From: Valentin Schneider <valentin.schneider@arm.com>
To: linux-kernel@vger.kernel.org
Cc: vincent.guittot@linaro.org, mgorman@suse.de, mingo@kernel.org,
peterz@infradead.org, dietmar.eggemann@arm.com,
morten.rasmussen@arm.com, linuxarm@openeuler.org,
xuwei5@huawei.com, liguozhu@hisilicon.com,
tiantao6@hisilicon.com, wanghuiqiang@huawei.com,
prime.zeng@hisilicon.com, jonathan.cameron@huawei.com,
guodong.xu@linaro.org, Barry Song <song.bao.hua@hisilicon.com>,
Meelis Roos <mroos@linux.ee>
Subject: [RFC PATCH 3/2] sched/topology: Alternative diameter >= 2 fixup
Date: Wed, 3 Feb 2021 15:54:32 +0000 [thread overview]
Message-ID: <20210203155432.10293-4-valentin.schneider@arm.com> (raw)
In-Reply-To: <20210203155432.10293-1-valentin.schneider@arm.com>
This follows Barry's approach, which yields fewer groups in the higher domains.
Signed-off-by: Valentin Schneider <valentin.schneider@arm.com>
---
kernel/sched/topology.c | 42 +++++++++++++++++++----------------------
1 file changed, 19 insertions(+), 23 deletions(-)
diff --git a/kernel/sched/topology.c b/kernel/sched/topology.c
index 0fa41aab74e0..e1bb97c06f53 100644
--- a/kernel/sched/topology.c
+++ b/kernel/sched/topology.c
@@ -949,20 +949,21 @@ static void init_overlap_sched_group(struct sched_domain *sd,
sg->sgc->max_capacity = SCHED_CAPACITY_SCALE;
}
-static struct sched_domain *find_node_domain(struct sched_domain *sd)
+static struct sched_domain *
+find_sibling_domain(struct sched_domain *sd, struct sched_domain *sibling)
{
- struct sched_domain *parent;
-
BUG_ON(!(sd->flags & SD_NUMA));
- /* Get to the level above NODE */
- while (sd && sd->child) {
- parent = sd;
- sd = sd->child;
+ /* Find a subset sibling */
+ while (sibling->child &&
+ !cpumask_subset(sched_domain_span(sibling->child),
+ sched_domain_span(sd)))
+ sibling = sibling->child;
+
+ /* If the above loop was a no-op, we're done */
+ if (sd->private == sibling->private)
+ return sibling;
- if (!(sd->flags & SD_NUMA))
- break;
- }
/*
* We're going to create cross topology level sched_group_capacity
* references. This can only work if the domains resulting from said
@@ -972,16 +973,14 @@ static struct sched_domain *find_node_domain(struct sched_domain *sd)
*
* Of course, groups aren't available yet, so we can't call the usual
* sd_degenerate(). Checking domain spans is the closest we get.
- * Start from NODE's parent, and keep going up until we get a domain
- * we're sure won't be degenerated.
+ * We can't go up as per the above subset check, so keep going down
+ * until we get a domain we're sure won't be degenerated.
*/
- while (sd->parent &&
- cpumask_equal(sched_domain_span(sd), sched_domain_span(parent))) {
- sd = parent;
- parent = sd->parent;
- }
+ while (sibling->child &&
+ cpumask_equal(sched_domain_span(sibling->child), sched_domain_span(sibling)))
+ sibling = sibling->child;
- return parent;
+ return sibling;
}
static int
@@ -1017,12 +1016,9 @@ build_overlap_sched_groups(struct sched_domain *sd, int cpu)
if (!cpumask_test_cpu(i, sched_domain_span(sibling)))
continue;
- /*
- * Local group is child domain's span, as is tradition.
- * Non-local groups will only span remote nodes.
- */
+ /* Ensure sibling domain's span is a subset of this domain */
if (first)
- sibling = find_node_domain(sibling);
+ sibling = find_sibling_domain(sd, sibling);
sg = build_group_from_child_sched_domain(sibling, cpu);
if (!sg)
--
2.27.0
prev parent reply other threads:[~2021-02-03 15:57 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-02-03 15:54 [RFC PATCH 0/2] sched/topology: Get rid of overlapping groups Valentin Schneider
2021-02-03 15:54 ` [RFC PATCH 1/2] sched/topology: Get rid of NUMA " Valentin Schneider
2021-02-08 10:04 ` Song Bao Hua (Barry Song)
2021-02-08 11:47 ` Valentin Schneider
2021-02-09 0:12 ` Song Bao Hua (Barry Song)
2021-02-09 10:41 ` Valentin Schneider
2021-02-03 15:54 ` [RFC PATCH 2/2] Revert "sched/topology: Warn when NUMA diameter > 2" Valentin Schneider
2021-02-08 10:27 ` Song Bao Hua (Barry Song)
2021-02-03 15:54 ` Valentin Schneider [this message]
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=20210203155432.10293-4-valentin.schneider@arm.com \
--to=valentin.schneider@arm.com \
--cc=dietmar.eggemann@arm.com \
--cc=guodong.xu@linaro.org \
--cc=jonathan.cameron@huawei.com \
--cc=liguozhu@hisilicon.com \
--cc=linux-kernel@vger.kernel.org \
--cc=linuxarm@openeuler.org \
--cc=mgorman@suse.de \
--cc=mingo@kernel.org \
--cc=morten.rasmussen@arm.com \
--cc=mroos@linux.ee \
--cc=peterz@infradead.org \
--cc=prime.zeng@hisilicon.com \
--cc=song.bao.hua@hisilicon.com \
--cc=tiantao6@hisilicon.com \
--cc=vincent.guittot@linaro.org \
--cc=wanghuiqiang@huawei.com \
--cc=xuwei5@huawei.com \
/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).