Browse Prior Art Database

Methods for reducing trace data required for system analysis using custom trace messages

IP.com Disclosure Number: IPCOM000241234D
Publication Date: 2015-Apr-07
Document File: 2 page(s) / 97K

Publishing Venue

The IP.com Prior Art Database

Abstract

Full program trace generation and extraction usually is intrusive and in some contexts, for example a system with 24 cores, impossible to obtain because of the increased amount of the data created. Using a special instrumentation mechanism we can obtain important size reduction for required trace data.

This text was extracted from a Microsoft Word document.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 57% of the total text.

Methods for reducing trace data required for system analysis using custom trace messages

Abstract

Full program trace generation and extraction usually is intrusive and in some contexts, for example a system with 24 cores, impossible to obtain because of the increased amount of the data created. Using a special instrumentation mechanism we can obtain important size reduction for required trace data.

Description

The amount of program trace necessary to generate hierarchical profiler reports for a complex application which is running on Linux for example, can reach absurd sizes. Using an efficient trace reduction mechanism that uses Ownership and Data Acquisition or ITM trace we can significantly reduce the amount of data required.

Hierarchical profiler is the most important tool used by developers to optimize their applications but generating this reports using program trace is difficult. The system proposed has the following components:

•         Instrumentation module which contains either the following functionalities:

ü  Source instrumentation algorithm - using the call graph generated based on static analysis of the sources. For each call site an unique ID will be assigned, information that will be stored at host level and will be used by analysis component.

ü  Binary instrumentation algorithm – parses the executable, detects all call sites and entry point for each function and inserts jump opcode based on configuration selected.

This component will allow two major configurat...