C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, Annapolis, 1996. resp. prefix GF: generator generator-family suffix C: completion  Home/Search   Document Details and Download   Summary   Related Articles   Check

....reach of a traditional PC further into our everyday lives. For example, a small device can access a database on a web server; aPC can provide a user interface for examining and analyzing sensor data in a home . Database Server Radio tower Smart TAG Radio tower Tag READER Win2K [1] I m #2 [3] #2 is O Brian [4] OK [2] Who is #2 Recognition action Figure 1: SOAP messages in a smart tag scenario Consider the system in Figure 1. A person carrying an active badge approaches the entrance of a secure building, which is equipped with a tag reader. Access can be granted based ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In the proceedings of the International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis MD, 1996.

....unlikely that such approaches will scale, given the degree of cooperation required from the application and the level of sophistication this demands of the implementation. It should be clear that replacing the stack frame and restarting the procedure is not a viable option in general. Cowan et al. [15] describe an approach where code update for a function or method is blocked while a client thread is executing the old version code; this has obvious problems if the client thread is executing a long lived loop in the code. Another approach is to simply kill any threads running the code that is ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis, MD, 1996.

....tools are also discussed, since existing kernel instrumentation systems are few. 2.1. 1 Extensibility in Research Kernels Extensible operating systems such as SPIN [10, 11, 37, 68, 84] Exokernel [34, 35, 36, 47] VINO [81, 82, 83, 87] Cache Kernel [21] Scout [41, 59, 61, 62] and Synthetix [23, 24, 73, 74, 75] allow user level processes to download code into the kernel, offering a coarse grained means of changing kernel code dynamically. SPIN allows applications to change the operating system s interface and implementation through extensions. SPIN extensions assign a handler routine to a specific ....

....commodity kernel. These steps are installing measurement and trace gathering code at run time, simulating in situ a proposed new policy algorithm, and installing that algorithm in place of the existing one. 2.3. 4 Run time Optimizations and Evolving Code in Commodity Kernels The Synthetix project [23, 24, 73, 74, 75] is a follow up to the Synthesis project, performing specialization on a modified version of a commodity operating system, HP UX. Synthetix allows the implementation of an algorithm (the read system call in particular) to be specialized on a per file and per process basis. The specialization can ....

C. Cowan, T. Audrey, C. Krasic, C. Pu, and J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. International Conference on Configurable Distributed Systems (ICCDS), Annapolis, MD, May 1996.

....mechanisms suitable for implementing mutation can be divided into three groups: Mutual exclusion: Mutation cannot happen while workers are executing methods of the object to be mutated. The implementation can be a read write lock, disabling preemption on a uniprocessor, or a holding tank [Cowan96] with reference counting. Mutual exclusion is simple in that there is no worker state associated with the object when mutation is allowed to happen. Transactional: Roll back the workers that are affected by mutation. Mutators and workers operate on an object transactionally and can be aborted ....

....Chorus [Rozier88] is the only system we know of that can be configured to use either a page based or a segmentbased VM system. MMLite is the first one that can run with or without VM, and dynamically load and unload it unless, of course, we look at MS DOS in a very twisted way. Synthetix [Cowan96] employs a limited form of object mutation for specialization. We have generalized mutation and made it usable in a number of new contexts. The synchronization mechanism presented in [Cowan96] is a holding tank that keeps workers from entering an object that is being mutated. The holding tank can ....

[Article contains additional citation context not shown here]

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In the proceedings of the International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis MD, 1996.

....designs [Tunes98] A number of other researchers are interested in the general area of adaptive or extensible operating systems. While not using full reflection, these projects have contributed a number of interesting ideas. Examples of adaptive operating systems are Spin [Bershad95] Synthetix [Cowan96], Exokernel Aegis [Engler95] Spring [Mitchell94] Cache Kernel [Cheriton94] Choices [Campbell95] Kea [Veitch96] KTK [Gheith94] Scout [Montz94] and the more architectural work of Chris Maeda on service decomposition [Maeda93, Maeda94] There has also been some related work on incremental ....

Cowan, C., T. Autrey, C. Krasic, C. Pu, J. Walpole, "Fast Concurrent Dynamic Linking for an Adaptive Operating System", Proceedings of the International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis. Maryland, May 1996.

....and only partially automated. Our work differs from these primarily in the scope and completeness of our annotations, which describe only specific implementation properties instead of complete behaviors. Partial evaluation improves performance by specializing routines for specific inputs [4, 5, 7]. The technique combines inlining, constant propagation and constant folding to evaluate as much of the program as possible at compile time. Our work differs in two important ways. First, our technique supports a higher level form of specialization that applies to library operations, not just ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, May 1996.

