Real-time Linux: what is next?
summary
PREEMPT_RT is mainlined but work continues: a scheduling-latency formal model maps kernel events to variables and yields a pessimistic worst-case bound; a companion tool scheduling-latency measurement 2 captures those events in-kernel to attribute latency sources. Measured worst-case latencies stay under ~1 ms on some architectures and ~2–4 ms on more complex ones. Ongoing gaps include multi-core deadline protocols, priority/deadline inheritance corner cases, migrate_disable interactions, disabled transparent hugepages, and KVMrt integration (used successfully in telco).
AI-written summary · about
chapters (8)
- 0:00Intro
- 0:15What is realtime
- 3:43Why realtime
- 8:45Results
- 11:42Migration Disable
- 13:30Deadline
- 16:10Final remarks
- 18:40Questions
related talks
47mReal-Time Linux: What Is Next?
40mRT Linux in Safety-Critical Systems: The Potential and the Challenges
52mEffectively Measure and Reduce Kernel Latencies for Real-time Constraints
01h 08mRT Linux in Safety Critical Systems: the potential and the challenges
54mIntroduction to Realtime Linux
53mPREEMPT_RT - How Not to Break It.
50mInside the RT Patch
35mPreempt-RT Raspberry Pi Linux