WULF, W., LEVIN, R., AND HARBISON , S. An Experimental Computer System. McGraw-Hill, New York, 1981.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....the most privileged protection ring, whereas capability based systems have a single privilege level. Thus capability based systems facilitate the addition of user level policy to the operating system. Indeed it was one of the main aims of the Hydra project to support this policy mechanism split (Wulf et al. 1981). The protection of operating system code in capability systems uses the same mechanisms as user code protected procedures. There is no system call in capability systems. Software implementations such as Hydra do have a kernel which resides in a specific address space separate from user ....

....a master mode descriptor base register switching routine to point to the segment in the target ring. 16 The round trip ring crossing during a procedure call on Multics was of the order of 2 to 3 milliseconds (Organick, 1972) This contrasts to the 20 to 30 millisecond cross domain call on Hydra (Wulf et al. 1981) and, at the other extreme, to a hardware implemented domain switch on the Intel iAPX 432 of 65 microseconds (Kahn et al., 1981) 17 The cost of system calls is similar to the domain call in the iAPX 432, varying from a minimum of 70 microseconds on Sprite on a Sun 3 60 (Douglis et al. 1991) to ....

[Article contains additional citation context not shown here]

Wulf, W.A., R. Levin and S.P. Harbison (1981) HYDRA/C.mmp, an experimental computer system. McGraw-Hill.

....have fewer software faults over time than a traditional monolithic kernel. An orthogonal approach has been to partition the system into a microkernel and a set of services running in independent address spaces, making the system easier to test and debug. Examples of this approach include Hydra [WLH81], Mach [RJO 89] and Chorus [RAA 88] Finally, automated testing [SaH94] can increase the fraction of faults that are found and fixed during operating system development. Another area of work has been techniques for avoiding the reboot that is required to recover from software errors in standard ....

....C.mmp scaled to 16 processors, which was large from the perspective of reliability issues given the low integration technology from which it was constructed. The developers found it necessary to add novel reliability oriented features to the Hydra operating system to achieve adequate MTBF [WLH81]. Hydra includes mechanisms to tolerate both hardware and software faults. At the hardware level, it has a watchdog mechanism to detect fail stop processor faults and a resource exploration phase at reboot time that can avoid bad memory pages. It tolerates certain non fail stop hardware faults, ....

W. Wulf, R. Levin, and S. Harbison. HYDRA/C.mmp, an experimental computer system. New York: McGraw-Hill, 1981.

....a named address space and the contents of the address space. Since it does not contain a process, it is completely passive. Hence, unlike objects in some object based systems, a Clouds object is not associated with any server process. The first system to use passive objects was Hydra [WCC 74, WLH81] The contents of each virtual address space are protected from outside access so that memory (data) in an object is accessible only by the code in that object and the operating system. Each object is an encapsulated address space with entry points at which threads may commence execution. The ....

W. A. Wulf, R. Levin, and S. P. Harbison. HYDRA/C.mmp, An Experimental Computer System. McGraw-Hill, Inc., 1981.

No context found.

WULF, W., LEVIN, R., AND HARBISON , S. An Experimental Computer System. McGraw-Hill, New York, 1981.

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