SoftMon: programmable software monitoring with minimum overhead by helper-threading

Authors:
Yung-Pin Cheng;Han-Shu Chen
Affiliations:
National Taiwan Normal University, Taipei City, Taiwan;National Taiwan Normal University, Taipei City, Taiwan
Venue:
Proceedings of the 2008 ACM symposium on Applied computing
Year:
2008

Citing 7
Cited 0

Guaranteed Intrusion Removal from Monitored Distributed Applications

SPDP '96 Proceedings of the 8th IEEE Symposium on Parallel and Distributed Processing (SPDP '96)
Design and Implementation of a Compiler Framework for Helper Threading on Multi-core Processors

Proceedings of the 14th International Conference on Parallel Architectures and Compilation Techniques
Dynamic Helper Threaded Prefetching on the Sun UltraSPARC CMP Processor

Proceedings of the 38th annual IEEE/ACM International Symposium on Microarchitecture
Perracotta: mining temporal API rules from imperfect traces

Proceedings of the 28th international conference on Software engineering
HeapMon: a helper-thread approach to programmable, automatic, and low-overhead memory bug detection

IBM Journal of Research and Development
Using Valgrind to detect undefined value errors with bit-precision

ATEC '05 Proceedings of the annual conference on USENIX Annual Technical Conference
Perturbation Analysis: A Static Analysis Approach for the Non-Intrusive Monitoring of Distributed Programs

ICPP '94 Proceedings of the 1994 International Conference on Parallel Processing - Volume 02

Quantified Score

Hi-index	0.00

Visualization

Abstract

In practice, in-house testing is not guaranteed to detect all software errors. A plan for extensive testing is often constrained by resources like cost and time. So, the need for continuous monitoring software programs in production runs is always strong. Software monitoring, however, suffers from the performance overhead of the probing techniques. Unless the overhead for probing is tolerable, the code inserted for probing is typically cleaned from the released programs. In production runs, programmers can only rely on observable effects produced by the programs to examine problems. In this paper, a prototype tool called SoftMon is proposed to monitor a computer program with minimum overhead. SoftMon uses a helper thread to monitor programs on a separate CPU without inserting code in the middle of programs. The body of the helper thread is programmable in a language called MCML. By compiling the MCML script, a helper thread is instrumented into a program automatically and transparently.