Write a small research on a PREEMPT RT patch for the RPI (might be used on a blog)

Job Description:

RTLINUX

research the linux RT patch for the Raspberry PI kernel (kernel version >=5 so not 4.19) and make a setup with a Raspberry PI proving that the Realtime Patch is necessary to achieve a timing goal. Can you guarantee that an RPI with or without load always responds within 1msec after detecting an event, even if it is 100% busy with other tasks at that time?

Use public RT measurement tools to demonstrate that with the RT patch the system may or may not slow down (?) but can process a certain input with 100% certainty. Comparing Cyclictest results between RTLinux kernel and regular os kernel is enough.

Compile and configure the kernel source code yourself and change the local "uname" to show your name when you do "uname -a".

EXTERNAL INTERRUPT

Use of external interface: use one or two wires to event an RPI via GPIO pins. This can be done on the same RPI, for example from GPIOxx to GPIOyy or from an Arduino to RPI pin. Which reaction speed is feasible and why? This is similar to cyclictest but with external events. Thread: [login to view URL]

KERNEL CONFIG CODE

The Linux kernel is open source. The RealTime Patch for Linux as well. The kernel configuration is very extensive - a lot can be set or turned off. Choose 2 config values ​​and describe in plain English what these config parameters do and if it helps to make the system faster.

Many RT tests have already been done by others and you can just imitate them and check whether it is faster / better / slower in your case.

Make it 'testable', so also upload any RPI kernel image you used and all source code.

Explain which (RT TESTS) test tools you used and how they work.

Take pictures of any wires connected for external interrupts.

It doesn't have to be grand, you can keep everything at a minimum so long as uname shows the correct things e.g. the name I provide, preempt RT of a version >=5 etc

Skills: Linux, Raspberry Pi

Project ID: #35867877