Fault Tolerant Operating Systems
- 1 December 1976
- journal article
- Published by Association for Computing Machinery (ACM) in ACM Computing Surveys
- Vol. 8 (4), 359-389
- https://doi.org/10.1145/356678.356680
Abstract
This paper develops four related architectural principles which can guide the construction of error-tolerant operating systems. The fundamental principle, system closure, specifies that no action is permissible unless explicitly authorized. The capability based machine is the most efficient known embodiment of this principle: it allows efficient small access domains, multiple domain processes without a privileged mode of operation, and user and system descriptor information protected by the same mechanism. System closure implies a second principle, resource control, that prevents processes from exchanging information via residual values left in physical resource units. These two principles enable a third, decision verification by failure-independent processes. These principles enable prompt error detection and cost-effective recovery. Implementations of these principles are given for process management, interrupts and traps, store access through capabilities, protected procedure entry, and tagged architecture.Keywords
This publication has 21 references indexed in Scilit:
- A lattice model of secure information flowCommunications of the ACM, 1976
- A Computer Architecture for Level Structured SystemsIEEE Transactions on Computers, 1975
- Dynamic verification of operating system decisionsCommunications of the ACM, 1973
- On The Advantages of Tagged ArchitectureIEEE Transactions on Computers, 1973
- Process StructuringACM Computing Surveys, 1973
- Some Deadlock Properties of Computer SystemsACM Computing Surveys, 1972
- Third Generation Computer SystemsACM Computing Surveys, 1971
- Virtual MemoryACM Computing Surveys, 1970
- The structure of the “THE”-multiprogramming systemCommunications of the ACM, 1968
- Programming semantics for multiprogrammed computationsCommunications of the ACM, 1966