New trace-based method brings MC/DC testing to optimized embedded builds

Structural coverage in embedded projects is typically collected at unit level because instrumentation there is cheap, but the most representative completeness signal comes from integration and system tests on real devices. Classical instrumentation perturbs timing, memory footprint, and concurrency, while purely trace-reconstructed coverage loses reliability for decisions and conditions under aggressive compiler optimization. Weiss, Schulz, and Wittner address this mismatch with a process-side integration-first strategy: integration and system tests become the baseline measurement, and residual gaps are closed explicitly as covered or justified, not just covered.

Technically, they introduce hybrid runtime analysis (hRA), a minimal, semantics-preserving observability scaffolding that keeps decision and condition boundaries visible in the trace stream of an -O3 build while all coverage state stays off-target. This converts object-to-source mapping from heuristic reconstruction into reviewable evidence, making branch, condition, and MC/DC measurement practical on release-like binaries. They also present Hyper Coverage, a consolidation layer that merges coverage data across test levels, runs, variants, and configurations, exposing source lines untested in every relevant variant. The paper (arXiv:2608.13322) demonstrates a path toward more trustworthy completeness evidence for safety-critical embedded software.