....can define concepts, such as data distribution, that extend beyond those of the base language. However, unlike most hints and pragmas, the incorrect use of our annotations can lead to transformations that do not preserve the library s semantics. Our work is closely related to partial evaluation [5, 6, 10], which improves performance by specializing routines for specific inputs. Partial evaluation combines inlining, constant propagation and constant folding to evaluate as much of the program as possible at compile time. Recent work in program specialization has generalized partial evaluation to the ....

....contexts. For example, the Open GL graphics standard [21] can customize various matrix transformations to exploit particular properties of matrices and matrix operations. In operating systems, specialized file system I O routines can be produced that are optimized for specific system states [10]: a specialized read routine can be created for the common situation in which the file is known to be open and the file position is correctly positioned to the next unread byte. As a final example, most layered systems can benefit from passing state information across layers [1] providing ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, May 1996.

....at compile time to save run time instructions. The specialized function is discarded when the quasi invariant becomes false. An example of the use of specialization in Synthetix is a specialized read( function with no concurrency checks for use with files opened exclusively for reading [23, 8]. Franz has also proposed making run time code generation a central operating system service. In his system, mostly compiled code is compiled just in time when an application is launched, then recompiled for optimization as the program runs [13] Specialization is partial evaluation working in ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis, MD, May 1996.

....schemes provide varying degrees of flexibility and convenience but all of them require the user to explicit manage the linking. There has also been considerable work on efficient (and safe) implementation of dynamic linking, particularly in the context of shared libraries and kernel extensions [6, 10, 13, 14, 19]. These schemes focus on the linking procedure and the performance of the linked code. They do not address the issue of determining what needs to be linked. Dynamic linking issues for mobile programs, in particular the need to be able to name procedures that access resources local to an execution ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the Third International Conference on Configurable Distributed Systems, pages 108--15, May 1996.

....in the global lookup table. Code loading by multiple threads is serialized. Most published work on systems with code replacement has addressed the question of managing the evolution of code, rather than optimizing the changing code. The closest related work of which we are aware is Synthetix [4]. There, procedures are optimistically specialized to take advantage of current system settings, quasiinvariants [10] If an invariant is violated by some other process (e.g. a file becomes shared) then the system backs out of the optimized code by directing threads into a waiting area and ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In Proc. International Conference on Configurable Distributed Systems (ICCDS'96), IEEE Press, 1996.

....schemes provide varying degrees of flexibility and convenience but all of them require the user to explicit manage the linking. There has also been considerable work on efficient (and safe) implementation of dynamic linking, particularly in the context of shared libraries and kernel extensions [7, 12, 15, 16, 21]. These schemes focus on the linking procedure and the performance of the linked code. They do not address the issue of determining what needs to be linked. Dynamic linking issues for mobile programs, in particular the need to be able to name procedures that access resources local to an execution ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the Third International Conference on Configurable Distributed Systems, pages 108--15, May 1996.

....replugger, in which only four cycles is added to the invocation path. The four cycles consist of setting the flag, clearing the flag, and a branch not taken to wake up a waiting replugger. Further details about the operation and implementation of an HP UX version of the replugger are described in [12], and the code for the Linux version of the replugger is available at www.cse.ogi.edu sysl projects synthetix. 4 Experiments In order to evaluate the effectiveness of our specialization toolkit, we applied it to a wide range of operating system components. This section describes our experiences ....

Cowan, C., T. Autrey, C. Pu, and J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. in International conference on Configurable Distributed Systems (ICCDS'96). 1996. Annapolis, MD.

....mark replaceable components by directly annotating the program, using a few syntax extensions. The management of specialized blocks is done automatically. Cowan et al. describe some execution support for managing several versions of the same procedure, in the context of adaptive operating systems [15]. Their work focuses on a replugging algorithm able to deal with the concurrency issues of an operating system. Finally, let us note that our compiling scheme for generating the execution support generalizes some programming patterns found in object oriented applications dealing with forms of ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In International Conference on Configurable Distributed Systems, Annapolis, MD, May 1996.

....not amenable to replication, because replicas faithfully reproduce the failure [23] 2 Applications that wish to continue to offer some service despite non maskable failure must adapt to the loss of resources. Numerous projects have developed adaptive capabilities for various conditions [2, 4, 5, 10, 12, 14, 16, 20, 21, 24, 28, 29]. However, when one wishes to compose these adaptive techniques, one encounters a complex space of alternatives, especially when some of the adaptations affect the same resources, producing conflicting adaptive behavior. This paper presents adaptation spaces, a framework for navigating among ....

....values like average available bandwidth, using tools such as SWiFT [15] 5. The implementations of adaptation spaces may use a wider variety of mechanisms for switching between adaptation cases. It might be some form of dynamic replugging of code used for specialization of a monolithic kernel [10], or it might be adjustment of configuration parameters, replanning a query, externalizing a program s state then reinitializing a different program from that state, turning on a software feedback loop [5] and so forth. It might even be that no explicit change is needed when changing cases. The ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis, MD, May 1996.

....security faults. With an immunity based approach, our first goal is to increase the inherent variety of OS code, to create resilience against malicious attacks (e.g. viruses and worms) and to contain the damage after the attacks have partially succeeded. Our technique, based on specialization [4, 6, 7, 13, 14] is particularly useful due to its ability to cooperate with verification techniques through explicit definitions of quasi invariants, and to cooperate with security wrappers through meta interfaces. Concretely, we are concerned with operating system survivability under two kinds of attacks. The ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis, MD, May 1996.

....effective response and recovery methods to counteract network attacks represent areas in need of research. In particular, we need fast, flexible, and low overhead mechanisms to support the incorporation and activation of a variety of response strategies against attacks. In the Synthetix project [3, 12, 5], we have been developing a toolkit for systematic specialization of operating system (OS) kernels. The first purpose of the specialization toolkit interpreted as necessarily representing the official policies or endorsements, either expressed or implied, of the Defense Advanced Research ....

....dimensions and the extended toolkit can support both and resolve the trade offs between them. 6 Conclusion We have described our plans for increasing operating system (OS) resistance against security faults, e.g. virus and worm attacks. Our work builds on the existing Synthetix project [3, 12, 5] research on specialization. The basic idea is to increase the representational variety of OS components in a systematic way, by using the Synthetix specialization toolkit. The same way biological systems need variety to resist bacterial and virus attacks, we see the need to increase the variety ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, Maryland, May 1996.

....cases, such as common access patterns to files, to represent them using invariants and quasiinvariants, and then to place the appropriate guards to support automatic replacement of specialized components. It also requires support for dynamic replacement of kernel components that may be executing [9]. The performance improvements that result from the approach depend on the ability to move interpretation code out of the operating system s commonly accessed fast paths ; the necessary guard code is placed in other, less frequently accessed, code paths. Our experiments show that, in the case of ....

Crispin Cowan, Tito Autrey, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. Report CSE-95-019, Dept. of Computer Science and Engineering, Oregon Graduate Institute, Portland, OR, October 1995. Submitted for review.

....with top of stack jump if not zero to constant address .canary death handler add 4 to stack pointer normal return instructions here .canary death handler: Figure 4: Function Epilogue Code: Checking a Canary 3. 2 Preventing Return Address Changes With MemGuard The Synthetix project [18, 1, 2, 24] introduced a notion called quasi invariants. Quasi invariants are state properties that hold true for a while, but may change without notice. Quasi invariants are used to specify optimistic specializations: code optimizations that are valid only while the quasi invariants hold. We have ....

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In International Conference on Configurable Distributed Systems (ICCDS'96), Annapolis, MD,May 1996.

....and requires fast, safe, concurrent dynamic linking. The problem is to facilitate very low latency execution of a function via an indirect function pointer, while concurrently allowing the pointer to be changed. Locks could be used, but locks may also substantially degrade performance. In [7], we describe a portable algorithm that supports low latency invocation of replaceable functions while allowing concurrent update of pointers to those functions. 3 Translation Specialization Our previous efforts have manually applied our various specialization tools [7, 8, 16, 17] Automatic ....

....degrade performance. In [7] we describe a portable algorithm that supports low latency invocation of replaceable functions while allowing concurrent update of pointers to those functions. 3 Translation Specialization Our previous efforts have manually applied our various specialization tools [7, 8, 16, 17]. Automatic translation of specialization plans should convert the high level specification of how to specialize the system into running code that integrates the various components. 3.1 Specialization Plans The specialization plan describes all possible ways in which the facility can be ....

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. In International Conference on Configurable Distributed Systems (ICCDS '96), Annapolis, MD, May 1996.

No context found.

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, Annapolis, 1996. resp. prefix GF: generator generator-family suffix C: completion

No context found.

Crispin Cowan, Tito Autrey, Charles Krasic, Carlton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, Annapolis, 1996.

No context found.

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In International Conference on Con gurable Distributed Systems (ICCDS'96), pages 108-115, Annapolis MD, May 1996.

No context found.

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable

No context found.

Crispin Cowan, Tito Autrey, Charles Krasic, Carlton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Configurable Distributed Systems, Annapolis, 1996.

No context found.

C. Cowan, T. Autrey, C. Krasic, C. Pu, and J. Walpole. Fast Concurrent Dynamic Linking for an Adaptive Operating System. IFIP International Conference on Computer Security, 1996.

No context found.

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proceedings of the International Conference on Congurable Distributed Systems, Annapolis, 1996.

No context found.

Crispin Cowan, Tito Autrey, Charles Krasic, Calton Pu, and Jonathan Walpole. Fast concurrent dynamic linking for an adaptive operating system. In Proc. of the 3rd International Conference on Configurable Distributed Systems, pages 108--115. IEEE Computer Society Press, May 1996.

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