Virtual memory, processes, and sharing in MULTICS

Daley, Robert C.; Dennis, Jack B.

doi:10.1145/363095.363139

Cited by 118 publications

(34 citation statements)

References 3 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…To load a library, most loaders first search a list of file system locations until the target library is found, then they load the library into memory. This mechanism does not differ significantly from the method used by the MULTICS [6] operating system that introduced dynamic linking in 1964. While it was suitable for a single-node machine, it breaks down in parallel.…”

Section: Introductionmentioning

confidence: 89%

Massively parallel loading

Frings

Ahn

Legendre

et al. 2013

Proceedings of the 27th International ACM Conference on International Conference on Supercomputing

View full text Add to dashboard Cite

Dynamic linking has many advantages for managing large code bases, but dynamically linked applications have not typically scaled well on high performance computing systems. Splitting a monolithic executable into many dynamic shared object (DSO) files decreases compile time for large codes, reduces runtime memory requirements by allowing modules to be loaded and unloaded as needed, and allows common DSOs to be shared among many executables. However, launching an executable that depends on many DSOs causes a flood of file system operations at program start-up, when each process in the parallel application loads its dependencies. At large scales, this operation has an effect similar to a site-wide denial-of-service attack, as even large parallel file systems struggle to service so many simultaneous requests. In this paper, we present Spindle, a novel approach to parallel loading that coordinates simultaneous file system operations with a scalable network of cache server processes. Our approach is transparent to user applications. We extend the GNU loader, which is used in Linux as well as proprietary operating systems, to limit the number of simultaneous file system operations, quickly loading DSOs without thrashing the file system. Our experiments show that our prototype implementation has a low overhead and increases the scalability of Pynamic, a benchmark that stresses the dynamic loader, by a factor of 20.

show abstract

Section: Introductionmentioning

confidence: 89%

Massively parallel loading

Frings

Ahn

Legendre

et al. 2013

Proceedings of the 27th International ACM Conference on International Conference on Supercomputing

View full text Add to dashboard Cite

show abstract

“…Support for segmentation: Several past and present architectures supported a segmented address space. Generally, segments are either supported without paging, as in early Intel 8086 processors [21], or more commonly on top of paging as in MULTICS [14], PowerPC, and IA-32 [23]. Use of pure segmentation is incompatible with current software, while segmentation on top of paging does not reduce the address translation cost of page-based virtual memory.…”

Section: Related Workmentioning

confidence: 99%

Efficient virtual memory for big memory servers

Basu

Gandhi

Chang

et al. 2013

SIGARCH Comput. Archit. News

106

188

View full text Add to dashboard Cite

Our analysis shows that many "big-memory" server workloads, such as databases, in-memory caches, and graph analytics, pay a high cost for page-based virtual memory. They consume as much as 10% of execution cycles on TLB misses, even using large pages. On the other hand, we find that these workloads use read-write permission on most pages, are provisioned not to swap, and rarely benefit from the full flexibility of page-based virtual memory.To remove the TLB miss overhead for big-memory workloads, we propose mapping part of a process's linear virtual address space with a direct segment, while page mapping the rest of the virtual address space. Direct segments use minimal hardware-base, limit and offset registers per core-to map contiguous virtual memory regions directly to contiguous physical memory. They eliminate the possibility of TLB misses for key data structures such as database buffer pools and in-memory key-value stores. Memory mapped by a direct segment may be converted back to paging when needed.We prototype direct-segment software support for x86-64 in Linux and emulate direct-segment hardware. For our workloads, direct segments eliminate almost all TLB misses and reduce the execution time wasted on TLB misses to less than 0.5%.

show abstract

“…This paradigm had roots in the research UNIX community and its USENET, with some philosophical roots later added with the "Free Software" principles of Stallman. The mid-1960s MULTICS (Daley & Dennis, 1968;Organick, 1972) project, part of the U.S. Defense Advanced Research Projects Agency (DARPA)-supported Project MAC (Fano & David, 1965) at MIT, gave rise to the original UNIX system (Ritchie & Thompson, 1974, 1978Thompson, 1978) (the name UNIX is in fact a pun on MULTICS) as a reaction to MULTICS system complexity. Unfortunately, in rejecting much of MULTICS, the UNIX system was not able to avail itself of the extensive effort devoted to developing protection models and security kernels (Schroeder, 1975;Schroder, Clark, & Saltzer, 1977) for MULTICS.…”

Section: The Open Source Alternativementioning

confidence: 99%

Global Information Technologies

2008

View full text Add to dashboard Cite

Global information technologies : concepts, methodologies, tools and applications / Felix Tan, editor.v. cm. Summary: "This collection compiles research in all areas of the global information domain. It examines culture in information systems, IT in developing countries, global e-business, and the worldwide information society, providing critical knowledge to fuel the future work of researchers, academicians and practitioners in fields such as information science, political science, international relations, sociology, and many more"--Provided by publisher.Includes AbstrActThis chapter reports on our experiences with POSSE, a project studying "Portable Open Source Security Elements" as part of the larger DARPA effort on Composable High Assurance Trusted Systems. We describe the organization created to manage POSSE and the significant acceleration in producing widely used secure software that has resulted. POSSE's two main goals were, first, to increase security in open source systems and, second, to more broadly disseminate security knowledge, "best practices," and working code that reflects these practices. POSSE achieved these goals through careful study of systems ("audit") and starting from a well-positioned technology base (OpenBSD). We hope to illustrate the advantages of applying OpenBSD-style methodology to secure, open-source projects, and the pitfalls of melding multiple open-source efforts in a single project.

show abstract

Virtual memory, processes, and sharing in MULTICS

Cited by 118 publications

References 3 publications

Massively parallel loading

Massively parallel loading

Efficient virtual memory for big memory servers

Global Information Technologies

Contact Info

Product

Resources

About