Re: RT task scheduling

[hui]

On Fri, Apr 07, 2006 at 01:39:31AM -0700, Bill Huey wrote:
[Me and Ingo's comments about creation of a new run class and thread
binding...]
> The RT rebalancing discussion should be oriented toward manual techniques
> for dealing with this on an app basis and not automatic load balancing
> stuff or anything like that. IMO, going down this direction is basically
> trying to solve a problem with the wrong tool set.

If some kind of automatic load balance is the focus, then extending the
notion of a CPU package for multicore processors sharing the same cache
controller and memory would be a better track to take. I saw this in the
-mm tree over a year ago an I haven't looked at the scheduler code recently
to see if it made into the mainline. If the RT load balancing is to be
extended, it should take into consideration whether the migration of a
thread should go to a core that's closer or farther away in terms of
memory hierarchy instead of just grabbing the first non-RT task running
CPU and hijacking it to run that RT task.

bill

-
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/