Efficient Processing of Naval Computer System State using DWARF

PHILIP S. QUINN, ZACHERY R. CAPELL, RUSSELL L. BINACO, STEPHEN S. MACDONALD, MATTHEW
MOORE, and JOSEPH N. LAGROSSA

When naval ships experience computer systems malfunction, critical information is logged in a highly coded fashion in Debugging Information Entry (DIE) object files in the DWARF format. The DWARF format has evolved markedly over the years to handle data compression and separation of debugging data from executable information. An efficient way to decode DIE files does not exist. Our research was to ascertain whether a Symbol Table Reader could remove much of the work out of deciphering the DIE Object files. This can be accomplished according to the specified DWARF format and allow inspection into different data structures of the program’s state at time of error. Human readable formats of such processed data could be represented in various formats such as MATLAB and XML. MATLAB, in particular allows for data visualization and iterative analysis, whereas XML is a standard interchange format. The research of the team explored combining these aspects of data analysis with an intuitive user interface.