[tip:sched/core] sched/fair: Let minimally loaded cpu balance the group

[tip:sched/core] sched/fair: Let minimally loaded cpu balance the group

[tip-bot for Peter Zijlstra]

Commit-ID:  c22402a2f76e88b04b7a8b6c0597ad9ba6fd71de
Gitweb:     http://git.kernel.org/tip/c22402a2f76e88b04b7a8b6c0597ad9ba6fd71de
Author:     Peter Zijlstra <a.p.zijlstra@chello.nl>
AuthorDate: Fri, 20 Apr 2012 16:57:22 +0200
Committer:  Ingo Molnar <mingo@kernel.org>
CommitDate: Wed, 9 May 2012 15:00:51 +0200

sched/fair: Let minimally loaded cpu balance the group

Currently we let the leftmost (or first idle) cpu ascend the
sched_domain tree and perform load-balancing. The result is that the
busiest cpu in the group might be performing this function and pull
more load to itself. The next load balance pass will then try to
equalize this again.

Change this to pick the least loaded cpu to perform higher domain
balancing.

Signed-off-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/n/tip-v8zlrmgmkne3bkcy9dej1fvm@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
---
 kernel/sched/fair.c |   10 +++++-----
 1 files changed, 5 insertions(+), 5 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 678966c..968ffee 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -3781,7 +3781,8 @@ static inline void update_sg_lb_stats(struct sched_domain *sd,
 {
 	unsigned long load, max_cpu_load, min_cpu_load, max_nr_running;
 	int i;
-	unsigned int balance_cpu = -1, first_idle_cpu = 0;
+	unsigned int balance_cpu = -1;
+	unsigned long balance_load = ~0UL;
 	unsigned long avg_load_per_task = 0;
 
 	if (local_group)
@@ -3797,12 +3798,11 @@ static inline void update_sg_lb_stats(struct sched_domain *sd,
 
 		/* Bias balancing toward cpus of our domain */
 		if (local_group) {
-			if (idle_cpu(i) && !first_idle_cpu) {
-				first_idle_cpu = 1;
+			load = target_load(i, load_idx);
+			if (load < balance_load || idle_cpu(i)) {
+				balance_load = load;
 				balance_cpu = i;
 			}
-
-			load = target_load(i, load_idx);
 		} else {
 			load = source_load(i, load_idx);
 			if (load > max_cpu_load) {

Re: [tip:sched/core] sched/fair: Let minimally loaded cpu balance the group

[Peter Zijlstra]

On Wed, 2012-05-09 at 07:26 -0700, tip-bot for Peter Zijlstra wrote:
> sched/fair: Let minimally loaded cpu balance the group
> 
> Currently we let the leftmost (or first idle) cpu ascend the
> sched_domain tree and perform load-balancing. The result is that the
> busiest cpu in the group might be performing this function and pull
> more load to itself. The next load balance pass will then try to
> equalize this again.
> 
> Change this to pick the least loaded cpu to perform higher domain
> balancing. 

OK, so while this all sounds great its horrible broken.. there's no
guarantee the selected cpu actually gets to balance the domain for it
might never reach as high in its own balance pass since by then it might
not be the least loaded anymore.

Will have to poke at this more tomorrow, but from what my tired brain
makes of it there's nothing but a straight revert (and probably similar
for the next patch).