Hi everyone,
I am currently drafting my GSoC proposal for the Controller Area Network (CAN) Stack Improvements, which involves porting standalone CAN test applications into the native RTEMS testsuites infrastructure to be run via rtems-test in QEMU.
Because these specific tests require a QEMU instance with the CTU CAN FD PCI emulation model, I want to ensure they don’t break the “run everywhere” expectation on other BSPs.
What is the RTEMS standard practice for isolating hardware-specific tests? Should the test code dynamically check for the PCI hardware and skip itself cleanly if missing, or is it strictly preferred to exclude these tests at the build/configuration level using .tcfg files?