From: Valentin Schneider <valentin.schneider@arm.com>
To: linux-kernel@vger.kernel.org
Cc: Peter Zijlstra <peterz@infradead.org>,
Ingo Molnar <mingo@kernel.org>,
Vincent Guittot <vincent.guittot@linaro.org>,
Dietmar Eggemann <dietmar.eggemann@arm.com>,
Morten Rasmussen <morten.rasmussen@arm.com>,
Qais Yousef <qais.yousef@arm.com>,
Quentin Perret <qperret@google.com>,
Pavan Kondeti <pkondeti@codeaurora.org>,
Rik van Riel <riel@surriel.com>
Subject: [PATCH 3/8] sched/fair: Tweak misfit-related capacity checks
Date: Thu, 28 Jan 2021 18:31:36 +0000 [thread overview]
Message-ID: <20210128183141.28097-4-valentin.schneider@arm.com> (raw)
In-Reply-To: <20210128183141.28097-1-valentin.schneider@arm.com>
During load-balance, groups classified as group_misfit_task are filtered
out if they do not pass
group_smaller_max_cpu_capacity(<candidate group>, <local group>);
which itself employs fits_capacity() to compare the sgc->max_capacity of
both groups.
Due to the underlying margin, fits_capacity(X, 1024) will return false for
any X > 819. Tough luck, the capacity_orig's on e.g. the Pixel 4 are
{261, 871, 1024}. If a CPU-bound task ends up on one of those "medium"
CPUs, misfit migration will never intentionally upmigrate it to a CPU of
higher capacity due to the aforementioned margin.
One may argue the 20% margin of fits_capacity() is excessive in the advent
of counter-enhanced load tracking (APERF/MPERF, AMUs), but one point here
is that fits_capacity() is meant to compare a utilization value to a
capacity value, whereas here it is being used to compare two capacity
values. As CPU capacity and task utilization have different dynamics, a
sensible approach here would be to add a new helper dedicated to comparing
CPU capacities.
Introduce capacity_greater(), which uses a 5% margin. Use it to replace the
existing capacity checks. Note that check_cpu_capacity() uses yet another
margin (sd->imbalance_pct), and is left alone for now.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
---
kernel/sched/fair.c | 15 +++++++++++----
1 file changed, 11 insertions(+), 4 deletions(-)
diff --git a/kernel/sched/fair.c b/kernel/sched/fair.c
index 7d01fa0bfc7e..58ce0b22fcb0 100644
--- a/kernel/sched/fair.c
+++ b/kernel/sched/fair.c
@@ -113,6 +113,13 @@ int __weak arch_asym_cpu_priority(int cpu)
*/
#define fits_capacity(cap, max) ((cap) * 1280 < (max) * 1024)
+/*
+ * The margin used when comparing CPU capacities.
+ * is 'cap' noticeably greater than 'ref'
+ *
+ * (default: ~5%)
+ */
+#define capacity_greater(cap, ref) ((ref) * 1078 < (cap) * 1024)
#endif
#ifdef CONFIG_CFS_BANDWIDTH
@@ -8253,7 +8260,7 @@ check_cpu_capacity(struct rq *rq, struct sched_domain *sd)
static inline int check_misfit_status(struct rq *rq, struct sched_domain *sd)
{
return rq->misfit_task_load &&
- (rq->cpu_capacity_orig < rq->rd->max_cpu_capacity ||
+ (capacity_greater(rq->rd->max_cpu_capacity, rq->cpu_capacity_orig) ||
check_cpu_capacity(rq, sd));
}
@@ -8352,7 +8359,7 @@ group_is_overloaded(unsigned int imbalance_pct, struct sg_lb_stats *sgs)
static inline bool
group_smaller_min_cpu_capacity(struct sched_group *sg, struct sched_group *ref)
{
- return fits_capacity(sg->sgc->min_capacity, ref->sgc->min_capacity);
+ return capacity_greater(ref->sgc->min_capacity, sg->sgc->min_capacity);
}
/*
@@ -8362,7 +8369,7 @@ group_smaller_min_cpu_capacity(struct sched_group *sg, struct sched_group *ref)
static inline bool
group_smaller_max_cpu_capacity(struct sched_group *sg, struct sched_group *ref)
{
- return fits_capacity(sg->sgc->max_capacity, ref->sgc->max_capacity);
+ return capacity_greater(ref->sgc->max_capacity, sg->sgc->max_capacity);
}
static inline enum
@@ -9421,7 +9428,7 @@ static struct rq *find_busiest_queue(struct lb_env *env,
* average load.
