[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20061031155819.29621173.pj@sgi.com>
Date: Tue, 31 Oct 2006 15:58:19 -0800
From: Paul Jackson <pj@....com>
To: Paul Jackson <pj@....com>
Cc: dino@...ibm.com, nickpiggin@...oo.com.au, akpm@...l.org,
mbligh@...gle.com, menage@...gle.com, Simon.Derr@...l.net,
linux-kernel@...r.kernel.org, rohitseth@...gle.com, holt@....com,
dipankar@...ibm.com, suresh.b.siddha@...el.com
Subject: Re: [RFC] cpuset: Explicit dynamic sched domain cpuset flag
pj wrote:
> 9) Finally, we really do need a way, on a production system, for user
> space to ask the kernel where load balancing is limited.
>
> For example one possible interface would have user space pass a
> cpumask to the kernel, and get back a Boolean value, indicating
> whether or not there are any limitations on load balancing between
> any two CPUs specified in that cpumask.
Ah - a simpler API, more user friendly, more "cpuset API style"
friendly:
A read-only, per-cpuset Boolean flag that indicates whether the
scheduler is fully load balancing across all 'cpus' of that cpuset.
Internally, the kernel would answer this by seeing whether or
not the cpusets cpus_allowed cpumask was a subset of one of the
members of the scheduler domains partition.
Call this per-cpuset flag something such as:
sched_is_fully_load_balanced # read-only Boolean
This goes along with having the control flag named something such as:
sched_ok_not_to_load_balance # read-write Boolean
If a task was in a cpuset that had 'sched_is_fully_load_balanced'
False, then it might not get load balanced.
--
I won't rest till it's the best ...
Programmer, Linux Scalability
Paul Jackson <pj@....com> 1.925.600.0401
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists