Messages in this thread | | | Date | Thu, 22 Apr 2010 15:50:52 +0200 | Subject | Re: Considerations on sched APIs under RT patch | From | Primiano Tucci <> |
| |
> It's been a while since I've used SMP VxWorks, but back then what it did > was to copy an image for ever CPU separately. It was not really SMP but > instead a separate OS for each CPU. Things may have changed since then > (it was around 2002 when I saw this). > > There's projects to do the same for Linux, and I feel it may give you > the most control of the system. But the hardware is still shared, so the > contention does not go away, it just gets moved to the hardware > resources.
I knew this kind of solution based on OS-partitioning, but my group and I are currently working on a Global-EDF scheduler, a unique scheduler (and therefore a unique OS/Kernel) that is able to migrate tasks between CPUs in order to maximize the global CPU usage. In order to to this we have a unique "super"-process (a Meta-Scheduler) that needs to be able to control priority and affinity of the managed tasks, without losing the control while doing this. We are able to make it work under VXWorks 6.7 SMP (that seems to be really-smp), now we are trying to port the same infrastructure on a Linux-RT Kernel to compare the performances, but we found the issue with the sefaffinity syscall as described into my first mail. I will try to update to the last kernel version and see if it works correctly.
Thanks, Primiano
-- Primiano Tucci http://www.primianotucci.com
| |