RTEMS related articles where I would be happy for some opinion/sense from people stronger than me in scheduling theory
Hello everybody,
there are three scheduling in real-time system related articles and two proposed changes for operating system which can have significant impact on RTEMS executive behavior. And the flawless behavior of RTEMS is important for future ESA and NASA missions for example.
I do not have good feeling about the first proposed change described in next article. I think that it mixes the things. It does not consider/validate options to select EDF and CBS scheduler options implementation available in RTEMS mainline (code developed at our department) and I am not sure if burst of task to re-catch timing is good idea at all and it is common solution in theory and RTOSes
Overrun Handling for Mixed-Criticality Support in RTEMS Kuan-Hsun Chen, Georg von der Bruggen, and Jian-Jia Chen Department of Informatics, TU Dortmund University, Germany http://ls12-www.cs.tu-dortmund.de/daes/media/documents/publications/download...
Proposal for system changes https://devel.rtems.org/ticket/2795
The other change proposal is ticket https://devel.rtems.org/ticket/2556 Implement the O(m) Independence-Preserving Protocol (OMIP)
Related articles
A Fully Preemptive Multiprocessor Semaphore Protocol for Latency-Sensitive Real-Time Applications Bjorn B. Brandenburg Max Planck Institute for Software Systems (MPI-SWS) https://people.mpi-sws.org/~bbb/papers/pdf/ecrts13b.pdf
A Schedulability Compatible Multiprocessor Resource Sharing Protocol - MrsP A. Burns and A.J. Wellings Department of Computer Science,University of York,York, YO10 5GH, UK http://www-users.cs.york.ac.uk/~burns/MRSPpaper.pdf
I have no clue if it is change in good direction or not. I would expect that it is good idea because of theory authors and changes implementor but I like to be more convinced by other people stronger in theory.
Best wishes,
Pavel Pisa
On Thu, Dec 01 2016, Pavel Pisa wrote:
Hello everybody,
there are three scheduling in real-time system related articles and two proposed changes for operating system which can have significant impact on RTEMS executive behavior. And the flawless behavior of RTEMS is important for future ESA and NASA missions for example.
I do not have good feeling about the first proposed change described in next article. I think that it mixes the things.
Do I understand correctly that you propose to present one of the referenced papers next Friday (Dec 9)? If yes, either select which one you want to present or we can vote, which one we find interesting. Tell us what you prefer.
-Michal
Hello all
On Thursday 01 of December 2016 09:13:04 Michal Sojka wrote:
On Thu, Dec 01 2016, Pavel Pisa wrote:
Hello everybody,
there are three scheduling in real-time system related articles and two proposed changes for operating system which can have significant impact on RTEMS executive behavior. And the flawless behavior of RTEMS is important for future ESA and NASA missions for example.
I do not have good feeling about the first proposed change described in next article. I think that it mixes the things.
Do I understand correctly that you propose to present one of the referenced papers next Friday (Dec 9)? If yes, either select which one you want to present or we can vote, which one we find interesting. Tell us what you prefer.
I can present some of the articles on December 9
I have not good feeling about proposal direction in the first one article listed in previous e-mail. So I for it I would be happy only to get some feedback from somebody with more knowledge in the area if suggestion makes sense. I think that it does not worth public presentation and reading as example of well written research article.
I think that next two articles are of more value for public presentation
A Fully Preemptive Multiprocessor Semaphore Protocol for Latency-Sensitive Real-Time Applications Bjorn B. Brandenburg Max Planck Institute for Software Systems (MPI-SWS) https://people.mpi-sws.org/~bbb/papers/pdf/ecrts13b.pdf
A Schedulability Compatible Multiprocessor Resource Sharing Protocol - MrsP A. Burns and A.J. Wellings Department of Computer Science,University of York,York, YO10 5GH, UK http://www-users.cs.york.ac.uk/~burns/MRSPpaper.pdf
They both target related problems. May it be that best option is to present and discuss both in one round. I would study both for sure.
My main motivation is a discussion which helps me better understand given theory and make opinion about strengths and weaknesses which can be even used in future discussion with other developers implementing this feature.
As people often ask me for consultation on technical and implementation aspects of many of their works, I would be happy if there people more strong in the theory can help me to build feeling about these articles from theoretical point of view.
Best wishes,
Pavel Pisa
Hello to everybody,
I have selected next article for Friday, December 16 presentation.
A Fully Preemptive Multiprocessor Semaphore Protocol for Latency-Sensitive Real-Time Applications Bjorn B. Brandenburg Max Planck Institute for Software Systems (MPI-SWS) https://people.mpi-sws.org/~bbb/papers/pdf/ecrts13b.pdf
It is one of two offered articles. The simple parallel between these articles and singlecore systems scheduling policies is that this article represents solution based on priority inheritance, the other one on priority ceiling.
I plan to use authors slides for presentation/discussion
https://people.mpi-sws.org/~bbb/papers/talks/ecrts13b.pdf
Best wishes,
Pavel
reading-group@rtime.felk.cvut.cz
-
Michal Sojka -
Pavel Pisa