Re: [Announce] [patch] Modular Scheduler Core and Completely Fair Scheduler [CFS]

[Peter Williams <pwil3058@bigpond.net.au>]

Peter Williams wrote:
> Are your new patches available somewhere for easy download or do I have 
> to try to dig them out of the mailing list archive?  Or could you mail 
> them to me separately?  I'm keen to see how you new scheduler proposal 
> works.

Forget about this.  I found the patch.

After a quick look, I like a lot of what I see especially the removal of 
the dual arrays in the run queue.

Some minor suggestions:

1. having defined DEFAULT_PRIO in sched.h shouldn't you use it to 
initialize the task structure in init_task.h.
2. the on_rq field in the task structure is unnecessary as many years of 
experience with ingosched in plugsched indicates that 
!list_empty(&(p)->run_list does the job provided list_del_init() is used 
when dequeueing and there is no noticeable overhead incurred so there's 
no gain by caching the result.  Also it removes the possibility of 
errors creeping in due the value of on_rq being inconsistent with the 
task's actual state.
3. having modular load balancing is a good idea but it should be 
decoupled form the scheduler and provided as a separate interface.  This 
would enable different schedulers to use the same load balancer if they 
desired.
4. why rename SCHED_OTHER to SCHED_FAIR?  SCHED_OTHER's supposed to be 
fair(ish) anyway.

Peter
--
Peter Williams                                   pwil3058@bigpond.net.au

"Learning, n. The kind of ignorance distinguishing the studious."
 -- Ambrose Bierce
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/