CS 481 Group Design Project

Fall 2002 Projects

Last Updated January 02, 2004

Project Title: Code Health Audit & Reporting System (CARS)

Sponsor Troy Pearse Hewlett-Packard Co, Boise, ID 208-396-4557 Troy_Pearse@hp.com	Sponsor Bruce Mayes Hewlett-Packard Co, Boise, ID 208-396-2805 Bruce_Mayes@hp.com

Project Description:  The CARS project is a series of modules that collectively measure the "Code Health" of a set of software subsystems and then generates web and email reports with the test results. "Code Health" is a term the HP LaserJet Lab uses to describe certain aspects of the firmware related to coding standards and defect prevention. These code health criteria are typically things that we have learned can cause failures if not detected and fixed. Compiler warnings is a good example of a code health metric. Having many compiler warnings is indicative that a body of code has poor health and may experience unexpected failures.

The CARS project modules will include the following:

• Two Code Health Plugin Modules which follow a standard "Compcheck Plugin" interface and allow them to be run in our Compcheck environment (Compcheck is a script that can do a variety of Component Checks). One plugin will return compiler warnings for a subsystem (ie: set of files). The second plugin will return lint warnings for a subsystem.
• Each of these Code Health Plugin Modules will need to compare their results (ie: compiler warnings) to mastered test output to determine if the tests have Passed or Failed.
• There will be an UpdateTERS (Test Execution Report Status) Plugin that can parse the Code Health Plugin Test results and export them to a test database. To begin with the test database will be a simple comma separated file. We hope to be able to expand this to the "real" TERS database by the end of the project.
• There will be a CodeHealthAudit Plugin that can parse the Code Health Plugin Test results and then for any subsystem with failures, email a canned report with the details of the failures to the owners of the subsystem.
• There will be a Web Report for the Code Health information that will read in the UpdateTERS output and generate a Web report (dashboard) based on the data. There will need to be a series of web pages for different LaserJet internal projects under development, and each page will show the high level pass/fail type information plus be able to link in the detailed results for each of the failures.

Project Title: Automated Crash Reporting System

Project Description: HP uses early development testing units (printer hardware) to test printer firmware in the mid to late stages of its development. These test units occasionally crash, storing and sending useful debugging information (stack trace) to the unit's Control Panel display (if it has one) and down the serial port connection.  Currently there is no efficient method in place to monitor the frequency of the failures or capture stack trace information that would point to the cause of the failure.  This early testing, known as Delta and Beta Test attempt to closely emulate "real world" use of the complete printing system. If information regarding the cause and frequency of firmware crashes during this type of testing were available, it would be a valuable tool in determining the stability of the firmware under development.

Solution:  Since the firmware dumps stack trace information through the serial port when it encounters a fatal error, it's possible to automatically record and report events.  Client PCs would be used to monitor the serial ports on the printers under test in Software and System Test. Another PC, acting as a Server, would record, and report on, the crash  information from numerous Client PCs. Two software utilities are needed for this project:

1. On the Client PC, a software program is needed to monitor the data sent from the printer, to the PC via its serial port. This client utility would parse the serial data sent by the printer. When an error condition is detected, the utility would record the error, collect stack information, then send the information across the network to the Server PC.

2. On the Server PC, a software program is needed to collect, analyze, and report on the crash data that it receives from the Clients.