*/
if (sd_has_asym_cpucapacity(env->sd) &&
- capacity_of(env->dst_cpu) < capacity &&
+ !capacity_greater(capacity_of(env->dst_cpu), capacity) &&
nr_running == 1)
continue;
--
2.27.0
next prev parent reply other threads:[~2021-01-28 18:35 UTC|newest]
Thread overview: 46+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-01-28 18:31 [PATCH 0/8] sched/fair: misfit task load-balance tweaks Valentin Schneider
2021-01-28 18:31 ` [PATCH 1/8] sched/fair: Clean up active balance nr_balance_failed trickery Valentin Schneider
2021-02-03 15:14 ` Qais Yousef
2021-02-03 18:42 ` Valentin Schneider
2021-02-04 15:05 ` Qais Yousef
2021-02-05 13:51 ` Vincent Guittot
2021-02-05 14:05 ` Valentin Schneider
2021-02-05 14:34 ` Vincent Guittot
2021-01-28 18:31 ` [PATCH 2/8] sched/fair: Add more sched_asym_cpucapacity static branch checks Valentin Schneider
2021-02-03 15:14 ` Qais Yousef
2021-02-09 8:42 ` Vincent Guittot
2021-01-28 18:31 ` Valentin Schneider [this message]
2021-02-03 15:15 ` [PATCH 3/8] sched/fair: Tweak misfit-related capacity checks Qais Yousef
2021-02-03 18:42 ` Valentin Schneider
2021-02-05 14:31 ` Vincent Guittot
2021-02-05 16:59 ` Valentin Schneider
2021-02-05 17:17 ` Vincent Guittot
2021-02-05 20:07 ` Valentin Schneider
2021-02-08 15:29 ` Vincent Guittot
2021-02-08 17:49 ` Valentin Schneider
2021-01-28 18:31 ` [PATCH 4/8] sched/fair: Use dst_cpu's capacity rather than group {min, max} capacity Valentin Schneider
2021-02-03 15:15 ` Qais Yousef
2021-01-28 18:31 ` [PATCH 5/8] sched/fair: Make check_misfit_status() only compare dynamic capacities Valentin Schneider
2021-02-03 15:15 ` Qais Yousef
2021-02-04 10:49 ` Dietmar Eggemann
2021-02-04 11:34 ` Valentin Schneider
2021-02-04 14:57 ` Dietmar Eggemann
2021-01-28 18:31 ` [PATCH 6/8] sched/fair: Filter out locally-unsolvable misfit imbalances Valentin Schneider
2021-02-03 15:16 ` Qais Yousef
2021-02-03 18:43 ` Valentin Schneider
2021-01-28 18:31 ` [PATCH 7/8] sched/fair: Attempt misfit active balance when migration_type != migrate_misfit Valentin Schneider
2021-02-03 15:16 ` Qais Yousef
2021-02-03 18:43 ` Valentin Schneider
2021-02-04 11:44 ` Dietmar Eggemann
2021-02-04 12:22 ` Valentin Schneider
2021-02-09 8:58 ` Vincent Guittot
2021-02-09 18:19 ` Valentin Schneider
2021-01-28 18:31 ` [PATCH 8/8] sched/fair: Relax task_hot() for misfit tasks Valentin Schneider
2021-02-03 15:17 ` Qais Yousef
2021-02-08 16:21 ` Vincent Guittot
2021-02-08 18:24 ` Valentin Schneider
2021-02-09 8:56 ` Vincent Guittot
2021-02-03 15:14 ` [PATCH 0/8] sched/fair: misfit task load-balance tweaks Qais Yousef
2021-02-03 18:43 ` Valentin Schneider
2021-02-04 12:03 ` Dietmar Eggemann
2021-02-04 12:36 ` Valentin Schneider
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20210128183141.28097-4-valentin.schneider@arm.com \
--to=valentin.schneider@arm.com \
--cc=dietmar.eggemann@arm.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@kernel.org \
--cc=morten.rasmussen@arm.com \
--cc=peterz@infradead.org \
--cc=pkondeti@codeaurora.org \
--cc=qais.yousef@arm.com \
--cc=qperret@google.com \
--cc=riel@surriel.com \
--cc=vincent.guittot@linaro.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).