2014 IEEE 15th International Symposium on High-Assurance Systems Engineering 2014
DOI: 10.1109/hase.2014.13
|View full text |Cite
|
Sign up to set email alerts
|

Evaluating Distortion in Fault Injection Experiments

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
2
1
1

Citation Types

0
5
0

Year Published

2014
2014
2020
2020

Publication Types

Select...
4
2

Relationship

2
4

Authors

Journals

citations
Cited by 9 publications
(5 citation statements)
references
References 23 publications
0
5
0
Order By: Relevance
“…Unfortunately, the combined coverage reached is quite low, especially on Linux. While it would be desirable to reach higher coverage [22], it is hard to do so because we can only use features supported by both operating systems in our workloads. It may be possible to reach higher coverage in systems that are more similar, but then there is still the issue that much of the hardware support is not used, especially when run in an emulator.…”
Section: Resultsmentioning
confidence: 99%
See 2 more Smart Citations
“…Unfortunately, the combined coverage reached is quite low, especially on Linux. While it would be desirable to reach higher coverage [22], it is hard to do so because we can only use features supported by both operating systems in our workloads. It may be possible to reach higher coverage in systems that are more similar, but then there is still the issue that much of the hardware support is not used, especially when run in an emulator.…”
Section: Resultsmentioning
confidence: 99%
“…According to [22], the most representative way to select faults is to make the chance of selecting a certain location or fault type proportional to the number of fault candidates. This means that larger components have a proportionally larger chance of being selected (consistent with [1]) and fault types for which there are many opportunities to introduce are injected more often.…”
Section: B Fault Selectionmentioning
confidence: 99%
See 1 more Smart Citation
“…This effort is driven using the regression tests that accompany the targeted MPI implementations. This is a common method [32] of achieving high coverage without resorting to random API testing. MPI regression tests use either the MPI_ERRORS_RETURN or the MPI_ERRORS_ARE_FATAL modes.…”
Section: Reproducing Error Propagation Bugsmentioning
confidence: 99%
“…Please note that this is an extended version of the earlier conference paper (van der Kouwe et al 2014), which shares these contributions. Among others, this version has a more complete empirical evaluation and discusses the methodology and its limitations in more depth.…”
mentioning
confidence: 93%