHT95] and VINO [SESS94] are characterized by speci c features that allow dynamic integration of application level code into the protection domain of the kernel. To exploit adaptability features programmers are burdened with performing resource management tasks explicitly. This aggravates the programming of distributed ....

M. Seltzer, Y. Endo, C. Small, and K. Smith. An introduction to the architecture of the vino-kernel. Technical Report TR-34-94, Computer Science, Harvard University, 1994.

....AK is time consuming, as it will call IPC functions many times and need to load and unload many kernel object descriptors. Their design is adequate for embedded systems , but not general purpose, memory intensive applications. Other developing systems, the SPIN [3] Exokernel [11] and VINO kernel [23], have the ability to export the memory caching management to user applications. Specific applications can dynamic load the executable object codes into the operating system kernel to tailor the system service to match their needs. The system safety is based on the advanced compiling techniques ....

M. Seltzer and Y. Endo and C. Small and K. A. Smith, `An Introduction to the Architecture of the VINO Kernel', Technical Report TR-34-94, Harvard University, 1994.

....[11] There are two main techniques used for implementing runtime extensible systems. ReAEective systems [14] 10] are object oriented and de ne a Meta Object Protocol (MOP) on system objects which can be used to modify aspects of the objects implementation or environment. Ad hoc systems [2] [12], 6] simply rely on exporting an API which empowers applications to extend parts of the system with their own code sequences at any instance in time. 3 Architectural Overview 3.1 Object Model DEIMOS adopts a mature and well understood object based model for the structuring of its system ....

Seltzer, M., Endo, Y., Small, C., Smith, K., "An Introduction to the Architecture of the VINO Kernel". Harvard University Computer Science Technical Report 34-94, 1994.

....this, one needs to be able to hierarchically construct a child protection domain for executing untrusted code within the invoker s process. Layered protection is a characteristic feature in single address space operating systems like OPAL [20] and extensible operating systems like SPIN [12] VINO [99], and Exokernel [41] In such operating systems, untrusted code may be (dynamically) introduced into a privileged protection domain (e.g. the kernel) One wants to avoid the untrusted code 3 It is revealing to read a passage in a standard operating systems text [100, p 112] However, unlike ....

....because additional code is introduced by the transformation. Despite this overhead, native code that is instrumented this way can run at speeds comparable to the original code [120] although not as efficiently as a proof carrying code version [77, 79] In extensible operating systems VINO [99] and Exokernel [41] users are allowed to dynamically download untrusted extension code into the kernel s address space to modify the behavior of the operating systems. To avoid corruption of the kernel address space, untrusted extension code units are subject to SFI transformation before ....

Margo Seltzer, Yasuhiro Endo, Christopher Small, and Keith A. Smith. An introduction to the architecture of the vino kernel. Technical Report 34-94, Harvard Computer Center for Research in Computing Technology, 1994. Also available at ftp://das-ftp.harvard.edu/techreports/tr-34-94.ps.gz.

....and the ability to recover from the errors. However, there is the trade off between fail safety and performance, and it is difficult to achieve both sufficient fail safety and good performance. Many microkernel operating systems like Mach [1] and many extensible operating systems like VINO [43] sacrifice the performance for the fail safety more or less. On the other hand, an approach using loadable kernel modules gives up the fail safety for the performance. To avoid this dilemma, we propose a new fail safe mechanism called multi level protection [17, 18, 20, 21, 19] It enables users ....

....LKM is implemented as a part of the kernel and runs very efficiently after linked with the kernel. However, the errors due to the module can make the whole operating system crash because fail safety is not considered at all. To solve the problem of LKM, some extensible operating systems like VINO [43, 44] and DECADE [29] protect the extension modules downloaded into the kernel from the rest of the operating system. VINO use software fault isolation [49, 45] to protect the kernel from illegal memory accesses due to the downloaded extension modules. It also limits the maximum amount of resources ....

Seltzer, M. I., Y. Endo, C. Small, and K. A. Smith, "An Introduction to the Architecture of the VINO Kernel," Tech. Rep. TR--34--94, Harvard University Computer Science, 1994.

....to individual projects in this type of system and include software fault isolation (SFI) Wahbe,93] safe languages e.g. Modula 3, proof carrying code (PCC) Necula,96] and interpreted languages e.g. Java. Classic examples of dynamically extensible systems include SPIN [Bershad,95] and VINO [Seltzer,94] 11. Summary and conclusions DEIMOS is a minimally prescriptive framework for the construction of dynamically extensible operating systems. The framework is based on a well understood object based computational model borrowed from the distributed systems community that supports fully general ....

Seltzer, M., Endo, Y., Small, C., Smith, K., "An Introduction to the Architecture of the VINO Kernel". Harvard University Computer Science Technical Report 34-94, 1994.

....for example, working at CMU (integrated multi server concept, project Bridge [Lucco94] Wahbe 93] and at the University of Washington (Spin [Bershad95] These projects mainly explore secure ways to bring new code into the operating system kernel in order to adapt its behavior. The VINO system [Seltzer] tries to give the applications a far less restricted access to kernel primitives in order to encourage reuse and efficiency. 8 FUTURE WORK 10 The Lipto system [Druschel92] tries to overcome the congruence between the module borders motivated by software technology and the borders of security ....

