Re: Considerations on sched APIs under RT patch

[Peter Zijlstra]

On Thu, 2010-04-22 at 15:50 +0200, Primiano Tucci wrote:
> 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.

I would hardly call a global-edf scheduler unique. Its a well studied
algorithm and even available in commercial SMP operating systems
(hopefully soon Linux too, see SCHED_DEADLINE, which will approximate
global-edf, much like the current SCHED_FIFO approximates global-fifo).

> 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. 

Implementing this as userspace/middleware seems daft. But if your
controlling process has a global affinity mask and runs as the highest
available userspace process priority its still all valid.



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/