Tutorial: Leveraging the OS CPU Scheduler to Write Real-Time MT Apps – Kaiwan Billimoria, kaiwanTECH
The Linux OS is powerful; in this tutorial session, you’ll learn some aspects of this power, particularly, how the kernel scheduler can be leveraged to support the writing of (soft) real-time multithreaded applications (with C). This tutorial first covers the basics: the (real!) meaning of real-time, the state machine of a Linux process or thread, and then, at an overview level, the meaning of the various POSIX scheduling policies available on the system. The meat of this talk (and live demo) will be how you, as a systems / embedded app developer, can leverage the kernel by using appropriate Pthread and/or system call APIs to query and set the CPU scheduling policy and priority, at a *per thread* level of granularity! Powerful stuff… We conclude with an overview of the LF Real-Time Linux (RTL) kernel, how to apply relevant patch(es), configure the kernel for RT, and deploy it for use (with appropriately written RT apps), thus enabling the ability to use Linux as an RTOS. The agenda in brief: – What does real-time actually mean – The Linux process state machine – How Linux schedules processes and threads – The POSIX Scheduling Policies and what they mean – Setting policy and priority on an application thread – Demo MT app – Overview: making Linux an RTOS – Q&A.