Seltzer, M., Y. Endo, C. Small, K. A. Smith: 'An Introduction to the Architecture of the VINO Kernel'. Harvard University: http://www.eecs.harvard.edu/~chris/papers/arch-intro.ps

....is the explicit view that the kernel should not provide high level core abstractions. In other systems, the effective operating system interface is much higher level (e.g. page tables are implemented by the kernel) Current extensible OS projects include Scout [20] Bridge [30] and Vino [41]. Some of the techniques used in these systems, such as type safe languages [7, 32, 37] and software fault isolation [16, 46] are also applicable to exokernels. These systems are just beginning to be constructed, so it is difficult to determine their relationship to exokernels in general and ....

Margo Seltzer et al. An introduction to the architecture of the VINO kernel, November 1994.

....implemented as a user process, the errors due to the module are not propagated to the rest of the operating system. However, the performance is sacrificed because the overheads of interprocess communication and context switches are large. The extensible operating systems such as SPIN [2] and VINO [8] alleviate the difficulty of debugging distributed file systems by allowing the users to install an extension module into the kernel on demand. These systems provide a certain fixed level of fail safety using type safe language Modula 3 [4] and software fault isolation [9] respectively, but it is ....

M. I. Seltzer, Y. Endo, C. Small, and K. A. Smith. An Introduction to the Architecture of the VINO Kernel. Technical Report TR--34--94, Harvard University Computer Science, 1994.

....places SPIN in the kernel level, extensible, software protected class. Extensions have kernel lifetimes and are limited procedural extensible, in that they are restricted to those interfaces explicitly exported by the domains that implement particular services. VINO The goal of the VINO project [18] is to provide resource intensive applications greater control over resource management. VINO supports the downloading of kernel extensions, which are written in C and protected using software fault isolation [22] To facilitate graceful recovery from an extension failure, VINO runs each ....

Seltzer, M., Endo, Y., Small, C., Smith, C. "An Introduction to the Architecture of the VINO Kernel," Harvard University Center for Research in Computing Technology Technical Report 34-94, 1994.

....application specific ones. We call this architecture procedural extensibility because extensions can be defined in arbitrarily small units. A special case of procedural extensibility occurs when the entry points with which extensions can be associated is limited. For example, in the VINO system [13], extensions may be associated only with designated kernel entry points. We call these architectures limited procedural extensible. Arbitration Between Conflicting Extensions The final issue we consider is that of arbitrating between conflicting kernel extensions. Once applications are given the ....

.... not persist across system reboots (and hence must be reloaded at boot time or as needed) The kernel controls resource allocation (e.g. determines when to allocate the CPU to each process) each process can then control resource use (e.g. which thread to schedule during its time slice) VINO VINO [13] is a downloadable system, that uses sandboxing to enforce trust and security. Extensions are written in C and are compiled by a trusted compiler that prevents the extension from accessing text or data outside its own address range, except for permitted entry points. This trusted code is then ....

Seltzer, M., Endo, Y., Small, C., Smith, C. "An Introduction to the Architecture of the VINO Kernel," Harvard University Center for Research in Computing Technology Technical Report 34-94, 1994.

....for a particular workload on a particular machine. By targeting a particular kernel for a particular workload, Scout can take advantage of advanced compiler optimization techniques, efficient kernel code paths, and a smaller system footprint. The extensible systems, such as SPIN [4] and VINO [15] allow applications to download code into the kernel to provide improved functionality and performance. Synthetix [13] provides improved flexibility and functionality by identifying commonly executed paths and producing optimized versions of them, but does not allow applications to modify or ....

Seltzer, M., Endo, Y., Small, C., Smith, K., "An Introduction to the Architecture of the VINO Kernel," Harvard University Computer Science Technical Report 34-94, 1994.

No context found.

M. Seltzer, Y. Endo, C. Small, and K. Smith. An introduction to the architecture of the VINO kernel. Technical Report TR-34-94, EECS Department, Harvard University, 1994.

No context found.

Margo I. Seltzer, Yasuhiro Endo, Christopher Small, and Keith A. Smith. An introduction to the architecture of the VINO kernel. Technical Report 34-94, Harvard University, Dept. of Computer Science, 1994. 167

No context found.

M. Seltzer, Y. Endo, C. Small, and K. Smith. An introduction to the architecture of the VINO kernel. Technical Report TR-34-94, EECS Department, Harvard University, 1994.

No context found.

M. Seltzer, Y. Endo, C. Small, and K. Smith. An introduction to the architecture of the VINO kernel. Technical Report TR-34-94, EECS Department, Harvard University, 1994.

No context found.

M. Seltzer, Y. Endo, C. Small, and K. A. Smith. An introduction to the architecture of the VINO kernel. Technical report. 1994.

No context found.

Seltzer, M., Endo, Y., Small, C., Smith, K., "An Introduction to the Architecture of the VINO Kernel". Harvard University Computer Science Technical Report 34-94, 1994.

First 50 documents

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC