Real-Time Linux: What Is N...
summary
PREEMPT_RT is largely upstream now after about $5.7M raised and years of backports work; high-resolution timers, tracing and threaded IRQs already mainline and only a few core patches (printk) remain. The persistent throttling problem is tackled by boosting CFS via a SCHED_DEADLINE reservation rather than simple throttling to avoid starvation. Diagnostic tooling like rtla (and formal RV models, timerlot prototypes) finally lands and proves useful; tuned isolated-CPU setups report <10 μs latency in measurements.
AI-written summary · about
chapters (20)
- 0:00Introduction
- 1:26Real-time community motivation
- 2:09PREEMPT_RT upstreaming progress
- 6:23Why avoid shortcuts
- 7:07Real-time throttling explained
- 10:38PREEMPT_RT impact on kernel
- 12:45Architecture benefits
- 13:06PREEMPT_RT benefits beyond real-time
- 15:11Real-time community next steps
- 17:57Priority inheritance explained
- 20:30Real-time in industrial safety
- 21:36SCHED_DEADLINE status
- 23:46Getting involved and improving documentation
- 26:16rtla and tracing tools journey
- 29:07Need for diagnostics guides
- 31:34Messaging challenges
- 35:48Documentation gaps
- 40:49Dynamic kernel parameters
- 43:14Kernel thread priorities
- 45:41AI tools for real-time
related talks
42mWhen the OS gets in the way
33mLinux Kernel Live Patching
24mContinuously Integrating the Upstream Linux Kernel on Hardware
37mAccelerating Linux Boot Time: Techniques and Strategies for Optimal Performance
37mKernel and userspace tracing with LTTng and friends
28mHyperscale Kernel Live Patching at Meta
10mTowards a More Efficiently Synchronization in KVM
39mThe End of the Paved Road: Maintaining Linux Kernel 4.4 Beyond LTS