DocumentCode
1952775
Title
Tracing with a minimal number of probes
Author
Baca, David
Author_Institution
Electr. Eng., Czech Tech. Univ., Prague, Czech Republic
fYear
2013
fDate
22-23 Sept. 2013
Firstpage
74
Lastpage
83
Abstract
When a program execution must be observed, probes are injected to trace its execution. Probes are intrusive, causing execution overhead and modifying a program´s real time properties. It is therefore desirable to place them efficiently, so that their negative effects are mitigated. This paper presents a new algorithm for efficiently allocating probes at vertices in the control flow graph of a program in a way that makes it possible to reconstruct program traces. Despite of the intractability of finding minimal vertex probe placement, a fact proven in this paper, the presented algorithm is capable of identifying a locally minimal set of vertices, which reduces the set of needed probes beyond the current state-of-the-art. First, the paper shows the advantages of placing probes at vertices, rather than edges. Next, the known methods for reducing the set of probes needed to profile execution are proven to be applicable to reduce the set needed to trace execution. Further, the new algorithm is described along with the algorithm for reconstructing the executed path from a trace file. In addition, the new algorithm is combined with methods for placing profiling probes to reduce the execution overhead even further. Finally, the new probe placement algorithm is compared with known algorithms to elaborate on its benefits.
Keywords
program diagnostics; control flow graph; execution overhead; minimal vertex probe placement; probe placement algorithm; profile execution; program execution; program trace reconstruction; Conferences; Frequency measurement; Probes; Radiation detectors; Real-time systems; Resource management;
fLanguage
English
Publisher
ieee
Conference_Titel
Source Code Analysis and Manipulation (SCAM), 2013 IEEE 13th International Working Conference on
Conference_Location
Eindhoven
Type
conf
DOI
10.1109/SCAM.2013.6648187
Filename
6648187
Link To Document