RTEMS | Locking Protocols for Multiprocessors (#4612)

[Karthikey Kadati (@karthikey_kadati)]

Karthikey Kadati commented: https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/4612#note_137648


Hi everyone,

I have verified the SMP environment using the `smpmrsp01` test suite on the SIS/Leon3 simulator. As requested, I modified the taskset in `init.c` to observe how it affects the execution results.

**What I changed:** I modified the initial priority of the main task to 1 and adjusted the MrsP priority ceiling to 5. This created a scenario where the executing task's priority was higher than the resource's ceiling.

**Analysis of Results:** This change actually caused the test to fail with an assertion at line 1411 in `test_mrsp_obtain_and_release_with_help`. By checking the switch logs, I confirmed that the failure occurred because the task priorities no longer matched the specific validation logic for the helping mechanism. This was a great exercise to see how strictly the RTEMS 7 MrsP implementation and its test suite monitor priority elevation and task-to-processor affinity.

Now that I have a solid understanding of how these protocols are structured in the modern SuperCore, I am moving forward with porting FMLP. I’ll be mapping the TU Dortmund implementation to the modern `_Thread_queue` and `_ISR_lock` primitives found in `mrspimpl.h`.

I'll share my progress on the FMLP port soon. Thanks!

-- 
View it on GitLab: https://gitlab.rtems.org/rtems/rtos/rtems/-/issues/4612#note_137648
You're receiving this email because of your account on gitlab.rtems.org.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.rtems.org/pipermail/bugs/attachments/20251219/edfe5959/attachment-0001.htm>