[tip:sched/core] sched/fair: Force balancing on NOHZ balance if local group has capacity

[tip-bot for Brendan Jackman]

Commit-ID:  583ffd99d7657755736d831bbc182612d1d2697d
Gitweb:     https://git.kernel.org/tip/583ffd99d7657755736d831bbc182612d1d2697d
Author:     Brendan Jackman <brendan.jackman@xxxxxxx>
AuthorDate: Thu, 5 Oct 2017 11:58:54 +0100
Committer:  Ingo Molnar <mingo@xxxxxxxxxx>
CommitDate: Tue, 10 Oct 2017 11:45:32 +0200

sched/fair: Force balancing on NOHZ balance if local group has capacity

The "goto force_balance" here is intended to mitigate the fact that
avg_load calculations can result in bad placement decisions when
priority is asymmetrical.

The original commit that adds it:

  fab476228ba3 ("sched: Force balancing on newidle balance if local group has capacity")

explains:

    Under certain situations, such as a niced down task (i.e. nice =
    -15) in the presence of nr_cpus NICE0 tasks, the niced task lands
    on a sched group and kicks away other tasks because of its large
    weight. This leads to sub-optimal utilization of the
    machine. Even though the sched group has capacity, it does not
    pull tasks because sds.this_load >> sds.max_load, and f_b_g()
    returns NULL.

A similar but inverted issue also affects ARM big.LITTLE (asymmetrical CPU
capacity) systems - consider 8 always-running, same-priority tasks on a
system with 4 "big" and 4 "little" CPUs. Suppose that 5 of them end up on
the "big" CPUs (which will be represented by one sched_group in the DIE
sched_domain) and 3 on the "little" (the other sched_group in DIE), leaving
one CPU unused. Because the "big" group has a higher group_capacity its
avg_load may not present an imbalance that would cause migrating a
task to the idle "little".

The force_balance case here solves the problem but currently only for
CPU_NEWLY_IDLE balances, which in theory might never happen on the
unused CPU. Including CPU_IDLE in the force_balance case means
there's an upper bound on the time before we can attempt to solve the
underutilization: after DIE's sd->balance_interval has passed the
next nohz balance kick will help us out.

Signed-off-by: Brendan Jackman <brendan.jackman@xxxxxxx>
Signed-off-by: Peter Zijlstra (Intel) <peterz@xxxxxxxxxxxxx>
Cc: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Cc: Mike Galbraith <efault@xxxxxx>
Cc: Morten Rasmussen <morten.rasmussen@xxxxxxx>
Cc: Paul Turner <pjt@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Link: http://lkml.kernel.org/r/20170807163900.25180-1-brendan.jackman@xxxxxxx
Signed-off-by: Ingo Molnar <mingo@xxxxxxxxxx>
---
 kernel/sched/fair.c | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index c04a425..cf3e816 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -8186,8 +8186,11 @@ static struct sched_group *find_busiest_group(struct lb_env *env)
 	if (busiest->group_type == group_imbalanced)
 		goto force_balance;
 
-	/* SD_BALANCE_NEWIDLE trumps SMP nice when underutilized */
-	if (env->idle == CPU_NEWLY_IDLE && group_has_capacity(env, local) &&
+	/*
+	 * When dst_cpu is idle, prevent SMP nice and/or asymmetric group
+	 * capacities from resulting in underutilization due to avg_load.
+	 */
+	if (env->idle != CPU_NOT_IDLE && group_has_capacity(env, local) &&
 	    busiest->group_no_capacity)
 		goto force_balance;