Summary

Miro Samek challenges the assumption that RTOS-based threads are inherently real-time, examining how shared-state concurrency and blocking obscure worst-case execution time (WCET). The blog explains why conventional RTOS primitives can invalidate timing guarantees and outlines design patterns and analysis steps to make thread timing provable.

Key Takeaways

• Identify common sources of non-determinism in RTOS systems (blocking calls, shared locks, dynamic memory, and priority inversion).
• Measure and bound worst-case execution time (WCET) by isolating interrupt paths, constraining execution regions, and eliminating unbounded primitives.
• Prefer message-passing, non-blocking designs, and priority-aware synchronization to minimize hidden latencies.
• Design and apply system-level analysis (response-time analysis, admission control, and timing budgets) to prove deadlines under realistic load.

Who Should Read This

Experienced embedded firmware engineers and system architects building latency-sensitive or safety-critical systems who need to understand and prove RTOS timing behavior.

TimelessAdvanced

Topics