The complexity involved in simulating and injecting hardware failures e.g PSU, DIMM, CPU errors, HDD failures, poses significant challenges for OpenBMC validation. Limited availability of faulty hardware & the high cost of physical test setups, including specific hardware, worsen these issues. Additionally, module-based code development increases the risk of missing existing or domain-specific functional scenarios during new code development. To address these challenges, we propose a unit & integration test framework that reduces the need for actual hardware. This framework uses stubs to simulate firmware code, converts C++ code to Python modules using the SWIG framework & writes function test cases for each module by simulating hardware faults. Beyond traditional testing, we will use AI for code coverage analysis, calculate Cyclomatic Complexity, perform automated patch analysis, vulnerability classification, and fault prediction using historical data and machine learning models.

This session was presented in OCP Global Summit 2025 in San Jose. You may find the session schedule here https://2025ocpglobal.fnvirtual.app/a/schedule/.